Improved Multi-objective Particle Swarm Optimization Algorithm And Its Application

The multi-objective optimization problems originated in designing， planning andmodeling real complex system. There are many real life important decisions are relative withmulti-objective optimization. This kind of multi-objective optimization algorithm provides amethod and way to solve multi-objective optimization problems. The development ofmulti-objective optimization algorithm has experienced the conventional multi-objectiveoptimization algorithm, multi-objective algorithm based on evolutionary algorithm, achievedgood results. There are many features in common between the particle swarm algorithm andthe traditional evolutionary algorithms, But the particle swarm algorithm has simpler andeasier operation than evolutionary algorithm. A multi-objective particle swarm algorithmavailable for multi-objective optimization problem is a hot topic this year. To further improvethe optimization performance, achieve better solution for some nonlinear, complicatedmulti-objective optimization problems. This paper has proposed an improved particle swarmalgorithm available for multi-objective optimization problems, and presents a constrainedmulti-objective particle swarm algorithm with improved penalty function on the basis of it.Optimization design and improvement of algorithm is finally to solve practical optimizationproblems, and his paper therefore selected two typical nonlinear multi-objective optimizationproblems for applied research, which has achieved satisfying optimal effect, also furtherverified the performance of the improved algorithm and application.Firstly this paper proposed an improved multi-objective particle swarm algorithm basedon proposed non-dominated sorting and density range selection which are effective in solvingnonlinear and complex optimization problems. What is more, it improved selecting individualand global optimal value by bidding competition strategy. Through the simulation study offunction test, the improved algorithm performed an outstanding uniform distributionperformance of Pareto optimal solution set. In order to further verify the performance of thealgorithm and do some application research, the paper designed a PID controller based onimproved algorithm, selecting a kind of inverted pendulum system to simulated, and obtainedsatisfactory results.Based on the improved multi-objective particle swarm algorithm, this paper improved aconstrained particle swarm optimization algorithm which can adjust to punish function thelevel of punishment. The detailed steps of algorithm are given, and through the study ofclassical test functions of simulation, the Pareto front is close to the true front and is superiorto the existing similar algorithms. In order to further verify the algorithm performance andapplication, the SMB simulation model of simulated moving bed were simulated to search foroptimization parameters based on improved constrained algorithm and achieved very goodresults.