Research On Many-objective Particle Swarm Optimization Algorithm

Many engineering system design,modeling and planning problems are related to multi-objective optimization problems,such as industrial scheduling,software engineering,resource allocation,etc.Almost all the important production practice decisions include multi-objective optimization problems,and many problems involve more than three objectives.It is known that effective solutions to such problems have important practical significance.However,the dimensionality and scale of the solution increase with the increase of the objective number,which leads to high complexity and difficulty in solving the problem.It is necessary to design an effective multi-objective optimization algorithm,improve its convergence and diversity,and maintain the balance between them.For constrained optimization problems,improving the feasibility of solutions is also one of the challenging research contents.Aiming at the above-mentioned difficulties and bottlenecks involved in multi/many-objective optimization,this dissertation studies multi/many-objective particle swarm optimization algorithms based on different dominance relationships to improve the convergence,diversity,and feasibility of the proposed algorithms.The main research contents of this dissertation are as follows:(1)A multi/many-objective particle swarm optimization algorithm based on intuitionistic fuzzy dominance is proposed.By combining intuitionistic fuzzy sets with Pareto dominance,an intuitionistic fuzzy dominance is proposed to enhance the convergence of the algorithm.The particle swarm optimization algorithm with dual search strategy is adopted to update the population to improve the search ability of particles in the objective space.The uniform reference points in the objective space are used to balance the convergence and diversity of the algorithm.The experimental results show that the algorithm is better than the comparison algorithms on most of the DTLZ and WFG test problems,indicating that the algorithm is advanced.(2)A multi/many-objective particle swarm optimization algorithm based on competition mechanism is proposed.The individual competition mechanism is improved to enhance the search ability of particles in high-dimensional objective space.The-dominance is adopted to improve the convergence of the algorithm to increase the selection pressure on the elite individuals.In order to ensure the diversity of the algorithm,the individuals with better distribution are selected by the maximum and minimum angle between the unselected and the selected individuals.The experimental results show that the algorithm can effectively solve most of the continuous and complex test problems of DTLZ,WFG,UF,which reflects the feasibility of the algorithm.(3)A multi/many-objective particle swarm optimization algorithm with convergence assisted strategy is proposed.The random strategy is used to combine the shift distance and the vector angle to select the population,which improves the diversity of the algorithm.An auxiliary strategy with strong global convergence is designed to enhance the convergence of the algorithm.Experimental results show that the algorithm can effectively handle the multi/many-objective optimization problems of DTLZ,WFG,Ma F,and has obvious advantages.(4)A constrained multi/many-objective particle swarm optimization algorithm based on two-level balance strategy is proposed.In order to enhance the convergence and diversity of the algorithm,the penalty-based boundary crossover operator is used as the utility function,and the corresponding environment selection strategy is designed on the basis of ranking the fitness of each objective.Based on the constraint dominance principle,the calculation of the constraint violation degree of the solution is used to enhance the performance of the algorithm in handling constrained multi/many-objective optimization problems.Individuals are selected based on the angle between vectors to maintain the diversity of the algorithm.Experimental results show that the algorithm can effectively deal with unconstrained and constrained multi/many-objective optimization problems on the DTLZ,WFG,CF,DOC,C-DTLZ,DOC-DTLZ,and its performance is better than other comparative algorithms.