多步预测自校正控制器

A Multi-step Prediction Self-tuning Controller

本文在作者提出的最小预测误差自校正控制器的基础上,又提出了一种多步预测自校正控制器的新算法。该算法运用动态规划法的思想严格推导出使目标函数真正达到最优的控制律,具有预测误差小、占用存贮少、计算简单的优点,而且便于理论分析与研究,是一种对模型要求较低、鲁棒性能好的控制算法。

In this paper the multi-step prediction self-tuning controller is shown, which is based on the 'A Minimum Predictive Error Self-tuning Controller' appoach proposed by the authors. The dynamic programming is applied to derive the optimal control. The predictive error is minimized at each prediction step. The controller needs less calculation and memory. The algorithm is also facilitates analysing and studying the controller theoretically. It has been shown that the controller has a good robust performance.