Research On Ultrasonic Nondestructive Characterization And Classification Techniques For Wood-Based Materials

Wood-based materials are widely used in various fields which are crucial to the modern industry and national economy. The characters and performance of wood-based materials becomes more and more diversified with the development of the wood composites. This trend makes it difficult to find a proper way to quantitatively evaluate the quality of wood product and to rationally use different kinds of wood-based materials. The propagation properties of ultrasonic wave are closely linked with the physical characters of wood-based materials, so that it is one of the most effective ways to evaluate the products and the materials under use. In this dissertation, based on the ultrasonic nondestructive testing technique, several acoustical parameters in the multi-layer model of wood materials were simultaneously optimized. A classification algorithm was also developed which obtained a high accuracy of greater than90%. Finally, a portable characterization and classification system of wood-based materials was designed in a distributive system structure, the feasibility and effectiveness of this system were proofed by experiments. The details of this dissertation were presented as follows:Chapter1, the importance of ultrasonic nondestructive characterization and classification of wood-based materials was analyzed. The research status and development trends of this technique were summarized to clarify the main issues waiting to be addressed in this dissertation.Chapter2, the propagation properties of sound wave in orthotropic materials and materials’interfaces were systematically discussed. The goals and the diagram of the characterization and classification system were also proposed.Chapter3, the ultrasonic propagation properties in multilayer wood-based materials were given. Then, a frequency domain model of wood-based material testing system was established. By employing the Nonlinear Least Square (NLS) algorithm, the acoustical parameters of the model, including the transmission coefficients, phase velocity and the attenuation coefficients, were simultaneously optimized.Chapter4, digital signal processing techniques was applied to extract the features of the ultrasonic transmission signals’envelop and the signals’spectrums. A multi-class classifier based on Support Vector Machine (SVM) was also constructed, by using the time domain features, frequency domain features and the acoustic parameters all together, the classifier obtained the accuracy of greater than90%.Chapter5, a potable ultrasonic non-destructive characterization and classification system was developed. The experiments result carried out by this system on several kinds of wood-based materials verified the feasibility and effectiveness of its design, proofing that the techniques developed in this dissertation were very practical.Chapter6, the main work and conclusions were summarized and the prospect for further research was put forward.