基于神经网络逼近器的液压支架多缸协同控制

Multi Cylinder Coordinated Control of Hydraulic Support Based on Neural Network Approximator

针对当前综采工作面液压支架定位精度低、直线度控制困难等不足。利用神经网络学习训练速度快、逼近能力强等优势,构建了RBF神经网络逼近器。通过高阶滑模状态观测器对液压缸状态进行估计,利用神经网络逼近器来逼近液压缸扰动非线性函数,从而完成液压支架多缸控制。MATLAB仿真结果表明,基于神经网络逼近器的液压支架多缸协同控制方法响应速度快,控制精度高,且具备协同作业的能力,能够进一步提升液压支架的控制效率。

In view of the low positioning accuracy and difficult straightness control of the hydraulic support in the current fully mechanized mining face.Using the advantages of fast learning and training speed and strong approximation ability of neural network,a RBF neural network approximator is constructed.The state of hydraulic cylinder is estimated by high-order sliding mode state observer,and the neural network approximator is used to approximate the nonlinear function of hydraulic cylinder disturbance,so as to complete the multi cylinder control of hydraulic support.MATLAB simulation results show that the multi cylinder collaborative control method of hydraulic support based on neural network approximator has fast response speed,high control accuracy,and the ability of collaborative operation,which can further improve the control efficiency of hydraulic support.