摘要
提出了一种基于遗传算法和模拟退火算法的混合策略,用于单包传输下的线性网络控制系统中远程PID控制器的设计。这种新的混合策略被称为GASA算法。由于网络导致的延迟存在,GASA算法被用于确定最佳的PID增益。在求解过程中,GASA算法在各温度下依次进行GA和SA搜索,其中SA的初始解来自GA的进化结果,SA经过Metropolis抽样过程得到的解又成为了GA进一步进化的初始种群,而温度的更新由退温操作来控制。对设定的性能指标如超调量和调整时间的违背量将作为GASA适应度函数中的惩罚项。本文对基于Profibus-DP网络的网络控制系统进行测试,并与修正Ziegler-Nichols法的优化结果进行比较,结果表明在对网络导致的延迟没有任何先验知识的情况下,用于网络控制系统中远程控制器设计的GASA算法能够获得满意的控制效果。
A hybrid scheme based on genetic algorithm (GA) and simulated annealing (SA) is proposed in this paper to design remote PID controller in the linear networked control systems (NCSs) with single-packet transmission. This novel hybrid scheme is called GASA.GASA is used to select the best PID gains due to the network-induced delay. In the solution process, GASA executes GA and SA searches sequentially at different temperatures. The initial solutions of SA come from the evolution results of GA, while the solutions obtained by the Metropolis sampling in SA are the initial population of further evolution in GA. Temperature is updated by temperature-lowering operation. The violations of specifications such as percent overshoot and settling time are penalized in the fitness function of GASA. Numerical results on a test system using Profibus-DP network are presented and compared with results of the modified Ziegler-Nichols method. The results show the effectiveness of GASA in designing remote controller for networked control systems without any prior knowledge on network-induced delay.
出处
《系统仿真学报》
CAS
CSCD
2004年第7期1571-1575,1588,共6页
Journal of System Simulation
基金
国家自然科学基金(60175015)
