Cell load prediction method based on improved clustering and long-term and short-term memory deep learning

Abstract

The invention discloses a cell load prediction method based on improved clustering and long-term and short-term memory deep learning. The method is characterized by dividing the residence categories through an improved clustering algorithm and according to the influence factors of different types of residential areas; using an LSTM algorithm to establish a corresponding prediction model for each type of residential area, and carrying out Dropout processing on the LSTM to avoid the local optimization, thereby obtaining the predicted load value, reducing the power utilization difference betweenthe self installation capacity of the residential area and the actual load and realizing the reasonable planning of the transformers in a court. By carrying out the improved clustering analysis according to each attribute value of the newly-built cell to obtain a cell category, and carrying out the load prediction by utilizing the prediction model of the corresponding category, the business expansion installation capacity is estimated, and the court construction is guided.

Description

technical field [0001] The invention belongs to the field of distribution network industry expansion and installation auxiliary construction, and in particular relates to a community load prediction method based on improved clustering and long-short-term memory deep learning. Background technique [0002] With the continuous acceleration of urbanization and the vigorous implementation of supply-side structural reforms, the power consumption in various regions has repeatedly hit new highs, but the distribution of social power resources is uneven, and the difference in power consumption in different communities is large, resulting in the load of most residential radio stations. Frequent heavy overloads are accompanied by unfavorable phenomena such as light loads, no loads or even idle stations in some areas. Therefore, reasonable load planning and capacity expansion in the station area are particularly critical for urban power grid planning, and distribution load forecasting i...

AI Technical Summary

Problems solved by technology

[0004] In order to solve practical problems such as the fact that few influencing factors are considered in the medium- and long-term load forecasting of the new community for the station area, and the artificial estimation error is large, which leads to unreasonable construction planning of the station area in the later stage, the present invention proposes a method based on improved clustering and long-term and short-term load forecasting. Residential load forecasting method based on memory deep learning: According to the impact factors of different types of residential areas, residential categories are divided, corresponding prediction models are established for each type of residential area, and the predicted load value is obtained, so as to realize the reasonable planning of transformers in the station area

Images