摘要
采用遗传算法对小车倒立摆PID控制器的8个控制参数进行寻优,从而避免手工整定的繁琐过程。传统的遗传算法将多个参数对应的元素位串连接在一起构成一个长的位串,而开发的寻优算法提出将8个参数构成一个2×4的矩阵,进而再将每个参数展开成相同长度的位串,所以每个个体即是一个三维2×4×L的二进制矩阵,由8个染色体组成,其优点是8个参数的交叉操作可以相对独立进行。实验证明,开发的算法迭代20次可以达到全局最优值,控制系统性能良好。
A GA-based approach is proposed to find the global optima of the eight control parameters for a PID controller of an inverted pendulum. Different from the classical GAs which combines the substrings of several parameters into a complete string as a chromosome, the proposed approach formulates the eight parameters into a 2×4 matrix, whose entries are the substrings of those parameters with the same length. Thus, in the approach one individual, i.e. a binary matrix of 2×4×L, consists of eight chromosomes. Its advantage lies upon that the crossover operation among different chromosomes can be performed independently. Experimental results show that the approach can achieve the global optima after 20 generations and that the control system performance is promising.
出处
《计算机测量与控制》
CSCD
2004年第12期1171-1174,共4页
Computer Measurement &Control
