克服广义预测控制模型参数失配的PID反馈校正策略

虽然广义预测控制(GPC)能够通过不断的在线辨识克服模型失配的不利影响,但它计算耗时,不利于在线实时的情况.为了克服在线辨识计算量大的缺点,提出了一种简便易行,计算量小的PID反馈校正策略,可有效地消除模型参数失配对控制性能的不利影响.由于所提出的方法把基于未来预测信息的超前控制和基于过去误差的事后控制有机地结合起来,从而呈现出良好的动态特性.计算机仿真表明,该方法能够有效克服模型与过程之间的增益失配问题,改善控制效果.

PID feedback correction strategy overcoming model parametric mismatch of GPC

Although repeated online identification, which is commonly used by a generalized predictive control(GPC) algorithm, can effectively overcome model mismatch, it is extraordinarily adverse to real time situations due to heavy computational burden. Therefore, a PID based feedback correction strategy is proposed in this paper to attenuate the detrimental influence on control performance induced by model mismatch of GPC algorithms. This correction strategy is very simple and its computational burden is also very light. Since this method incorporates a past error based PID method into a predictive information based GPC strategy, the system performance is significantly superior to a sole PID or GPC approach. A simulation example shows that the presented method can effectively overcome modeling errors and improve control performance.