Multi-level information technology for technical system diagnostics based on monitoring data

The article develops and investigates a multi-level information technology for technical diagnostics of dynamic systems based on monitoring data. The relevance of the research is determined by the complexity of degradation processes in modern industrial facilities, such as turbofan engines, and the need to improve the accuracy of forecasting their technical condition to prevent economic losses and safety risks. The proposed information technology is based on the principles of complementarity of diagnostic approaches, multi-level organization, modularity, ensemble of solutions, combination of forecasting and classification, as well as the use of differentiated time representations. Experimental verification was carried out on NASA C-MAPSS reference datasets (FD001 and FD004). The results demonstrated high efficiency. The F1-score of the state classification exceeded 92% for the complex multi-mode FD004 dataset. A comparative analysis showed that the developed technology provides the highest resistance to changes in operating modes. The drop in quality when transitioning from simple to complex operating conditions is minimal compared to existing analogues. Fig.: 2. Tabl.: 3. Refs.: 39 items.