Detection Method of Water Wall Defect of Power Plant Boiler by Computer Vision
Boiler inspection of the power plant can effectively avoid safety accidents.In the process of on-site inspection,the in-spection area of the boiler water-wall is large and some areas is difficult.A water-wall defect detection system based on the YO-LOv3model is developed.The UAVcarries a visual collection device to collect images of the boiler water-wall of the Henan Yu ne-ng Company.The picture is compressed and wirelessly transmitted to the detection terminal device in real time.The YOLOV3 al-gorithm is used to analyze the data of the water-wall.Adjust the important parameters of the model and make sample enlarge-ment and balance improvement to improve the detection effect.A total of 106 failure parts such as wear,cracks,and oxidation were measured.Compared with manual detection,the success rate of failure part detection reached 77.9%.This method solves the problem of inspection difficulty in large inspection area and part of inspection area,and effectively reduces the actual cost of wa-ter wall inspection for large power plant boiler.
Power Station BoilerWater-WallUAVYOLOv3Defect Detection
万方数据
维普
