Spectral data processing methods for aviation oil condition monitoring

Problem statement. The paper considers the urgent problem of improving the reliability of aircraft engines through the development of an approach to spectral analysis of used oils. In contrast to traditional methods, which require complex sample preparation and do not allow carrying out rapid measurements, the proposed approach is based on digital processing of spectral data. Particular attention is paid to the development of a comprehensive technique including adaptive noise filtering, background component correction and normalisation of the obtained data. Modern aircraft engines are complex technical systems, the reliability of which directly depends on the condition of friction units. One of the key methods of their technical condition assessment is spectral analysis of aviation oils for the presence of metal-containing particles. Low signal-to-noise ratio due to complex oil composition, overlapping of spectral lines of different chemical elements and lack of standardised methods of results interpretation significantly reduce the accuracy of measurements. These problems require the creation of specialised data processing algorithms capable of ensuring the reliability and reproducibility of the analysis. Objective. The aim of the work was to develop an automated system for analysing spectral data for diagnostics of aircraft engine condition based on the composition of exhaust oils. An important component of the study was the development of spectra normalisation techniques, including the use of reference peaks and principal component methods, to ensure reproducibility of results under conditions of changing measurement parameters. Results. In the course of this research, a comprehensive approach to spectral data analysis has been developed, including an optimised algorithm for spectra preprocessing and an automated system for element identification. Practical significance. The developed methodology is essential for the aviation industry, making it possible to implement operational monitoring of engine condition, to switch from routine maintenance to a predictive model based on real wear of friction units, as well as to reduce operating costs by optimising service intervals and preventing catastrophic failures.