Regional Electricity Sales Forecasting Research Based on Big Data Application Service Platform

Regional monthly electricity sales forecast is an important basis for regional power grid planning and construction, evaluation of regional economic development and operation, and protection of residents' lives. It is also an important work of regional power regulation and management, decision-making of power generation and purchase, improvement of power supply equipment utilization rate and deepening of power system reform. Based on the current situation of power supply enterprise information development, distribution network business status and characteristics, this paper analyzes the factors affecting electricity sales. According to the characteristics of annual changes in electricity sales and data quality factors, the recurrent neural network model is selected based on the big data application service platform. The long short term memory neural network model performs multi-step multivariate prediction on time series, and uses the attention mechanism to combine two independent models for prediction. Experiments conducted on the historical electricity sales data set of a power supply company show that compared with traditional machine learning methods, this method has advantages in accuracy and efficiency.