Intelligent power grid electrical load forecasting system

Abstract

The invention provides an intelligent power grid electrical load forecasting system and load forecasting method thereof.. The system comprises edge computing nodes, a cloud server, a distributed computing platform, and an adaptive optimization module. The edge computing nodes acquire electrical parameters and associated data, and upload the data to the cloud server after preprocessing. The distributed computing platform is responsible for data storage and operation of complex models, while the adaptive optimization module verifies forecasting results, adjusts parameters, and issues alarms. The method adopts a fusion strategy of traditional algorithms and deep learning to realize multi-time-horizon forecasting, wherein an attention mechanism is introduced for short-term forecasting, an extreme gradient boosting algorithm is combined with time series decomposition for medium-term forecasting, and policy data and economic data are incorporated into long-term forecasting. The invention improves forecasting accuracy and system response speed, provides reliable support for power grid dispatching, and ensures stable power supply.