In this article, the main methods and algorithms for spectral data analysis for vibroacoustic diagnostics of diesel-generator sets at nuclear power plants are considered. To collect the diagnostic data, an experimental setup was developed, thanks to which the sound signals of the diesel generator were obtained under various operating conditions. The recording and processing of signals was carried out using the application package and MATLAB programming language. The article describes the application of correlation and spectral analysis for data processing and analysis. Also, the authors apply regression analysis to find the dependence of the speed of the diesel engine on the frequency of acoustic oscillations. The prediction of the number of revolutions from the frequency of sound vibrations makes it possible in the future to build a more accurate mathematical model of engine operation, and also to find diagnostic features for detecting malfunctions and anomalies in the operation of a diesel generator.