Handling multi-objective optimization problems with a comprehensive indicator and layered particle swarm optimizer

The multi-objective particle swarm optimization algorithm has several drawbacks, such as premature convergence, inadequate convergence, and inadequate diversity. This is particularly true for complex, high-dimensional, multi-objective problems, where it is easy to fall into a local optimum. To address these issues, this paper proposes a novel algorithm called IMOPSOCE. The innovations for the proposed algorithm mainly contain three crucial factors: 1) an external archive maintenance strategy based on the inflection point distance and distribution coefficient is designed, and the comprehensive indicator (CM) is used to remove the non-dominated solutions with poor comprehensive performance to improve the convergence of the algorithm and diversity of the swarm; 2) using the random inertia weight strategy to efficiently control the movement of particles, balance the exploration and exploitation capabilities of the swarm, and avoid excessive local and global searches; and 3) offering different flight modes for particles at different levels after each update to further enhance the optimization capacity. Finally, the algorithm is tested on 22 typical test functions and compared with 10 other algorithms, demonstrating its competitiveness and outperformance on the majority of test functions.