基于遗传编程的逆变器拓扑进化设计

Evolutionary design of inverter circuit topology based on genetic programming

模拟电路的优化设计多受制于设计者的经验,尚未有比较系统性的优化设计方法。而使用遗传编程进行电路拓扑优化设计多见于连续电路,像模拟滤波器电路等。对于逆变器这样含有多开关非线性混杂电路,其物理拓扑和运行都有严格的条件限制,如三极管必须对称且成对通断等,对遗传进化操作以及仿真都有严格要求,要实现完全计算机自动化设计比较困难。针对逆变电路的特点,将遗传编程运用到逆变器混杂电路设计中,实现比较系统性的逆变器优化设计方法。另外在仿真方法上采用Matlab编程、Simulink命令生成模型、simpowersystems仿真,实现一体化仿真,避免了多平台仿真下程序接口的问题,提高了仿真速度。最后实现了一个五电平逆变器的自动生成,验证了方法的可行性。

Optimization design of analog circuit is always subjected to the designer＇s experience, and has not been a systematic and theoretical approach so far. The use of genetic programming for circuit topology optimization design is always seen in continuous circuit, such as analog filter circuit, etc. The inverter circuit contains many nonlinear hybrid switches. The physical topology and operation have strict requirements. The triode must be symmetrical and paired on and off. It is hard to achieve a complete automation design by computer. Based on the characteristics of inverter circuit, the genetic programming was applied into inverter hybrid circuit designing, and a systematic method of inverter optimization design was carried out. In addition, Matlab/Simulink command line simulation technology was used to avoid the program interface problem, and improve the simulation speed. Finally, a five-level inverter was generated which verified the validity of the method proposed.