Power distribution network electricity peak load prediction method and system based on multiple models

By constructing a serially connected prediction model using a multi-model approach and correcting the model parameters using a loss function, the problem of poor generalization performance in traditional distribution network load forecasting is solved, achieving higher load forecasting accuracy and stability.

CN115660153BActive Publication Date: 2026-07-24ANHUI JIYUAN SOFTWARE CO LTD +1
View PDF 1 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
ANHUI JIYUAN SOFTWARE CO LTD
Filing Date
2022-10-11
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Traditional load forecasting methods for distribution networks based on machine learning or deep learning suffer from poor generalization performance due to large assumption spaces and the potential for inadequate accuracy and stability of load forecasting when using a single model.

Method used

A multi-model approach is adopted. By obtaining the prediction error distribution of preset decision modules, the decision module with the largest prediction error is determined as the first decision module, and the remaining decision modules are sequentially ordered to construct a serially connected prediction model. The model parameters are corrected using a loss function, and the load prediction is performed by combining the output results of multiple decision modules.

Benefits of technology

It improves the accuracy and stability of peak load forecasting for power distribution networks, enhances the generalization performance of the model, and is applicable to load forecasting for various application scenarios and raw data characteristics.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115660153B_ABST
    Figure CN115660153B_ABST
Patent Text Reader

Abstract

The application discloses a power distribution network electricity peak load prediction method and system based on multiple models, which comprises the following steps: obtaining the number of preset decision modules; determining the decision module with the maximum prediction error as the first decision module based on the prediction error distribution of all preset decision modules, and sequentially sorting the remaining decision modules; connecting the output of the previous decision module and the input end of the next decision module of the two adjacent decision modules to build a first prediction model for predicting the electricity peak load; training the first prediction model based on the original training sample set; inputting the to-be-predicted data into the trained first prediction model to obtain the electricity peak load prediction result, and determining the prediction result of the first prediction model based on the sum of the prediction results output by all decision modules. The application is based on the series connection of multiple decision modules, shortens the model parameter fitting efficiency of the first prediction model, and improves the prediction accuracy of the first prediction model.
Need to check novelty before this filing date? Find Prior Art