«上一篇/Previous Article|本期目录/Table of Contents|下一篇/Next Article»

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:

34

期数:

2024年06期

页码:

201-206

栏目:

新型计算应用系统

出版日期:

2024-06-10

文章信息/Info

Title:

Short-term Electricity Load Forecasting with BiGRU-Attention Based on Bayesian Optimization

文章编号:

1673-629X(2024)06-0201-06

作者:

包广斌; 张 瑞*; 彭 璐; 李 明; 赵怀森

兰州理工大学 计算机与通信学院,甘肃 兰州 730050

Author(s):

BAO Guang-bin; ZHANG Rui*; PENG Lu; LI Ming; ZHAO Huai-sen

School of Computer and Communication,Lanzhou University of Technology,Lanzhou 730050,China

关键词:

电力系统; 负荷预测; 贝叶斯优化算法; 双向门控循坏单元; 注意力机制

Keywords:

power system; load forecasting; Bayesian optimization algorithm; bi-directional gated bad-cycle unit; attention mechanism

分类号:

TP39

DOI:

10.20165/j.cnki.ISSN1673-629X.2024.0079

摘要:

电力系统的可靠供应对于工业、商业和居民的生活至关重要。 为了满足电力需求并维持电力系统的稳定运行,提高短期电力负荷预测的准确性和可靠性尤为关键;针对负荷数据存在复杂的非线性特性,该文提出一种基于贝叶斯优化算法的双向门控循环单元和注意力机制(BO-BiGRU-Attention)的混合预测模型对短期电力负荷进行精准预测。 首先,使用 Min-Max Normalization 方法对负荷数据进行归一化处理。 其次,利用 BiGRU 网络捕获序列中的长期依赖关系和上下文信息,结合注意力机制,通过在输入序列的不同部分给予不同的权重,从而突出关键特征。 最后,针对 BiGRU-Attention 模型的超参数难以选取最优解的问题,引入贝叶斯优化算法对 BiGRU-Attention 模型的超参数进行寻优,完成短期电力负荷的预测。 采用印度北部某地区的电力负荷数据进行预测分析,仿真结果表明,BO-BiGRU-Attention 网络表现优于其他模型,各误差评价指标最小,其中 MAE、RMSE 和 MAPE 分别为 56. 67,73. 49 和 1. 16% ,预测精度达到了 99. 47% 。

Abstract:

The reliable supply of the power system is essential for industry,commerce and the life of residents. In order to meet the power demand and maintain the stable operation of the power system,it is especially critical to improve the accuracy and reliability of short-term power load forecasting. In view of the complex nonlinear characteristics of the load data,we propose a hybrid forecasting model based on Bayesian optimization algorithm with bidirectional gated recurrent unit and attention mechanism (BO-BiGRU-Attention) to accurately forecast the short- term power load. Firstly,the load data are normalized using the Min - Max Normalization method. Secondly, the BiGRU network is used to capture the long - term dependencies and contextual information in the sequences,and combined with the attention mechanism,the key features are highlighted by giving different weights in different parts of the input sequences. Finally,to address the problem that it is difficult to select the optimal solution for the hyperparameters of the BiGRU-Attention model,a Bayesian optimization algorithm is introduced to optimize the hyperparameters of the BiGRU-Attention model to complete the prediction of short-term electricity load. Electricity load data of a region in north India is used for forecasting analysis,and the simulation results show that the BO-BiGRU-Attention network outperforms the other models,with the minimum of each error evaluation index,in which the MAE,RMSE,and MAPE are 56. 67,73. 49,and 1. 16% ,respectively,and the prediction accuracy reaches 99. 47% .

常用功能

更新日期/Last Update: 2024-06-10