Optimization of random amplified polymorphic DNA protocol for molecular ident. of L. gastrophysus
易腐商品最优订货批量与定价及其粒子群优化解
2005年3月 系统工程理论与实践 第3期文章编号：1000-6788(2005)03-0046-06易腐商品最优订货批量与定价及其粒子群优化解田志友，蒋录全，吴瑞明（上海交通大学管理学院，上海 200030）摘要：对易腐商品的订货批量与定价问题进行了研究.基于一种负二项分布的离散需求函数，推导了易腐品利润最大化模型.由于模型中涉及多个随机变量的概率分布，常规函数极值法对此具有极大局限性，故首次将粒子群优化算法引入该领域，并提出两种不同的求解思路：1)枚举法.利用粒子群算法依次计算不同订货批量下的最大化利润，然后根据边际分析法确定最优订货批量及相应定价；2)二维寻优法.将利润视为订货量与定价的二维函数，利用粒子群算法对其进行二维演化寻优.算例分析表明：两种方法均可有效获得问题的满意解，当订货量波动范围较小时，枚举法效果更优.关键词：易腐商品；订货批量；定价；需求分布；粒子群优化算法中图分类号：F830 文献标识码：AOptimal Order Quantity and Pricing for Perishable Commodities and Solutions with Particle Swarm OptimizationTIAN Zhi-you, JIANG Lu-quan, WU Rui-ming(School of Management, Shanghai Jiaotong University, Shanghai 200030, China)Abstract: The problem of ordering policies and optimal pricing for perishable commodities ismainly studied. According to a kind of demand distribution, which can be represented as a negativebinomial distribution, the profit maximization model of those products is deduced. Since themodel involves several different stochastic distributions, which are difficult for the normalfunction optimization methods to solve, the particle swarm optimization (PSO) algorithm isintroduced for the first time to settle it, and two different solving processes are proposed,one can be called enumerative method, which will calculate the optimal price and maximum profitfor each possible orders, and then find the ultimate optimal solution by marginal analysis. Theother is two-dimensional search, which can determine the optimal order quantity and pricesimultaneously with PSO technique. At the end a numerical example is studied and the two methodsare compared, the results indicate that: both can obtain satisfactory solutions effectively,and when the bounds for possible orders are relatively small, the first is preferred.Key words: perishable commodities； order quantity; pricing；demand distribution; particleswarm optimization1 引言易腐商品是指那些必须在有限时间内售出，否则将发生质变，必须清仓处理的商品.狭义的易腐商品主要是指生鲜食品，广义来说，凡是超过正常销售期后市场价值有明显降低的商品均可归属易腐品的类别，如时装服饰，电子消费品，客房,机票等服务.由于这些商品保质期或市场需求周期比较短，或具收稿日期：2004-04-26资助项目：国家自然科学基金（70371075）作者简介：田志友(1974—)，男，河北石家庄人，博士研究生，主要研究方向：指数化评价，系统复杂性, Email: totzy@第3期 易腐商品最优订货批量与定价及其粒子群优化解 47 有较高的保存成本，持续时间越长则利润损失越大，因此，销售者需要在销售期初，综合考虑市场需求的波动、顾客的消费偏好，以及此类商品的销售期长短，制定合理的采购批量和售价，以确保实现其利润最大化目的.关于随机需求条件下易腐商品的订货量与定价问题，文献[1-4]给出了较为详尽的评述和一些共性结论，如：不同时段内顾客的到达服从不同质的随机分布，一般设为泊松分布；在同一时段内，顾客的感知价值是相互独立的，并服从某种同质概率分布；这种感知价值将随时间延续而不断降低，并且不同时段内的需求分布有可能发生质变.在满足上述假设条件下，通过经验数据可以获得销售期内的期望需求，进而可以在利润最大化原则下制定出相应的最优定价和最优订货批量.由于在订货与定价过程中，顾客的到达、对商品价值的感知等均具有不同形式的概率分布，往往导致有效需求的分布形式比较复杂，常规函数极值算法不易获得问题的解析解.本文将在文献[4]研究基础上，重点针对单一时段条件下最优订货批量与定价问题，推导易腐商品的利润最大化模型，并首次将粒子群优化算法引入该模型的求解过程，提出了两种不同的求解思路，以便互相印证，有效地获得单阶段最优订货批量与最优定价.2 建模2.1变量定义在对易腐商品最优订货批量与定价问题研究中，所涉及的若干变量及其含义如下：s ——易腐商品的订货量；t——正常销售期；ω——正常销售期内单位商品的定价；φ——超过正常销售期后的单位商品处理价；c——单位商品的综合成本(包括存储、运输、采购、处理等费用)；πs (ω)——订货量为s ，定价为ω时销售者的期望收入；R s (ω)——订货量为s ，定价为ω时销售者的期望利润，R s (ω)= πs (ω)-sc ；n——正常销售期内到访顾客的总人次，∞=,,2,1,0"n ;m——正常销售期内，定价为ω时的商品需求；P ω(m )——正常销售期内，定价为ω时，商品需求为m 的概率分布.2.2 需求分布与常见商品的需求分布相似，易腐商品的需求函数也应该满足如下两个条件.1) 需求与价格呈反方向变动，即00()(), 0,1,...,; 0k km m P m P m k ωω+∆==≤=∞∆>∑∑.2) 当价格趋向于无穷时，期望需求趋向于0，即lim (|)0E m ωωω→∞=，其中，(|)E m ω表示定价为ω时的期望需求.此外，影响易腐商品需求量的主要因素还包括：销售期t 内的到访顾客人数n ，以及所有到访顾客中可能会发生购买行为的人数m 等.关于顾客的到达，常见研究中均假设销售期t 内到访顾客的总人次n 属于系统外生变量，与定价无关，并服从参数为λ的泊松分布[3]，即：()(|), 0,1,2,...,!n tt e P n n n λλλ−==∞ (1) 考虑到不同时段内顾客到达率λ具有较大波动性，可假设λ服从参数为(α, β)的gamma 分布，即：(1)/1(), 0()g e αλβαλλλαβ−−=≤<∞Γ (2) 之所以选择gamma 分布，是因为到达率λ是一个非负取值的随机变量，并且，当参数α或β取某固定值时，原来的gamma 分布将相应地转化为2χ分布或指数分布.因此，选择gamma 分布可以涵盖较多的λ的变动情况[4].在所有到访顾客中，只有那些对商品的感知价值超过定价的顾客才会发生购买行为.设第i 位到访顾48 系统工程理论与实践 2005年3月 客的感知价值为X i ，0≤X i <∞，根据Gallego 等人的研究[3]，可以认为所有到访顾客的感知价值(X 1, X 2, …, X n )均为独立同分布的连续随机变量，概率密度设为f (x ).当定价为ω时，顾客感知价值的累积分布函数为F (ω)，并且满足：0()1 ; lim ()1F F ωωω→∞≤≤=.我们可以把销售期t 内的商品需求m 定义为：到达并愿意以当前定价购买一单位商品的潜在人次，即所有感知价值X i ≥ω的顾客人数.则潜在需求m 可以表示为一个服从二项分布的随机变量，分布概率为：()(|)[1()][()] , 0,1,2,...,.n m n m m P m n C F F m n ωωω−=×−×= (3)根据公式(1)-(3)，销售期t 内潜在需求m 的最终概率分布可以表示如下：001()(|)(|)()[1()]1 , 0,1,...,; 0,1,,;1[1()]1[1()]n m m m P m P m n P n g d t F C n m t F t F ωωλααλλλβωβωβω∞∞==+−=⎡⎤⎡⎤−=××=∞=∞⎢⎥⎢⎥+−+−⎣⎦⎣⎦∑∫" (4)可以看出，最终所得销售期内的需求m 服从一种负二项分布，其期望值为：()(1())E m t F w αβ=−.2.3 利润最大化模型当期初订货量为s 时，如果销售期t 内的潜在需求m ≥s ，则销售收入πs (ω)=s ω；如果m <s ，意味着部分商品将在销售期过后按处理价φ进行低价清仓，则此时的销售收入为：πs (ω)=m ω+(s -m )φ.综合两种情况，可得s 单位易腐商品在定价为ω时的期望收入为：10110010()()[()]() ()[()()()] ()()()s s m s m s s m m s m s P m m s m P m s s P m m P m s P m m P m s s m P m ωωωωωωωπωωωϕωωωϕϕωωϕ∞−==−−==−==++−=−++−=−−−∑∑∑∑∑ (5) 对应的销售利润为：1(1)0[1()]1()()()()()1[1()]1[1()]m s m s s m m t F R sc s c s m C t F t F ααβωωπωωωϕβωβω−+−=⎧⎫⎡⎤⎡⎤−⎪⎪=−=−−−−×⎨⎬⎢⎥⎢⎥+−+−⎣⎦⎣⎦⎪⎪⎩⎭∑ (6) 则最优订货量s *和最优定价ω*就是如下最大化模型的解：10max ()()()()().. >; 0;s s m R s c s m P m s t s ωωωωϕωϕ−==−−−−>∑ (7)由于潜在需求m 和顾客达到人次n 均为离散取值随机变量，顾客对商品的感知价值则为连续分布随机变量，如采用求偏导数等常规函数极值法，将很难获得问题的解析解.下面我们选用粒子群优化算法，对模型(7)进行演化求解.3 粒子群优化求解粒子群优化算法(Particle Swarm Optimization, PSO )是一种较新的全局优化方法，最早由Eberhart和Kennedy 博士于1995年提出[5].与遗传算法相比，粒子群算法没有交叉、变异等遗传操作，可调参数少，具有结构简单、运行速度快等特点，尤其适用于实数编码问题的求解.将其引入易腐商品最优订货批量与定价的求解过程，有利于快速有效地获得满意解.3.1 算法描述在粒子群算法中，待优化问题的每个潜在解均称为搜索空间中的一个粒子，每个粒子都用位置向量和速度向量来表示.其中，位置代表参数取值，速度表示各参数改进的方向和步长.算法首先通过随机初第3期 易腐商品最优订货批量与定价及其粒子群优化解 49 始化产生一群粒子，然后进行叠代寻优.在每一演化代中，粒子通过跟踪两个极值来不断更新自己：一个是粒子本身所找到的最优解，称为个体极值pBest ，另一个是整个种群目前为止所找到的最优解，称为全局极值gBest .然后根据下列公式来不断更新速度与位置：1()(Pr )2()(Pr )V w V c rand pBest esent c rand gBest esent =×+××−+××− (8) Pr Pr esent esent V =+ (9) 其中，V 是粒子速度，Present 是粒子当前位置，rand ()表示是(0, 1) 之间的随机数，c 1和c 2 被称作学习因子，通常，c 1 = c 2 = 2. w 表示加权系数，一般取值在0.1到0.9之间.叠代过程中，算法将根据粒子的适应度值不断更新pBest 与gBest 的取值，粒子在不断向全局最优转移的同时也不断向个体最优靠拢.当满足既定的终止规则，如达到预定演化代数、出现满足要求的满意解，或全局极值的改进步长小于指定阈值时，搜索过程结束，最后得到的gBest 就是最终的满意解[6].3.2 求解在利用粒子群算法求解模型(7)时，可以有两种不同的解题思路, 分别命名为枚举法和二维寻优法.3.2.1枚举法根据边际分析法，当边际收益不小于边际成本时，即：**11()()s s s s c πωπω−−−≥时，多订购一单位商品将增加销售商的净利润.因此，可以把利润视为价格ω的一元连续函数，利用粒子群算法依次计算不同订货量情况下所对应的最优定价与最大化利润，并将满足上述不等式的最大订货批量s *及其对应的定价*s ω作为最终解.设订货量的波动范围是：[s l , s u ]，求解流程描述如下.1) 令s =s l ;2) 利用粒子群算法求解一元连续函数10()()()()s s m R s s m P m ωωωωϕ−==−−−∑的最优订价*s ω与最大化利润*()s sR ω; 3) 判断：如果s<s u ，令s =s +1，转入步骤2)，否则继续;4) 令**()()s s s s R sc πωω=+，**11()()s s s s πωπω−−∆=−，然后确定满足c ∆≥的最大订货量s *及对应的*s ω; 5) 输出最终结果：[s *, *sω, *()s s πω, *()s s R ω]. 二维寻优法将利润R s (ω)视为订货量s 和定价ω的二元函数，利用粒子群算法在二维解空间进行演化寻优，从而同时确定最优订货量s *及最优定价ω*.求解过程如下：1) 个体解编码.每个粒子编码方式如下：p=[s, ω, v s , v ω, f ].其中，v s 和v ω,分别为s 和ω的运动速度，f 为该粒子的适应度值，即当订货量为s 、价格为ω时的销售利润;2) 粒子群初始化.设种群规模为popsize ，在订货量s 和价格ω的波动范围内随机取值popsize 组，作为初始种群，并随机初始化粒子的速度;3) 根据公式(6)，计算粒子的目标函数值;4) 根据公式(8)和(9)，更新粒子运动速度和所在位置;5) 更新当前演化代中的个体极值和全局极值;6) 检验终止规则.当满足指定规则时输出最终结果.否则，转入步骤3);7) 输出最优结果.最后一代种群中的gBest 即为满足利润最大化条件的最优解.4 算例分析设某种易腐商品的销售期t =1，单位成本c =6，超过正常销售期后的处理价φ=5.根据以往销售数据，销售期内顾客的到达率λ服从gamma 分布，参数为：α=3，β=2；顾客的感知价值X i 服从正态分布，分布参数为：µ=10，σ=1.则当订货量为s 、价格为ω时，利润函数为：10()(6)(5)()()s s m R s s m P m ωωωω−==−−−−∑ (10)50 系统工程理论与实践 2005年3月首先利用枚举法求解最优订货量与定价.算法参数设定如下：种群规模popsize=100；学习因子c1=c2=2；加权系数取固定值w=0.2.根据销售经验，设订货量s取值范围为[1，20]，售价ω取值范围为[c, 2c]，即[6, 12]；s和ω的速度取值范围分别为[0, 0.3]和[0，0.1]；算法终止规则为：当全局最优解改善程度小于0.01时终止运行.枚举法所得结果见表1.表 1 不同订货量情况下所对应的最优定价及最大化利润订货量s 1 2 3 4 5 6 7 8 9 10最优定价ω10.08 9.803 9.6039.4529.3359.2449.1719.114 9.069 9.033最大化利润R 3.38 5.877 7.6938.9449.72310.10910.1759.986 9.595 9.048订货量s11 12 13 14 15 16 17 18 19 20最优定价ω9.005 8.982 8.9658.9528.9418.9338.9278.923 8.92 8.917最大化利润R8.382 7.625 6.801 5.927 5.017 4.08 3.125 2.156 1.178 0.193从表1可以看出，当需求分布已知时，最大化利润将随订货量s增加而呈现先增后减趋势，而边际收益则持续下降.当边际收益等于边际成本时，所对应的最优订货量为7，最优定价为9.171，最大化利润为10.175.从图1中也可看出各变量的变动趋势.在二维寻优过程中，随着种群规模的增加，算法将逐渐收敛于全局最优，即订货批量为7，定价为9.17，这与枚举法所得结论是一致的.通过对算例的求解可以看出：枚举法的实质是在订货量s既定情况下，将利润视为价格ω的一元连续函数，利用粒子群算法进行一维优化，从而得到不同订货量下的最大利润与最优定价，然后根据边际分析，确定最优订货批量与定价.其特点是：准确度高，但效率较低，主要适用于订货量s波动范围较小的情况.二维寻优法的实质是将利润视为订货量和定价的二维函数，利用粒子群算法同时在s和ω所构成的二维解空间内寻找全局最优解,这种方法的特点是运行速度快，但容易停留在局部最优.因此，在实际操作中，建议同时采用两种方法，分别计算最优解，并互相印证，以便帮助销售者制定更为合理的采购与定价决策.5 结论1)在易腐商品订货量与定价研究中，最为关键的是准确估计销售期内的需求分布，而影响需求的主要因素包括：商品定价，顾客的到达情况，以及顾客对商品的感知价值.2)当其他参数不变时，对粒子群算法演化结果影响最大的参数就是种群规模.这是因为：粒子群算法是一种有导向的概率寻优方法，其探索未知空间的主要方式是通过跟踪全局极值和个体极值来实现的，增加种群规模有利于尽快发现全局最优.3)本文将整个销售期t视为一个完整阶段，并重点解决了最优订货批量与最优定价问题.由于顾客对易腐商品的感知价值将随时间的延续而不断缩减，不同时段内的需求分布有可能发生质变，而销售者也可以根据需求的波动、剩余库存量的多少等动态调整定价，以实现利润最大化目的.因此，对不同时段内、不同质需求条件下的最优定价与最优订货批量等问题，都还需要进一步的详细研究.第3期 易腐商品最优订货批量与定价及其粒子群优化解 51[1] Weatherford L R, Bodily S E. A taxonomy and research overview of perishable-asset revenue management: Yieldmanagement, overbooking, and pricing [J]. Operations Research,1992,40(5):831-844.[2] Wen Zhao, Yu-Sheng Zheng. Optimal dynamic pricing for perishable assets with nonhomogeneous demand [J].Management Science,2000,46(3):375-389.[3] Gallego G, Van Ryzin G. Optimal dynamic pricing of inventories with stochastic demand over finite horizons[J]. Management Science,1994,40(2):999-1020.