广义预测与递推最小二乘法相结合的预测控制

PREDICTIVE CONTROL COMBINING GENERALISED PREDICTION AND RECURSIVE LEAST SQUARE

针对网络控制系统的时延具有随机、时变等特性,提出一种广义预测与递推最小二乘法相结合的预测控制策略。首先针对网络随机时延性,采用广义预测控制算法对随机时延进行估计,并采用网络延迟补偿器根据当前时刻的最新控制信号对网络系统实现有效控制,然后采用递推最小二乘法对被控对象模型参数进行在线辨识,并采用仿真实验测试其与其他控制策略的优劣。仿真结果表明,该控制策略较好地解决了网络控制系统的随机时延难题,改善了网络控制系统在参数突变情况的性能。

Aiming at the characteristics of random and time-varying the time delay of network control system has,in this paper we present a novel predictive control strategy which combines the generalised prediction with recursive least square. First,for network random delay property,it uses generalised predictive control algorithm to estimate the random delay,and uses network delay compensator to realise effective control of network system based on current latest control signals,then it adopts the recursive least squares method to make online identification on the parameters of the object model to be controlled,and uses simulation experiments to test its advantages and disadvantages over other control strategies. Simulation results show that the proposed control strategy well solves the problem of random delay of network control system,improves the performance of the network control system in the circumstance of sudden change in parameters.