A novel intelligent coordination control system for a unit power plant

Based on the integration of neuron model-free controller with traditional PID controller, a novel intelligent control method is proposed for multivariable systems in this paper. This new control strategy not only retains the characteristics of the traditional PID, such as its simple structure and practicable control algorithm, etc., but also divorces the generation of control signal from the plant model. A PID controller is placed in each main input-output path. Between the PID controller and the controlled plant, a neural network composed of the model-free neuron is set up to compute the control signal according to the online learning from the system error and the outputs of each PID controller. The simulation test results show its good performances and strong robustness in a unit power plant with big uncertainties of the load. This new multivariable intelligent control method is very simple. The parameters of the controller can be chosen roughly. The method can be used conveniently in practice.