Research On Multi-objective Optimization Algorithm Based On Population Environment Information And Optimization Of Rolling Schedule

Multi-objective problems widely exist in scientific research and industrial production,and multiple conflicting objective functions need to be optimized at the same time.Without too much prior information,a set of compromise solutions can be obtained in a single operation through intelligent optimization algorithm.Therefore,it has become an effective way to solve the multi-objective and many-objective optimization problems.However,when dealing with optimization problems with different characteristics,it is difficult for multi-objective intelligent optimization algorithm to effectively search the decision space,and the maintenance of convergence and diversity is also an obstacle.In addition,the increasing demand in engineering practice has led to an increase in the objective dimension of optimization problems,and the problems have become increasingly complex.In the high-dimensional objective space,the convergence of the population is degraded,and the proportion of non-dominated solutions is increasing.It is difficult to compare the pros and cons of the solutions,which leads to the difficulty of solving the problem further.In the face of complex optimization problems,the environmental information of the population can often reflect the characteristics of the problem and the current operating status.Therefore,this paper makes full use of the feedback information of the population to carry out research around the multi-objective optimization problem.The main work is as follows:Aiming at the problem that the population convergence and diversity are difficult to balance in the multi-objective optimization algorithm,a completion degree detection factor is proposed to guide individual evolution.Through the feedback information of the population in the algorithm process,the velocity update formula of the particle swarm optimization algorithm is adaptively selected to increase the effectiveness of the population search;personal optimal and global optimal solutions are selected through decomposition method,and dominance-based methods is employed to select elite solution sets,so as to achieve the balance of population convergence and diversity.Next,for the multi-objective optimization problems with complex characteristics,in order to improve the population distribution,the shape of adaptive reference vector is used to guide the population evolution.In the initialization stage,the reference vector sets with different shapes are generated in advance.During the operation of the algorithm,the reference vector sets with high consistency with the population shape are adaptively selected,in order to guide the search direction of the population and ensure the accurate and effective evolution of the population.In the early stage of the algorithm,the self-learning strategy is used to maintain the population convergence;in the middle and late stage of the algorithm,the reference vector is used to select the solutions to improve the population diversity.Simulation results show that the proposed algorithm can effectively deal with multi-objective optimization problems and balance convergence and diversity.Aiming at the problem of poor diversity in many-objective or irregular Pareto front problems,a many-objective decomposition evolutionary algorithm based on neighborhood adaptation adjustment is proposed.By analyzing the surrounding environment of each individual in the current population,the neighborhood information of each individual is used to adapt the position of the reference vector.By this method,the calculation resources of the reference vector of the sparse or discrete area are allocated to the neighboring area to enhance the adaptability of the population to the Pareto front;the pioneer dynamic population strategy is introduced to quickly search the objective space with fewer individuals,and non-dominated individuals are accumulated generation by generation to quickly reduce the search range.Finally,the dominance method and the decomposition method are combined to select the environment.Multiple criteria are adopted to increase the selection pressure,refine the information of each generation population and balance convergence and diversity.The simulation results show that the proposed algorithm can handle many-objective optimization problems and effectively enhance the diversity of the population.Aiming at the problems of large search space,insufficient selection pressure and difficult to converge of many-objective optimization problem,a multi-stage many-objective evolutionary algorithm based on sampling points is proposed.Clustering method is hired to identify and eliminate the dominance inhibition solution in the population,so as to determine the extreme solutions of the population,narrow the search area of the population,and speed up the convergence speed.The change information of the sampled population is fully considered to determine the environmental status of the population,and the population search status is changed by adjusting the offspring matching pool.Then,the matching pool guide the population to explore or develop the search space and improve the convergence and diversity of the population in stages.In the environmental selection mechanism,the solution,which is different from the selected population,is chosen on the basis of non-dominated solution to increase the population diversity.The simulation results show that the proposed algorithm can increase the selection pressure in the high-dimensional objective space and balance the overall performance of the population.The rolling schedule optimization of a tandem cold rolling mill is a typical multi-objective optimization problem.With the goal of reducing production energy consumption,improving equipment utilization,and ensuring product quality,the two multi-objective optimization algorithms proposed in this paper are applied to the study of rolling schedule optimization.The simulation experiment results show that in the2-objective and 4-objective rolling schedule optimization problems,the optimized rolling schedule candidate can make full use of the rolling equipment and effectively reduce the probability of slipping.It has certain guiding significance for the formulation of process regulations in practical production.