Desert photovoltaic power station carbon emission management system and method based on large model
The desert photovoltaic power station carbon emission management system, trained by a large model, accurately predicts the impact of sandstorms and outputs the optimal cleaning strategy, solving the problem of untimely or excessive cleaning of desert photovoltaic power stations and achieving low-carbon operation and maintenance and improved power generation revenue.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- INST OF WATER RESOURCES FOR PASTERAL AREA MINIST OF WATER RESOURCES P R C
- Filing Date
- 2026-05-26
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies, carbon emission management for desert photovoltaic power plants cannot accurately predict the impact of sandstorms on dust accumulation on photovoltaic panels and power generation, leading to untimely or excessive cleaning, resulting in power generation loss and increased carbon emissions, and failing to meet personalized low-carbon operation and maintenance needs.
A carbon emission management system based on large models is adopted. By collecting and processing historical operating data, three types of models are trained: the intensity of sandstorm weather and dust accumulation parameters, the thickness of dust accumulation and the decline in power generation, and the cleaning strategy and carbon emission balance. Combined with edge computing and cloud storage, the system can accurately predict the timing and scope of cleaning and output the optimal cleaning strategy.
This has reduced carbon emissions during the dust storm cleaning process, decreased power generation losses and energy waste, and achieved a synergistic improvement in low-carbon operation and maintenance and power generation revenue.
Smart Images

Figure CN122414567A_ABST
Abstract
Citation Information
Patent Citations
Regional electric carbon emission factor prediction method and system based on multi-modal neural network
CN120317434B