基于线性判别分析的电力负荷预测  

Power load forecasting based on linear discriminant analysis

作　　者：杨怡菲 高岩[1] YANG Yifei;GAO Yan(Business School,University of Shanghai for Science and Technology,Shanghai 200093,China)

机构地区：[1]上海理工大学管理学院,上海200093

出　　处：《电子设计工程》2023年第11期102-106,共5页Electronic Design Engineering

摘　　要：为了更加有效地减少电力负荷预测模型的输入变量以提高模型运行速度和预测精度,利用线性判别分析法(LDA)对输入变量进行降维处理,并与主成分分析法(PCA)在BP神经网络、支持向量机(SVM)和决策树梯度提升算法(LGBM)三种模型上进行实例分析对比,结果显示,使用LDA降维的三种模型的RMSE分别降低了42 MWh、62 MWh、55 MWh左右,MAPE分别降低了0.6%、0.9%、0.7%左右,预测精确度更高,在其中两种模型上还能显著缩短运行时间,模型的整体性能得到有效提高。In order to reduce the input variables of the power load prediction model more effectively to improve the model operation speed and forecasting accuracy,the Linear Discriminant Analysis(LDA)is used to reduce the dimension of the input variables.The comparison of LDA method and PCA method on BP model,SVM model and LGBM model shows that compared with PCA method,the RMSE of the three models using LDA method is reduced by about 42 MWh,62 MWh,and 55 MWh,and the MAPE is reduced by about 0.6%,0.9%,and 0.7%.The prediction accuracy improved significantly,and can significantly shorten the running time on two of them,so that the overall performance of the model is effectively improved.

关 键 词：电力负荷预测 线性判别分析法 主成分分析法 降维 

分 类 号：TN98[电子电信—信息与通信工程]