LSTM Power Load Peak Prediction Method Based on Bayesian Network

Authors

  • Liyuan Sun Yunnan Power Grid Co., Ltd, China Southern Power Grid Co., Ltd, Kunming 650200, China
  • Yuan Ai Yunnan Power Grid Co., Ltd, China Southern Power Grid Co., Ltd, Kunming 650200, China
  • Yiming Zhang Yunnan Power Grid Co., Ltd, China Southern Power Grid Co., Ltd, Kunming 650200, China
  • Jianyu Ren Yunnan Power Grid Co., Ltd, China Southern Power Grid Co., Ltd, Kunming 650200, China

Abstract

In recent years, the development of the power sector has progressed rapidly, the peak power load is increasing, and the imbalance between power supply and demand is becoming more and more serious. Therefore, this research was conducted to predict the peak power load in order to provide effective response and solution measures. The current study used big data technology, Long Short-Term Memory (LSTM) network and Bayesian network as its theoretical basis, regularizes the design of loss function, optimizes the network weights, and finally obtains the LSTM electric load peak prediction model based on Bayesian network. Experiments were conducted to compare the performance of the proposed model with that of the traditional LSTM
model. The model proposed in this study has a training time of 44 s for the integrated station data, and the prediction accuracy reaches 96.54% for the load peak percentage of 95.5% data. The performance results of Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Relative Error (MRE), and Mean Absolute Error (MAE) are 115.947, 10.161, 0.027, and 9.656, respectively. The experimental results indicated that the LSTM peak power load prediction model based on the Bayesian network has high prediction accuracy, which confirms the validity and feasibility of the proposed model.

Keywords: Bayesian network; forbidden search algorithm; great likelihood estimation; LSTM; load peaks; prediction

Cite As

L. Sun, Y. Ai, Y. Zhang, J. Ren, "LSTM Power Load Peak Prediction Method Based on Bayesian
Network", Engineering Intelligent Systems, vol. 32 no. 2, pp. 155-164, 2024.



Published

2024-03-01