The double-headed screw rotor is a common and critical power generating component and the accuracy of its profile has a direct bearing on the mechanical properties and service life of various products. The current profile accuracy measurement methods for double-headed screw rotors are all contact-type, and are divided into two type. One type is to manually measure the profile accuracy with a single-parameter instrument such as a micrometer. Although measurement accuracy and efficiency are both low, this method can be used for on-machine measurement. The second type refers to automatic measurement of the profile accuracy with a coordinate measuring machine (CMM). Despite its high measurement accuracy, this method consumes significant auxiliary time, increases the positioning error in off-machine measurement and cannot guarantee the post-correction accuracy of a nonconforming screw. To measure the profile of double-headed screw rotors with high precision and efficiency, an on-machine measuring (OMM) system based on a laser triangulation displacement sensor (LTDS) is designed and implemented by considering a four-axis whirlwind milling machine as the carrier.
To improve the measurement accuracy of the system, the generalized variable-structural-element morphological method, polynomial interpolation algorithm, and ellipse fitting method are combined to realize micron-level centroid extraction from a noise-containing spot image. Then, the hybrid method is experimentally verified. Subsequently, a smoothing algorithm for point cloud data is devised based on the Lagrange multiplier method to avoid the defects associated with the piecewise curve fitting method, that is, function continuity and differentiability cannot be satisfied at piecewise points. Finally, the profile parameters are calculated in real time according to the data reconstruction results, and the machining quality is assessed.
The measuring results are shown in Fig.9. In comparison with that of the traditional LTDS-based measuring method, the profile measured through the OMM method is closer to the CMM result. This outcome indicates that the LTDS data acquisition accuracy can be improved with the proposed method in Section 3.1, that is, the generalized variable-structural-element morphological edge extraction algorithm is applied first to realize effective edge measurement of the spot image and then the polynomial interpolating subpixel edge positioning algorithm is applied to realize rapid subpixel positioning. Ultimately, centroid extraction is conducted through the ellipse fitting algorithm. As shown in Table 3, all difference values are within the acceptable tolerance zone for all three standard measuring methods compared with their corresponding nominal values, indicating that the screw rotor part is acceptable after machining. However, the proposed spot centroid extraction method can improve the measurement accuracy of traditional LTDS for a free-form surface compared to OMM. The profile accuracy of the screw rotor obtained through the proposed LTDS-based on-machine measuring system is within ±9 μm from the difference values between OMM and CMM results.
Such whirlwind milling machines configured with four-axis screw rotors have been placed in operation. Actual results indicate that the measurement accuracy using the proposed method is ±9 μm, in which the measurement uncertainties of major axis, minor axis, and screw pitch are 0.72, 0.69, and 1.65 μm, respectively, and measuring one screw pitch consumes 39.7 s. Therefore, the results indicate that the proposed on-machine measuring system can satisfy the requirements for accurate control and rapid measurement of large workpieces in actual operation.
