Development Of Mechanical Wear Expert System Based On Abrasive Wear Mechanism

Friction and damage of mechanical equipment are generated in the process of its operation,which will not only cause the loss of parts and materials,but also reduce the accuracy of equipment,increase the failure rate,and shorten the life of equipment.The type of abrasive particles produced by equipment wear is closely related to the wear parts of equipment.The abrasive particles can be used to diagnose the fault parts of equipment for timely maintenance and repair,which plays an important role in normal production activities and the service life of equipment.At present,intelligent identification based on the combination of ferrography analysis technology and computer technology is the main way to study the types of abrasive particles.Its main content is to use the image of abrasive particles to obtain characteristic parameters for statistical analysis,and use the intelligent algorithm to identify them.In this paper,the expert system of mechanical wear based on abrasive wear mechanism is established by combining image processing,pattern recognition and expert system,which can quickly and accurately identify the types of abrasive particles,and then determine the wear state of equipment,and solve the disadvantages of manual recognition.The main research contents of this paper are as follows:1.Obtain target abrasive particles.Image preprocessing is carried out by gaussian filtering and threshold segmentation is carried out by otsujin threshold method,so as to separate the target abrasive particles from the complex background.Moreover,the acquired burrs on the edge of the abrasive particles and the holes in the target area can be well solved by morphological processing.2.Extract feature parameters.On the basis of analyzing the geometrical and textural features of abrasive grains,the method of identifying spherical abrasive grains and cutting abrasive grains by shape features and identifying fatigue abrasive grains and severe sliding abrasive grains by combining shape and texture features is determined.3.Grinding particle identification.Through a comparative study of various recognition algorithms,it is found that the support vector machine(SVM)algorithmbased on statistical learning theory is most suitable for the recognition of abrasive particle types in this paper.After selecting the hierarchical method as the classification method and the radial basis function as the kernel function,a classifier model is established to realize the classification and recognition of abrasive particle images.4.Built the system platform with HALCON software and C# programming,and conducted simulation experiments to verify the feasibility of the system.The experimental results show that the accuracy of the system is 93%.Although the types of abrasive particles are not all correctly identified,the error is within the acceptable range.It can be seen that the system can be used to recognize the type of abrasive image.