Abstract: The Elman network are used to identify model of the quadruped robot hydraulic driving position unit since the mathematical model of linear differential equations that can not represent the actual system. In order to reduce the error between Elman network output and expected output, BFGS and GDX are used to correct the weight of the network and Mean Square Error (MSE) and Normalized Mean Square Error (NMSE) are used to correct the error function. BP neural network is designed in order to on-line adjust PID parameters based on identification model. The experimental results show that the fitting accuracy between the identification model data and the experimental data is high, and the BP neural network PID algorithm based on the identification model is effective, which further verifies the validity of the identification model.

Key words:  Hydraulic driving, Elman neural network identification, BP neural network control, position unit

摘要： 针对线性微分方程数学模型不能反映实际系统的问题，采用Elman网络对四足机器人液压位置驱动单元进行动态神经网络辨识研究。为了减小Elman网络输出与期望输出之间的误差，采用拟牛顿算法BFGS和自适应学习率算法GDX对网络的权值进行修正，并采用均方误差(Mean Square Error, MSE)与归一化均方误差 (Normalized Mean Square Error, NMSE)修正误差函数。基于辨识模型设计BP神经网络来修正PID参数。实验结果表明，辨识模型数据与实验数据拟合精度高，且基于辨识模型的BP神经网络PID算法控制有效，进一步验证辨识模型的有效性。

关键词: 液压驱动, Elman神经网络辨识, BP神经网络控制, 位置单元