Abstract
Obtaining high convergence and uniform distributions remains a major challenge in most metaheuristic multi-objective optimization problems. In this article, a novel multi-objective particle swarm optimization (PSO) algorithm is proposed based on Gaussian mutation and an improved learning strategy. The approach adopts a Gaussian mutation strategy to improve the uniformity of external archives and current populations. To improve the global optimal solution, different learning strategies are proposed for non-dominated and dominated solutions. An indicator is presented to measure the distribution width of the non-dominated solution set, which is produced by various algorithms. Experiments were performed using eight benchmark test functions. The results illustrate that the multi-objective improved PSO algorithm (MOIPSO) yields better convergence and distributions than the other two algorithms, and the distance width indicator is reasonable and effective.
Highlights
Multi-objective optimization problems (MOPs) are very common in engineering and other areas of research, such as economics, finance, production scheduling, and aerospace engineering
We introduce a new multi-objective particle swarm optimization (PSO) algorithm based on Gaussian mutation and an improved learning strategy to solve MOPs
Unlike other MOPSOs, that often randomly select a solution from the external archive as the global optimal solution gbest, we present different learning strategies to update the individual positions of the non-dominated and dominated solutions; (3) To further measure the distribution width, the indicator Distribution width (DW) is proposed
Summary
Multi-objective optimization problems (MOPs) are very common in engineering and other areas of research, such as economics, finance, production scheduling, and aerospace engineering. Reddy and Kumar [9] proposed an elitist-mutation multi-objective PSO (EM-MOPSO) algorithm with a strategic mechanism that effectively explores the feasible search space and speeds up the search for the true Pareto optimal region. Cheng et al [17] presented a hybrid multi-objective particle swarm optimization that combines the canonical PSO search with a teaching–learning-based optimization (TLBO) algorithm to promote diversity and improve the search ability. We introduce a new multi-objective PSO algorithm based on Gaussian mutation and an improved learning strategy to solve MOPs. The main new contributions of this work can be summarized as: (1) Gaussian mutation throw points strategy to improve the uniformity of external archives and current populations; (2) For MOPs, it is difficult to select the gbest value of velocity and update the formula.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
