The Study Of Strictly Constrained Optimization Based On Multi-Objective Optimization Algorithm

Optimization problems are widespread in the control,network communication,water conservancy and hydropower,computer engineering.However,the engineering practice problems often have high complexity,difficulty and strong constraints.The traditional optimization algorithm cannot meet the requirements of calculation speed,convergence and initial sensitivity when solving these problems.Intelligent optimization algorithm relies on a population and randomly generated new solutions in the form of "generate + detection" on the basis of a certain policy or rules in the solution space,and it make population gradually approach the optimal solution of problems and finally converges by the evaluation and comparison with new solutions.So it has become an effective way to solve the optimization problem with strong constraints.Therefore,this paper applies "swarm intelligence algorithm +constraint processing technology" to solve the optimization problem with strong constraints.The main research contents are as follows:(1)According to large computational complexity of classical multi-objective particle swarm optimization algorithm(CMOPSO),we use the crowded degree operator replace the original algorithm update mechanism.And adopting a mutation operator,the number of mutation particles decrease with the increase of the number of iterations to ensure the algorithm's development capabilities in the late iterations.(2)We transform constrained optimization problem into multi-objective optimization problem.Aiming at the different essence of the multi-objective optimization problem and the traditional multi-objective optimization problem,the external file maintenance and updating mechanism is improved,and an improved multi-objective particle swarm algorithm M?IMOPSO is proposed.(3)On the basis of the improved algorithm,this paper aims to improve the external archive maintenance and update mechanism according to the different essence between the multi-objective problem after transformation and the traditional multi-objective optimization problem,(3)Aiming at the defect of low quality of random initial population,this paper added greedy algorithm based on M?IMOPSO algorithm and selected appropriate greedy strategy to initialize the population.In addition,using constraints violation and crowded degree to maintain external archive to improve the performance of the algorithm.