基于改进粒子群算法的约束隐式广义预测控制

A Kind of Constrained Implicit Generalized Predictive Control Based on Improved Particle Swarm Optimization

针对大多数工业系统的控制输入输出都存在约束的情况,提出了一种基于改进粒子群算法的广义预测控制算法。广义预测控制算法采用隐式算法,用最小二乘法直接辨识控制增量表达式中的参数,避免求解丢番图方程;为避免粒子群算法陷入早熟,提高精度,在广义预测控制算法的滚动优化环节采用速度变异的粒子群优化算法,克服了受约束优化问题处理的缺陷,更快更准确地寻到最优目标函数值。仿真结果表明了该方法的有效性和良好的控制性能。

With respect to constrained control input and output of most industrial systems,this paper proposes a kind of generalized predictive control based on improved particle swarm optimization.Generalized predictive control is an implicit algorithm to identify directly parameters in control increment expression by least square method,to avoid solving Diophantine equations;to avoid particle swarm optimization getting into local convergence and improve precision,speed variant particle swarm optimization is used in rolling optimization of generalized predictive control to overcome defect in processing constrained optimization problem and seek optimal objective function value more quickly and precisely.Simulation results show effectiveness and good control performance of this method.