| The glass furnace charged object is a seriously nonlinear, time-variant, complex process system. The process has inherence characteristics of large uncertainty, delay time, time constant and interference factors. So it is hard to build the precise mathematical model. Since the traditional control strategies have to depend on a precise mathematical model, when the delay time and parameters change, the control system doesn’t work very well.Fuzzy logic control and neural network control have the characteristic of special nonlinearity and need not establish the precise mathematical model. Fuzzy control system is good at expressing knowledge and its logical reasoning is similar to man’s thought. But this system lacks capacity of adaptation and learning. Neural network has the characters of self-organization and self-learning. But its network parameters lack the physical meaning capacity. In this paper, after combining these two system for absorbing their advantages, the author proposed a controller built with compensatory fuzzy neural networks to control the glass furnace temperature. Through the introduction of compensatory fuzzy inference and rapid learning algorithm, the compensatory fuzzy neural networks can work much better than the common fuzzy neural networks.The thesis introduces the glass furnace structure, technological process, according to the requirement of the field control, presents the requirements of temperature control system and analysis of the glass furnace system dynamic characteristic. Through certain simplification and assumption, this paper gives the mathematical model of glass furnace temperature control process.Based on the simulation, the structure of the network whose object is the glass furnace temperature is illustrated and compared with PID and fuzzy control method. The simulation results show that the neural network controller based on the compensation fuzzy logic overcomes the disadvantages of the traditional PID control and fuzzy control with large overshoot and slow response. It has very good robustness and adaptive ability and self-organization ability. So it can work very well in the glass furnace temperature control system. |
