Research On Ultrasonic Testing Of Eccentric Shaft And Intelligent Defect Recognition Method

The shaft is a very common parts,these bearing parts are used in production,but they are easy to produce the cracks,holes and other defects on the surface.At present,most of the defect detection relies on manual detection,and the automatic detection system is mainly used in the detection of some simple-shaped workpieces.Aiming at the problems of low detection efficiency and simple appearance of the detected objects in the existing ultrasonic detection methods,an automated ultrasonic detection system has been developed,and the detected defects have been intelligently identified and classified.Based on the research of ultrasonic automatic inspection domestic and overseas for the past few years,combining mechanical technology,automatic control technology and ultrasonic flaw detection technology,an automatic ultrasonic inspection system based on eccentric shaft defect detection is designed.The system mainly includes a five-axis manipulator,a water tank inspection platform and a multi-channel ultrasonic flaw detection card.Aiming at the eccentric shaft,the special structure of the revolving body adopts a robot hand to clamp the ultrasonic probe for automatic detection.The scanning trajectory of the eccentric shaft is divided,combined with the kinematics model of the manipulator,to determine the position and posture of a certain point on the spatial trajectory of the ultrasonic probe,and realize the precise and stable motion control of the manipulator holding the probe through the fifth-order polynomial trajectory planning.To extract the feature signal of the collected information.First,because of collected information contains a lot of noise.Proposing a soft and hard threshold tradeoff theory to analysis the collected information by wavelet denoising.Then the influence of the threshold and the number of decomposition layers on the noise reduction of the defect signal is discussed,and the experimental analysis is carried out.The simulation results show that the four-layer wavelet denoising using Sqtwolog threshold for the collected defect signals has a good denoising effect.Then extract the feature information of the denoising ultrasonic signal by the wavelet packet energy spectrum theory,and use the energy values in the eight sub-frequency bands as the feature vector of the eccentric shaft defect,which is used for training of models such as neural networks.Finally,the eccentric shaft is realized Intelligent recognition and classification of defects.In order to realize the classification of the four kinds of defects of the eccentric shaft,a probabilistic neural network is used to identify the defects.An improved probabilistic neural network based on saprrow algorithm is proposed on the basis of probabilistic neural network.The sparrow search algorithm is used to optimize the smoothing factor of the probabilistic neural network and improve the accuracy of its recognition.Through experimental simulation,it is concluded that the improved probabilistic neural network based on the sparrow search algorithm has better recognition and classification performance than the unimproved probabilistic neural network and radial basis function neural network.Finally,through experimental tests,wavelet analysis and pattern recognition are carried out on a certain number of defect samples.The experiment shows that the method is effective and feasible in the recognition of eccentric shaft defects.Therefore,the theoretical analysis and experimental research methods adopted for the ultrasonic detection of eccentric shaft defects are feasible,which improves the reliability and accuracy of defect recognition.