Abstract: Automobile crankshaft roundness error and cylindricity error detection is necessary, and there exists low efficiency, low accuracy presently. The tolerance evaluation model of plane curve and space curve is established based the error conversion principle. The roundness and cylindricity error evaluation mathematical model are also established according on the minimum zone method, the genetic algorithm is used to calculate the form and position errors of the crankshaft with minimum assessment requirement, the method solve the problem of ideal contain elements of the pose parameters being not accurate. Meanwhile the automatic detection scheme of automobile crankshaft shape error is designed based on image domain. The connecting rod shaft is revolute along the main shaft neck, so the measurement data in the image domain is not in the same coordinate system. In order to solve the problem, the holes in the crankshaft flange end are chose as datum feature and through template matching and the hole edge extraction, the rotation angle of the connecting rod shaft has gotten. A certain type of engine crankshaft as an example is taken to have the large number of error detection, and the measured results are compared with the results of the three coordinate measuring machine. Data show that the crankshaft shape error precision of the proposed detection method is 1 µm, and duplicate detection error less than 0.1 µm, so the this method is correct in theory and feasible in practice.

Key words: automobile crankshaft, cylindricity error, error conversion, roundness error