A novel robust adaptive controller for EAF electrode regulator system based on approximate model method  

作　　者：李磊 毛志忠 

机构地区：[1]School of Information Science and Engineering,Northeastern University

出　　处：《Journal of Central South University》2012年第8期2158-2166,共9页中南大学学报（英文版）

基　　金：Project(N100604002) supported by the Fundamental Research Funds for Central Universities of China;Project(61074074) supported by the National Natural Science Foundation of China

摘　　要：The electrode regulator system is a complex system with many variables, strong coupling and strong nonlinearity, while conventional control methods such as proportional integral derivative （PID） can not meet the requirements. A robust adaptive neural network controller （RANNC） for electrode regulator system was proposed. Artificial neural networks were established to learn the system dynamics. The nonlinear control law was derived directly based on an input-output approximating method via the Taylor expansion, which avoids complex control development and intensive computation. The stability of the closed-loop system was established by the Lyapunov method. The current fluctuation relative percentage is less than ±8% and heating rate is up to 6.32 ℃/min when the proposed controller is used. The experiment results show that the proposed control scheme is better than inverse neural network controller （INNC） and PID controller （PIDC）.The electrode regulator system is a complex system with many variables, strong coupling and strong nonlinearity, while conventional control methods such as proportional integral derivative (PID) can not meet the requirements. A robust adaptive neural network controller (RANNC) for electrode regulator system was proposed. Artificial neural networks were established to learn the system dynamics. The nonlinear control law was derived directly based on an input-output approximating method via the Taylor expansion, which avoids complex control development and intensive computation. The stability of the closed-loop system was established by the Lyapunov method. The current fluctuation relative percentage is less than ±8% and heating rate is up to 6.32 °C/min when the proposed controller is used. The experiment results show that the proposed control scheme is better than inverse neural network controller (INNC) and PID controller (PIDC).

关 键 词：approximate model electric arc furnaces nonlinear control normalized radial basis function neural network （NRBFNN） 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TM924.4[自动化与计算机技术—控制科学与工程]