Load Forecasting Based on Multiple Load Features and TCN-GRU Neural Network

Traditional load forecasting does not deeply consider the impact of load sequence on model forecasting accuracy. To improve the prediction accuracy, a multi-load feature combination (MLFC) is proposed, and a load prediction framework is constructed by combining Temporal Convolutional Network (TCN) a...

Full description

Saved in:
Bibliographic Details
Main Authors: Haofeng ZHENG, Guohua YANG, Wenjun KANG, Zhiyuan LIU, Shitao LIU, Hong WU, Honghao ZHANG
Format: Article
Language:zho
Published: State Grid Energy Research Institute 2022-11-01
Series:Zhongguo dianli
Subjects:
Online Access:https://www.electricpower.com.cn/CN/10.11930/j.issn.1004-9649.202012107
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Traditional load forecasting does not deeply consider the impact of load sequence on model forecasting accuracy. To improve the prediction accuracy, a multi-load feature combination (MLFC) is proposed, and a load prediction framework is constructed by combining Temporal Convolutional Network (TCN) and Gated Recurrent Unit (GRU). Firstly, the load change rate feature and the load component feature based on the ensemble empirical mode decomposition are introduced, and the feature combination MLFC is formed with the load and date features; Second, select TCN and GRU for feature extraction and prediction, and build an MLFC-TCN-GRU prediction framework based on MLFC; Finally, using different models to verify the proposed method, the results show that MLFC helps to improve the prediction accuracy and is suitable for different models; at the same time, MLFC-TCN-GRU has the highest prediction accuracy compared with other models.
ISSN:1004-9649