[4] Young H Chun. Optimal pricing and ordering policies for perishable commodities [J]. European Journal ofOperational Research.2003,144(1):68-82.[5] Kennedy, R C Eberhart. Particle swarm optimization [A]. Proceedings of the 1995 IEEE InternationalConference on Neural Networks [C].Perth,Australia, 1942-1948.[6] 侯志荣, 吕振肃. 基于MATLAB的粒子群优化算法及其应用[J].计算机仿真,2003,20(10):68-70.Hou Zhirong, Lv Zhensu. Particle swarm optimization with application based on matlab [J]. Computers Simulation, 2003, 20(10): 68-70. (In Chinese)﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡﹡ (上接第6页)[7] Engle R F, Manganelli S. CAViaR: Conditional autoregressive value at risk by regression quantiles[R]. NBER,Working Paper 7341, 1999.[8] Hamilton J D. Time Series Analysis[M].Princeton University Press, New Jersey,1994.[9] Hans F and Alexander S. Stochastic Finance, An Introduction in Discrete Time[M].Walter de Gruyter,2002.[10]Hansen B E.Testing for parameter instability in linear models[J]. Journal of PolicyModeling,1992,14(4):517-533.[11] Inclan C and Tiao G C. Use of cumulative sums of squares for retrospective detection of changes ofvariances[J]. J. Amer. Statist. Assoc,1994, 89(3):913-923.[12] Jorion P. Value at Risk: The New Benchmark for Controlling Market Risk[M].Chicago: Irwin ProfessionalPublishing,1997.[13] Morgan J P. Risk Metrics Monitor[M]. 2nd Quarter,1996.[14] Koenker R, Bassett J. Regression quantiles[J]. Econometrica,1978,46(1):33-50.[15] Manganelli S, Engle R F. Value at risk models in finance[R]. working paper(75), European Central Bank,2001.[16] Schwert G W. Why does stock market volatility change over time?[J]. Journal of Finance,1988,XLIV(5):1115-1153.[17] Taylor S. Modeling Financial Time Series[M]. John Wiley and Sons, London, 1986.。
论文简介模板定稿
大青叶SRAP扩增体系的建立与优化专业名称：制药工程学号：08580208 姓名：王亮指导教师：杨中铎职称：副教授摘要目的：建立并优化大青叶的SRAP-PCR扩增体系。
方法:先对单因素（酶浓度，dNTP浓度，引物浓度，Mg2+浓度）进行筛选，然后用正交实验确立最优体系。
结果：在50μL的反应体系中，,Taq 酶的浓度为 4 U, dNTP的浓度为 0.20mmol/L，引物的浓度为 0.60μmol/L，Mg2+的浓度为1.0 mmol/L 时SRAP-PCR反应体系带型清晰，稳定性好。
本研究建立的反应体系将为大青叶的种质资源鉴定及筛选与4（3H）喹唑酮含量相关的基因奠定基础。
关键词：大青叶；基因组DNA； SRAP分子标记Abstract：Objective: To establish and optimize the SRAP amplification system of Folium . methods : First, to design the single factor experiments ( enzyme concentration , dNTP concentration , primer concentration of Mg2 + concentration) , and then using the orthogonal experiment to establish the optimal system . Results: in total 50μL reaction system , the Taq enzyme concentration is 4 U , dNTP concentration is 0.20mmol / L , the concentration of the Primer is 0.60μmol / L and Mg2 + concentration is 1.0 mmol / L . Under this condition , the amplification pattern with rich polymorphism and clearband is obtained, and have a good stability. Establish the the Folium Optimizational amplification system , and the success of this experiment wil make the foundation for Folium the germplasm and screening and 4 ( 3H) quinazolinone high content related gene.Keywords: Folium Genomic DNA SRAP markers Orthogonal Optimization一、综述1.1 实验研究的背景大青叶是我国的传统中药，其味苦、性寒、归心、胃经。
乳酸菌及其产生的细菌素
表 １ 乳酸菌的分子鉴别方法及其特性 [１]
Ｔａｂｌｅ １ Ｍｏｌｅｃｕｌａｒ ｔｙｐｉｎｇ ｔｏｏｌｓ ｏｆ ＬＡＢ ａｎｄ ｔｈｅｉｒ ｃｈａｒａｃｔｅｒｉｓｔｉｃｓ
分子工具
核糖体基因分型( ＲＴ)
扩增核糖体 ＤＮＡ 限制性分析( ＡＲＤＲＡ)
随机扩增多态性 ＤＮＡ( ＲＡＰＤ)
扩增片段长度多态性( ＡＦＬＰ)
脉冲场凝胶电泳( ＰＦＧＥ)
重复序列 ＰＣＲ( Ｒｅｐ￣ＰＣＲ)
变性 / 温度梯度凝胶电泳( ＤＧＧＥ / ＴＧＧＥ)
末端限制性片段长度多态性( Ｔ￣ＲＦＬＰ)
细菌多位点序列分型( ＭＬＳＴ)
实时 ＰＣＲ ( Ｒｅａｌ￣ｔｉｍｅ￣ＰＣＲ)
微阵列( Ｍｉｃｒｏａｒｒａｙ)
质谱( ＭＳ)
全基因组测序( ＷＧＳ)
化的主要产物是乳酸ꎬ而异型发酵中除乳酸外ꎬ还产
串珠 菌 属 ( Ｌｅｕｃｏｎｏｓｔｏｃ ) 、 肉 食 杆 菌 属 ( Ｃａｒｎｏｂａｃｔｅｒｉ￣
差异分为同型发酵或异型发酵ꎬ同型发酵中葡萄糖转
收稿日期:２０２２￣０４￣２０
作者简介:唐俊妮(１９７１￣) ꎬ女ꎬ博士ꎬ教授ꎬ研究方向:食品安全与食品微生物ꎬＥ￣ｍａｉｌ:ｊｕｎｎｅｙｔａｎｇ＠ ａｌｉｙｕｎ ｃｏｍꎬＯＲＣＩＤ ｉＤ:ｈｔｔｐｓ: / /
乳酸菌及其产生的细菌素
唐俊妮
( 西南民族大学食品科学与技术学院ꎬ四川 成都 ６１００４１)
摘 要:乳酸菌是用于生产发酵食品的革兰阳性和过氧化氢酶阴性的一类微生物ꎻ细菌素是由细菌核糖体合成的抗菌
肽ꎻ乳酸菌及其产生的细菌素已被广泛用作益生菌和抗菌剂ꎬ成为食品化学防腐剂和抗生素的潜在可行替代品. 本文
ｐｈｉｓｍꎬＴ￣ＲＦＬＰ) ꎬ细菌多位点序列分型( Ｍｕｌｔｉｌｏｃｕｓ ｓｅ￣
分子生物学技术鉴定药材
学术讨论 分子生物学技术鉴定药材王建云 熊丽娟1(昆明650011云南省药品检验所;1昆明650011昆明市中医院)摘要 目的:介绍分子生物学技术鉴定中药材的研究进展,供药学工作者参考。
方法:从DN A提取,D NA的R AP D分析及DN A序列测定三个方面进行综述,讨论了这些方法的优缺点。
结果与结论:认为分子生物学技术解决了动物药材鉴定的难题,将是今后中药材真伪鉴定技术发展的新方向。
关键词 D N A序列测定;RA P D技术;鉴定药材中药材的真伪优劣直接影响着中药的质量和疗效。
为此人们作了大量的研究,建立了一些确实有效的评价药材质量的方法[1],其鉴定标准也从最初仅有性状,逐渐增加了显微、理化、色谱、光谱等方法。
但是,对一部分药材,仅以上述方法是难以达到鉴定的目的,例如,鹿鞭[2]、前胡[3]、木蓝[4]等。
因此人们不断研究更完善、更准确、更具科学性的方法来解决药材鉴定问题。
本世纪80年代,分子分类学的问世,把生物分类的手段从形态表征,扩展到了生物的遗传物质DN A。
由于同种生物体具相同的DN A系列,不同种生物具不同的D N A系列,这便使人们可依据DN A系列的差异来鉴定生物物种。
药材除矿物类外,大多数来源于动物和植物。
但由于动植物药材多数是死亡的干燥生物体或生物的组织器官,在生物死亡的过程中,细胞会产生核酸酶大量降解DN A,这给药材D N A的分析带来困难。
随着分子生物学技术的飞速发展,特别是微量D NA提取技术[5~7]和多聚酶链式反应(polymer ase chain reactio n,P CR)技术[8]的发展,使人们能够从药材中提取微量的DN A进行分析,这就为用D NA分析技术鉴定药材提供了可能。
近4年来,这方面的研究集中在药材DN A提取、随机扩增多态DN A(random amplified polym orphic D N A,PA PD)和DN A序列测定几个方面,现简述如下。
基于随机森林与粒子群算法的隧道掘进机操作参数地质类型自适应决策
第 54 卷第 4 期2023 年 4 月中南大学学报(自然科学版)Journal of Central South University (Science and Technology)V ol.54 No.4Apr. 2023基于随机森林与粒子群算法的隧道掘进机操作参数地质类型自适应决策刘明阳，陶建峰，覃程锦，余宏淦，刘成良(上海交通大学 机械与动力工程学院，上海，200240)摘要：考虑到隧道掘进机的性能对地质条件比较敏感且其操作依赖于司机经验，提出基于随机森林和粒子群算法的隧道掘进机操作参数地质条件自适应决策方法。
利用随机森林(RF)分别建立地质类型、操作参数与推进速度、刀盘转矩的映射关系模型；结合映射关系模型，构建以盾构机推进速度最大为目标，以刀盘转速、螺旋输送机转速、总推力、土仓压力4个操作参数为控制变量的优化方程；利用粒子群算法(PSO)求解各地质类型地层中的最优操作参数决策结果。
通过新加坡某地铁工程施工数据验证所提方法的有效性和优越性。
研究结果表明：建立的随机森林模型中推进速度和刀盘转矩预测的决定系数R 2分别达到0.936和0.961，均大于adaboost 、多元线性回归、岭回归、支持向量回归和深度神经网络模型中相应的R 2；基于粒子群算法的操作参数决策方法能够准确求解操作参数最优解，寻优用时均比遗传算法、蚁群算法和穷举法的短。
本文所提决策方法使隧道掘进机在该施工段的福康宁卵石地层、句容地层IV 、句容地层V 、海洋黏土地层中的推进速度分别提升了67.2%、41.8%、53.6%和15.0%。
关键词：隧道掘进机；操作参数决策；随机森林；粒子群优化中图分类号：TH17；TU62 文献标志码：A 文章编号：1672-7207（2023）04-1311-14Geological adaptive TBM operation parameter decision based onrandom forest and particle swarm optimizationLIU Mingyang, TAO Jianfeng, QIN Chengjin, YU Honggan, LIU Chengliang(School of Mechanical Engineering, Shanghai Jiaotong University, Shanghai 200240, China)Abstract: Considering that the performance of TBM is affected by geological condition and driver experience, a geological adaptive TBM operation parameter decision based on random forest(RF) and particle swarm optimization algorithm(PSO) was proposed. RF was used to establish the mapping relation model between geological types, operating parameters and thrust speed, cutter head torque. An optimization equation was established using the mapping relationship model in which the maximum TBM thrust speed was taken as the target, and cutterhead speed, screw conveyor speed, total thrust and earth pressure were taken as control variables.收稿日期： 2022 −06 −19； 修回日期： 2022 −08 −21基金项目(Foundation item)：国家重点研发计划项目(2018YFB1702503) (Project(2018YFB1702503) supported by the National KeyR&D Program of China)通信作者：陶建峰，博士，教授，从事机械电子工程研究；E-mail ：**************.cnDOI: 10.11817/j.issn.1672-7207.2023.04.010引用格式： 刘明阳, 陶建峰, 覃程锦, 等. 基于随机森林与粒子群算法的隧道掘进机操作参数地质类型自适应决策[J]. 中南大学学报(自然科学版), 2023, 54(4): 1311−1324.Citation: LIU Mingyang, TAO Jianfeng, QIN Chengjin, et al. Geological adaptive TBM operation parameter decision based on random forest and particle swarm optimization[J]. Journal of Central South University(Science and Technology), 2023, 54(4): 1311−1324.第 54 卷中南大学学报(自然科学版)PSO was used to solve the optimal combination of operating parameters for each geological type. The validity and superiority of the proposed method were verified by the construction data of a subway project in Singapore. The results show that the R2 of the driving speed and cutter head torque predicted by random forest model reaches 0.936 and 0.961, which are greater than those of adaboost, multiple linear regression, ridge regression, SVR and DNN. PSO can accurately solve the optimal solution of operating parameters, and the time consumption is shorter than that of genetic algorithm, ant colony algorithm and exhaustive algorithm. By using the proposed method, the TBM thrust speed increases by 67.2%, 41.8%, 53.6%, 15.0% in the strata of Fokonnen Pebble Formation, Jurong Formation IV, Jurong Formation V and Marine Clay Formation in this construction section, respectively.Key words: tunnel boring machine; operating parameter decision; random forest; particle swarm optimization隧道掘进机是一种大型隧道掘进装备，具有开挖速度快、自动化程度高、施工质量好的优点，广泛地被应用于地铁、铁路、公路等隧道工程中[1]。
PCR及相关技术
Pwo DNA聚合酶 从喜温性古细菌中发现，现在市场销售的Pwo DNA聚合酶是在E.coli中表达后提取的产物。 1) 具有强的5’-3’聚合酶活性，兼有3’-5’外切酶活性。 2) 具有高的耐热性，100度下2小时后仍有一半的活性。 3) 可扩增5kb的片段，扩增DNA的准确度比Taq DNA聚合酶高约10倍。 4) 且其扩增产物为平末端，可直接用于平末端连接。 C.therm.聚合酶 1) 是一种以Mg2+为辅助因子的逆转录酶，最适温度在60-70度之间。在RT-PCR系统中催化逆转录反应。 2) 具有3’-5’外切酶活性，合成cDNA准确度比Tth DNA聚合酶高约2倍。
与形态标记、细胞学标记相比，生化标记具两个方面的优点： 一是表现近中性，对植物经济性状一般没有大的不良影响； 二是直接反映了基因产物差异，受环境影响较小。但目前可使用的生化标记数量还相当有限，且有些酶的染色方法和电泳技术有一定难度，因此其实际应用受到一定限制。
四、分子标记（DNA标记） 在遗传学研究中广泛应用的DNA分子标记已经发展了很多种，一般依其所用的分子生物学技术大致可以分为两大类：
Taq DNA聚合酶的特性
