Short-term load forecasting based on depth measurement information

Abstract

According to the predicted time span, the power load forecasting problem can be roughly divided into long-term and short-term forecasting. Among them, ultra-short-term load forecasting generally outputs load changes in the next few minutes to for on-line monitoring of power equipment, while short-term load forecasting mainly refers to pre-day load forecasting and pre-week load forecasting. It provides reference for hydropower dispatching, unit start and stop, water-fire coordination and so on, and is the basic work needed for the daily operation of power grid. In this paper, long and short memory networks network is used to predict short-term load, and depth measurement information is used as neural network input to construct mapping relationship based on load sequence timing.