CAJ | 학술논문

针对多目标粒子群算法进行了收敛性和分布性分析，提出了一种应用概率分配的自适应调整惯性因子的粒子群优化算法。该算法通过粒子非劣排序的支配等级，设定个体的适应度数值，为增强最优解集的分散性，采用拥挤距离对适应度进行惩罚，进而根据概率选择比较获取相应的最优个体；同时算法根据粒子个体所处位置以及相应的迭代次数，对惯性因子进行了自适应调整，增强了算法的收敛性。最后通过测试函数对改进算法进行了效果验证，表明了算法的有效性。
침대다목표입자군산법진행료수렴성화분포성분석，제출료일충응용개솔분배적자괄응조정관성인자적입자군우화산법。해산법통과입자비렬배서적지배등급，설정개체적괄응도수치，위증강최우해집적분산성，채용옹제거리대괄응도진행징벌，진이근거개솔선택비교획취상응적최우개체；동시산법근거입자개체소처위치이급상응적질대차수，대관성인자진행료자괄응조정，증강료산법적수렴성。최후통과측시함수대개진산법진행료효과험증，표명료산법적유효성。
After analyzing convergence and diversity of the multi-objective Particle Swarm Optimization(PSO), an improved multi-objective PSO, which is introduced the method of probability assignment, is proposed. The arithmetic calculates the fitness of the particles basis of the dominative rank using the Pareto sorting. In order to increase the diversity of the opti-mal solution, a penalty function including the crowding distance is added to the fitness. Then the best particle is selected according to the probability comparing. Also the inertia coefficients are adjusted adaptively based on the information of the particles and the iterative number performances. So the convergence rate of the arithmetic can be accelerated. At last the validity of the arithmetic is validated via the testing function.