Research On Many-objective Knee Point Pigeon-inspired Optimization Algorithm

Pigeon-inspired optimization algorithm as a swarm intelligence algorithm proposed in recent years.Because of its good performance in multi-objective optimization problems,it has been widely recognized by academia and industry.In recent years,practical engineering problems have become more and more complex.When solving many-objective optimization problems that are not favored by decision-makers,with the increase of the number of objectives,the algorithm has problems of insufficient performance,intensified conflict between diversity and convergence,and lack of selection pressure.Aiming at the above problems,the following research work has been done to address the above issues:1.When the existing many-objective pigeon-inspired algorithm solves the many-objective optimization problems that are not preferred by decision-makers,although other strategies are added to the traditional Pareto mechanism to improve the solution selection ability of the algorithm,with the increase of the number of objectives.The solutions in the population are almost non-dominated to each other,so improving the algorithm selection pressure and convergence ability has become an urgent problem to be solved.Aiming at the above problems,considering that the knee point is a bias for a larger hypervolume,its performance is inherently better than other individuals in the population.This paper proposes a knee point-driven many-objective pigeon-inspired algorithm.By introducing the environment selection strategy based on knee-oriented dominance,the selection pressure of the algorithm is effectively improved;by combining the velocity and position update strategies with different distributions,the population evolution direction is dynamically adjusted in different stages of iteration,thereby improving the overall search performance of the algorithm.Experimental results show that the proposed algorithm has good performance in individual selection and approximation to the true Pareto Front.2.In view of the deficiency of the existing many-objective pigeon-inspired algorithm in balancing convergence and diversity when the number of objectives increases sharply,and to further improve the solution performance of the algorithm when dealing with many-objective optimization problems.This paper proposes a KIGD indicator based many-objective pigeon-inspired algorithm.The algorithm realizes the identification of the solution of the knee point and its surrounding area through the environmental selection strategy based on KIGD indicator,thereby improving the diversity of the solution set covering the knee point region.A population renewal strategy based on competition is introduced to improve the selection pressure of the algorithm and ensure its convergence performance.The simulation results also prove the superior performance of the proposed algorithm based on KIGD indicator.3.In order to further solve the problem of algorithm convergence and diversity conflict caused by the increase of the number of objectives in the many-objective pigeon-inspired algorithm,this paper proposes a many-objective pigeon-inspired based on multiple selection strategies.From the perspective of concurrent integration,an elite retention strategy based on multiple selection strategies is designed.By integrating a variety of excellent selection strategies,different operators can be selected according to the characteristics of different problems to improve the solution efficiency,and at the same time,convergence and diversity of population can be guaranteed to a certain extent.In addition,an external archive set is also used to store the elite non-dominated individuals.The experimental analysis shows that the many-objective pigeon-inspired optimization algorithm based on multiple selection strategy proposed in this paper has relatively good comprehensive performance.