Tth DNA 聚合酶的特点： 从喜温性真菌中分离获得 1) 具有强的5’-3’聚合酶活性，缺3’-5’核酸外切酶活性。 2) 反应pH值为9，最适温度为75度。 3) 在Mn2+存在时，具有很强的反转录酶活性。 4) 当系统同时存在Mg2+时，反转录产生的cDNA在这种酶的作用下还可以进行PCR扩增。 故可用于RT-PCR。
英文缩写
英文名称
中文名称
RFLP
Restriction fragment length polymorphism
分子标记技术及应用
分子标记技术和应用一、基于DNA杂交技术的分子标记RFLP (Restriction Fragment Length Polymorphism,DNA限制片段长度多态性) RFLP是以Southern杂交为核心,应用最早的分子标记技术。
RFLP首先是在人类基因组研究中发展起来的,主要用于遗传疾病诊断和法医鉴定，RFLP的概念由人类遗传学家Botstein等首次提出，其原理为:碱基的改变与染色体结构的变化导致生物个体或种群之间DNA片段酶切位点的变化,用限制性内切酶切割改变的DNA,将产生长短、种类、数目不同的限制性片段,这些片段经聚丙烯酰胺凝胶电泳分离后就会呈现出不同的带状分布,而具有差异的DNA片段就可通过Southern杂交检测出来。
利用RFLP技术可进行遗传图谱构建、基因定位、数量性状基因座定位(QTL)及遗传多态性分析等。
RFLP标记具有下列优点：结果可靠，这是由于限制性内切核酸酶识别序列的专一性决定的。
结果稳定，RFLP标记无表型效应，其检测不受外界条件、性别及发育阶段的影响。
RFLP标记的等位基因间是共显性的，对选择隐形基因极为有利。
RFLP标记的非等位基因之间不存在基因互作，标记互不干扰。
RFLP起源于基因组DNA的自然变异，这些变异在数量上几乎不受限制，而且可利用的探针很多，可以检测到很多遗传位点。
但RFLP标记也有自身的不足：需要大量高纯度的DNA （5-10μg）。
所需仪器设备较多、检测步骤多、技术较复杂，周期长、成本高。
通常都用到同位素，对人体有一定的伤害。
具有种属特异性，且只适合单拷贝和低拷贝基因。
多态性产生的基础是限制性酶切位点的丢失或获得，所以RFLP多态位点数仅1或2个，多态信息含量低。
二、基于PCR技术的分子标记技术1.基于随机引物PCR的分子标记技术在聚合酶链式反应（PCR）技术发明后（1987年，Mullis和Faloona），由于PCR技术操作简单，成功率较高，出现了一大批以PCR技术为基础的分子标记。
一种联合优化的球形解码预处理方法
是 nR ∋ nR 的正交 矩阵, Q1 和 Q2 分别为 Q 的 前 nT 列和 后 nR
- nT 列。在半径为 r 超球内的格点满足如下关系:
# y& - Rx # 2 + # QT2 y # 2 ∀ r2
# y& - Rx # 2 ∀ r&2
(5)
这里, r&2 = r2 - # QT2 y # 2 。设 R 的元素为 r ij , ( i ∀ j ) , 则不等
计算机科学 2007V ol 34 7
一种联合优化的球形解码预处理方法* )
程 波 杨宗凯 刘 威 ( 华中科技大学电子与信息工程系 武汉 430074)
摘 要 在多输入 多输出( M IM O) 系统中, 球形 解码( SD) 是一种 非常有 吸引力 的解码 方法。 本文提 出了 一种针 对 SD 的联合优化预处理方法, 这种优 化方法的核心是以一种联合的方式, 通过一个 排序迫零 反馈均衡( O ZF DF E) 检测 器来获得 SD 优化的初始搜索半径和优化的解 码顺序。这个 O ZF DF E 区别 于其它 的预处 理方法 的一个 显著特 点是 它内嵌在通用 SD 中。本文同时也提出了对 SD 本身的一种改进 方法。仿真结果 表明这种 联合优 化的策 略在复 杂度 上相比其他优化方法, 具有更优的性能。 关键词 预处理, 球形解码, 多输入多输出系统, 无 线通信
y&- !i + r ii
dc- Ti
} , Si = 0} , g oto 步骤 4} 。
nT
步骤 6. ( 减少 i: 移 入上一层) IF i ) 1 T HEN { !i- 1 = ( j= 1
r i- 1,j x j , T i- 1 = T i + # y&i - !i - rii x i # 2 , i= i+ 1, g oto 步 骤 3}。
一种稳健恒定束宽宽波束形成算法
6DIGITCW2024.020 引言宽带数字波束形成技术是阵列信号处理中的关键技术之一，其在声呐、目标识别、导航等诸多领域之中都有着非常广泛的应用[1]。
目前，宽带信号的波束形成方式主要有两种，分别是频域波束形成以及时域波束形成[2]。
频域波束形成首先对接收数据进行离散傅里叶变换（），将信号变换至频域上，再分成多个窄带信号进行子带波束形成后进行宽带综合。
由于分段DFT 仅选择有限频带做子带窄带波束形成，因此分段DFT 波束形成输出的时间序列会出现不连续的情况，因此会出现波形失真的情况。
近年来，为保证在波束主瓣宽度内不失真地接收信号，研究学者提出恒定束宽波束形成技术[3]，即通过设计权系数值，保证主瓣宽度随频率的变化保持恒定，以保证主瓣区间内入射的不同频率下的信号经过波束形成之后不发生频谱失真[4]。
在雷达波速扫描的过程中，为了可以获得恒定的主瓣宽度并且确保尽可能低的旁瓣电平，文中提出了一种无约束的方向不变恒定束宽波束形成算法。
经仿真结果验证，这种算法可以满足优化后的不同频率的波束主瓣逼近生成的参考波束主瓣，同时尽量保持波束的低旁瓣特性。
1 信号模型与广义线性组合算法理论1.1 宽带基阵信号模型本文研究了由M 个阵元组成的间距为d 的均匀线性阵列（），每一个阵元后接阶数是L 的FIR 滤波器。
假设现在有D +1个远场宽带点源信号从D +1个方作者简介：张远驰（1998-），男，汉族，湖北宜昌人，硕士研究生，研究方向为阵列信号处理。
一种稳健恒定束宽宽波束形成算法张远驰，胡 进（中国船舶集团有限公司第七二四研究所，江苏 南京 210000）摘要：传统宽带波束形成算法在导向矢量失配时输出性能下降，为解决该问题，文章提出一种稳健恒定束宽波束形成算法。
该算法首先构造与快拍数相关的对角加载函数；其次，基于空域积分思想，结合入射信号的方向误差范围估计期望信号的实际入射方向，并结合构造的对角加载系数生成优化波束加权系数；最后，联合优化后的波束权值与FIR滤波器系数完成宽带波束响应的全局优化设计。
基于全离散粒子群优化的纳电子MPRM电路面积优化算法
基于全离散粒子群优化的纳电子MPRM电路面积优化算法作者：卜登立来源：《现代电子技术》2018年第04期摘要：针对具有较多输入数的可编程阵列结构纳电子混合极性Reed⁃Muller电路的面积优化问题，提出一种全离散粒子群优化算法。
通过将粒子速度合并到位置更新方程，充分挖掘粒子群优化中的学习因素得到全离散化的粒子更新方程，在此基础之上设计FDPSO算法，并使用探索概率作为算法参数控制算法全局探索与局部开拓间的平衡。
对一组输入数大于20的MCNC电路进行优化的实验结果表明，与其他能够用于可编程阵列结构纳电子混合极性Reed⁃Muller电路面积优化的智能算法相比，全离散粒子群优化算法具有较强的全局收敛能力和结果稳定性，能够以较高时间效率获得较好的优化结果。
关键词：纳电子； MPRM电路；面积优化；粒子群优化；更新方程；算法参数中图分类号： TN431.2⁃34； TP391.72 文献标识码： A 文章编号： 1004⁃373X（2018）04⁃0078⁃05Abstract： In allusion to the area optimization problem of programmable array⁃structured nanoelectronic mixed⁃polarity Reed⁃Muller （MPRM） circuit with many input numbers， a fully discretized particle swarm optimization （FDPSO） algorithm is proposed. The learning factors in particle swarm optimization are fully mined by integrating particle velocity into the location update equation to obtain the fully discretized particle update equation. On this basis， FDPSO algorithm is designed and the exploration probability is used as the algorithm parameter to control the balance between global exploration and local exploitation of the algorithm. The optimization experiment for microelectronics center of North Carolina （MCNC） circuit whose input numbers are larger than 20 was carried out. The results show that， in comparison with other intelligent algorithms that can be applied to area optimization of programmable array⁃structured nanoelectronic MPRM circuit，FDPSO algorithm has stronger global convergence capability and result stability， and can achieve better optimization results with higher time efficiency.Keywords： nanoelectronic； MPRM circuit； area optimization； particle swarm optimization； update equation； algorithm parameter0 引言混合极性Reed⁃Muller（Mixed⁃Polarity RM，MPRM）逻辑是乘积项的异或和表示[1]，因其逻辑表示的紧凑性及其电路实现的良好可测试性[2]，在算术电路、通信电路以及校验电路中得到了广泛应用[1]。
基于弹性网和直方图相交的非负局部稀疏编码
DOI： 10． 11772 / j． issn． 1001-9081． 2018071483
基于弹性网和直方图相交的非负局部稀疏编码
*பைடு நூலகம்
万 源，张景会 ，陈治平，孟晓静
( 武汉理工大学 理学院，武汉 430070) ( * 通信作者电子邮箱 Jingzhang@ whut． edu． cn)
摘 要： 针对稀疏编码模型在字典基的选择时忽略了群效应，且欧氏距离不能有效度量特征与字典基之间距离 的问题，提出基于弹性网和直方图相交的非负局部稀疏编码方法（ EH-NLSC） 。首先，在优化函数中引入弹性网模型， 消除字典基选择数目的限制，能够选择多组相关特征而排除冗余特征，提高了编码的判别性和有效性。然后，在局部 性约束中引入直方图相交，重新定义特征与字典基之间的距离，确保相似的特征可以共享其局部的基。最后采用多 类线性支持向量机进行分类。在 4 个公共数据集上的实验结果表明，与局部线性约束的编码算法（ LLC） 和基于非负 弹性网的稀疏编码算法（ NENSC） 相比，EH-NLSC 的分类准确率分别平均提升了 10 个百分点和 9 个百分点，充分体现 了其在图像表示和分类中的有效性。
Key words: sparse coding; elastic net model; locality; histogram intersection; image classification
0 引言
图像分类是计算机视觉领域的一个重要研究方向，广泛 应用于生物特征识别、网络图像检索和机器人视觉等领域，其 关键在于如何提取特征对图像有效表示。稀疏编码是图像特 征表示 的 有 效 方 法。考 虑 到 词 袋 （ Bag of Words，BoW） 模 型［1］和空 间 金 字 塔 匹 配 （ Spatial Pyramid Matching，SPM） 模 型［2］容易造成量化误差，Yang 等［3］ 结合 SPM 模型提出利用 稀疏编 码 的 空 间 金 字 塔 的 图 像 分 类 算 法 （ Spatial Pyramid Matching using Sparse Coding，ScSPM） ，在图像的不同尺度上 进行稀疏编码，取得了较好的分类效果。在稀疏编码模型中， 由于 1 范数在字典基选择时只考虑稀疏性而忽略了群体效 应，Zou 等［4］提出一种新的正则化方法，将弹性网作为正则项 和变量选择方法。Zhang 等［5］提出判别式弹性网正则化线性
多模态PSO自适应变异
Particle swarm optimization with adaptive mutation for multimodaloptimizationHui Wang a ,⇑,Wenjun Wang b ,Zhijian Wu caSchool of Information Engineering,Nanchang Institute of Technology,Nanchang 330099,PR China bSchool of Business Administration,Nanchang Institute of Technology,Nanchang 330099,PR China cState Key Laboratory of Software Engineering,Wuhan University,Wuhan 430072,PR Chinaa r t i c l e i n f o Keywords:Particle swarm optimization (PSO)Adaptive mutationMultimodal optimization Global optimizationa b s t r a c tParticle swarm optimization (PSO)is a population-based stochastic search algorithm,which has shown a good performance over many benchmark and real-world optimization problem.Like other stochastic algorithms,PSO also easily falls into local optima in solving complex multimodal problems.To help trapped particles escape from local minima,this paper presents a new PSO variant,called AMPSO,by employing an adaptive mutation strat-egy.To verify the performance of AMPSO,a set of well-known complex multimodal bench-marks are used in the experiments.Simulation results demonstrate that the proposed mutation strategy can efficiently improve the performance of PSO.Ó2013Elsevier Inc.All rights reserved.1.IntroductionParticle swarm optimization (PSO)[1]is a population-based stochastic search algorithm inspired by the social behaviors of fish schooling and birds flocking.For PSO’s simple concept,easy implementation yet effectiveness,it has been applied in many optimization areas [2].Many experiments have shown that the basic PSO algorithm easily suffers from premature convergence in solving com-plex multimodal problems [3].The main reason is that particles are dominated by their previous best particles and the global best particle.Once these best particles are stagnant,all particles in the swarm will quickly converge to the direction along the previous best particles and the global best particle.The velocity will rapidly decrease to 0.If the current convergence is not the global optimum,all particles will hardly escape from the local minima.Under this circumstance,the position updat-ing model does not work because the velocity is 0.To tackle this problem,some mutation techniques have been proposed to help trapped particles jump to better positions [4–7].However,most existing mutation techniques are not suitable for all kinds of problems.Because they are problem-oriented.For example,the Cauchy mutation is beneficial only when the global optimum is sufficiently far away from the current search point.To improve the flexibility of mutation in PSO,this paper presents an adaptive mutation strategy.In the new approach,three different mutation operators including Gaussian,Cauchy and Lévy,are utilized.To determine which mutation is the best choice for the current evolution,we calculate the selection ratio of each mutation operator.A mutation operator is cho-sen to be conducted on the pbest and gbest based on the