Efficient solution method for power facility relocation planning based on SVM-PSO optimization
Shichang Zhao, Zhonghao Zhang, Shuai Wang
Abstract
• Proposed an SVM-PSO method to optimize power facility relocation planning efficiency. • Integrated GIS, weather, and date factors for database creation and feature selection. • Achieved a minimum RMSE of 0.1356 and solution time of 2.52 seconds in experiments. • Introduced multi-stage optimization with parallel computing for reduced computational time. • Suggested adaptive strategies to enhance SVM-PSO performance in complex scenarios. In the context of large - scale power facilities relocation planning, existing methodologies exhibit high computational complexity. Consequently, the solution time is extended, rendering it challenging to meet real - time requirements. To address this problem, an efficient solution method for power facility relocation planning based on SVM - PSO (Support Vector Machine - Particle Swarm Optimization) optimization has been developed. The load data is normalized by incorporating weather and date characteristics. Furthermore, neighborhood component analysis is introduced as a non - parametric embedded method for feature selection and nearest - neighbor classification optimization. GIS (Geographic Information System) technology is employed to establish the database for power facilities relocation planning, thus optimizing the layout of power facilities relocation planning. By integrating the SVM model with the particle swarm optimization algorithm, the PSO algorithm is used to determine two key parameters of the SVM model. An efficient solution approach of pre - optimization - coarse optimization - fine optimization for multi - stage power facilities relocation planning is proposed. Experimental results demonstrate that the load ratio of power facilities designed by this method lies within the optimal range. The minimum root - mean - square error (RMSE) is 0.1356, and the shortest solution time is 2.52 seconds. The average relative error is 1.149%, indicating a significant improvement in efficiency.