Research And Application Of Electricity Consumption Prediction Based On Big Data Mining

As one of the important energy sources for national economic operation,social production and people’s life,electric power is of self-evident importance.Under the situation that the proportion of new energy generation is still not high and the research and industrialization of energy storage equipment have encountered bottlenecks,accurate prediction of electricity consumption and avoidance of resource waste have been studied by many scholars for many years.In addition,accurate prediction of electricity consumption plays an important role in ensuring the economy,security and stability of power system.On the other hand,with the continuous development of smart grid construction,the informatization degree of power grid data is gradually enhanced,and the massive power data brings new challenges to the power system,as well as new opportunities to the power grid operation and management.With the advent of the era of big data,power data,an important factor of production,should keep pace with the development of big data and serve the power production.This paper starts from the actual demand of a power supply company of State Grid in Chongqing,and makes a qualitative analysis of the electricity consumption structure and influencing factors in this area by combining with its actual electricity marketing data.Through the research of machine learning algorithms,and puts forward the electricity consumption forecasting method based on PCA-GBDT combination model,using SparkML machine learning tools to complete the industrial categories and residents in the region category of electricity consumption forecast,make up the GBDT algorithm under the condition of high dimension data limitations,improve the training efficiency and prediction accuracy;Then,the algorithm parameters of PCA-GBDT were further optimized,and the experimental results show that the improved algorithm model has better prediction results.Finally,a distributed cluster with Spark as the core is built,and a power consumption prediction and data analysis system based on B/S architecture is designed to help the Marketing Department of State Grid to predict and analyze power consumption by using big data tools,so as to optimize marketing strategies.At the same time,assist the development department to arrange the electricity consumption plan,adjust the operation mode of power system appropriately to reduce the cost.The system has been put into use in a power supply company of State Grid Chongqing in 2021.The comparison between the prediction results and the actual situation proves that the system has high prediction accuracy,good use experience,and significantly improves the work efficiency of enterprises.Through the research of this paper,the big data mining method is reasonably applied in practical work and production,so that the electric power enterprises can quickly carry out comprehensive data display and analysis in the face of massive business data,and complete accurate electricity consumption forecast.It is of great significance to formulate marketing plan,rationally arrange the operation mode of power system,correctly plan capital allocation and investment,and do a good job in the layout of power development planning.