黑箱系统的一种简便自适应预测控制策略

A simple adaptive predictive control strategy of black box systems

采用一种输入输出增量式一元线性回归模型作为黑箱系统的预测模型,应用投影算法估计模型参数.该模型将对象输出增量分解为2个分量:一个为非零控制增量作用下的强制分量,另一个为控制增量为零时的自由分量,是一种非齐次时变线性模型.此外,应用广义预测控制理论,提出了一种基于该模型的自适应多步预测控制策略,导出了基于该模型的多步最优预测算式和最优控制律.该控制策略不仅具有广义预测控制的强自适应能力和强鲁棒性,且模型参数少,算法简单,适用于黑箱系统的控制.仿真结果表明该控制策略是有效的.

In this paper, an inputoutput incremental simple linear regression model is used as a predictive model of black box systems, and a projection algorithm is adopted to estimate the model parameters. This model divides the output increment into two components, i.e., one is the forced component caused by the nonzero input increment, the other is the free component corresponding to zero input increment. It is an inhomogeneous timevarying linear model. Then based on the model and the parameter estimation, an adaptive multistep predictive control strategy is proposed by using the adaptive generalized predictive control theorem, the multistep optimal predictive algorithm and the optimal control low are deduced. Not only does the control strategy inherit the strong adaptability and the robustness of the adaptive generalized predictive control, but also the control algorithm is simple because of the less model parameters. The control strategy proposed in this paper is suitable for the control of black box systems. Simulation results confirm the effectiveness of the control strategy.