A quantum particle swarm multi-objective optimization method

Abstract

The invention discloses a quantum particle swarm multi-objective optimization method. The method comprises the following steps: S1, establishing a quantum double-delta potential well model based on double-potential well simplification; S2, establishing a particle position updating model based on the double potential well model; S3, constructing a shared learning strategy of the particles; And S4,constructing a quantum particle swarm multi-objective optimization algorithm. According to the quantum double-potential well model, a particle position updating formula is established, an inner localattractor and an outer local attractor are introduced, the local optimization precision of the algorithm is improved, and solution distribution is more uniform; According to the shared learning mechanism provided by the invention, the searching range of particles can be expanded, the diversity of solutions is increased, and the tendency that a quantum particle swarm algorithm is easy to converge to a boundary solution is avoided; When the method is used for processing the high-dimensional target optimization problem, the good convergence performance and distribution performance can still be kept, and the multi-target optimization problem in engineering application can be solved more practically.

Description

Technical field [0001] The invention belongs to the technical field of multi-objective optimization of industrial design, and in particular relates to a multi-objective optimization method of quantum particle swarms. Background technique [0002] Unlike single-objective optimization that converges to a single solution, multi-objective optimization obtains an optimal solution set. Therefore, the Pareto optimal solution found by each generation of particles needs to be stored in an external file, and this external file follows the movement of the particles. Continuous updates and maintenance have finally reached the frontier of Pareto. Multi-objective optimization expects that the final Pareto frontier is as close as possible to the real Pareto frontier, that is, it has good convergence. Secondly, the Pareto frontier is required to be evenly distributed and the distribution range is as wide as possible. [0003] The quantum particle swarm algorithm exhibits good global search capabi...

AI Technical Summary

Problems solved by technology

First, a single local attractor P is used in the quantum particle swarm optimization algorithm i (t), causing particles to easily gather near a single solution in the late iteration; secondly, the characteristic length L of the particle aggregation state in the quantum particle swarm optimization algorithm i (t) adopts the average best position c(t), causing each particle to refer to the information of the same c(t), and in multi-objective optimization, each particle should find the Pareto optimal solution, and the particles use the same c (t) Limiting the search space and the diversity of groups, resulting in a decrease in the accuracy of the algorithm when dealing with high-dimensional target optimization problems in engineering applications

Images