The Rail Surface Defects Detection Based On PMP

Since the grating projection 3D optical detection technology has the advantages of non-contact, low loss, and capable of automatic, tc, it has been widely used in industry, the field of computer recognition and physical model test.Based on the study of rail profile detection and surface defects testing at home and abroad, by the comparative analysis of optical three dimensional measurement method and the theoretical derivation of Phase Measurement Profilometry(PMP), practicability of using PMP in rail profile detection and surface defects testing is analyzed and experimental verified. A systematic research of in-service rail profile detection is operated, especially for the one with the feature of strong reflections. According to the effect of polarization of light, the strong reflections on the rail surface can be weakened, with the analysis of the reflections field through the rotation of the polarizer and the proper pictures taken from a window. Testing system is designed and experiment platform is established. Following the analysis of PMP’s four-step and five-step phase-shifting algorithm, an improvement of the traditional Stoilov phase-shifting algorithm is operated. Based on statistical approximation method singular points are removed, the errors are lowered and the wrapped phase accuracy is improved. In the step of phase unwrapping, several typical phase unwrapping methods are comparative analyzed, and then Branch-cut method is selected. With the analysis of PMP optical path, the theory of real rail profile detection in the testing system is deduced. With the combination of calibration theory, the calibration of the measuring system is studied. The theory and experiment of Zhang Zhengyou’s calibration theory and the calibration method of Z direction are analyzed and verified respectively.In this thesis taking the 71MN rail as a sample, with the method of PMP, a simulation and reconstruction of the rail simulated by the 3 dimensional data generated by a computer are operated. The results show that the phase error of the corresponding point is 0.06rad. With the above discussed method and the data of the profile and defects of a real rail captured in the laboratory, a 3D reconstruction is operated with a micron accuracy, which is far better than the light-section method as a popular way used in the profile detection and defect testing.