Research Of The Automatic Detection Of Engine Crankshaft Journal’s Form Error Based On Image Domain

As one of the core parts composing the engine, the quality and performance of the crankshaft decide and restrict the performance of almost each aspect of the whole engine in a seriously influenced manner. In the structure of the crankshaft, the form error left in the main journal and rod journal will influence the smoothness of operation of the reciprocating engine, therefore, it is very important to detect the form error of the crankshaft journal. For solving the common problems of engine crankshaft journal detection methods with low efficiency and low accuracy nowadays, this dissertation studied into a new detection method of journal’s form error based on image domain and theory with experiments. The main content and results of the research are as follows:Based on the principle of error transformation, the programming models, which are applicable to data extracted from image domain, of roundness and cylindricity errors have been established. Also have solved the imprecise pose parameters estimates of ideal envelope feathers.The core techniques of image processing needed to collect the sampling data from image domain, such as image preprocessing, edge extraction and the template matching have been researched deeply. The problems involved such as smoothing, automatic threshold segmentation, and et al. have been resolved.The detection scheme of automatic detection system evaluating the journal’s form error has been designed. The optical components and the camera motion control equipment have been selected according to the requirements of the scheme. Then constructed an experimental equipment of journal’s error detection system with the developed man-machine interface.Analyzed the sources of error produced by the detection method. The problems involved in the experiment such as camera calibration, crankshaft reset, and et al. have been solved.Based on the above research, conducted a large number of experimental measurements on the type4102engine crankshaft from Wuxi Diesel, and compared the data with the measured results gained by coordinate measuring machine. The results showed that the accuracy of the detection method proposed in this paper is1μm and repeatability within0.1μm, so the detection method of the crankshaft journal’s form error is both correct in theory and feasible in practice.