一种监控场景下光伏板异常检测方法
By constructing a feature extraction network in a monitoring scenario and combining a random crossover mechanism and multiple loss function optimization, the problem of lack of labeled data in photovoltaic panel anomaly detection is solved, and efficient and accurate photovoltaic panel anomaly detection is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- INNOVATION & INNOVATION CENT OF STATE GRID ZHEJIANG ELECTRIC POWER CO LTD
- Filing Date
- 2024-05-15
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies suffer from poor training performance of photovoltaic panel anomaly detection models due to a lack of accurate and reliable large amounts of labeled data, which affects detection accuracy.
An anomaly detection method for photovoltaic panels in a monitoring scenario is adopted. By acquiring abnormal images of photovoltaic panels and monitoring images as test and training samples, a feature extraction network is constructed using deep learning. The network is then trained and optimized using a random crossover mechanism and various loss functions to achieve anomaly detection without the need for labeled samples.
It improves the accuracy and efficiency of photovoltaic panel anomaly detection, reduces computational complexity, enhances the model's generalization ability and detection reliability, and achieves real-time anomaly detection and environmental adaptability.
Smart Images
Figure CN118552892B_ABST