Research on a wiring detection method of physical terminal blocks in converter station based on artificial intelligence

In order to solve the time-consuming and laborious problems of manual on-site verification of terminal block wiring, this paper proposes a method of artificial intelligence-based terminal block physical wiring detection in transformer stations. After shooting videos on the terminal blocks, the videos are extracted key frames based on the difference and sharpness. According to the objects to be identified in the extracted pictures, different recognition partitions are established. In this paper, the improved YoloV5 algorithm is used to cut the effective areas of the terminal blocks, and then DBNet and CRNN are used to detect and recognize the texts in the recognition area respectively. Finally, the terminal block wiring detection information model can be established. The experimental results show that: The recognition method put forward by this paper achieves the accuracy of more than 90% in extracting key frames in videos, and the accuracy of effective area recognition can reach 95%. The accuracy of DBNet detection and CRNN recognition on texts is about 85% and 80% respectively. Since every step of the detection method is of high accuracy and feasibility, it is helpful to solve the problems of manual checking terminal blocks’ wiring.