Gerasymov V. Information technology of eddy current inspection in the conditions of action of high-intensity noises of complex structure

Target of the research is the process of eddy current defectoscopy of materials in the conditions of high-intensity interferences of a complex structure. Subject of the study is the models, methods and information technologies for the processing and analysis of defect signals under conditions of quasideterministic interference with random appearance time and random amplitude, and high intensity noise. Research methods are statistical, correlation, spectral, Fourier basis and wavelet, methods, the theory of statistical tests, methods and program of simulation. For the first time, spectral methods, Fourier bases and wavelets, are proposed for identifying defect signals under conditions of quasideterministic interference (signals of tilts of the sensor relative to the material surface), and high intensity noise caused by significant roughness of the material surface, using the proposed criteria. For the first time, using the proposed spectral methods of identification and statistical testing theory, a probabilistic analysis of the identification of defect signals at different values of the probability of false alarms and different noise intensities was performed; a comparison of proposed spectral identification methods based on probabilistic analysis was performed; the probabilities of detecting and separating signals of different sizes of surface cracks from signals of quasideterministic interference, depending on signal amplitudes and noise intensity are determined. By simulating the process of scanning the monolayer surface from the strands of reinforcing fibers in cases where the diameters of the harnesses are distributed according to Gaussian and uniform laws, the noise model due to the roughness of the surface of the composite materials is refined. Based on statistical and correlation processing of information obtained during the simulation of the scanning process the defective material surface, further development of high-speed combined processing methods, which are optimal for the signal / noise criterion for various variants of scanning modes, have been further developed. For the first time an information technology was proposed that combines the developed processing methods of defectoscopic information both in time and in the spectral area and provides probabilistic optimization of the identification process of defects under conditions of quasideterministic interference and high intensity noise. The results of the dissertation work have been implemented in the part of methods and algorithms of defectoscopy data processing at LLC "NVP SKB Nerkon", in LLC "NVO DniprotekhTrans" in the development of computer information systems for controlling the operation of transport, in the educational process of the Oles Gonchar Dniprovsky National University.