基于扰动观测器的双容液位系统RBF神经网络滑模控制  

RBF Neural Network Sliding Mode Control of Two-tank Liquid Level System Based on Disturbance Observer

作　　者：张克 于海生[1,2] 孟祥祥 颜克甲 ZHANG Ke;YU Haisheng;MENG Xiangxiang;YAN Kejia(College of Automation,Qingdao University,Qingdao 266071,China;Shandong Province Key Laboratory of Industrial Control Technology,Qingdao 266071,China)

机构地区：[1]青岛大学自动化学院,山东青岛266071 [2]山东省工业控制技术重点实验室,山东青岛266071

出　　处：《控制工程》2024年第5期954-960,共7页Control Engineering of China

基　　金：国家自然科学基金资助项目(61573203);山东省自然科学基金资助项目(ZR2021MF005)。

摘　　要：针对双容液位系统存在的外部扰动、模型参数不确定等问题,设计了一种基于非线性扰动观测器(nonlinear disturbance observer,NDOB)的径向基函数神经网络滑模控制(radial basis function neural network sliding mode control,RNNSMC)方法。建立双容液位系统数学模型,采用积分型滑模面来提高系统的鲁棒性,在常规积分滑模控制的基础上,通过RBF神经网络(RBF neural network,RNN)对系统的非线性函数进行逼近,并设计非线性扰动观测器估计外部扰动,选用Lyapunov稳定性判据证明了控制策略的闭环稳定性。仿真结果表明,所提控制策略与积分滑模控制(integral sliding mode control,ISMC)方法相比,不需要精确的数学模型,且控制精度更高,抗干扰能力更强。For the problems of external disturbance and model parameter uncertainty in two-tank liquid level system,a RBF neural network sliding mode control(RNNSMC)method based on disturbance observer(NDOB)is designed.The mathematical model of two-tank liquid level system is established,the integral sliding mode surface is used to improve the robustness of the system.Based on the conventional integral sliding mode control,the nonlinear function of the system is approximated by RBF neural network(RNN),and a nonlinear disturbance observer is designed to estimate the external disturbance,the closed-loop stability of the control strategy is proved by using Lyapunov stability criterion.The simulation results show that compared with the integral sliding mode control(ISMC)method,the proposed control strategy does not need an accurate mathematical model,and has higher control accuracy and stronger anti-interference ability.

关 键 词：RBF神经网络 滑模控制 双容液位系统 非线性扰动观测器 外部扰动 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]