A photovoltaic panel defect detection method based on an RTMDet framework

By using a photovoltaic panel defect detection method based on the RTMDet framework, and leveraging dynamic activation gating mechanism and axial long-range context modeling, combined with topology-aware feature enhancement, the problems of insufficient accuracy and poor anti-interference ability in photovoltaic panel defect detection are solved, achieving high-precision and robust defect detection.

CN122115365APending Publication Date: 2026-05-29SICHUAN ENERGY INVESTMENT XINGWEN ELECTRIC POWER CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SICHUAN ENERGY INVESTMENT XINGWEN ELECTRIC POWER CO LTD
Filing Date
2026-02-06
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing photovoltaic panel defect detection technologies lack sufficient accuracy in complex outdoor environments, especially in identifying minute and extended defects. Furthermore, they fail to effectively utilize the prior knowledge of the photovoltaic panel's physical structure, resulting in poor anti-interference capabilities and limited generalization performance.

Method used

A photovoltaic panel defect detection method based on the RTMDet framework is adopted. Feature selection and enhancement are performed through dynamic activation gating mechanism. Feature enhancement is combined with axial long-range context modeling and topology awareness. Hierarchical feature fusion is then performed, and finally joint classification and regression processing are carried out.

Benefits of technology

It significantly improves the detection accuracy, robustness, and engineering practical value of multi-scale photovoltaic panel defects in complex backgrounds, while maintaining the lightweight characteristics of the model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122115365A_ABST
    Figure CN122115365A_ABST
Patent Text Reader

Abstract

The application discloses a photovoltaic panel defect detection method based on an RTMDet framework. Through adaptive feature screening and strengthening of the input image by using a dynamic activation gating mechanism, a preliminary feature map is obtained, and then the feature map is subjected to axial long-range context modeling, direction-sensitive long-distance dependence features are extracted by using strip convolution, and channel weighting aggregation is performed based on learnable routing parameters constructed based on topological priors of a photovoltaic panel grid line structure, so as to generate a strengthened feature map. Then, through feature pyramid decomposition, multi-level features are obtained, attention-guided local enhancement is performed on shallow detail features, and the enhanced shallow features and deep semantic features are cross-level fused by using the topological perception routing parameters, so as to generate a multi-scale fusion feature map. Based on the feature map, a detection head is used for target classification and positioning, and a defect detection result is output. Thus, the detection precision, robustness and engineering practical value of the multi-scale photovoltaic panel defect under a complex background are effectively improved.
Need to check novelty before this filing date? Find Prior Art