Three-phase Electrode Submerged Arc Furnace Lift The Automatic Control System Research And Application

Submerged arc furnace is the core device of silicon-manganese production. Through three-phase electrode to import current into the furnace, the furnace can produce arcing resistance thermal to heat up raw materials, thus chemical reduction reactions can take place to produce silicon-manganese under certain temperature. Three-phase lifting system control performance directly affects furnace temperature and the balance of three-phase power pool, which determines smelting efficiency and quality of silicon-manganese alloy.Taking automated reconstruction project of large-scale closed submerged arc furnace in a ferroalloy factory as study background, in order to realize process optimization of silicon-manganese alloy production, enhance productivity and reduce energy consumption, hardware platform of three-phase lifting automatic control system is constructed. And researches on submerged furnace system analysis, system modeling and control strategy are conducted.The diversity of the furnace conductive method, the complexity of furnace smelting reactions and coupling of furnace three-phase current determine that submerged arc furnace had time-varying, strong coupling and multi-variable non-linear characteristics.According to characteristics of submerged arc furnace, a kind of location-current model which take electrode location as input and electrode current as output is established. The parameters of submerged arc furnace model are identified by least squared method. The model can truly reflect function mapping relationship between submerged arc furnace location and current, which is significant for studying control strategy.Aiming at characteristics of submerged furnace arc system, a kind of neural adaptive PSD (proportion, sum and differential) decoupling control algorithm is designed. Comparing to traditional PID control algorithm, the proposed control algorithm has characteristics as follows: not relying on mathematic model of subject; simple structure with clear physical meaning; with on-line learning function which can overcome control system affects caused by time-varying parameters; short learning time, rapid parameters convergence, strong adaptive ability and good decoupling effect. Simulation experiments show that the control algorithm has better dynamical characteristics which can greatly improve control performance of phase lifting system and enhance production efficiency.