Identification Of Micro Defects For China Ⅵ Emission Standard Piston Throat Based On Eddy Current Detection Signals

With the implementation of the national VI emission standard,the piston will bear about 270bar explosive pressure,and more and more high quality are required for piston surface.When there is a crack source consisting of a micro-defect in the piston throat,the high explosive pressure will easily lead to the cracking failure of the throat and endanger the safety of the vehicle.Therefore,the accurate identification of the micro-defect in the piston throat has important practical significance.Aiming at the problems of small size of piston throat,complex surface structure and narrow space,a vertical retractable probe clamping rod is designed by using a small eddy current probe with diameter of 4 mm.A 4-axis linkage testing scheme is proposed.The eddy current testing equipment for micro-defect of piston throat is developed.The probe trajectory is retrieved from the curve of piston throat,and the 3-DOF control of the probe is realized.In order to improve the signal-to-noise ratio(SNR)of the detected signal,an improved wavelet threshold denoising algorithm is used to improve the SNR of the detected signal.The denoising effect of 16 threshold criteria and combination of threshold functions is analyzed by using MATLAB software.The results show that the combination of Heursure criterion and exponential threshold function can improve the signal-to-noise ratio by 16.4835 dB,and the denoising effect is good.The method is used to reduce the noise of groove defect,hole defect and normal eddy current signal.Aiming at the characterization of micro-defect grooves and holes in piston throat,12 signal features are extracted from time domain and frequency domain as traditional time-frequency domain feature sets,and 48 features of spectrum energy distribution are extracted by EMD decomposition as time-frequency domain features,which constitute 60 comprehensive feature sets.PCA algorithm is used for feature set reduction and compression comparison.The results show that the compression rate of the comprehensive feature set is 16.67%higher than that of the traditional time-frequency feature set,and its ability to represent signal difference is better than that of the traditional time-frequency feature set.Based on the PCA dimension reduction defect synthesis feature set,the kernel function parameters and penalty factors of SVM classifier are optimized by using the method of five fold cross validation.The training of three SVM classifiers was completed by collecting 240 groups of three kinds of defect signals through experiments.SVM classification algorithm is used to classify and recognize the defect features.The results show that 21 groove defects,19 groove defects,20 normal signals and 1 groove type are misjudged as groove signals in 60 test sets.Therefore,the detection rate of defects is 100%,and the recognition accuracy of defect types is 98.33%.The accuracy of PCA-SVM classification algorithm for the identification of micro-defects in the throat of national VI emission standard is verified.