selection ration.If a mutation operator is more suitable than others,it will obtain larger selection ratio.The rest of the paper is organized as follows.In Section 2,the basic PSO algorithm is introduced.In Section 3,a brief review of related work is presented.Section 4presents our proposed approach.The benchmark functions,parameter 0096-3003/$-see front matter Ó2013Elsevier Inc.All rights reserved./10.1016/j.amc.2013.06.074⇑Corresponding author.E-mail addresses:huiwang@ (H.Wang),wangwenjun@ (W.Wang),zhijianwu@ (Z.Wu).H.Wang et al./Applied Mathematics and Computation221(2013)296–305297 settings,experimental results and discussions on AMPSO are given in Section5.Finally,Section6concludes with a summary and further work.2.Particle swarm optimizationLike other evolutionary algorithms(EAs),PSO is also a population-based search algorithm and starts with an initial pop-ulation of randomly generated solutions called particles.Each particle in PSO has a velocity and a position.PSO remembers both the best position found by all particles and the best positions found by each particle in the search process.For a search problem in an D-dimensional space,a particle represents a potential solution.The velocity v ij and position x ij of the j th dimension of the i th particle are updated according to Eqs.(1)and(2)[8]:v ijðtþ1Þ¼wÁv ijðtÞþc1Árand1ijÁðpbestðtÞÀx ijðtÞÞþc2Árand2ijÁðgbest jðtÞÀx ijðtÞÞ;ð1Þijx ijðtþ1Þ¼x ijðtÞþv ijðtþ1Þ;ð2Þis the previous best position where X i is the position of the i th particle,V i represents the velocity of the i th particle,the pbestiof the i th particle,and gbest is the global best particle found by all particles so far.The inertia factor w was proposed by Shi and Eberhart[8],rand1ij and rand2ij are two random numbers independently generated within the range of[0,1],c1and c2 are two learning factors,and t¼1;2;...indicates the iterations.There are two main models of the PSO algorithm,cabled gbest(global best)and lbest(local best),respectively.The two models differ in the way of defining the neighborhood of each particle.Some studies demonstrated that these two models show different performances[2,9].For the attraction of the self-best particle,the gbest model has fast convergence rate but run a high risk of falling into local minima.For the neighborhood topology,the lbest model does not easily fall into local min-ima but show a slow convergence rate.By the suggestions of[10],the lbest model with a ring population topology is used in this paper.3.Related workSince the introduction of PSO,it has attracted many researchers to work on improving its performance.In the last decade, many variants of PSO have been proposed.A brief overview of these improved approaches is presented as follows.Shi and Eberhart[8]introduced a parameter called inertia weight w into the original PSO.The inertia weight is used to balance the global and local search abilities.From the analysis of[8],a linearly decreasing w over the evolutionary process is a good choice.Stacey et al.[4]has added a Cauchy random number in each dimension with probability1=D,where D is the number of dimensions.Higashi and Iba[5]has used a parameter based on the search space in one dimension to control Gaussian random numbers.Krohling[6]has proposed a new approach by means of applying Cauchy probability distribution. When a particle has no change in afixed number of generations,the particle should jump to a new point by adding a random number generated by Cauchy distribution.Cruz-Cortes et al.[11]has introduced Gaussian mutation and Cauchy mutation in an artificial immune system.The two mutation operators have simply added random numbers,generated by Gaussian dis-tribution and Cauchy distribution,to the solution in each dimension.Wang et al.[7,12]presented a novel adaptive Cauchy mutation,which used an average velocity of swarm to generate the step size of mutation.Cui et al.[13]presented a fast PSO algorithm(FPSO)which does not evaluate all new positions owning afitness and asso-ciated reliability value of each particle of the swarm and the reliability value is only evaluated using the truefitness function if the reliability value is below a threshold.Xi et al.[14]proposed an improved quantum-behaved PSO with weighted mean best position.Jie et al.[15]developed a knowledge-based cooperative PSO.Cui et al.[16]combinedfitness uniform selection strategy(FUSS)with random walk strategy(RWS)to solve high dimensional function optimization problems.Wang et al.[17]developed a new crossover operator to recombine the current position and its next position.The presented results show that the proposed approach can maintain the diversity of swarm by adding the dissimilarities among particles.Bergh and Engelbrecht[18]proposed a cooperative approach to PSO(CPSO-H)for solving multimodal problems.Liang et al.[3]introduced a comprehensive learning PSO(CLPSO)to learn other particles’experiences in different dimensions. Li and Yang[19]presented an adaptive learning PSO(ALPSO)for function optimization,in which the learning mechanism of each particle is separated into three parts:its own historical best position,the closest neighbor and the global best one.By using this individual level adaptive technique,a particle can well guide its behavior of exploration and exploitation. Hsieh et al.[20]presented an efficient population utilization strategy for PSO(EPUS-PSO),which introduced a population manager and a solution sharing strategy.In EPUS-PSO,the population size is variable.The population manager can increase or decrease particle numbers according to the searching status.Zhan et al.[21]presented an adaptive PSO(APSO)by employ-ing two strategies.Thefirst one evaluates the population distribution and particlefitness and identifies the current search status.The second one utilizes an elitist learning strategy to help the global best particle jump out of the likely local optima. Sun et al.[22]proposed an improved vector PSO(IVPSO)to solve constrained optimization problems.Wang et al.[23]pro-posed another improved CLPSO by employing a generalized opposition-based learning(GOBL).Li et al.[24]proposed a new adaptive PSO algorithm(SLPSO),in which each particle have a set of four strategies to cope with different situations in thesearch space.Experimental study on45benchmark functions show that SLPSO has a superior performance in comparison with several other peer algorithms.4.PSO with adaptive mutation4.1.Why PSO needs mutationIn the basic PSO,each particleflies to its previous best particle pbest and the global best particle gbest.With increasing of generations,particles become similar to pbest and gbest.It means that the values of pbestÀX and gbestÀX become small. Once these best particles get stuck in local optima,all particles in the current swarm will quickly converge to the local min-ima.To illustrate this phenomenon,Fig.1presents the changes of pbestÀX and gbestÀX on Ackley’s function during the evolution.Due to the limitation,this paper only presents the changes of the1st dimension of a particle in the swarm(the changes in other dimensions are similar to the1st dimension).From the results of Fig.1,the changes of pbestÀX and gbestÀX are very large at the beginning of evolution.That is help-ful to push the particles forward.With increasing of generations,the changes of these two parts become small.After about 100generations,the values of pbestÀX and gbestÀX tend to0.When these two parts are near to0,the velocity tends to0.If the current gbest is not the global optimum,the swarm will be stagnant for Xðtþ1Þ¼XðtÞþ0.Under this case,the trapped particles are difficult to jump to other positions for the zero velocity.To tackle this problem,we should break the equalities gbest¼¼X or pbest¼¼X and guarantee a non-zero velocity.To break the above equalities,we usually use some mutation techniques to generate mutant particles instead of gbest;X or pbest[4–7].It is to hope that the mutation could push the trapped particles forward.4.2.Proposed approachAs mentioned before,it is possible to help trapped particles jump to other positions by breaking the equalities gbest¼¼X or pbest¼¼X.The main idea behind this is to generate a non-zero velocity and keep the trapped particles move.To accom-plish this objective,this paper presents a novel mutation strategy to conduct mutation on pbest and gbest.If thefitness value of pbest or gbest does not improve in m generations(m is a predefined constant number),it demonstrates that the pbest or gbest maybe trapped into local minima.To help the trapped particles jump to other