基于序值与拥挤度的拟态物理学多目标算法

摘要

根据约束多目标优化问题的特点，在拟态物理学优化（aritificial physics optimization，APO）算法的基础上，将无约束多目标APO（multi-objective APO，MOAPO）算法引入到约束多目标优化领域中。提出约束违反度的判断准则，并采取一种更为有效的约束处理技术，从而构造出一种解决约束多目标优化问题的基于序值与拥挤度的拟态物理学多目标优化（improved constrained rank multi-objective aritificial physics optimization，ICRMOAPO）算法。在随机搜索过程中动态调整引力因子与惯性权重，增强了非劣解集的多样性。实验结果说明了该算法的有效性，通过与序值约束多目标 APO（constrained rank multi-objective APO，CRMOAPO）算法、非支配排序遗传（non-dominated sorting genetic algorithm，NSGA）算法、多目标遗传（multi-objective genetic algorithm，MOGA）算法的对比实验，表明了该算法具有较好的分布性能，为约束多目标优化问题的求解提供了一种新的思路与方法。%According to the characteristics of the constrained multi-objective optimization problem,the un-constrained multi-obj ective aritificial physics optimization algorithm (MOAPO )is introduced into the field of constrained multi-obj ective optimization on the basis of aritificial physics optimization algorithm (APO ).The judgment criterion of constraint violation degree is put forward,and a more effective constraint processing tech-nology is taken.Then a swarm intelligence improved constrained rank multi-obj ective aritificial physics optimi-zation (ICRMOAPO)algorithm based on sequence value and crowding degree applied to solve the problem of constrained multi-objective optimization is constructed.In random searching,the factor of gravity and inertia weight are adjusted dynamically,to enhance the diversity of the non-inferior solution set.The comparative ex-periments between constrained rank multi-objective aritificial physics optimization(CRMOAPO),non-dominated sorting genetic algorithm(NSGA),and multi-obj ective genetic algorithm(MOGA)show that the effectiveness of the proposed algorithm has better performance of distribution and convergence,thereby providing a new train of thought and a method for solving the constrained multi-objective optimization problem.