反重力铸造装备的模糊建模与预测控制  

Fuzzy Modeling and Predictive Control for the Counter-gravity Casting Equipment

作　　者：曹兴华[1] 韩勇[1] 杨煜普[1] 

机构地区：[1]上海交通大学电子信息与电气工程学院,上海200240

出　　处：《自动化仪表》2011年第5期49-52,共4页Process Automation Instrumentation

基　　金：国家"973"计划基金资助项目(编号:2004CB720703)

摘　　要：工业中的反重力装备在铸造过程中具有时滞性、强非线性和时变性的特点,传统的控制算法难以达到满意的效果,由此提出了一种基于T-S模型的模糊预测控制方法。该方法根据T-S模型的描述来设计神经网络结构,利用误差反传算法离线辨识前件参数和后件参数,使用加权最小二乘递推方法进一步在线修正模型的后件参数,并将T-S模型转化为状态空间模型,使用丢番图方程推导出预测控制律。仿真结果表明此方法具有良好的动态特性。Due to the features of the counter-gravity equipment in casting industrial process,e.g.,time delay,strong nonlinear and time varying,satisfactory effects are difficult to be reached by using traditional control algorithms.Thus the fuzzy predictive control method based on T-S model is proposed.In accordance with the description of T-S model,the structure of neural network is designed.By adopting error back propagation algorithm,the antecedent and consequent parameters are identified offline,while the consequent parameter of the model is modified online with the weight recursive least square method in further,and the T-S model is transformed into state space model,the predictive control law is deducted from Diophantine equation.The result of simulation shows that this method offers excellent dynamic characteristics.

关 键 词：神经网络 预测控制 最小二乘法 T-S模型 误差反传 铸造业 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]