Abstract: An analog filter’s design used to need modification, computation and debugging in many times by manual work. An evolutionary hardware structure and multi-objective parallel evolvable design method for an analogue transconductance filter is presented. A parallel genetic algorithm with migration scheme based on penetration theory is applied, and parameters optimization of the transconductance filter is realized. Genetic operations to multi-objective parallel genetic algorithm are improved, and a multi-object fitness function for the filter parameter has been built. In the course of high Q value transconductance filter designing, structure of four-order bandpass transconductance filter is built by cascade method, which is able to meet performance request in its stop-band, passing band, transition band, and the practicality has been simulated and verified. Values of evolutionary parameters are in line with theoretical values exceedingly.

Key words: evolvable hardware, transconductance filter, multi-object parallel genetic algorithm, evolve, simulate

摘要： 以往模拟滤波器的设计，需要经过人工多次修改、计算和调试。提出一种模拟跨导滤波器的硬件进化结构及多目标并行进化的设计方法。应用一种基于渗透原理迁移策略的多目标并行遗传算法，实现跨导滤波器的参数优化。对多目标并行遗传算法的遗传操作进行了改进，构造了适合滤波器参数进化的多目标适应度函数。在高[Q]值的跨导滤波器的设计中，采用级联法构成的四阶带通跨导滤波器能够满足其在阻带、通带以及过渡带方面的性能要求，对其实效性进行了仿真和验证，进化的参数数值与理论结果符合得非常好。

关键词: 进化硬件, 跨导滤波器, 多目标并行遗传算法, 进化, 仿真