Research On Detection Algorithm Of Wheel Tread Defect Of Sintering Machine Trolley

The wheel is the main part of the sintering machine trolley,the state of the wheel is related to the safety of the entire sintering machine,the wheel tread feature is a basis for measuring whether the wheel is running normally,if the wheel fails during operation,it will usually lead to the wheel tread,defects such as scratches and peelings are generated,the existence of these defects will have a certain impact on the working efficiency of the sintering machine trolley,Therefore,it is very necessary to detect the wheel tread defects of the sintering machine trolley.At present,the detection of wheel tread defects of sintering machine trolleys is mostly judged by manual visual observation,and there are often problems such as false detection,missed detection and inconsistent detection standards in manual detection.Aiming at the above problems,this thesis proposes a detection scheme for wheel tread defects of sintering machine trolleys based on the YOLOv5 algorithm.First,the images of wheel tread defects are obtained through the camera equipment,and the YOLOv5 detection algorithm data set is constructed by feature labeling of the images that meet the detection standards.Since the number of real wheel tread defect data sets collected is insufficient to meet the training of the YOLOv5 model,the thesis uses generative adversarial network and Poisson fusion algorithm to enrich and expand the wheel tread defect data.The test results of the YOLOv5 training model show that the training model with the expanded data set is6.5% more accurate in detecting wheel tread defects than the training model without the expanded data set,thus providing a solution program for the detection of wheel tread defects on sintering trolleys.In this thesis,on the basis of using YOLOv5 detection algorithm to detect the wheel tread defects of sintering machine trolley,further design the wheel tread defect detection system and data analysis system of sintering machine trolley,and the system interface is designed with Py Qt.The detection system includes functions such as picture detection,video detection,camera real-time detection,system alarm,and We Chat message push.The system detection results can be saved to the My SQL database in real time.The data analysis system is to edit and analyze the data in the database,and realize the visualization of the system detection data.The detection system tested the running video of the wheel tread of the sintering machine trolley of Baotou Steel.The experimental results showed that the recall rate of the wheel tread defect detection system of the sintering machine trolley was more than 92%,and the precision rate could reach 100%.The target of wheel tread defects missed by the analysis system is small,which basically does not affect the normal operation of the wheels of the sintering machine trolley,so the system can be applied to the detection of wheel tread defects of the sintering machine trolley.