positions(which may not be located at local optima),a mutation operator is conducted on the pbest and gbest.The new strategy differs from the previous work [6,25],which mainly monitor the changes offitness value of each particle,while our proposed method monitors the changes of the pbest and gbest.To assure the pbest or gbest jump to better positions,an elitist selection is used.If the pbest or gbest jumps to a better position,then update the position of the old particle with the new position;otherwise keep the old particle unchangeable.Although many mutation techniques have been introduced to the basic PSO algorithm[4–6],these strategies are usually problem-oriented.For instance,the Gaussian mutation is suitable for local search.When the current search position is suf-ficiently far away from the global optimum,the Gaussian mutation may not work.To enhance theflexibility of mutation in PSO,the proposed approach employs multiple mutation strategies,including Gaussian,Cauchy and Lévy.The former one is good at local search,while the latter two are beneficial for global search.For a given problem,it is difficult to determine which mutation is more suitable during different evolutionary stages.At the initial stage,the Cauchy or Lévy mutation is better than the Gaussian for their large mutation step sizes.As the iterations increase,the algorithm gradually converges to the neighborhood of the global optimum.Therefore,the algorithm should switch from the Cauchy or Lévy mutation to the Gaussian to enhance the neighborhood search.However,it is not easy to determine when the algorithm conducts the switch.In this paper,an adaptive mutation strategy is proposed to dynamically choose different mutation operators during the evolution.Gaussian Mutation–the Gaussian mutation is defined by:pbest0ij¼pbest ijþgaussian jðÞ;ð3Þgbest0j¼gbest jþgaussian jðÞ;ð4Þwhere pbesij is the j th vector of pbesti;gbestjis the j th vector of gbest,and gaussianjðÞis a random number generated by Gauss-ian distribution.Cauchy Mutation–the Cauchy mutation is defined by:pbest0ij¼pbest ijþcauchy jðÞ;ð5Þgbest0j¼gbest jþcauchy jðÞ;ð6Þwhere cauchyðÞj is a random number based on Cauchy distribution with a scale parameter t¼1[26].298H.Wang et al./Applied Mathematics and Computation221(2013)296–305H.Wang et al./Applied Mathematics and Computation221(2013)296–305299Lévy Mutation–the Lévy mutation is defined by:¼pbest ijþle v y jðÞ;ð7Þpbest0ij¼gbest jþle v y jðÞ;ð8Þgbest0jwhere le v yðÞj is a random number generated by Lévy distribution with a parameter a¼1:3[27].Algorithm1.PSO with Adaptive Mutation(AMPSO)1Randomly initialize each particle2FEs¼N3for i¼1to N do4pbest i¼P i5lastpbest i¼pbest i6monitor[i]=07end8gbest¼f P i j min fj fðP iÞgg9lastgbest¼gbest10monitor[N+1]=011while FEs6MAX FEs do12Standard PSO(gbest model)13FEs¼FEsþN14for i¼1to N do15if fðpbest iÞ¼¼fðlastpbest iÞthen16monitor[i]++;17end18else19monitor[i]=020lastpbest i¼pbest i21end22if monitor[i]P m then23Determine which mutation to be conducted according to the selection ration pg;pc and pl24Conduct a mutation on the pbest i according to the selected mutation operator25FEs++is better than pbest i then26if pbest0i27Replace pbest i with pbest0i28end29Update pbest i and lastpbest i if needed30end31end32if fðgbestÞ¼¼fðlastgbestÞthen33monitor[N+1]++34end35else36monitor[N+1]=037lastgbest¼gbest38end39if monitor[N+1]P m then40Determine which mutation to be conducted according to the selection ration pg;pc and pl41Conduct a mutation on the gbest according to the selected mutation operator42FEs++43if gbest0is better than gbest then44Replace gbest with gbest045end46Update gbest and lastgbest if needed47end48endIn our approach,each mutation operator has an independent selection ratio.For Gaussian,Cauchy and Lévy mutation,their selection ratios are defined by pg ;pc and pl ,respectively (pg þpc þpl ¼1).At the initial stage,pg ;pc and pl are set to 13.It means that each mutation operator has the same chance to be selected.During the evolution,the selection ratio of each mutation is related to its successful probability.The pg ;pc and pl are defined as follows.pg ¼c þð1À3Ác ÞÁsuc gsuc m ;ð9Þpc ¼c þð1À3Ác ÞÁsuc csuc m;ð10Þpl ¼c þð1À3Ác ÞÁsuc lsuc m;ð11Þwhere suc g is the successful number of Gaussian mutations,suc c is the successful number of Cauchy mutations,suc l is the successful number of Lévy mutations,suc m is the number of successful mutations,the parameter c is the minimum selec-tion ration for each mutation operator,and suc g þsuc c þsuc l ¼suc m .The pg ;pc and pl are updated after each generation.In this paper,c is set to 0.05based on empirical studies.To choose the best mutation operator,a roulette wheel selection mechanism is employed.A mutation operator with larger selection ratio wins more chances to be selected.For example,the Schwefel function has deep local optima being far from the global optimum.For this problem,the Gaussian mutation with small step size can hardly help trapped par-ticles jump out of the deep local optima,while the Cauchy or Lévy mutation is possible to do for their large mutation sizes.Therefore,the pg will be smaller than pc and pl .During the evolution,the Gaussian mutation will have smaller chances to be selected.The main steps of our approach AMPSO are described in Algorithm 1,where N is the population size,P i is the i th particle in the swarm,lastpbest i indicates the last fitness value of the pbest i ;lastgbest records the last fitness vale of the gbest ,monitor [i ]records the successive number of iterations where the fitness value of pbest i does not change,monitor [N +1]re-cords the successive number of iterations where the fitness value of gbest does not change,m is the maximum value of monitor ½i and monitor [N +1],FEs is the number of fitness evaluations,and MAX FEs is the maximum number of fitness evaluations.5.Experimental studies 5.1.Test functionsIn order to verify the performance of our approach,a set of well-known benchmark functions is used in the experiments.These functions were early considered in [3,28].All test functions are to be minimized.The description of the benchmark functions and their global optimum (s)are listed in Table 1.300H.Wang et al./Applied Mathematics and Computation 221(2013)296–305H.Wang et al./Applied Mathematics and Computation221(2013)296–305301Table1The test functions used in the experiments,where D is the dimension of the functions,and X2R D is the definition domain.F Name D Xf1Rosenbrock’s function30½À2:048;2:048 f2Ackley’s function30½À32:768;32:768 f3Griewanks’s function30½À600;600 f4Weierstrass function30½À0:5;0:5 f5Rastrigin’s function30½À5:12;5:12 f6Noncontinuous Rastrigin’s function30½À5:12;5:12 f7Schwefel’s function30½À500;500 f8Rotated Ackley’s function30½À32:768;32:768 f9Rotated Griewanks’s function30½À600;600 f10Rotated Weierstrass function30½À0:5;0:5 f11Rotated Rastrigin’s function30½À5:12;5:125.2.Effects of different mutation operatorsTo investigate the effects of different mutation operators used in PSO,three PSO variants are compared with AMPSO in the experiments.The related algorithms and their parameter settings are listed as follows.PSO with Gaussian mutation(GPSO).PSO with Cauchy mutation(CPSO).PSO with Lévy mutation(LPSO).PSO with adaptive mutation(AMPSO).For the sake of a fair comparison,the same parameter settings are used for all algorithms.By the suggestions of[29], w¼0:7298;c1¼c2¼1:49618.The population size is set to40.The maximum numberfitness evaluations(MAX FEs)is set to100,000.For AMPSO,the parameters m and c are set to10and0.05,respectively based on empirical studies.All the experiments are conducted30times,and the mean error values fðxÞÀfðx oÞ(fðx oÞis the global optimum of fðxÞ)are recorded.Results for PSO with different mutation strategies are listed in Table2.The comparison results among AMPSO and other algorithms are summarized as‘‘w=t=l’’in the last row of the table,which means that AMPSO wins in w functions,ties in t functions and loses in l functions,compared with its competitors.As seen,AMPSO surpasses GPSO,CPSO and LPSO on all test functions.Among GPSO,CPSO and LPSO,GPSO only performs better then other two algorithms on f1.CPSO achieves better results than GPSO and LPSO on three functions(f3;f5and f9).For the rest7functions,LPSO obtains better results than GPSO and CPSO.It demonstrates that their is no afixed mutation oper-ator for a given problem.For Rosenbrock function(f1),Gaussian mutation is more suitable for Cauchy and Lévy,while it does not show any advantage for the rest functions.By hybridization of these three