Application Of Integrated Intelligent Optimize Control Strategies In Electric Arc Furnace

Electric arc furnace (EAF) system is a kind of industrial production process, which possesses considerable nonlinear, time varying and strong coupled. The performance of electrode regulating system is an important factor in production efficiency. Presently, the constant impedance control method which based on classical control theory is widely used. This paper analyzes the deficiencies of the EAF constant impedance control method, on this basis for the system characteristics and control requirements, refer to domestic and foreign research and development direction of EAF control, study a neural network intelligent control method of EAF, in other word, to identify EAF model and establish EAF controller model.Due to the neural network training algorithm determines the non-linear model identification capability largely whether strong or not. and PSO as the global optimal algorithm owns a unique effect to resolve the local optimal problems when neural network approaching much-variable and complex model. Therefore, this control system adopts PSO to train the neural network weights mainly according to sample data collected from actual scene, which greatly improved the neural network mapping capability and network training speed, and further optimize the controller model to ensure high quality control performance.In addition, based on the neural network control module, the constant impedance expert control system is also designed to ensure safe and reliable in steel-making process, thus makes up for the shortage of single intelligent control strategy. This system can better adapt to load variety and external disturbances; its control performance is superior to conventional control system of EAF. so saves energy and improves production efficiency.To verify the control scheme in this paper, a software & hardware mixed experimental platform is built which based on computer,data acquisition board and hardware circuit. By using the VC++6.0 development tool, realizes the basic system software which contained training algorithms, control algorithms and control interface, then does simulation study on the mixed experiment platform. The result verifies the correctness and validity of control scheme and algorithms, and it has some practical value..