Ferrography Quantitative Analysis Methods And Applied Research

Ferrography is based on wear particle analysis and widely used in many fields as tribology research, condition monitoring and fault diagnosis. This thesis is committed to applying image analysis to ferrography based on the status quo of ferrography. A program on VC++ 6.0 applied to wear particle image analysis is completed. Researche on different fields of ferrografic quantification with the tool is done.The combination of image processing methods is optimized and applied to wear particle image processing, including such methods as image smoothing, sharpening, filtering, enhancement and Mathematical Morphology procecing.The method of image segmentation for the speciality of wear particle image is developed. Image segmentation is very important in image analysis. Two new segmentation methods based on Maximum Classes Square Error are developed. The tool of manual segmentation is supplied. The tool of counting pixels of different areas and deleting small areas valueless to ferrography analysis is supplied. The combination of different methods has the segmentation quality ensured. Wear particle image is better for analysis after doing image processing and segmentation.The algorithm of wear particles' texture and shape characteristics is applied. The combination of Moment Invariants and Gray-grads Level Co-occurrence Matrix is taken to quantifying wear particles for expanding the fields of single particle's quantification methods. Some better parameters are chosen after measuring many wear particles. BP neural network is set up, developed and applied to wear particle automatic classification based on quantifying single wear particle for improving the intelligence and automation level of ferrography analysis.Quantitative analysis is an important method for ferrography analysis. Traditionaly, the parameter of quantification is based on measuring the coverage of wear particles. The method of measuring the ratio of wear particles' coverage on image is applied. The effectiveness of the method is proved by applying the method to real analysis. Instruments like Direct Reading Ferrograph and densitometer are not required in the method.The size of wear particle has corresponding relation with wear type. The statistic and survey of wear particles' size is very important on the survey of wear condition. Currently there is still no effective method to separating different particles on image, so the size of particles can't be measured. Based on the status quo of image processing technology, the definition of wear particles on sediment chain and unusual particle is suggested. The tool of manually designating boundaries of unusual particles and measuring the size of sediment chain and unusual particle is supplied. Quantitative analysis of images taken by experimentation is completed and the effects of different parameters are compared. A new model of wear pattern is suggested based on the size of sediment chain and unsual particles.