mutation operators,AMPSO obtains better performance than PSO with a single mutation operator.Fig.2shows the convergence characteristics of GPSO,CPSO,LPSO and AMPSO on selected functions.It is obvious that AMPSO converges faster than other three algorithms.Fig.3presents the results of selection ratio of Gaussian,Cauchy,and Lévy mutation operations in AMPSO.It can be seen that the frequencies of three mutation operators being selected are quite different for each test function.For Rosenbrock function(f1),the Gaussian mutation wins more chances of being selected than Cauchy and Lévy during the evolution.ForTable2Results for PSO with different mutation strategies on the test suite,where‘‘Mean’’indicates the mean errorfitness value,and‘‘w=t=l’’means that AMPSO wins in w functions,ties in t functions and loses in l functions,compared with its competitors.The best results among the comparison are shown in bold.Functions GPSO mean CPSO mean LPSO mean AMPSO meanf1 2.37E+01 2.39E+01 2.75E+01 2.23E+01 f2 1.62EÀ07 1.09EÀ07 5.89EÀ08 2.34EÀ08 f3 5.74EÀ10 1.33EÀ117.86EÀ11 4.71EÀ12 f49.18EÀ05 6.23EÀ05 1.89EÀ05 5.62EÀ06 f5 3.68E+01 2.69E+01 3.98E+01 2.29E+01 f6 1.02E+017.15E+00 4.02E+00 5.43EÀ02 f7 3.47E+03 2.39E+03 1.93E+03 1.87E+03 f87.52EÀ07 2.78EÀ07 2.33EÀ07 3.84EÀ08 f9 1.97EÀ027.40EÀ03 1.23EÀ02 2.81EÀ10 f109.68E+008.37E+007.56E+00 5.82E+00 f11 6.57E+01 5.88E+01 5.49E+01 5.13E+01 w=t=l11=0=011=0=011=0=0–Griewank function(f3),the selection ratios of Cauchy and Lévy mutations keep at a high level during the evolution.For non-continuous Rastrigin function(f6),the Lévy mutation obtains higher selection ratio than Gaussian and Cauchy.For Schwefel function(f7),the Gaussian mutation keeps at a very low level during the evolution.The above results confirm that one mutation operator may perform better than other ones on a certain type of problems, but worse than another type of problems.For a given problem,different mutation operator maybe only suitable for different evolutionary stages.By utilizing adaptive mutation operators,the proposed mutation strategy is suitable for more types of problems.parison of AMPSO with other PSO variantsIn this section,we compare the performance of AMPSO with three well-known PSO variants on the test suite.The in-volved algorithms and parameter settings are listed as follows.Cooperative PSO(CPSO-H)[18].Comprehensive Learning PSO(CLPSO)[3].Adaptive PSO(APSO)[21].Our approach AMPSO.The parameter settings of CLPSO-H and CLPSO are described in[3].By the suggestions of[21],the same parameter settings of APSO are used.For AMPSO,w¼0:7298,c1¼c2¼1:49618.The parameters m and c are set to10and0.05,respectively based on empirical studies.The above four PSO algorithms use the same population size(N¼40)and maximum number offitness evaluations(MAX FEs¼200;000)[3].All the experiments are conducted30times,and the mean error values are recorded.H.Wang et al./Applied Mathematics and Computation221(2013)296–305303Table3Comparison among four PSO algorithms on the test suite,where‘‘Mean’’indicates the mean errorfitness value,and‘‘w=t=l’’means that AMPSO wins in w functions,ties in t functions and loses in l functions,compared with its competitors.The best results among the comparison are shown in bold.Functions CPSO-H[18]mean CLPSO[3]mean APSO[21]mean AMPSO meanf1 1.37E+01 2.08E+01 1.83E+01 1.76E+01 f2 2.25EÀ14 1.85EÀ07 1.09EÀ147.63EÀ15 f3 1.90EÀ02 4.37EÀ09 1.20EÀ020.00E+00 f4 4.74EÀ15 5.62EÀ07 4.77EÀ020.00E+00 f5 3.32E+00 1.50EÀ04 6.27E+00 1.86E+01 f6 6.67EÀ01 1.93EÀ03 2.27E+00 1.24EÀ11 f7 2.65E+03 3.88EÀ04 3.74E+02 1.71E+03 f8 1.82EÀ01 1.07EÀ05 1.22E+00 1.12EÀ14 f9 2.30EÀ02 6.49EÀ05 1.38EÀ02 1.11EÀ16 f108.33E+00 2.99E+008.40E+00 3.89E+00 f117.33E+01 5.48E+017.09E+01 4.78E+01 w=t=l9=0=28=0=39=0=2–Results of mean error values achieved by the four PSO algorithms are listed in Table3.The comparison results among AMPSO and other algorithms are summarized as‘‘w=t=l’’in the last row of the table,which means that AMPSO wins in w functions,ties in t functions and loses in l functions,compared with its competitors.304H.Wang et al./Applied Mathematics and Computation221(2013)296–305Table4Average rankings achieved by Friedman test for the four PSO algorithms.The highest ranking is shown in bold.Algorithms RankingsAMPSO 3.36CLPSO 2.82CPSOH 2.00APSO 1.82Table5Wilcoxon test between AMPSO with other PSO algorithms on functionsf1–f11.The p-values below0.05are shown in bold.DNSPSO vs.p-valuesCPSO-H 1.55EÀ01CLPSO 6.57EÀ01APSO 2.85EÀ01From the results of Table3,AMPSO outperforms CPSO-H and APSO on9functions,while CPSO-H and APSO only achieves better results on2functions.CLPSO obtains better performance than AMPSO on3functions,while AMPSO surpasses CLPSO for the rest8functions.All algorithms fall int local minima on Rosenbrock function(f1),which is a non-convex problem and its global optimum is inside a long,narrow,parabolic shapedflat valley.Most PSO algorithms can easilyfind the valley,but hardly converge to the global optimum.The rotation seriously affects the performance of most algorithms.For Ackley function(f2),all algorithms obtain promising solutions,while CPSO-H and APSO hardly achieve good results on its rotated problem f8.For Griewank function(f3),its rotation does not affect the performance of all algorithms.For Rastrigin and Weierstrass functions(f4and f5),all algorithms are seriously affected by the rotation.In order to compare the performance of multiple algorithms on the test suite,we conducted Friedman and Wilcoxon tests according to the suggestions of[30].Table4shows the average ranking of CPSO-H,CLPSO,APSO and AMPSO.The highest ranking is shown in bold.As seen,the performance of the four algorithms can be sorted by average ranking into the following order:AMPSO,CLPSO,CPSO-H,and APSO.The highest average ranking is obtained by the AMPSO algorithm.It demonstrates that AMPSO is the best one among the four PSO algorithms.To compare the performance differences between AMPSO and the other three PSO algorithms,we conduct a Wilcoxon signed-rank test[31].Table5shows the resultant p-values when comparing between AMPSO and the other three algorithms. The p-values below0.05are shown in bold.From the results,it can be seen that AMPSO is not significantly better than other algorithms,AMPSO performs better according to the average rankings shown in Table4.6.ConclusionsThe movement of PSO is dominated by pbest and gbest.Once these particles fall into local minima,all particles will quickly converge to the direction along these best particles.Under this case,the velocity of each particle will tend to zero.The posi-tion updating model of particles does not work.To help trapped particles jump to another position which may not be located at the local optima,this paper presents an adaptive mutation strategy to dynamically adjust the selection rations of three different mutation operators.Experimental verifications on a set of well-known complex multimodal functions show that our approach could achieve promising solutions on the majority of test problems.The results also confirm that one mutation operator is not suitable for all kinds of problems.By utilizing adaptive mutation operators,our approach can be suitable for more kinds of problems.For the parameters m and c,this paper employs empirical values,but they may affect the performance of AMPSO.How to choose the best m and c will be investigated in the future work.AcknowledgmentsThe authors thank the editor and anonymous reviewers for their detailed and constructive comments that help us to in-crease the quality of this work.This work is supported by the Humanity and Social Science Foundation of Ministry of Edu-cation of China(No.13YJCZH174),the Science and Technology Plan Project of Jiangxi Provincial Education Department(No. GJJ13744),and the National Natural Science Foundation of China(Nos.61070008,61175127,61261039).References[1]J.Kennedy,R.C.Eberhart,Particle swarm optimization,in:Proceedings of IEEE International Conference on Neural Networks,1995,pp.1942–1948.[2]R.Poli,J.Kennedy,T.Blackwell,Particle swarm optimization:an overview,Swarm Intelligence1(1)(2007)33–58.。
辣椒抗PMMoV基因L4连锁标记的验证分析
ｈ ｔｔｐ: / / ｊｓ ｎ ｙ ｘ ｂ.ｊａ ａ ｓ.ａ ｃ.ｃ ｎ
１６２７
朱雪梅ꎬ郭广君ꎬ潘宝贵ꎬ等. 辣椒抗 ＰＭＭｏＶ 基因 Ｌ４ 连锁标记的验证分析[ Ｊ] .江苏农业学报ꎬ２０２２ꎬ３８(６) :１６２７￣１６３６.
收稿日期:２０２２￣０４￣０７
基金项目:国家自然科学基金面上项目( ３２０７２５９７) ꎻ江苏省种业振
兴揭榜挂帅项目[ ＪＢＧＳ( ２０２１) ０６５] ꎻ国家现代农业产业
技术体系建设专项( ＣＡＲＳ￣２３￣Ｇ４２)
作者简介:朱雪梅 (１９９６－) ꎬ女ꎬ山东滨州人ꎬ硕士研究生ꎬ主要研究
方 向 为 辣 椒 遗 传 育 种 和 分 子 生 物 学ꎮ ( Ｅ￣ｍａｉｌ )
ｕｎｓｔａｂｌｅ ａｎｄ ｃｏｕｌｄ ｅａｓｉｌｙ ｃａｕｓｅ ｆａｌｓｅ ｐｏｓｉｔｉｖｅｓꎬ ｓｏ Ｌ４ＳＣ３４０ ｗａｓ ｕｎｓｕｉｔａｂｌｅ ｆｏｒ ｔｈｅ ａｓｓｉｓｔｅｄ ｓｃｒｅｅｎｉｎｇ ｏｆ Ｌ４ ｒｅｓｉｓｔａｎｔ ｇｅｎｅ.
