Application of correlation-regression and cluster analysis to study data based on thermographic images of cooling system pipelines

Context and relevance. Among the various methods of non-destructive testing, thermography has been actively developing in recent years. Its use allows recording defects associated with a violation of wall thickness, the presence of cracks, potholes, and corrosion. Thermography is often used to inspect buildings, structures, machines and mechanisms, which is due, among other things, to the possibility of continuing their operation even during diagnostic activities. One of such devices, the shutdown of which is difficult to assess the technical condition, is the cooling system of marine engines. Objective. To conduct a correlation-regression and cluster analysis of a dataset of pipes thermograms of a marine engine cooling system and to determine the nature of the relationship between various parameters. Hypothesis. It seems likely that in the course of performing the correlation-regression and cluster analysis, dependencies will be established between the temperature characteristics of the base metal of the cooling system pipes and the temperature of the zones of intra-pipe defects, as well as their geometry. Methods and materials. A dataset of 210 thermographic images of cooling system pipes, cluster analysis methods, correlation and regression analysis methods, Loginom software package. Results. As a result, dependencies were established between the type of defects (detected by the thermographic method), their geometric features and changes in the temperature characteristics of damaged and intact pipe zones used in ship engine cooling systems. The possibility of effectively using correlation and regression and cluster analyses as tools for processing data based on thermographic images of pipes was confirmed. Conclusions. Organizations responsible for regulatory and methodological support for the use of non-destructive testing methods should use the obtained results to update the requirements for thermographic research procedures. In particular, it is necessary to update GOST R ISO 18434-1-2013. Condition monitoring and diagnostics of machines. Thermography. General methods. The updated terminology and methodological parts of this standard should include references to the possibility of using correlation and regression and cluster analyses as tools for processing thermographic data.