摘要
在分析液压 AGC的组成元件及其动态特性的基础上 ,利用神经网络具有逼近任何非线性函数且具有自学习和自适应的能力 ,建立基于时间序列的前馈动态模型辨识结构 ,应用扩展 BP算法对轧机液压 AGC力控制系统进行非线性预测 ,将预测结果应用最小二乘辨识方法进行线性系统的特征参数辨识 ,仿真及实测结果表明此方法行之有效 ,为轧机液压 AGC的辨识提供了新途径。
Based on studying the components of hydraulic AGC system, and analyzing the dynamic peculiarities of the components, a new adaptive identification method was proposed for a nonliner hydraulic AGC press system in stripe mill, in which a feedforward and dynamic neural network structure was built.Using enlarged backpropagation algorithm,the nonlinear performance of force control system of the hydraulic AGC system can be predicted.Based on the predicted results the characteristic parameters of linear system were identified by least square method. Finally, the applicability of the adaptive identification method was illustrated and verified by simulation results.
出处
《机械工程》
CAS
CSCD
北大核心
2004年第5期450-453,共4页
基金
国家自然科学基金资助项目 (60 0 740 2 2 )