DNA遗传算法及其在燃料电池中的应用研究

发布时间：2018-03-09 15:24

本文选题：DNA遗传算法　切入点：质子交换膜燃料电池　出处：《浙江工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：遗传算法具有操作简单,搜索能力强,自适应能力强等优点,是智能优化算法的典型代表,但是遗传算法仍存在着搜索效率低,易早熟收敛等缺点。DNA遗传算法结合了DNA计算和遗传算法发展而来,它采用四进制编码,并在遗传算法的基础上对种群进行遗传操作,从而更好的模拟了生物体遗传和进化的本质,提高了算法的性能。本文结合其他算法的优点,对DNA遗传算法进行了改进,并将改进的DNA遗传算法应用到燃料电池的模型优化的问题中。本文主要工作和成果如下:1.在DNA遗传算法的研究基础上研究并比较了不同数目的个体间交叉操作对算法的进化速率和收敛速度的影响。根据DNA的编码方式在多个个体间设计了新型的交叉方式,从而保证了在进化当中种群的多样性。利用几个经典的测试函数验证了所提出的交叉操作的有效性,并比较得出了当交叉个体数目为三时算法得到的结果最优。2.为了简化算法提高算法效率,同时又保持算法的有效性,提出了选择性交叉变异操作的DNA遗传算法。该算法依照一定概率只执行交叉操作和变异操作中的一种操作,而且交叉操作选择在多个不同种类的个体间进行,变异操作中的变异概率随遗传代数的增加而变化。四个测试函数最值的寻优结果验证了所提算法的有效性。3.受DNA遗传算法多个体交叉操作的启发,提出了改进的多种群进化的DNA遗传算法。每个子种群根据自身不同的侧重点进行独立的遗传操作,当种群进化到一定代数时还可进行种群间的信息交换,并采用种群融合策略对种群进行融合再分,从而很好的提高了算法的全局搜索能力和局部搜索能力。测试函数计算的结果表明所提出的多种群DNA遗传算法的优越性。4.准确的燃料电池模型的参数估计受到学者们的广泛关注。将所提出的选择性交叉变异操作的DNA遗传算法及改进的多种群进化的遗传算法应用到质子交换膜燃料电池(Proton Exchange Membrane Fuel Cell,PEMFC)模型的参数辨识中,在给定的PEMFC的输出电压与负载电流间的关系式的基础上,分析不同验证策略下本文提出的这两种算法的辨识性能,并分别与已有的四种智能算法进行对比,分析结果验证了所提两种算法在PEMFC模型参数优化上的高效性。将改进的多种群进化的遗传算法应用到高温固体氧化物燃料电池(SOFC)模型的参数辨识中,分析比较不同温度条件下得到的电压-负载电流特性,与已有算法进行对比得出了该算法的有效性。
[Abstract]:Genetic algorithm (GA) has the advantages of simple operation, strong searching ability, strong adaptive ability and so on. It is a typical representative of intelligent optimization algorithm, but it still has low search efficiency. The DNA genetic algorithm is developed by combining DNA computation with genetic algorithm. It uses quaternary coding and carries on the genetic operation to the population on the basis of genetic algorithm. In order to better simulate the nature of biological genetics and evolution, improve the performance of the algorithm. This paper combines the advantages of other algorithms, improve the DNA genetic algorithm, And the improved DNA genetic algorithm is applied to the optimization of fuel cell model. The main work and results of this paper are as follows: 1. Based on the research of DNA genetic algorithm, we study and compare the number of cross-operation pairs between individuals. The influence of the evolutionary rate and convergence rate of the algorithm. According to the coding method of DNA, a new intersecting method is designed among several individuals. Thus ensuring the diversity of the population in evolution. The effectiveness of the proposed crossover operation is verified by several classical test functions. In order to simplify the algorithm and improve the efficiency of the algorithm, it is found that when the number of crossover individuals is three, the algorithm is optimal .2.In order to simplify the algorithm, the efficiency of the algorithm can be maintained at the same time. In this paper, DNA genetic algorithm for selective crossover mutation operation is proposed. The algorithm performs only one of the crossover and mutation operations according to certain probability, and the crossover operation is selected between several different kinds of individuals. The mutation probability in mutation operation varies with the increase of genetic algebra. The optimization results of the four test functions verify the effectiveness of the proposed algorithm. 3. Inspired by the multi-individual crossover operation of DNA genetic algorithm, An improved DNA genetic algorithm for multi-population evolution is proposed. Each sub-population performs independent genetic operations according to its own emphasis. When the population evolves to a certain algebra, it can exchange information between populations. And the population fusion strategy is used to subdivide the population. The result of test function calculation shows the superiority of the proposed multi-population DNA genetic algorithm .4.The accurate parameter estimation of fuel cell model is studied. The proposed DNA genetic algorithm for selective crossover mutation operation and the improved multi-population evolution genetic algorithm are applied to the parameter identification of Proton Exchange Membrane Fuel cell PEMFCmodel for proton exchange membrane fuel cell (PEMFC). On the basis of the relationship between output voltage and load current of PEMFC, the identification performance of the two algorithms proposed in this paper under different verification strategies is analyzed, and compared with four existing intelligent algorithms. The results show that the proposed two algorithms are efficient in the parameter optimization of PEMFC model. The improved genetic algorithm of multi-population evolution is applied to the parameter identification of high temperature solid oxide fuel cell (SOFC) model. The characteristics of voltage-load current obtained at different temperatures are analyzed and compared with the existing algorithms, and the effectiveness of the algorithm is obtained.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TM911.4;TP18

【相似文献】