用GA优化模糊控制器及其应用

Optimizing Fuzzy Controller Using GA and its Application

用遗传算法优化模糊控制器的隶属度函数,得到优化的模糊控制器。在遗传算法设计方面,采用十进制基因编码,减少了译码的麻烦,避免了常规情况下以二进制形式编码的染色体物理意义不明显和在交叉、变异操作中容易破坏本身特性的缺点。给出了一种比较实用的交叉、变异和选择的方法,不仅使得种群进化的操作计算变得简单,而且既保留了染色体中好的特性又优化了不良的染色体。全部程序由Matlab编程实现。针对某货船,将优化后的模糊控制器在Matlab的Simulink中进行了多种情况下的仿真研究。此外,对于种群规模、进化代数对模糊控制器性能的影响也做了对比的仿真研究。由仿真结果可知,遗传算法能够有效地提高模糊控制器的性能。

The membership functions of FLC are optimized using GA and an optimized FLC is got.In the design of GA,for avoiding many shortcomings derived from the binary encoding by which chromosomes have no obvious physical meaning and apt to destroy itself in crossover or mutation,the decimal encoding system of the gene is used to reduce the trouble of the decoding.In addition,a relatively practical method of crossover,mutation and selection is given to make the population evolve more simple and not only the good chromosomes keep those good characters but also the bad chromosomes stand the chances to evolve.All programs are realized by Matlab.Using a cargo ship as a controlled plant ,the simulation is carried out in many kinds of situations,the system performance through comparing with changes of population size and evolving generation is also given.GA can improve the performance of FLC effectively by the sim-ulation results.