Design Of The Controller For Electric Arc Furnace Based On Neural Network

The electrode-regulator control system is the key link of electric arc furnace steelmaking, its rapid and accurate electrode position control is one of the key factors of saving electric energy, shortening smelting period, reducing energy consumption and prolonging lining life. However, Electrode control system is a complex system with highly nonlinear, parameters time-varying, variable coupling and strong random disturbance. Therefore, electrode regulating system control method becomes the major target for the study of the electric arc furnace control, which by adjusting the electrode position to realize the control of arc length, to achieve the goal of control smelting power.With the continuous development of industrial computer technology, represented by artificial neural network control of intelligent control technology arises at the historic moment, brought new opportunity to solve the problem of traditional control problem.At first, this dissertation introduces the electric arc furnace steelmaking process, equipment and working principle, and expounds the research background and significance. Then on the basis of the understanding of the arc characteristics and working mechanism of electrode regulating system, establishes alternating current arc, the three-phase power supply system and hydraulic system model, and make the corresponding simulation analysis, to verify the rationality of the model.Then, on the basis of consulting a large number of literatures, this dissertation summarizes the current research status of the electrode control method at home and abroad. Combining the requirement of the electrode regulating system to the controller, analyzed the existing theory and method of control, determined to design the electrode regulating system controller based on neural network.Finally, using the linearization method by Taylor series expansion, establishes the approximation model of generalized controlled object of electrode regulating system. Using the echo state network (ESN) to design the electrode regulating system adaptive approximation model controller. In the research of the basic features about ESN and its learning algorithm and according to the characteristics of the electric arc furnace electrode regulating system, put forward to improve the ESN learning algorithm. Designed based on improved the ESN of electric arc furnace electrode adjustment controller, computer simulation verify the rationality and effectiveness of the proposed method.