基于GEP的模拟电路系统辨识技术

System Identification of Analog Circuits Based on Gene Expression Programming

根据模拟电路系统的输入输出数据辨识,其数学模型是基于模型的装备快速测试方法中的关键技术之一.针对模拟电路系统的特点,设计了基于R-square的适应度函数、基于精英主义和轮盘赌算法的选择策略,采用部分映射、启发式变异算子,提出了一种基于基因表达式编程算法的模拟电路系统辨识方法并进行了系统实现.与MATLAB系统辨识工具箱对比,该方法不仅能够辨识电路系统的参数,而且还能辨识系统的结构.实验表明,通过设置适当的辨识精度,可以得到与理论模型一致的系统模型.

Identifying the mathematical model of the analog circuit through the IO data is the key technology in the model- based fast equipment test method. According to the electrical characteristics of the analog circuit system, we designed R-square-based fitness function, and the selection strategy combined with elitism and roulette. The establishment method of analog circuit system identification based on gene expression prograrnming （GEP） was constituted with the partially mapped crossover operator, the heuristic mutation operator, and the sequence insertion operator. Compared with the MATLAB system identification toolbox, this method can identify not only the system parameter, but also the structure. If you set the suitable identifica- tion precision, we can get the theoretical system model via the SISO and MIMO tests.