Anomaly Detection for Solar Panel Joint Clamps Using Deep Learning Based Detection and Recognition Techniques

Expectations for solar power generation, one of the renewable energy sources, are increasing in order to reduce greenhouse gas emissions. However, solar panels are industrial products, and there is always a risk of local failure of the products. Several automatic anomaly detection systems have been developed for solar panels and their roofs. However, there is no precedent for automatic anomaly detection of joint clamps supporting roofs and panels, and visual inspection by technicians is still performed as a regular inspection. Thus, accurate and quick evaluation of the joint clamps of solar panels will contribute to reducing the labor and time required for inspections. In this study, we attempted to develop an automatic anomaly detection system for joint clamps, using object detection and recognition technologies based on deep learning. In the first stage, the joint clamp is detected by object detection technology. In the second stage, the system judges the abnormality from the detected the joint clamp by recognition technology. The effective of the proposed system was evaluated through the field experiments. © 2022 The Society of Materials Science, Japan.