Research On Coil Edge Face Defect Detection System Based On 3D Machine Vision

With the rapid development of China’s economy and increasingly fierce competition with foreign markets,the demand for strip steel is higher and higher,and the requirements for strip steel quality are also more and more strict.Strip steel is one of the main end products in iron and steel industry.In many cases,its product quality directly affects the quality and performance of follow-up products.The quality of strip steel products has not only become a key index for domestic and foreign iron and steel enterprises to compete,but also related to the economic benefits,survival and development of enterprises.In the production process of strip steel,it is usually necessary to roll the strip steel and transport it by rail or crane.In the process of rolling the strip steel,due to various problems,defects such as tower shape,edge crack,edge loss,edge protrusion,serrated edge and so on will appear at the end of the coil.However,in the process of transportation,because the end face of the coil is uneven,the crane is in use,In the process of loading and unloading on the track,the end face of the steel coil may be damaged by teeth,saws,rings and other transportation tools.These defects not only bring serious quality problems to the products,but also cause a large number of return materials,so that the production cost is too high,lose the advantage of competition with the international market,and even lead to serious belt breaking accidents in the downstream production process.Therefore,it has important research and commercial value to carry out the research and development of coil end online detection system based on machine vision in China.At present,most of the steel production lines in our country use manual detection method to detect the end defects of steel coil.The efficiency of traditional manual detection method is low,workers work for a long time easy to fatigue,false detection rate,missing detection rate is high,the speed is slow;At the same time,the coil temperature of hot rolling coiler is very high,which also has a big safety hazard to workers.To solve these problems,this paper designs a set of coil end surface defect detection system based on 3D vision,which can realize the automation of coil end surface defect detection,replace the manual detection,and improve the detection efficiency,replace the manual detection,improve the detection efficiency,and improve the economic benefits of enterprises.The main contents of this paper are as follows:(1)The hardware system of coil end surface defect detection is built.Through the demand analysis,the camera pan tilt is designed,the image acquisition scheme is determined,and the selection and construction of the system is finally completed.Through the experimental debugging in the steel plant of Shougang Qiangang,the requirements of the enterprise for the detection system are met.In the hardware system,a new image acquisition method of rotary ring scanning is designed,and a clear and complete image of the coil end face is obtained.(2)An improved optimal Gabor filter is proposed for defect detection of steel coil end face image.At the same time,an improved optimal Gabor filter is proposed to ensure the detection accuracy and improve the detection speed.The method based on SVM is used to classify the defects.(3)In view of the fact that there is no evaluation method for the defect grade of coil end face in China,a set of basic evaluation methods for the defect grade of coil end face is established.Specifically,the weight of each class is determined by analytic hierarchy process(AHP),the defect matrix is determined by TOPSIS,and the severity evaluation value of each defect is obtained after normalization.(4)Based on visual studio.Net development tool and C#language programming,the software system of coil end face defect detection is developed.The six modules of software system design are discussed in detail.The experimental results show that the design of coil end defect detection system can be stable and effective for coil end defect real-time,non-contact detection classification and defect severity evaluation,which has important use value for enterprises to realize automatic detection production line.