The Study On Adaptive PID Temperature Control Algorithm Based On RBF Neural Network For Baking Furnace

Anode baking is one of the important steps in the aluminum industry. The qualities of anode directly affect the electricity efficiency and energy consumption of aluminum electrolysis production. Therefore, how to improve the baking technology in order to improve the quantity of anode has become a very important research issue. The only approach to solve this problem is to do systematic research of the fundamental theory and control algorithm of the anode baking furnace. At the same time, it is an urgent requirement to develop and to design new baking furnaces control system independently in China. The research of temperature control algorithm is based on the anode baking furnace of Baiyin Aluminum Ltd in this thesis.Firstly, the thesis depicts baking theory of anode baking furnace in detail, introduces the work principle and structure constitute, and summarizes the production craft of anode baking furnace. It also reviews the domestic and international anode baking technique development condition at present, and analyzes the realistic meaning of the anode baking furnace control algorithm. The thesis proposes a synthetic research method to further the temperature control algorithm of anode baking furnace in China, Based on this idea, the relevant research work follows.An adaptive PID control strategy based on Radial Basis Function (RBF) neural network(NN) is proposed in this thesis for the temperature control system of anode pre-baking furnace, which is a complex system with the characteristics of time-delay, nonlinear and no precise mathematical model, it has immeasurable disturbance and strong temperature coupling between baking furnace room. The parameters of PID controller are tuned on-line using the self-learning ability of RBFNN. So the proposed control strategy has the advantages of neural network and conventional PID control, and the ability of making correspondingly the decisions quickly according to the current characteristic of the object and overcoming the inconsistency of the steady and veracity. A distributed decoupling control of neural network is adopted for decoupling control of anode pre-baking furnace, and the corresponding decoupling control law is achieved by the self-learning, nonlinear mapping and fault-tolerant ability of neural network. The structure of control algorithm is simple, and easy to achieve parallel distributed real-time processing, it will simplify the design problem in project. So it is suitable for the anode pre-baking furnace temperature control.Based on the two algorithms above, an adaptive PID online decoupling control algorithm based on Radial Basis Function (RBF) neural network(NN) is proposed in this thesis, which is designed to meet the high targets of control requirements and complex conditions of anode pre-baking furnace. The algorithm not only avoids the problem of poor quality control with multivariable strong coupling plant when using parameters adaptive PID control algorithm individually. But also solves the problem of original parameters of the controller can not adapt to changes in the controlled plant. It was applied to the temperature system of the anode pre-baking furnace, the results show that the proposed controller has the adaptability, strong robustness and satisfactory control performance.It is proved to be a useful research method in the design and process optimization of the baking furnace of aluminum industry. At the same time, it will promote the development of intelligent control technology and its extensive application in industrial process control.