基于CPSO优化的改进多核支持向量机的短期电力负荷预测

Short-term power load forecasting based on improved Muti-Kernel Support Vector Machine optimized by CPSO

摘要

短期电力负荷预测是保障电力系统调度优化与安全运行的关键环节，面对负荷数据中普遍存在的非线性、周期性及扰动性特征，传统单核支持向量机模型在表达能力与预测精度方面存在一定局限。为此，本文提出一种融合混沌粒子群优化与自适应核参数调整机制的多核支持向量机模型用于提升短期负荷预测的精度与鲁棒性。该方法首先构建线性核、周期核与RBF核的加权组合核函数，通过CPSO对核权重进行全局优化，以提高多尺度特征建模能力；同时引入自适应机制，根据验证误差反馈动态调整核参数，增强模型对非平稳输入的响应能力。预测结果表明，所提CPSO-AMK-SVM模型在预测精度、稳定性及泛化能力方面均优于传统模型，展现出良好的实际应用前景。

Abstract

Short-term power load forecasting is a critical task for ensuring optimal scheduling and secure operation of power systems. Given the prevalent nonlinear, periodic, and fluctuating characteristics in load data, traditional single-kernel Support Vector Machine models exhibit limitations in both representational capacity and prediction accuracy. To address this issue, this paper proposes an Adaptive Multi-Kernel Support Vector Machine model that integrates Chaotic Particle Swarm Optimization and an adaptive kernel parameter adjustment mechanism, aiming to improve forecasting accuracy and robustness. Specifically, a weighted composite kernel is constructed by combining a linear kernel, a periodic kernel, and a radial basis function (RBF) kernel, wherein the kernel weights are globally optimized using CPSO to enhance multi-scale feature modeling capabilities. Meanwhile, an adaptive mechanism is introduced to dynamically update kernel parameters based on validation error feedback, enabling the model to better respond to nonstationary inputs. The prediction results show that the proposed CPSO-AMK-SVM model significantly outperforms traditional models in terms of prediction accuracy, stability, and generalization ability, indicating strong potential for practical applications.

关键词

多核支持向量机 / 混沌粒子群 / 自适应机制 / 短期负荷预测

Key words

Multi-Kernel Support Vector Machine / CPSO / adaptive adjustment mechanism / short-term load forecasting

参考文献

[1] 李晨， 尹常永， 李奇洁. 电力系统负荷预测研究综述[J]. 电子世界， 2021(16): 81-82.

[2] WANG Z X, KU Y Y, LIU J. The power load forecasting model of combined SaDE—ELM and FA—CAWOA—SVM based on CSSA[J]. IEEE Access, 2024, 12: 41870-41882.

[3] AL—MUSAYLH M S, DEO R C, ADAMOWSKI J F, et al. Short—term electricity demand forecasting with MARS, SVR and ARIMA models using aggregated demand data in Queensland, Australia[J]. Advanced Engineering Informatics, 2018, 35: 1-16.

[4] DONG Q, HUANG R B, CUI C H, et al. Short—Term Electricity—Load Forecasting by deep learning: A comprehensive survey[J]. Engineering Applications of Artificial Intelligence, 2025, 154: 110980.

[5] LI C, SHI J R. A novel CNN—LSTM—based forecasting model for household electricity load by merging mode decomposition, self—attention and autoencoder[J]. Energy, 2025, 330: 136883.

[6] SAXENA N, KUMAR R, RAO Y K S S, et al. Hybrid KNN—SVM machine learning approach for solar power forecasting[J]. Environmental Challenges, 2024, 14: 100838.

[7] ZULFIQAR M, KAMRAN M, RASHEED M B, et al. Hyperparameter optimization of support vector machine using adaptive differential evolution for electricity load forecasting[J]. Energy Reports, 2022, 8: 13333-13352.

[8] YIN C, MAO S H. Fractional multivariate grey Bernoulli model combined with improved grey wolf algorithm: Application in short—term power load forecasting[J]. Energy, 2023, 269: 126844.

[9] WANG L, WANG X Y, ZHAO Z C. Mid—term electricity demand forecasting using improved multi—mode reconstruction and particle swarm—enhanced support vector regression[J]. Energy, 2024, 304: 132021.

[10] BARMAN M, DEV CHOUDHURY N B. A similarity based hybrid GWO—SVM method of power system load forecasting for regional special event days in anomalous load situations in Assam, India[J]. Sustainable Cities and Society, 2020, 61: 102311.

[11] WANG Z X, ZHAO G J, SUI J X, et al. IPFOA—MKSVM and BA—MLP models for predicting closed bus bar temperatures in high voltage nuclear power plants in different vacuum environments[J]. Vacuum, 2025, 232: 113825.

基金资助

山东省重点研发计划(YDZX2024138)

山东省重点研发计划(2024CXGC010809)

山东省重点研发计划(2024TSGG0509)