Brief-Term Electricity Load Forecasting Based on Improved PSO-Optimized BP Neural Network
- 24 March 2023
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 404-408
Abstract
To address the issue of low accuracy of temporary-term power demand forecasting, a method to optimize BP (Back Propagation, BP) neural network forecasting modeling based on the Improved Particle Swarm Optimization (IPSO) algorithm is proposed. Firstly, the chaotic sequence theory is incorporated into the population initialization of PSO, and the chaotic regime is used to generate a uniformly distributed population to enhance its convergence speed; Secondly, with the introduction of nonlinear inertia weights and adaptive learning factors, the global optimization-seeking ability and iterative convergence speed of the PSO algorithm are substantially improved, and the ability of the particles to jump out of the local optimum is enhanced; Then, the parameters of the BP prediction model are optimized by IPSO (inertial weights, triggers), which can better address the problem of inaccurate manual selection of arguments; Finally, the simulation results prove that the load prediction accuracy can be greatly improved by this paper compared with random forest, BP, and GA-BP method.Keywords
This publication has 3 references indexed in Scilit:
- Oblique and rotation double random forestNeural Networks, 2022
- Short-term power load forecasting based on BiGRU-Attention-SENet modelEnergy Sources, Part A: Recovery, Utilization, and Environmental Effects, 2022
- Particle swarm optimization with crossover: a review and empirical analysisArtificial Intelligence Review, 2015