1. Electricity Power Load Forecast via Long Short-Term Memory Recurrent Neural Networks
- Author
-
Qiang Jiang, Haiyin Qing, Min Li, and Jia-Xiong Zhu
- Subjects
Support vector machine ,Smart grid ,Recurrent neural network ,Computer science ,business.industry ,Supervised learning ,Electricity ,Time series ,business ,Electrical grid ,Reliability engineering ,National Grid - Abstract
This paper focuses on electricity load forecasting of large-scale electrical grid. The smart grid is developing rapidly in the half-decade; it becomes the new goal of electricity industrial construction. Stability, reliable, flexible and self-cure are the main electricity characters, and accuracy forecasting power load play an important role in this process of smart construction. In this paper, we take Estonia country power load as a case, convert the load data to supervised learning data, and then use long short-term memory (LSTM) recurrent network do training model and forecasting. The experiment demonstrates the method is efficient, and compared with support vector regressive (SVR), LSTM could extract the feature of power load better accuracy and obtain the better performance of forecasting. The result can help national grid planning.
- Published
- 2018