Design of algorithms for predicting the technical condition of electromechanical drives

Forecasting the technical condition of electromechanical drives requires the timely detection of faults whose progression may lead to accidents or the inability of a system to fulfill its intended function. This paper proposes a methodology for designing algorithms to predict the technical condition of electromechanical drives, integrating mathematical modeling techniques with regression, spectral, and correlation analyses. The novelty of this work lies in the adaptation and integration of established stationary signal processing methods, tailored to the specific characteristics of diagnostic objects comprising interacting electrical and mechanical subsystems. The presented results are based on data generated using a developed mathematical model that accounts for typical mechanical faults. To improve the accuracy of trendbased models, diagnostic features were identified and extracted. Relationships were established between these features and parameters describing the degree of degradation associated with the consumption of the drive's service life. Forecast accuracy of technical condition parameters and model complexity, defined by the number of diagnostic features employed, were adopted as objective functions. An example is provided illustrating the design and application of a fault prediction algorithm for the electromechanical drive of a fixed-wing unmanned aerial vehicle, addressing faults associated with variations in backlash, viscous friction, and dry friction.