燕山大学学报2024，Vol.48Issue(1)：54-61,8.DOI:10.3969/j.issn.1007-791X.2024.01.007
基于PLESN和LESQRN概率预测模型的短期电力负荷预测
Short-term power load forecasting based on probabilistic forecasting model of PLESN and LESQRN
摘要
Abstract
In view of the fact that the current load forecasting can not reflect the periodicity and trend of load data and the volatility of residual,a short-term power load forecasting method based on the probability forecasting model of leaky integrator echo state network based on periodicity(PLESN)and leaky integrator echo state quantile regression network(LESQRN)is proposed.Firstly,in order to capture the multiple characteristics of the load,periodic and trend loss functions are defined to assist the optimization of the point prediction the PLESN model.Then,in order to overcome the fluctuation of residual error,the probabilistic prediction model is used to predict residual error for modeling prediction.Finally,the point prediction value and the residual prediction interval are integrated to obtain the result of the probability prediction model.The results of actual calculation examples show that,compared with other models,the proposed model not only effectively suppresses the tip oscillation phenomenon,but also can generates reasonable probability density distribution.关键词
短期电力负荷预测/周期性建模/泄露积分型回声状态网络/分位数回归Key words
short-term power load forecasting/periodic modeling/leaky integrator echo state network/quantile regression分类
信息技术与安全科学引用本文复制引用
樊江川,于昊正,王冬生,安佳坤,杨丽君..基于PLESN和LESQRN概率预测模型的短期电力负荷预测[J].燕山大学学报,2024,48(1):54-61,8.基金项目
河北省自然科学基金资助项目(E2019203514) （E2019203514）
