电液负载模拟器神经网络辨识器及控制器设计

针对传统控制方法无法解决飞机舵机电液负载模拟器受多余力等非线性因素严重干扰的问题,给出了一种基于神经网络辨识器及控制器的复合控制结构,结合了神经网络系统辨识与自适应实时控制的工作特点。根据电液负载模拟器控制结构及工作原理,采用BP神经网络辨识器在线辨识,获得系统辨识模型以替代理论数学模型。然后,采用Adaline神经网络控制器实时控制,利用系统误差信号与BP神经网络反向递归计算Adaline网络权值调整信息,获得系统控制参数,实现复合控制器的有效监督与智能控制。最后利用MATLAB进行实验验证,仿真结果表明：该方法能够提高系统控制精度,多余力消扰率达92%；并且可以有效模拟飞机舵机所受力载荷的变化情况,实现系统指令信号快速、准确、稳定的加载。

Neural Network Identifier and Neural Network Controller on Electro-hydraulic Load Simulator

In view of the problem that is caused by surplus force and other nonlinear factors of aircraft rudder electro-hydraulic load simulator is not easy to solve by traditional control method, compound controller is proposed that based on the working characteristics of neural network system identification and adaptive real-time control. According to the control structure and working principle of the electro-hydraulic load simulator, the theoretical mathematical model is replaced with identification model by BP neural network online identification. Secondly, the Adaline network controller can realize real-time control, with the help of systematic error signal and BP neural network reverse recursive to calculate network weight adjustment information to obtain system control parameters. Thus effective supervision and intelligent control are achieved through the compound controller. Finally, by the means of MATLAB to verify the method, the results show that this method not only can improve control accuracy and suppress surplus force by 92%, but also can effectively simulate changes in the power loads experienced by the aircraft rudder, and achieve rapid, accurate and stable loading for system command signal.