A low-voltage distributed photovoltaic power prediction method, system and medium

CN122246681APending Publication Date: 2026-06-19STATE GRID HUBEI MARKETING SERVICE CENT (MEASUREMENT CENT)

Patent Information

Authority / Receiving Office: CN · China
Patent Type: Applications(China)
Current Assignee / Owner: STATE GRID HUBEI MARKETING SERVICE CENT (MEASUREMENT CENT)
Filing Date: 2026-02-25
Publication Date: 2026-06-19

AI Technical Summary

Technical Problem

Existing photovoltaic power generation forecasting methods suffer from insufficient forecasting capabilities due to incomplete consideration of environmental adaptability and characteristics, as well as poor data robustness, which fails to meet the stability and reliability requirements of power systems.

Method used

An improved Harris Eagle optimization algorithm is used for variational mode decomposition, combined with LSTM and random forest models for deterministic prediction, and quantile regression and kernel density estimation for probabilistic prediction. The probabilistic prediction mechanism integrating quantile regression and kernel density estimation is used to quantify prediction uncertainty.

Benefits of technology

It improves the accuracy and stability of photovoltaic power generation forecasts, reduces computational costs, generates forecast interval distributions that are more realistic, adapts to the nonlinear and non-stationary characteristics of photovoltaic power generation, and enhances the operating efficiency and reliability of the power system.

✦ Generated by Eureka AI based on patent content.

Smart Images

Figure CN122246681A_ABST

Patent Text Reader

Abstract

This application provides a low-voltage distributed photovoltaic (PV) power prediction method, system, and medium. The method includes the following specific steps: collecting and preprocessing PV prediction data, collecting historical PV power data and related meteorological factors, performing data cleaning and normalization, then processing PV power and key meteorological data to decompose them into multiple stationary intrinsic mode function (IMF) components, proposing an improved Harris Eagle optimization algorithm to achieve adaptive global optimization of variational mode decomposition parameters; screening key input features through correlation analysis; constructing a dual deterministic prediction model to provide diverse high-precision deterministic prediction schemes; integrating quantile regression and kernel density estimation probabilistic prediction mechanisms to achieve comprehensive quantification of prediction uncertainty; and conducting system verification based on actual datasets to verify the superiority of the proposed framework in terms of deterministic prediction accuracy and probabilistic prediction reliability. This application can effectively reduce the charging cost of electric vehicles and stabilize distribution network load fluctuations.

Need to check novelty before this filing date? Find Prior Art