Research Of Type Identification And Reconstruction Of Defects Within The Cylinder Based On The Ultrasonic Detection

Cylinder components are widely used in industrial fields. Due to some uncontrollable factors in the process of production which is usually caused component internal defects, we need to test the components in advance, avoiding some defect components into the subsequent production. Aiming at cylinder component ultrasonic automatic testing, this paper studies the defect echo feature extraction technology, types of defects recognition technique and defect reconstruction technique.The paper through analysis the advantages and disadvantages of wavelet packet decomposition and empirical mode decomposition, finally adopted to decomposition of flaw echoes from multi-angle using the ensemble empirical mode decomposition method, and extracts the energy distribution and the information entropy as feature in six IMF. Using the support vector machine(SVM) as classifier and the particle swarm algorithm for classifier parameters automatic optimization, recognition accuracy is reach to 93%. Aiming at the problem of eigenvector redundancy, the feature evaluation technique which based on the average distance was introduced to calculate sensitivity of feature and sort feature from high sensitivity to low sensitivity. Through experiment contrast analysis, optimized input freature which in classifier is not only increases the recognition accuracy to 97.78% but also reduces the running time. Through calculate delay time of defect echo,circular cross section was reconstructed by using the method of transit time and border fitting. Then, three- dimensional reconstruction of all cross section by volume rendering, the effect of the reconstruction is intuitive and accurate.The research results and methods in this paper can be applied to internal defects recognition and reconstruction of cylinder components, and provide the reference basis for this type of component detection.