Application Of Improved Multi-Objective Particle Swarm Optimization Algorithm In Ethylene Cracking Furnace

The performance of the multi-objective intelligent optimization algorithm mainly includes the global convergence ability,local search ability,convergence speed and running speed of the algorithm,but the intelligent optimization algorithm can not take into account the optimization speed and good quality optimization results.Therefore,in order to improve the convergence speed,running speed,global convergence ability and local search ability of the multi-objective intelligent optimization algorithm,this paper studies and proposes two improved multi-objective intelligent optimization algorithms.The improved multi-objective intelligent optimization algorithm is applied to the production model of ethylene cracking furnace to verify the optimization effect of the improved algorithm in the production model of ethylene cracking furnace.The specific work is as follows.(1)Aiming at the problems that multi-objective intelligent optimization algorithm has long running time,slow convergence speed and cannot get the optimal solution set quickly,the Quantum behavior multi-objective particle swarm optimization algorithm based on improved population variation strategy(IPV-MOQPSO)was proposed.The global optimal solution and individual historical optimal solution are used to guide the variation of particles,and the size and distribution of random range are adjusted according to the progress of algorithm search,so that the variation of particles can be directional and self-adaptive,and the effective variation can be increased and the convergence rate can be improved.By comparing ZDT and UF test functions with the Quantum behavior multi-objective particle swarm optimization algorith(MOQPSO),The results show that the proposed IPV-MOQPSO algorithm has short running time and fast convergence speed.(2)Aiming at the problem that the global convergence and local search ability of the multi-objective intelligent optimization algorithm are weak,resulting in poor distribution and convergence of the optimal solution set,A Quantum behavior multi-objective particle swarm optimization algorithm based on improved δ well and triple dynamic search is proposed(IPT-MOQPSO).By improving the δ potential well model,the probability of particles occurring far away from the center point and the dynamic particle search method are improved,the particle search scope is expanded,the IPT-MOQPSO algorithm is prevented from falling into local optimal,the mobility and flexibility of particle search are enhanced,and the accuracy of the final result is prevented from missing the optimal solution,and the local search ability of the IPT-MOQPSO algorithm is improved.The proposed IPT-MOQPSO algorithm is compared with the MOPSO,the PESA-II,the NSGA-II,the MOEA/D and the MOQPSO on the test functions of ZDT and UF,and it is verified that the IPT-MOQPSO algorithm has stronger global convergence and local search ability.(3)In view of the complex structure of ethylene cracking furnace,long production cycle,unable to quickly get the best production scheme and the diversity of production schemes and low accuracy,the IPV-MOQPSO algorithm and the IPT-MOQPSO algorithm proposed in this paper were applied to the ethylene cracking furnace model.The MOPSO,the PESA-II,the NSGA-II and the MOEA/D algorithms were used as comparison algorithms,and the experimental results were analyzed to verify that the IPV-MOQPSO algorithm has faster optimization speed on complex chemical production models,and the IPT-MOQPSO algorithm has stronger global convergence and local search ability on ethylene cracking furnace.