基于逐维黄金分割法的PID控制器参数优化

PID Controller Parameter Optimization Based on Dimension-by-dimension Golden Section

针对群体智能算法应用于PID控制器参数优化时随机性强、收敛慢、耗时长的不足,文中提出一种确定性的寻优方法。首先仿真验证了PID参数优化问题中目标函数随各个参数变化呈单峰性的特点。然后将坐标轮换法和黄金分割法相结合,形成了逐维黄金分割法。最后将该算法分别应用于有自衡、无自衡、串级系统的PID参数优化。基于热力发电厂典型控制系统的仿真应用表明,该方法的寻优速度明显快于粒子群算法,更适合于PID控制系统的在线实时参数优化。

Aiming at the deficiencies of strong randomness, slow convergence and time-consuming of swarm intelligence algorithm in its application in PID controller parameter optimization,this paper presents a deterministic optimization method.At first,simulation verifies the characteristic of objective function in PID parameter optimization problem,i.e.showing a single peak with each parametric variation.And then combine cyclic coordinate method and golden section to form dimension-by-dimension golden section.Finally,the algorithm is applied separately to PID parameter optimization with self-regulation, without self-regulation, and with cascade system.The simulation application based on typical control system of thermal power plants shows the optimizing speed of this method is significantly faster than particle swarm algorithm and more suitable for online real-time parameter optimization of PID control system.