基于数据驱动的CSTR自适应准滑模控制  

Data-driven Based Adaptive Sliding Mode Like Control for CSTR

作　　者：丰杰华 官星辰 马鲁宁[1] 赵东亚[1] 张贝贝[1] FENG Jie-hua;GUAN Xing-chen;MA Lu-ning;ZHAO Dong-ya;ZHANG Bei-bei(College of New Energy,China University of Petroleum,Qingdao 266580,China)

机构地区：[1]中国石油大学(华东)新能源学院,山东青岛266580

出　　处：《控制工程》2020年第8期1305-1309,共5页Control Engineering of China

基　　金：国家自然科学基金项目(61473312,61973315)。

摘　　要：连续搅拌反应釜(Continuous Stirred Tank Reactor,CSTR)作为化工生产中的核心设备之一,具有典型的非线性特征。针对这种二阶非线性不确定系统,基于积分滑模的控制特性和RBF神经网络的逼近能力设计了一种基于数据驱动的自适应准滑模控制器。仿真结果表明,该算法能很好地使CSTR跟踪给定的温度信号,与普通滑模控制器相比,不需要动力学模型,放松了设计条件;与基于RBF神经网络整定的PID控制器相比,表现出较强的鲁棒性。As one of the core equipment of chemical production,the continuous stirred tank reactor(CSTR)has typical nonlinear characteristics.According to the control characteristic of the integral sliding mode and the approximation ability of RBF neural network,a novel data-driven based adaptive sliding mode controller is developed for this second-order nonlinear uncertain system.The simulation results show that the proposed algorithm can make the CSTR to track the given temperature signal well.Compared with the ordinary sliding mode controller,the proposed control algorithm does not require the system dynamic model and then relax the design conditions;compared with the RBF based PID,the proposed controller has stronger robustness.

关 键 词：数据驱动 自适应滑模控制 CSTR RBF神经网络 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]