The Particle Swarm Optimization Algorithm Improved Based on Multiple Strategies Including Adaptive Nonlinear Weights
Jisheng Huang,Dalang Guo,Junwei Zhang
DOI: https://doi.org/10.1145/3652628.3652658
2024-01-01
Abstract:The benefits of Particle Swarm Optimization (PSO) include its straightforwardness, simplicity, and powerful parallelism, so it is widely used to solve nonlinear, non-convex, high-dimensional problems. However, the typical particle swarm algorithm is inclined to settle for solutions that are best suited to the local context, the phenomenon of "early maturity", slow convergence speed, unable to achieve accurate convergence, and other problems in managing complex engineering problems. To address the deficiencies of the initial algorithm, the Improved particle swarm optimization algorithm (IPSO) is proposed to improve the standard particle swarm optimization algorithm. In IPSO, four strategies of Tent chaotic mapping, adaptive nonlinear weights, the center of gravity reverse learning and Cauchy perturbation are introduced to improve the original algorithm. To verify the model effect, the standard PSO algorithm is tested against the improved IPSO algorithm as well as several swarm intelligent optimization algorithms, which are widely used at present, for 10 sets of basic functions. The IPSO algorithm, which has been tested multiple times, has been demonstrated to possess superior global optimization-seeking capability and a quicker convergence rate in ten distinct fundamental functions, and the convergence curve is smooth, which allows the model to be applied to complex practical problems such as performance prediction.