基于自适应并行遗传算法优化设计的有源滤波器

Optimization design of active filters based on adaptive parallel genetic programming

提出一种用于有源滤波器的改进自适应并行遗传算法设计.引入了两个自适应算子:其一根据进化过程实现交叉和变异概率的自动调节;其二通过设计随机个体集和健壮个体集,实现种群个体的多样性和保护适应度高的个体不被破坏.采用基于岛屿的交换模型实现多种群间信息交换,扩大了种群的规模和相应的搜索空间.给出了利用该方法设计四阶切比雪夫低通滤波器的设计结果,并与基本遗传算法进行了比较实验,结果表明该算法收敛速度快、精度高,有效地克服了早熟现象.为大规模有源滤波器设计提供了方法上的支持.

This paper introduces a method of active filters design using an improved adaptive parallel genetic algorithm. The algorithm includes two adaptive operators. One operator is used to change the probabilities of crossover and mutation automatically according to the evolution process. The other is used to adjust the individuals in a population by designing a random individual set and a robust individual set. The two sets are helpful for improving the diversity of the population and protecting the good individuals from being destroyed. The algorithm also realizes information exchanges between multi populations based on the island model, which is a parallel design model. The parallel algorithm can enlarge the size of the whole population and corresponding search space. In the paper this method was used to design a 4 orders Chebyshev low pass filter and compared other genetic algorithms with this method. The results show this method has a higher convergence speed and can solve the problem of prematurity effectively and the method is a good choice for large scale higher order active filters design.