基于DRNN神经网络的挖掘机伺服系统参数辨识

为有效分析挖掘机电液伺服系统，提高依据模型设计控制器的精度，建立了系统状态空间模型。针对模型中的不确定参数，提出了基于对角回归神经网络的系统辨识策略。通过神经网络在线学习得到系统Jacobian信息，将实测信息代入含Jacobian信息与待辨识参数的线性方程，利用最小二乘法求得未知参数。实验表明，辨识模型能从初始阶段的微小误差逐渐地逼近实际系统，所提出的方法能有效辨识系统参数。

Parameter identification of servo system for excavator based on DRNN

To analyze the servo system of the excavator effectively and design the controller precisely, the state space model was built. According to the characteristics of the parameter uncertainties, an identification method based on diagonal recurrent neural network (DRNN) was developed. The Jacobian information was achieved by on line learning through neural networks, and least square method was proposed to identify the unknown parameters by using the achieved information. The comparison experiment demonstrated that the identified model can approximate actual system gradually,and that the proposed method in this paper can meet the need for identifying the unknown parameters.