Multi-level data correlation fusion-based short-term load forecasting method for optical storage and direct flexible system

By constructing a multi-source heterogeneous data acquisition and preprocessing system and a multimodal attention fusion prediction model, the problems of insufficient data acquisition and inaccurate prediction in the optical-storage-direct-flexible system were solved, achieving efficient and stable load prediction and decision support, and improving the system's operational stability and accuracy.

CN120127625BActive Publication Date: 2026-07-10CHINA CONSTR SECOND ENG BUREAU LTD
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Patents(China)
Current Assignee / Owner
CHINA CONSTR SECOND ENG BUREAU LTD
Filing Date
2025-02-17
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Existing technologies in photovoltaic-storage-direct-drive-flexible systems suffer from insufficient data acquisition, inadequate data processing, inaccurate load forecasting, and insufficient decision support, leading to unreasonable energy allocation and unstable system operation.

Method used

Construct a multi-source heterogeneous data acquisition and preprocessing system, implement dynamic weighted fusion and correlation analysis, develop a multimodal attention fusion prediction model, and deploy an intelligent load prediction and decision support platform, using edge computing and Transformer architecture for data processing and prediction.

Benefits of technology

It enables real-time and accurate acquisition and processing of various uncertain load data from photovoltaic-storage-DC-flexible systems, improving the accuracy and adaptability of short-term electricity load forecasting, and generating comprehensive reports to support the optimized operation of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120127625B_ABST
    Figure CN120127625B_ABST
Patent Text Reader

Abstract

The multi-level data correlation fusion photovoltaic and energy storage flexible system short-term load forecasting method constructs a multi-source heterogeneous data acquisition and preprocessing system to collect and process uncertain load data such as photovoltaic output power and building electricity load. Dynamic weighted fusion and correlation analysis are used to calculate the dynamic correlation weight between multi-source data, and multi-scale feature extraction is performed to construct a fusion data set. A multi-modal attention fusion forecasting model is developed, and a spatio-temporal joint attention mechanism is designed to realize explicit modeling of long-range correlation and nonlinear interaction features in high-dimensional data. An intelligent load forecasting and decision support platform is deployed, and the forecasting model is integrated into the energy management system. Combined with dynamic pricing and energy storage control strategies, comprehensive reports are generated, and real-time interaction and policy issuance are realized through the OPC UA protocol and the SCADA system. The present application effectively improves the accuracy of load forecasting and the stability of system operation, and realizes the optimization management and efficient utilization of energy.
Need to check novelty before this filing date? Find Prior Art