| Gear vision measurement is a real-time gear parallel measurement technology based on the optical imaging method,which is compatible with the development trend of precision,integration and intelligence of measuring instruments.Edge positioning is one of the key factors that determine the measurement accuracy of gear vision measurement system.On the premise of strictly cleaning the tooth profile surface,the measurement environment,operation mode and other factors will also make the tooth profile surface contaminated by dust,oil and other impurities.These pollutions cause changes in the position of the tooth profile edge of the gear to be measured obtained by the gear vision measurement,resulting in gross errors,which have a direct impact on the accuracy of tooth profile edge extraction,and further affect the visual measurement accuracy of tooth pitch deviation.How to process the original information obtained by the gear vision measurement,better eliminate the influence of external interference on the gear vision measurement information,and improve the environmental resistance and measurement authenticity of the gear vision measurement method is an urgent problem in the field of gear vision measurement,which in the machinery manufacturing industry.In order to eliminate the influence of the edge distortion of the gear profile image on the measurement precision of gear pitch deviation,the paper conducts a research on the edge distortion compensation technology of the gear vision measurement image.First of all,the paper converts the edge signal of the two-dimensional gear under test vision measurement image into a one-dimensional signal of the normal deviation of the tooth profile edge.Then,carried on the edge distortion area discrimination algorithm,the edge distortion classification algorithm,the local distortion area signal compensation algorithm research work.The following tasks are mainly completed:(1)According to the involute characteristic of the involute spur gear,the data fusion of the visual measurement image position information and gray information of the gear to be tested is performed,and the image segmentation and the information processing of the pixel points of the edge transition zone of the tooth profile image were completed.It established the mapping relationship between the normal offset of the pixel point in the transition zone of the tooth profile and the polar diameter,and converted the edge signal of the two-dimensional visual measurement image of the gear to be measured into a one-dimensional signal of the normal offset of the tooth profile.(2)According to the accuracy characteristics of the involute spur gear itself,the wavelet packet decomposition and reconstruction method was improved,and the variable weight wavelet packet decomposition and reconstruction algorithm of the tooth profile edge signal was proposed to realize the extraction of the tooth profile edge signal μ_k.The tooth profile edge signal μ_k was decomposed into the tooth profile edge dynamic component signal μ_kand the tooth profile average trace signal ε_k.Designed a variable threshold iterative approximation algorithm based on the variable weight wavelet packet decomposition and reconstruction algorithm of the tooth profile edge signal to realize the identification of the edge distortion area of the tooth profile image.It satisfied the positioning accuracy requirements of the edge distortion signal compensation of the tooth profile image.(3)According to the need of distortion compensation,the edge distortion signal of the tooth profile image was defined as five types:normal type,neglect type,alarm type,rejection type and compensation type.A partial binary tree twin support vector machine multi-classification algorithm based on the optimal classification feature was proposed,which according to the high-dimensional,small sample and inseparable characteristics of the non-stationary transient signal of the tooth profile image distortion.It realized the classification of the edge distortion signal of the tooth profile image,and meeted the requirement of"compensation on demand"for the edge distortion signal of the tooth profile image.(4)The edge distortion signal compensation algorithm was studied based on the similarity analysis of adjacent tooth profile with the same name.According to the similarity of the adjacent tooth profile of the involute spur gear with the same name and the principle of support vector machine regression analysis,two kinds of compensation algorithms for the edge distortion area of the tooth profile visual measurement image were proposed:the distortion area compensation algorithm based on the mean value of the adjacent tooth profile with the same name and Distorted area support vector machine regression compensation algorithm.The tooth profile edge correction signal constructed after the signal compensation of the distortion area was used for the relative method indexing circle pitch measurement,which solved the problem that the edge distortion signal affected the relative method indexing circle pitch measurement accuracy,thereby improved the subsequent measurement of the tooth pitch deviation accuracy.Gear vision measurement experiments show that the variable weight wavelet packet decomposition and signal reconstruction algorithm,which fitting the edge position of the tooth profile.the variable threshold iterative approximation algorithm based on it can quickly and automatically identify the image distortion area.The radial positioning accuracy of the distortion area boundary reaches 2.5 pixels(50μm),which can meet the positioning accuracy requirements of image edge distortion compensation;Under the condition of a small sample of tooth profile image distortion and non-stationary transient signal data,the classification accuracy of the tooth profile image edge distortion of the OCF-PBT-TWSVM multi-classification algorithm is 96.96%.It can meet the distortion compensation requirements of subsequent"compensation on demand";Comparing the edge compensation signal of the distortion area obtained by the distortion area compensation algorithm based on the mean value of the adjacent tooth profile and the distortion area support vector machine regression compensation algorithm with the measured signal when the tooth profile is cleaned without edge distortion,the error is within 3μm;Used the tooth profile edge correction signal constructed after the distortion area signal compensation to measure the gear pitch.Compared with the measurement result of the tooth profile edge signal without edge distortion,the error is within 2μm;It solves the problem that the edge distortion of tooth profile vision measurement image affects the accuracy of pitch deviation vision measurement,improves the real-time performance,environment resistance and authenticity of measurement data of gear vision measurement,and can be used for production inspection and finished product test in the process of gear manufacturing or product acceptance. |
PDF Full Text Request