Research On Improved Multi-objective Particle Swarm Optimization Algorithm And Its Application In Power Curve Of Electric Arc Furnace

The heat source of Electric Arc Furnace steelmaking process is electric energy, the electric energy converted into heat energy by the arc discharging between the electrode and the burden, the heat energy melted metal and furnace slag, and it can smelt various components of steel and alloy. The reasonable power supply curve is the premise of the stable operation, fast-paced and low-cost smelting, it is very essential to the steel and iron industry.Traditional formulation power strategy method is simplified to determine the arc current. But the strategy of SR-EAF (Shunt Reactor Electric Arc Furnace), the reasonable electrical energy input, working current, voltage and reactance value are needed for the purpose of shortening smelting cycle, saving electrical energy, reducing electrode consumption and prolonging lining life in different stages among the smelting process. So the EAF power supply curve optimization is a typical mutil-objective optimization problem (MOP). Swarm intelligence optimization algorithm for multi-objective optimization problem is a hot topic in the current scientific research for its fast convergence and simplicity, it attracted more and more attention of researchers, and widely used in function optimization, neural networks, fuzzy systems control and other applications.According to the problems when PSO (Particle Swarm Optimization) solved the multi-objective optimization, series improved algorithms are presented to solve the optimization model. The dissertation consists of five parts as follows:1. The basic concepts of MOP and the MOPSO algorithms are summarized, which is the preparation for the further research on the MOPSO algorithms.2. A multi-objective EAF power curve optimization model is established, in consideration of cost and factor. The electric power consumption per ton steel, smelting time, electrode consumption and lining life are considered simultaneously in the optimal model. Under the smelting process allowable conditions, the optimal reactance value, working current and voltage set to enhance comprehensive economic efficiency.3. A dynamic swarm multiobjective optimization PSO-based maximin function is proposed. The method aimed at electric arc furnace supply optimization model with a multi-peak closed interval search space, in which the swarm size is adjusted dynamically to avoid converging at a false Pareto Front strengthened the algorithm’s local search ability, in order to balance calculation cost and algorithm performance, meanwhile, search more Pareto solutiuons. The relative fitness variance of particle is introduced into the formula of Maximin to get rid of the effect of population diversity reducing which guarantee for convergence to the Pareto front.4. A chaos variable regional multiobjective optimization PSO algorithm is proposed. The method aimed at electric arc furnace supply optimization model with constraints, accroding to the constraints MOP. In which transform strategy will convert the constraint conditions to one preparative optimization objective, by weighted summing the satisfactory function of constraint conditions. A variable domain acceleration operator is introduced to expedite the convergence speed of the algorithm. Meanwhile, chaotic operator is introduced to prevent the algorithm from prematurity, so as to enhance the algorithm’s searching capability around local optimal solutions. ARCS operator is used to guarantee the diversity of populations.5. An improved multi-objective particle swarm optimization algorithm based artificial immune network (AIN) is proposed. The method aimed at electric arc furnace supply optimization model with multi-peak closed interval search space and constraints, in which in order to give simultaneously attention to convergence performance and solutions quality, improve the golbal search ability, AIN algorithm is used. An improved migration method of particle position is improved to accelerate convergence speed; an improved adaptive variance mutation method and clustering immune network are proposed in order to enhance the function of MOPSO and AIN. The global convergence properties, convergence rate and time complexity of the improved algorithm are analyzed and described in theory. Finally, the improved PSO algorithm solved the electric arc furnace supply optimization model and the power supply strategy is used to guide the steelmaking process. Compared with the traditional strategy, power and electrode consumption has a better performace. The control of the electric arc furnace process indicators and economic indicators have been realized.