Research On Key Technologies Of Defect Detection System For Large Ring Forgings Based On Machine Vision

Nowadays,with the continuous development of global industrialization,energy and environmental crisis has become a common problem.As a clean and renewable energy,wind energy has been widely used in wind power generation.The large yaw ring forging for wind power generation is a key component of wind power equipment,and its function is equivalent to large sliding bearing.The quality of ring forging affects the conversion efficiency of electric energy.At present,the detection of surface defects of ring forgings in domestic enterprises mainly depends on manual work,the detection accuracy and efficiency are difficult to meet the needs of actual production.The detection technology based on machine vision has many advantages such as non-contact,robustness and high efficiency,which can effectively avoid the problems in manual detection.Therefore,in this thesis,the key technologies of machine vision inspection system are studied,and the main research contents and conclusions are as follows:The hardware design of defect detection system for large ring forgings based on machine vision.Through the characteristics of surface defects of large ring forgings and the actual inspection requirements,the inspection indexes are determined;According to the detection requirements and indicators,the hardware of the system is selected,including camera,lens,light source and other key devices;And through Solid Works software,the 3D modeling of the rotating platform is carried out that according to the actual detection environment.The research on surface image processing method of large ring forging.The ROI of the image is extracted,and then the image surface noise is processed.Several filtering methods are analyzed and compared,and the experiment proves that the low-pass filtering method with the best performance is selected;Aiming at the image enhancement part,an improved Sobel algorithm is proposed,and its superiority is verified by experimental comparison;The QTSU segmentation algorithm is used to segment the image,and the morphological algorithm is used to remove the small interference areas.Experiments prove the superiority of this algorithm.The study on classification method of surface defects of large ring forgings.Analyze the features of the surface image,and extract the geometric features and texture features of the defective parts.Circular LBP algorithm is used to extract texture features,and the algorithm performs correlation calculation by interpolation estimation;PCA(Principal Component Analysis)is used to reduce the dimension of the image and improve the efficiency of the detection system;The theory of SVM(Support Vector Machine)is studied,and the appropriate kernel function and selection strategy of support vector machine are selected for the SVM model,which verifies the feasibility and accuracy of applying support vector machine to the classification and identification of surface defects of large ring forgings.The software design and experimental study of testing system for large ring forgings.According to the actual testing requirements,the specific functions of the software system are determined,and the relevant functions are actually tested.The feasibility and superiority of the algorithm in this thesis are verified by testing experiments with the designed hardware and software.The experimental data show that the accuracy rate of the system reaches 94.4%,the average false detection rate reaches 5.62%,and the average time of a single piece reaches 1.6min.All the indexes meet the requirements of the manufacturers.