基于α阶神经网络逆系统的液压活套解耦控制

Decoupling Control of Hydraulic Looper Based onα-order Neural Network Inverse System

针对液压活套系统的非线性和参数时变性,建立液压活套系统动力学模型,并通过Interactor算法对模型的可逆性进行分析。利用活套系统输入输出数据训练神经网络,将α阶神经网络逆系统与活套系统串联复合成伪线性系统,采用PSO-PID控制器构成闭环控制回路。结果表明:基于神经网络逆控制的方法能获得良好的解耦效果,并且能克服系统模型的不确定性和参数的时变性对解耦效果的影响。利用该方法可有效避免逆解耦控制法依赖系统模型精度、对系统模型变参数化敏感的缺点,具有较强的鲁棒性。

In view of the nonlinearity and parameter time-varying of the hydraulic looper system, the dynamic model of the hydraulic looper system was established and the reversibility of the model was analyzed by using the Interactor algorithm. The neural network was trained by using input and output data of the looper system. The inverse system of α-order neural network and the looper system were combined in series to form a pseudo linear system, and the PSO-PID controller was used to form a closed-loop control loop. The results show that the method based on neural network inverse control can achieve good decoupling effect, and avoid the influence of unmodeled of system model and time-varying parameters on decoupling effect. By using the method, it can effectively avoid the shortcomings that the inverse decoupling control depends on the system model and is sensitive to the system model’s variable parameters, and has strong robustness.