Increasing-Dimension Evolution: Make the evolutionary design of passive filters more efficient

Abstract

Evolutionary circuit design is a promising way to achieve the automatic synthesis of analog circuits, and passive filters are essential circuits that must be considered. For now, in the evolutionary design of passive filters, the size control of filters under practical difficulty is still an important issue to be studied. This paper proposed a search strategy called Increasing-Dimension Evolution (IDE). The strategy introduces the concepts of exon and intron to control gene expression levels in evolutionary search. Through this strategy, even if the length of the chromosome is fixed, the search engine can effectively search up multiple design spaces and find a small circuit size at low computing costs. We built an effective evolutionary filter design model using the IDE strategy. Experimental results show that the proposed strategy can obtain filter sizes comparable to advanced conventional design methods for several passive filter design tasks with practical difficulties. In addition, the strategy makes the search engine more effective in a given complex task, and this task is difficult to be directly solved using conventional design methods.