Ｍａｒｋｅｒｓ ０８７Ｈ３Ｔ７ ａｎｄ Ｌ４￣ＳＣＡＲ ｃｏｕｌｄ ｂｅ ｕｓｅｄ ｔｏ ｓｃｒｅｅｎ ｐｅｐｐｅｒ ｇｅｒｍｐｌａｓｍｓ ｃａｒｒｙｉｎｇ Ｌ３ ａｎｄ Ｌ４ ｒｅｓｉｓｔａｎｔ ｇｅｎｅｓꎬ ｂｕｔ ｔｈｅｙ
感病材料 Ｍｉｅ￣ｍｉｄｏｒｉ 构建 Ｆ２ 群体ꎬ鉴定出 １ 个随机
扩增 多 态 性 ＤＮＡ ( Ｒａｎｄｏｍ ａｍｐｌｉｆｉｅｄ ｐｏｌｙｍｏｒｐｈｉｓｍ
多 态 性 ( Ａｍｐｌｉｆｉｅｄ ｆｒａｇｍｅｎｔ ｌｅｎｇｔｈ ｐｏｌｙｍｏｒｐｈｉｓｍꎬ
ＡＦＬＰ) 标 记 Ｌ４￣ｃ 开 发 出 的 序 列 特 异 性 扩 增 区 域
月季ITS反应体系的优化
月季ITS反应体系的优化第28卷第6期2009年12月华中农业大学JournalofHuazhongAgriculturalUniversityV o【_28No.6Dec.2009,756～758月季lTS反应体系的优化张慧D唐开学邱显钦'蹇洪英王其刚张颢(云南农业大学园林园艺学院,昆明650201;云南省农业科学院花卉研究所,昆明650205)摘要以古老月季品种'玉玲珑'为试材,利用CTAB法对月季进行总DNA提取,对其PCR扩增条件中的主要因素进行了多梯度的优化,总反应体系为25L,其中引物浓度为0.4/~mol/I,M 浓度为2.0mmol/L,dNTP为0.2mmol/L,Taq酶用量为1.0U,DNA模板用量为70ng.利用该优化体系对部分月季品种和野生种进行PCR扩增和电泳检测,扩增条带清晰,结果稳定,测序结果理想,表明该体系适用于月季的ITS序列分析.关键词ITS;月季;优化中图法分类号S685.120.353文献标识码A文章编号1000—2421(2009)06—075603 月季(RosahybridaI.)是蔷薇科(Rosaceae)蔷薇属(RosaI.)植物,蔷薇属又分为Hulthemia,Hesperhodos,Platyrhodon和Eurosa等4个亚属,约有2OO多个种和变种,广泛分布于亚,欧,-IL:II~,北美等寒温带至亚热带地区l1].蔷薇属植物种类繁多,来源广泛,加上频繁的种内杂交,其种质资源复杂多样,使分类和鉴定工作难度加大l2..近年来,核糖体DNAITS序列已作为重要的分子性状用于种问的系统学研究.研究表明ITS在研究属内种问,属问关系都表现出较高的趋异率和信息位点百分率,为类群内部的系统重建提供了较好的支持[4].ITS是基于基因序列的分子标记, 基因组DNA的提取和纯化是整个工作的基础[6.]. 本试验对提取DNA的方法进行改良,同时对PCR 反应体系进行了优化,得到了稳定,高含量的PCR 扩增产物,旨在对蔷薇属植物来源的DNA模板扩增提供重要参考.1材料与方法1.1试验材料的采集与处理月季样品采自于云南省农业科学院花卉研究所资源圃,将新鲜叶片放人20℃冰箱备用.1.2方法1)DNA的提取.在总结前人对月季进行DNA提取的经验基础上,用改良的CTAB法l8.ll进行总DNA的提取,提取的DNA的质量和浓度采用琼脂糖胶电泳和紫外分光光度计测定,最后把样品稀释到20ng//xI.2)PCR扩增反应体系及程序.PCR扩增在PTC-100型DNA扩增仪上进行.所用引物序列为P4:5一TCCTCCGCTTATTGA TATGC一3;P5:5一GGAAGTAAAAGTCGTAACAAGG一3,Taq 酶,dNTP购自上海生T生物工程技术服务有限公司.反应体系为25L.反应程序为:第1阶段: 95℃,预变性,5min;1个循环;第2阶段:95℃,变性30S;55℃,退火3OS;72℃,延伸,1.5min;35个循环;第3个阶段:72℃后延伸10min;终止反应,4℃保存.3)ITS扩增片段的检测.扩增的PCR产物经1.0琼脂糖凝胶(含花青素)电泳检测,用全自动紫外与可见分析装置进行凝胶成像.2结果与分析2.1Mg浓度Mg.的浓度对PCR反应的特异性和扩增效率有很大影响,反应体系的Mg+用量是决定PCR扩增反应成败的关键因素之一.在25L反应体系中设6个Mg.浓度梯度:1.0,1.5,2.0,2.5,3.0,4.0mmol/I(图1).收稿日期:20081108;修回日期:20090622*国家自然科学基金项目(30660117),国家高技术研究发展计划项目(2006AA100109)和云南省科技计划项目(2006NG14)资助**通讯作者.E-mail:**********************.cn张慧,女,1984年生,云南农业大学园林园艺学院硕十研究生,昆明650201.E—mail:huihui一**************.cn第6期张慧等:月季ITS反应体系的优化图1不同Me..浓度的PcR扩增结果Fig1ElectrophoresisresultofdifferentMgconcentration泳道l～6:分别为Mg..浓度1.0,1.5,2.0,2.5,3.0,4.0mmol/LIanel～6:Mgconcentration:1.0,1.5,2.0,2.5,3.0,4.0mmol/[从图1可见,Mg计浓度在1.0～1.5mmol/L时扩增产物很少,Mg+浓度在2.0～2.5mmol/I时有特异性条带出现,且浓度在2.0mmol/I时扩增产物条带亮且清晰,在3.O～4.0mmol/L时扩增产物不清晰.试验结果表明,Mg浓度在2.0mmol/I时为宜.2.2dNTP浓度dNTP是提供扩增所需要的原料,设5个浓度梯度:1.0,1.5,2.0,2.5,3.0mmol/I(图2).从图2可见,dNTP浓度在1.0,1.5,2.5,3.0mmol/I时扩增产物很少,在2.0mmol/L时'增条带亮且清晰.试验结果表明,2.0mmol/I的dNTP浓度可满足反应要求.图2不同dNTP浓度的POR扩增结果Fig2Electr0DhOresisresultofdifferentdNTPconcentration 泳道1～6:分别为dNTP浓度1.0,l_5,2.0,2.5,3.0mmol/I Iane1～61dNTPconcentrationl_01.5,2.0,2.5,3.0mmol/l 2.3Taq酶的用量PCR所需Taq酶的用量对反应有一定的影响.在25肚I反应体系中,设了5个Taq酶的用量梯度:0.25,0.5,1.0,1.5,2.0U(图3).从图3可见,Taq酶用量在0.5,1.0,1.5,2.0U时均有扩增产物条带出现,但Taq酶用量在0.5U时出现非特异性条带;当增大q酶用量为1.5～2.0U时扩增条带产物反而减少;而在1.0U时,扩增条带最清晰.故本试验选择1.0U为反应的Taq酶用量.图3不同Taq酶浓度的PCR扩增结果Fig3EIectr0Dh0resisresultofdifferentFaqDNA polymeraseamount泳道l～5分别为:Taq酶0.25,0.5,1.0,1.5,2.0ULane 1～5:TaqDNApolymeraseamount0.25,0.5,1.0,1.5,2.0U 2.4DNA模板用量DNA模板使用量也是影响PCR产物扩增的重要因素,DNA浓度过低产物的量少或无扩增产物;浓度过高会产生非特异性条带.在25I反应体系中选用8个DNA模板用量梯度:30,4O,5O,6O,70,80,90,100ng.从图4可见,DNA用量在3O～40ng时几乎无扩增产物,在5O～70ng时有条带,但在70ng时条带最清晰,DNA用量在8O～100ng时也几乎无扩增产物.因此,选用的模板DNA用量以70ng为宜.图4不同DNA模板用量的PCR扩增结果Fig4ElectrophoresisresultofdifferentDNAtemplateamount泳道l～9分别为:DNA模板用量30,40,50,6O,70,80,90,100rigI.anel～9:DNAtemplateamount30,40,50,60,70,80,90.100ng2.5优化条件组合的PCR扩增结果采用上述优化条件对10份月季材料进行PCR扩增,以检测该月季ITS反应体系的适用性.结果如图5所示,在750bp处10份材料均有一明亮的DNA带,与预期大小一致.这表明优化确立的ITS-PCR反应体系稳定可靠,可用于月季的ITS序列分析.3讨论PCR扩增条件中Mg浓度,Taq酶用量,dNTPs浓度和样品DNA模板用量等均会影响758华中农业大学第28卷l234567891oM图5不同月季品种和野生种的POR扩增结果Fig.5EIectroDhOresisresultofdifferent rosecultivarsandspecies泳道1～1O分别为:玉玲珑,青莲学士,绿萼,映El荷花,木香花,黄刺枚,香水月季,好莱坞,花车,糖果条Lane1～10:Rosa'Yulin glong'.Rosa'Qinglianxueshi'.Rosa'Viridiflora'.Rosa'Yingrihe—hua',R.banksiae,R.xanthina,R.odorats,Hollywood,Hanaguruma, CandyStripePCR扩增结果.其中,Mg.的浓度影响着TaqDNA聚合酶的反应活性;Mg的另一个作用是和DNA模板结合,屏蔽DNA分子表面的电荷,防止其由于相互凝集而沉淀.Taq酶用量过多会导致非特异性产物增加,用量过少会导致扩增产物量不足.dNTPs提供扩增所需原料,dNTPs浓度过高,可能造成非靶序列启动和延伸时核苷酸的错误掺人,还会抑制Taq酶的活性;浓度过低,则会影响扩增产量.DNA模板的用量也是影响PCR扩增的重要因素,DNA浓度过高或过低都会使PCR扩增无条带产生.因此对反应体系中各因子进行优化组合是决定PCR扩增反应成败的关键.参考文献[1]中国科学院中国植物志编辑委员会.中国植物志第三十七卷[M].北京:科学出版社,1985:360—455.[2]TIRRESAM,MIIIANT,CUBEROJI.Identilyingroseeul—tivarsusingrandomamplifiedpoiymorphicDNAmarkersEJ]. HortScience,1993,28(4):333—334.E37REYNDERS-ALOISIS,BOLLEREAUP.Characterizati0nof geneticdiversityingenusRosabyrandomlyamplifiedpolymor—phicDNA[J].ActaHortculturae,1996,424:253—259.r4]SCHWARBACHAE,RICKLEFSRE.Systematicaffinitiesof Rhiz0phoraceaeandintergenericrelationshipswithinRhizo—phoraceae,basedonchloroplastDNA,nuclearribosomalDNA, andmorphology[J].AmericanJournalofBotany,2000(87): 547564.[5]STANFORDAM,HARDENR,PARKScRPhylogenyandbio—gepgraphyofJugland(Juglandaeeae)basedonmatKandITS8e—quence_J].AmericanJournalofBotany,2000(87):872—882.[6]邹利波,黄升谋.月季总DNA提取方法的比较研究[J].安徽农学通报,2007,13(1):5758.[7]余晓丽,范喜梅,曾万勇,等.黄刺玫基因组DNA提取方法的研究『J].西北农业,2007,14(4):272274.[87张英,黄明辉,杨梦苏,等.植物基因组DNA提取方法学评析与验证[J].药品评价,2004,1(4):292298.[9]裴杰萍,端青.DNA提取方法的研究进展口].微生物学免疫学进展,2004,32(3):76—78.[1O]罗玉兰,张冬梅,杨娅.红刺玫DNA提取及SSR引物筛选[J].同林科技,2007(1):15-16.[11]运江,伊华林,庞晓明,等.几种木本果树DNA的有效提取_J].华中农业大学,2001,20(5):481—483.OptimizationofITS—PCRinRoseZHANGHui"TANGKai—xueQ1UXian-qine'JIANHong-yinge'WANGQi-gangeZHANGHao2 ("CollegeofLandscapeandHorticulture,YunnanAgriculturalUniversity,Kunming65020 1,China;FlowerResearchInstitute,YunnanAcademyofAgriculturalSciences,Kunming650205,Ch ina)AbstractDNAofroseYulinglongwasextractedwithCIAB.PrincipalfactorsofPCRhaddiffe rentcon—centrationsandtheirvariationchangedtheresultofITS—PCRFactorsaffectingtheITSresultsofrosewere studiedandthebestreactionsystemofITS—PCRwgts:0.4umol/Lofeachprimer,2.0mmol/L+,0.2mmol/LofdNTPs,1.0UofTaqDNApolymeraseand70ngtemplateDNAin25ffLreactionsy ingabovePCRsystem,ITSfragmentsofsomerosecultivarsandspecieswereobtained.Thecleari tyandstabilityof amplificationindicatedthissystemwassuitableforanalyzingITSsequencesinroses. KeywordsITS:rose:optimization(责任编辑:杨锦莲)。
基于粒子滤波的混沌时间序列局域多步预测
基于粒子滤波的混沌时间序列局域多步预测姜娇娇;郭俊;杨淑莹【摘要】对混沌时间序列进行预测研究具有重要的价值和实用性,例如,进行股票预测,降雨量预测,温度预测.混沌时间序列预测的难点在于其不确定性和多步预测的困难性.一般利用最小二乘法求解模型参数,从而对混沌时间序列进行局域预测,但是预测精度不是很高.为了提高局域线性预测的精度,提出基于粒子滤波(PF)的混沌时间序列局域多步预测法,利用粒子滤波进行参数优化得到更准确的优化模型进行多步预测.仿真实验结果表明,该方法的单步和多步预测效果明显得到了提升.%It has important value and practicability(such as stock forecasting,rainfall forecasting and temperature fore-casting)to predict the chaotic time series. It is difficult to predict the chaotic time series due to its uncertainty and realization of multi-step prediction. The least square method is used to solve the model parameters and perform local prediction for the chaotic time series,but has low prediction accuracy. In order to improve the accuracy of local linear prediction,a local multi-step pre-diction method based on particle filtering(PF)is proposed for chaotic time series. The particle filtering is adopted to optimize the parameters to obtain more accurate optimization model for multi-step prediction. Simulation results show that the multi-step and single-step prediction effects of this method are improved significantly.【期刊名称】《现代电子技术》【年(卷),期】2018(041)001【总页数】4页(P43-46)【关键词】局域线性预测;混沌时间序列;粒子滤波;多步预测;邻近点;预测误差【作者】姜娇娇;郭俊;杨淑莹【作者单位】天津理工大学计算机与通信工程学院,天津300384;天津理工大学计算机与通信工程学院,天津300384;计算机视觉与系统教育部重点实验室,天津300384【正文语种】中文【中图分类】TN911.1-34;O415.5混沌是确定的非线性动力系统产生的复杂行为，将混沌理论与时间序列预测相结合的思想从一开始就得到了广大学者的关注。
基于自适应变异的混沌粒子群优化算法
基于自适应变异的混沌粒子群优化算法李建美;高兴宝【摘要】粒子群优化算法参数少,寻优速度快,但其寻优效率低且在寻优后期易早熟收敛.为改善其寻优性能,在标准粒子群优化算法中,通过引入混沌映射和自适应变异策略,提出具有自适应变异的混沌粒子群优化(ACPSO)算法,以增强种群的全局寻优性能和局部寻优效率.六个基准测试函数的仿真结果表明,ACPSO算法比已有的五个算法具有更好的寻优能力.【期刊名称】《计算机工程与应用》【年(卷),期】2016(052)010【总页数】6页(P44-49)【关键词】粒子群优化;自适应策略;混沌映射;数值优化【作者】李建美;高兴宝【作者单位】陕西师范大学数学与信息科学学院,西安710062;陕西师范大学数学与信息科学学院,西安710062【正文语种】中文【中图分类】TP301LI Jianmei,GAO Xingbao.Computer Engineering and Applications,2016,52（10）：44-49.通过对鸟群捕食行为的研究，Eherhart和Kennedy于1995年提出了一种基于随机种群的新型智能优化方法——粒子群优化（PSO）算法[1]。
PSO算法由于结构简单、实现容易、收敛速度快、待调整参数少等优点，已广泛应用于函数优化、神经网络训练、工业系统优化和模糊系统控制等领域[2-3]。
但作为一种新型的智能优化算法，PSO算法存在早熟收敛，进化后期收敛速度慢，精度较差等缺点。
