Short-term Load Forecasting with Deep Learning Techniques

Abstract Short-term load forecasting is essential to the modern power system, which basis guarantee equilibrium of supply and demand. It not only ensures secure stable operation grid, but also provides an important for dispatch, can effectively reduce economic loss caused by energy surplus or shortage in a short term. In previous research work, traditional linear regression models, typical machine learning models other have been explored, there are still problems low accuracy single influence factors. this paper, short-term model based on Temporal Convolutional Network (TCN) proposed. Historical data, weather conditions, date information their combinations separately used forecast different prediction models. Compared with evaluation metric TCN shows that its higher, it more suitable forecasting.