Power data sampling model based on multi-layer sensing and prediction

Abstract

Abstract With the development of smart grid and energy Internet, more and more smart sensing devices are installed and used in the power system, thus forming the advanced metering infrastructure (AMI). It makes the power system generate massive data all the time, which may come from smart meters, digital protection devices and so on. Behavior evaluation is to filter out the selected tags that meet the type of user portrait from the user system through the known partial tags, and these similar tags together determine the user’s behavior. How to make good use of the collected power big data is an important research topic in the field of power system. Using data mining technology to analyze power big data is a common research method to deal with power big data problems. To facilitate the staff to quickly understand the characteristics of the user, save service costs, improve user service satisfaction, and find the weak links of the power grid, this paper introduces the random forest algorithm, further researches on the basis of user portraits, combines database technology, machine learning algorithm and Logistics algorithm appropriately, constructs the user data sampling model based on big data, fully excavates the big data information of users and predicts the behavior of power users. According to the results of the forecast, it helps the power enterprises to correct and improve the relevant measures in the future. Experiments show that compared with the traditional Logistics model, the K-S value is increased by about 65.38%. The Gini coefficient is increased by about 12.60%, and the efficiency of the algorithm is improved by 5.26%. It has great advantages in stability and generalization. It has a better effect on power data sampling and forecasting, and helps power enterprises to improve market evaluation and corporate reputation.