The Identification Of Log Species And Hole Defect In Log Based On The Support Vector Machine (SVM)

The ultrasonic wave wood nondestructive test is one of the most widely used nondestructive test methods, but only one signal can be processed and analysed in one time, so it needs much time to process multiple signal data, and the result is not so correct. The support vector machine(SVM) can process many signal in a short time, if it is used in processing and analyzing ultrasonic wave signal, it will be faster and more correct. Two dimentional visualization is necessary to direct at the situation that the results are not so visual because of the processing and analyzing is only to the one dimentinal signal of ultrasonic wave.In this paper, based on the theory and method of ultrasonic wave nondestructive test in wood, the classification principle and method of support vector machine(SVM), and the drawing method of the MATLAB software, combined these things and tested the lumber. Through the extraction, processing and analysis of the ultrasonic time domain feature values, combined the application of the SVM, and time domain features of the ultrasonic wave were extacted, processed and analyzed. The support vector machine was used to classify data, and the MATLAB was used to draw.The main content of this paper are the propogation characteristic of ultrasonic wave spreading in the cross section of log, and defect two dimentional visualization in log cross section based on the classification and identification of support vector machine. First of all, effect of different spacies of log and hole in the cross section of log to the ultrasonic wave propogation characteristic were analyzed; then, the classification and identification to the different sizes of hole in log cross section based on the support vector machine were done; in the end, the hole defect two dimentional visualization in log cross section based on the result of classification and identification of support vector machine was done.Many significant conclusions were drawn like below:the velocity of ultrasonic wave in wood was smallest when the ultrasonic wave propogation direction was vertical to the tree axis, and the velocity of ultrasonic wave in wood was biggest when the ultrasonic wave propogation direction was parallel to the tree axis; the propogation velocity of ultrasonic wave in ribbed birch and linden has a significant positive correlation; the propogation distance of the ultrasonic wave did not have a significant effect to the cycle, the half cycle, the amplitude, the fall time and the rise time of the first wave; limiting the propogation distance between30cm and35cm, the propogation velocity of the ultrasonic wave in larix gmelinii, linden and ribbed birch separately was930m/s to1040m/s、770m/s to860m/s and860m/s to980m/s; the holes in linden made the range of the feature values above bigger; the size of the holes did not have a significant effect on the feature values; application of the SVM in classifying the trees sort, the holes and the size of holes was viable, and it had a high accuracy rate of more than80%; and based on the SVM and ultrasonic wave, the MATLAB software could draw two dimensional figures of the hole defect in the cross section in wood.