Research On Eddy Current Testing Of The Leading Electrode Crackle Of The Motor

The eddy current testing is one of the five normal non-destructive testing methods. In comparison with the others, it is easily to implement, can test in higher speed, is extremely sensitive to the surface and the near surface defect, as well as needs not remove the superficial coating, and so on, therefore it is specially suits to detect the defects of parts under the bad working conditions. Considering problem of the leading electrode of the Locomotive traction motor crack flaw, the eddy current testing method for leading electrode crack defect is researched in the paper.Firstly, given the definition of non-destructive testing, combined with the characteristics of the structure and the operating environment, eddy current testing used to identify the crack of leading electrode. Secondly, the principle of eddy current test is introduced. Based on electromagnetic field theory, some important formulas are deduced. Impedance analytical method is presented, through which we analyze some factors that alter testing coil impedance's value. Then, characteristic frequency of laid test coil is calculated, it is important to select the test frequency. And eddy current testing shows that the existing three effects: the edge effect, the skin effect, the effect of lift-off.On the base of analyzing and referencing the existing structure forms of the probes with better performance, of the main testing environment for a very, very surface of the leading electrode with the parcel, and other unfavorable factors, the paper choose the structure form of absolute probe. Paper introduced the selection principle of the related parameters of the probe, and the calculation formula of the parameters are given. The parameters of the probe determined by the design method of orthogonal experiment.In this paper first to identify the leading electrode crack detection system program, and then to the detection system hardware design. In order to improve the sensitivity of the sensor design is derived through the mathematical signal conversion circuit in the various parameters, and given the parameters of the formula to solution, given the detection system excitation, signal conditioning, A/D conversion and the collection of the hardware design. In software design: The modular software design, gives the software to realize the process flow chart.Based on the main crack to carry out a feature extraction and classification of defects of the purpose of the paper carried a principal component analysis (PCA) and principal component of nuclear analysis (KPCA) are two recent feature extraction methods and means classifier, K neighbor classifier, BP Neural network classifiers, as well as support vector machine algorithm, and other four categories. Through experiments, come to KPCA feature extraction method based on support vector machine classifier classification algorithm can effectively reduce the error rate in classification.