Optimal Tracking Neuro-Control of Continuous Stirred Tank Reactor Systems: A Dynamic Event-Driven Approach
Xiong Yang, Yingjiang Zhou
Abstract
It is often challenging to design an optimal tracking controller for the continuous stirred tank reactor (CSTR) system due to its nonlinearity nature and physical limitations. This paper presents a dynamic event-driven optimal tracking neruo-control scheme for the CSTR system with asymmetric input constrains. Initially, an improved nonquadratic cost function is introduced for the CSTR system to tackle asymmetric control restrictions. Then, a dynamic event-driven mechanism together with the event-driven Hamilton-Jacobi-Bellman equation (ED-HJBE) is proposed. To solve the ED-HJBE, a critic neural network (CNN) is constructed within the critic learning framework. The CNN's weight vector is tuned via combining the gradient descent method and the concurrent learning technique. After that, uniform ultimate boundedness of the tracking error and the CNN's weight estimation error is assured based on Lyapunov method. Finally, experiment studies are conducted to validate the present optimal tracking neuro-control strategy.