Short-term load forecasting method based on heterogeneous big data of power

With the integration of various renewable energy sources, the power system is transforming into a more intelligent, flexible, and interactive system. In the planning and operation of future power grids, load forecasting, especially short-term load forecasting for individual power customers, is becoming increasingly important. To address this issue, a short-term load forecasting method based on heterogeneous big data in the power system is proposed. The scheme collects data from smart meters and weather forecasts, preprocesses it, and loads it into a non-relational database for storage, and further processes the heterogeneous data. At the same time, a long short-term memory recursive neural network model is designed and implemented to determine load distribution and predict the electricity consumption of a residential community in the next 24 hours. Finally, a smart meter dataset from a residential community is used to test the proposed short-term load forecasting framework, and the performance of the prediction model is compared with two classical algorithms using two indicators: root mean square error and mean absolute percentage error, validating the effectiveness of the proposed model.