为克服算法的缺点，并加强其寻优性能，人们对PSO算法进行了改进。
例如：通过对PSO算法中的惯性权重、加速系数等参数做不同的改进，得到带有线性递减惯性权重的PSO（LDWPSO）算法[4]、具有时变加速系数的PSO（TVACPSO）算法[5]；将PSO算法与差分进化、量子进化等算法融合，得到DEPSO[6-7]和QPSO[8]等混合算法。
尽管上述算法在不同程度上改善了PSO算法的性能，但其不能同时保证快速收敛和好的寻优效果。
分子生物学基本技术—PCR
六、PCR反应特点
n 1.特异性强：
①引物与模板DNA特异正确的结合； ②碱基配对原则； ③Taq DNA聚合酶合成反应的忠实性； ④靶基因的特异性与保守性。
n 2.灵敏度高：PCR产物的生成量以指数方式增加
n 3.简便、快速：PCR反应一般在2～4 小时完成扩增反应。扩增
产物一般用电泳分析，不一定要用同位素，无放射性污染、易推广。
mRNA needed to be RT as cDNA模板核酸的量与纯化程度， 是PCR成败与否的关键环节之一。
n Primes（引物）：每条引物的浓度0.1～1umol或10～
100pmol，以最低引物量产生所需要的结果为好，引物浓 度偏高会引起错配和非特异性扩增，增加二聚体的机会。 纯化引物在25％乙腈溶液中4℃；冻干引物于-20℃可保存 1-2年，液体状态于-20℃可保存6个月。不用时应-20℃保 存。
•5’
PPT文档演模板
分子生物学基本技术—PCR
•回忆：DNA 的复制……
•DNA •解旋解链
•5•’ATCGCGATAGCGTAGCTGCGACCTAGC
•3’
•3’•TAGCGCTATCGCATCGACGCT
•DNA •聚合酶
•合成引物
•子链延长
•DNA •聚合酶
•ATAGCGTAGCTGCGACCTAGC
•模板DNA
PPT文档演模板
分子生物学基本技术—PCR
•2
•50℃
•引物2
•DNA引物
•引物1
PPT文档演模板
分子生物学基本技术—PCR
•3
•72℃
•Taq酶 •引物1
•DNA引物
•引物2 •Taq酶
PPT文档演模板
基于文化框架的随机粒子群优化算法
基于文化框架的随机粒子群优化算法王正帅;邓喀中【期刊名称】《计算机科学》【年(卷),期】2012(039)006【摘要】Bringing standard PSO and random particle swarm optimization(rPSO) proposed in the paper into the framework of cultural algorithm(CA) ,a novel optimization method named random particle swarm optimization based on cultural algorithm(CA-rPSO) was established. In CA-rPSO,the evolving algorithms of belief space and the population space were represented with rPSO and PSO respectively, forming independent and parallel "dual evolution-dual promotion" mechanism. 5 testing functions were selected to simulate and analyze CA-rPSO. The result shows that optimization performance of CA-rPSO is obviously promoted and the algorithm is simple and easy to carry out%提出了随机粒子群优化算法(rPSO),并将其与标准PSO纳入到文化算法(CA)框架中,建立了基于文化框架的随机粒子群优化算法(CA-rPSO).该算法以rPSO作为信念空间的进化算法,以PSO作为群体空间的进化算法,形成了两者独立并行进化的“双演化双促进”机制.选取5个测试函数进行了仿真实验分析并与其他算法进行了比较,结果表明CA-rPSO的寻优性能得到显著提高,且算法简单、易于实现.【总页数】3页(P198-200)【作者】王正帅;邓喀中【作者单位】徐州师范大学测绘学院徐州 221116;中国矿业大学环境与测绘学院徐州 221116【正文语种】中文【中图分类】TP18【相关文献】1.基于随机维度划分与学习的粒子群优化算法 [J], 张庆科;孟祥旭;张化祥;杨波;刘卫国2.基于随机惯性权重的简化粒子群优化算法 [J], ZHAO Zhi-gang;HUANG Shu-yun;WANG Wei-qian3.基于随机漂移粒子群优化算法的三维脑部磁共振图像分割 [J], 施佳佳;孙俊;范方云;王梦梅4.基于云自适应粒子群优化算法和随机森林回归(CAPSO-RFR)的负载均衡预测 [J], 李雨泰; 李伟良; 尚智婕; 王洋; 董希杰5.基于粒子群优化算法的BPSK信号随机共振研究 [J], 田万平;向亚丽;颜冰;王琪因版权原因，仅展示原文概要，查看原文内容请购买。
基于混沌随机滤波器的 CS-MIMO 雷达测量矩阵优化设计
基于混沌随机滤波器的 CS-MIMO 雷达测量矩阵优化设计彭珍妮;贲德;张弓【摘要】An optimized measurement matrix design method for compressive sensing-multiple input multiple output (CS-MIMO)radar is proposed by applying the chaotic dynamical system to random filter design.Most of the previous research takes the Gaussian random matrix as the measurement matrix.However,it cannot realize on-line optimization and is hard to be implemented in physical electric circuit.Considering that the basis matrix is obtained from the CS-MIMO radar signal model,we propose a new measurement matrix design method apply-ing the random filter.By constructing the filter coefficients with the chaotic sequence,the CS is achieved for the received signal.Moreover,an optimization method is performed on the equivalent measurement matrix of the random filter,by making the Gram matrix approach the diagonal matrix.The simulation results show that the proposed measurement matrix design and optimization method based on the chaotic random filter can effectively improve the direction of arrival (DOA)estimation accuracy of the CS-MIMO radar.%提出了一种在压缩感知多输入多输出(compressive sensing-multiple input multiple output,CS-MIMO)雷达中利用混沌非线性系统设计随机滤波器进而实现测量矩阵优化的方法。
- 1、下载文档前请自行甄别文档内容的完整性，平台不提供额外的编辑、内容补充、找答案等附加服务。
- 2、"仅部分预览"的文档,不可在线预览部分如存在完整性等问题,可反馈申请退款(可完整预览的文档不适用该条件!)。
- 3、如文档侵犯您的权益，请联系客服反馈,我们会尽快为您处理(人工客服工作时间：9:00-18:30)。
1
identification [3]. This technique can be used to differentiate individuals or breeding stocks with a given species but also to differentiate among different species. The conditions of the analysis can be optimized to reveal only species-specific differences, since all individuals belonging to a given species have more of their genetic material in common than with individuals from other species [7]. The Lophiidae family comprises four genus and twenty-five species, where the Lophius is the most important one, due to its economical potential. In Brazil, it is possible to find the L. gastrophysus species, which lives in depth between 100 and 500 meters, main target of the fishing fleet that operates above the isobath of 200 meters in the south and southeast region [9]. The blackfin goosefish has been detached by the high availability and very good acceptance in the international market, becoming one of the most valuable fisheries in Brazil. Several authors have demonstrated the discriminatory power of PCR analysis for animal species identification [2, 4, 5, 11, 12]. L. gastrophysus can be easily distinguished if it is sold as a whole fish by means of traditional taxonomic classification methods. In the case of transformed products, another technique for identifying fish species is needed to avoid fraud through species substitution. In this study, we described the optimization of RAPD protocol for identification of L. gastrophysus. Different conditions (annealing temperatures, MgCl2 concentrations, DNA quantity) were tested in order to find reproducible and adequate profiles.
RESUMO
Otimização do protocolo de amplificação RANDÔMICA de DNA polimórfico para identificação molecular de Lophius gastrophysus. Lophius gastrophysus apresenta importante valor comercial no Brasil, principalmente para a exportação. Neste estudo, descrevemos uma otimização do protocolo de amplificação aleatória de DNA polimórfico (RAPD) para identificação de L. gastrophysus. Diferentes condições (temperatura de recozimento, quantidade de DNA e concentração de MgCl2) foram testadas para obter perfis reprodutíveis. Os iniciadores A01, A02 e A03 geraram os melhores resultados de amplificação quando utilizados temperatura de recozimento de 36oC, 25 ng de DNA e 2,5 mM de MgCl2. A identificação exata de espécies e da origem dos produtos marinhos faz-se necessária e a RAPD é uma ferramenta rápida e precisa para expor fraudes comerciais. Palavras-chave: fraude, identificação de peixes, peixe sapo, RAPD-PCR.
Optimization of Random Amplified Polymorphic DNA protocol for molecular identification of Lophius gastrophysus1
Micheline S. RAMELLA2, Mariela A. KROTH2, Caroline TAGLIARI2, Ana Carolina M. ARISI2,* SUMMARY
1 - INTRODUCTION
Due to the increase of the international trade in seafood and seafood products, there is a need for suitable methods for the identification of fish species or part of it to ensure compliance with the labeling regulations, quality and pricing policies imposed by various countries and, thereby, to prevent the substitution of fish species [2, 8]. Identification might be required to support enforcement of regulations, for example restrictions on fishing for a particular species or enforcement of regulations relating to naming of fish offered for sale, or in testing for compliance with specifications. The use of such techniques will help to protect consumers’ rights by enabling the enforcement of labeling regulations. It will also help to regulate imports from producer countries and it will assist in the protection of endangered species [11]. Molecular biological methods represent the future in food control laboratories [8]. Methods based on comparison of DNA fragments have been developed. The polymerase chain reaction (PCR) method has been used successfully to identify fish species and to avoid fraudulent label [2, 4, 5, 11, 12]. The random amplified polymorphic DNA (RAPD) [13, 14] technique consists in the amplification, by PCR, of random segments of genomic DNA using a single short primer with arbitrary sequences, thus, one can expect to scan the genome more randomly than using conventional techniques [3] to produce a fingerprint of the DNA contained in the sample [7]. The relatively low cost of the technique and requirement of only nanograms of template DNA provide advantages in the use of RAPD in fish and fish products