Power load probability prediction method based on user power consumption behavior clustering

Abstract

The invention relates to a power load probability prediction method based on user power consumption behavior clustering. The method comprises the following steps: acquiring user historical power consumption fine-grained data and date type data; extracting quantile auto-covariance of historical power consumption fine-grained data of each user to represent power consumption behavior characteristics of the user; clustering the power consumption behavior characteristics of the users by using a hierarchical clustering algorithm, and marking a category label for each user; aggregating the electrical loads of the various users to obtain aggregated loads of the various users; constructing a data set based on the aggregation loads and date type data of the users; using the data set to establish a quantile regression forest prediction model for each type of the users, and performing prediction to obtain load prediction results of each type of the users; and summing the load prediction results of the users to obtain a user aggregate load probability prediction result. Compared with the prior art, the power load probability prediction can be accurately and quickly carried out, and the power load prediction precision of the user aggregate is effectively improved.

Description

technical field [0001] The present invention relates to the technical field of power load forecasting, in particular to a power load probabilistic forecasting method based on user power consumption behavior clustering. Background technique [0002] Power load forecasting is one of the important daily tasks of power companies. Through power load forecasting, it can provide a data basis for power system dispatching, planning, maintenance, and arranging power generation plans. Accurate short-term load forecasting is conducive to arranging daily power generation plans and determining power system economics. Safe operation mode and improvement of clean energy generation capacity, thereby reducing energy consumption. However, with the rapid development of the energy Internet, the connection of various clean energy power generation equipment to the power grid and the in-depth reform of the power trading market have made the changing law of power load more complicated. [0003] Mos...

AI Technical Summary

Problems solved by technology

In addition, the construction of smart grids in recent years has led to the rapid popularization of smart meters, so that the fine-grained data of electricity consumption of massive users can be collected. It has great intrinsic value and provides a new perspective for power load forecasting. At present, there are methods to establish a forecasting model for a single user, but there are problems such as large amount of calculation and low forecasting accuracy, and it is difficult to reflect the dynamics of the user's power consumption data, resulting in Unable to perform reliable and accurate probabilistic forecasting of electrical loads

Images