Research On Multi-objective Particle Swarm Optimization For Load Distribution Of Hot Finishing Mills

Rolling schedule calculation of hot finishing mills is the basis of continuous strip rolling operation,which is an important factor of product quality and stability.Moreover,the load distribution is the premise and basis of rolling schedule calculation,also is a key step of the rolling schedule calculation.The reasonable load distribution can effectively utilize equipment capacity,which not only can improve the yield and the quality of the product,but also can reduce the rolling energy consumption and improve the production efficiency.Therefore,it is of great practical significance to study the load distribution of hot finishing mills.As multi-objective optimization algorithms based on decomposition can keep solutions diversity effectively but they are hard to obtain the nadir points;rather,multi-objective optimization algorithms based on Pareto dominance can acquire the nadir point but their diversity is poor in the complex optimization problems,aiming at these problems,this dissertation firstly proposes a hybrid multi-objective particle swarm optimization,namely HMOPSO,which obtains Pareto front based on the Pareto dominance which can promote population convergence towards Pareto front,and uses the decomposition to maintain external archive by the method of objective space being normalized based on the nadir point of Pareto front and population being partitioned,which can improve the distribution performance of population.Through the research on the math model of the load distribution of hot finishing mills,a multi-objective optimization model based on traditional experience load distribution method is established with load balancing,good strip shape and minimum power as the goal,the exit-thickness of each mill as the decision variables and the process characteristics and actual production conditions as the constraints.Simulation results show that the convergence and distribution performance of the Pareto optimal solution set obtained by HMOPSO is competitive with respect to other two state-of-the-art algorithms,and the load distribution plan obtained by HMOPSO is more reasonable compared with the empirical load distribution.In order to address the problem that rolling process may be affected by uncertainties and lead to fluctuation of product performance,a robust optimization model considering the optimality,robustness and feasibility robustness is established.The model measures the optimality of the solution based on the expected performance of the objective function of the sampling points,the robustness of the solution based on the standard deviation of the objective function of the sampling points,and uses 6 ? to ensure the feasibility robustness of the solution.Simulation results show that the overall performance of HMOPSO is better;in addition compared with the load distribution plan of load distribution optimization problem,the load distribution plan of the robust optimization problem can guarantees optimality and robustness.