Ferrography Wear Image Pre-processing And Abrasive Recognition Technology

Recently, ferrography technique has been one of the effective methods for machine condition monitoring and fault diagnosis, which is based on the analysis of wear particle. In the ferrography, the recognition of the wear particle is the core , so the result of the recognition has direct relation to correctness of the wear condition monitoring for the equipments. With the rapid development of the computer technology, some methods were applied in the ferrography technique, such as computer vision technique,expert system,artificial neural network. Intellectualize of the recognition of wear particle is an important and difficult issue.Debris automatic identification and analysis system had been developed in the present research, which is based on Visual C++ 6.0 software platform by studying image processing technique of computer, pattern recognition method and tribology knowledge. In this system, firstly make pretreatments of the original color debris image, including geometric transformation,grey scale,enhance filter processing,image sharpening image segmentation,extraction of brim contour lines and so on. Wear particles are choosed directly from the ferrography images after pretreatments, and the wear particle is signed and selected though this method.Firstly, character parameters of wear particles(including four types: Layering, sliding, cutting and sphericity wear particles) were sorted into four types: size character parameters, contour shape character parameters, structure character parameters and colour character parameters; and every character parameter value was gained though the training of sample of wear particles. As so many character parameters, the original character parameters were standardized though the method of Principal Component Analysis in this text. Combining with the SAS statistical analysis method, datas of character parameter were standardized, and finally the five types of character parameters, called for the type recognition of wear particles, were confirmed according to the contribution rate of every character parameter.The first step of ferrography wear particle recognition is to select some standard wear particles from the gallery of ferrography wear particle as sample, and make training study. Based on the filtered character parameters, different character parameter's distributing ranges were received on different ferrography wear particles by the training of wear particles sample. By characteristic of the four types of ferrography wear particles, the priority order of every character parameter was confirmed in the recognition process. In the end, compute the distance between ferrography wear particles await recognition and the distributing range of standard character parameter, and identified the type of wear particles according to the principle of least distance. It was the ultimate objective of this text to make a qualitative class on the wear particles of ferrography piece.At the end of this text, joined with the credibility experimentation of engines and the wear condition of the main friction components, the trial analysis of the ferrography oil of this engine's lubricant was carried though as follows. Ferrography images from this trial were analysed and recognized by the developed software of ferrography wear particles pretreatments and Recognition, then compared the analysis conclusion with the ferrography qualitative analysis.