The Consistency Machining Quality Inspection Problem Of The Multi-holes Part Based On Wavelet And Bispectrum Features Differences Of Monitoring Signals

With the development of manufacturing technology, the high-precisionmulti-holes parts with the same drilling parameters are used widely and increasingly.The consistency quality inspection of drilling holes is the key step to guarantee themulti-holes parts quality. In this dissertation, surrounding the problem that must bestudied in the consistency quality inspection of drilling holes, we presents a detailedstudy on the aspects of drilling monitor platform set-up, signal preprocessing, featureextraction, feature fusion, pattern recognition and so on.(1) The drilling monitor platform set-up, the monitor signal acquisition andpre-processing. The three-axis acceleration sensors and the acoustic emission sensorsare used to build up the drilling monitor system and obtain the drilling monitoringsignals. And monitoring signal preprocessing techniques is employed to eliminatepower line interference.(2) The monitoring signal analysis and feature extraction. The wavelet analysis isused to extract the domain energy feature and the bispectrum method is used to extractthe bispectrum feature of monitoring signals. Then, we can be discovered thatbispectrum feature can be represented the Gaussian properties of signals, and there is aorganic connection between the hole system drilling process quality fluctuation andmonitoring signal features differences.(3) The consistency quality inspection of hole system drilling process based onfusion of clustering monitoring signals features. The domain energy of feature bandsand the mean of bispectrum amplitude are served as process features. Then, the fuzzyC-means clustering technique, the weighted clustering based on the ReliefF algorithm,the visual clustering based on radar chart are used to analysis the consistency ofdrilling quality.The consistency of24groups drilling process quality is analyzed by use of themethod successfully in this dissertation. Research results show that: the clustering ofmonitoring signals features can be used to analyze the consistency of holes drillingprocess quality efficiently, can realize the quality inspection of multi-holes parts to acertain extent. This method can overcome the problems of manual sampling detectionlike missing detection, destructive testing and so on. This new method offer a newmethod and the theoretical basis for realize the consistency quality testing and control problem of the multi-holes drilling process quality.