The Consistency Of Deep Hole Quality Inspection Based On Peck Drilling Monitoring Signals Similarity Measurement

In aerospace,automobile,electronics and other fields,deep hole parts are used widely and increasingly,and their drilling quality directly affects the performance of products.The traditional drilling process system has poor rigidity,chip removal and cutting heat dissipation difficulties,and the drilling process is in a closed or semi-enclosed environment,so it is difficult to achieve high-efficiency,high-precision,low-cost processing quality inspection requirements.In this dissertation,surrounding the problem that must be studied in the consistency quality inspection of deep hole,we present a detailed study on the aspects of acquisition and pre-processing of peck drilling monitoring signals,monitoring signals detection and segmentation,feature extraction and fusion,similarity measurement and experimental analysis of deep hole quality consistency detection and so on.(1)Acquisition and pre-processing of monitoring signals for deep hole pecking drilling.Considering the characteristics of deep hole drilling,sensor type,sensitivity and installation position of sensor,the vibration acceleration sensor and acoustic emission sensor are used to build up the deep hole peck drilling monitoring platform,and the monitoring signal is mathematically described and denoised.(2)Deep hole pecking signal adaptive detection and segmentation based on HHT envelope.The HHT envelope of the peck drilling signals is extracted,and the noise estimation of the peck drilling signals is realized by using statistical information.Two adaptive update thresholds are set according to the noise level to filter out the noise and segment accurately,and an incremental length control method is introduced to correct the misjudgment caused by burst interference.The effectiveness and accuracy of the proposed method for detecting and segmenting the peck drilling signal are evaluated by simulation and experiment.(3)Wavelet packet feature extraction and fusion of deep hole peck drilling monitoring signal.Five-layer wavelet packet decomposition of db5 is used to decompose the peck drilling signals,and the wavelet packet is reordered according to the energy.The appropriate number of wavelet packets is selected as feature packages based on the relationship between the reconstructed signal and the original signal.The wavelet coefficients,time domain,frequency domain and time series model parameters are used to index the feature packet.Finally,principal component analysis(PCA)method is used to realize the wavelet packet feature fusion of peck drilling monitoring signals.(4)Similarity measurement of peck drilling monitoring signals feature based on improved Dynamic Time Warping(DTW)algorithm.Aiming at the unequal length of pecking drilling signal feature,a similarity measurement method based on improved DTW method is proposed.Considering the application of online monitoring,the time series approximation method Bottom-Up is used to further reduce the signals feature.The improved DTW is used to regularly align the feature of the peck drilling monitoring signal,so as to reduce the time complexity and prevent ill-conditioned regularization.Based on the above research methods,9 groups of samples are selected from the orthogonal experimental data of spindle speed,tool condition and feed.Combined with the actual physical detection of the internal surface roughness values Ra and SEM analysis,the consistency detection of the deep hole peck drilling process quality is realized.The method can overcome the defects of traditional deep hole parts quality detection,such as difficult detection,high cost and low efficiency,and provide a new method and reference for achieving fast and consistent non-destructive testing of deep hole quality.