基于小波网络动态补偿的广义预测控制器  

GENERALIZED PREDICTIVE CONTROL WITH DYNAMIC COMPENSATION BASED UPON WAVELET NETWORK

作　　者：王群仙[1] 李少远[1] 李焕芝[1] 陈增强[2] 袁著祉[2] 

机构地区：[1]河北工业大学电气工程系,天津300130 [2]南开大学计算机与系统科学系,天津300071

出　　处：《自动化学报》1999年第5期701-704,共4页Acta Automatica Sinica

摘　　要：由于广义预测控制器依赖于表征过程的线性模型，没有考虑未建模误差的影响，随着预测长度的增加，其预测误差也增大．因此，提出用小波网络建立误差的预测模型，并对模型预测进行动态补偿，以抑制模型失配的影响，增强鲁棒性．仿真结果表明了算法的有效性．Generalized predictive control relies on the process linear model and doesn't consider the effect of unmodelled error. With a longer prediction length, the prediction error becomes greater. Therefore this paper presents an error prediction model using wavelet network and uses it to compensate the error dynamically. The adjusted GPC can reduce the influnce of unmodelling section and enhance its robustness. The simulation results have shown the method's validity.

关 键 词：小波网络 预测控制 鲁棒性 广义预测控制器 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]