Diagnosis for systems with multi-component wear interactions

Predicting remaining useful lifetime is key to improving operational efficiency, and increasing the reliability of machinery. This paper presents an approach for increasing the accuracy of diagnostics of systems with multiple components. We first discuss a degradation model for systems, where the deterioration process of a component is influenced by the state of deterioration of the other components. Then, we present a gearbox accelerated life testing platform, where we collect vibration data from accelerometers mounted over each gear supporting shaft. Next, we provide our methodology of extracting health indicators, from systems with such complex wear interactions and noisy signals, using data pre-processing for denoising, and Short Time Fourier Transform (STFT). Finally, by using the approach introduced in this paper and the experimental results, we demonstrate the need for monitoring and modelling wear interdependencies in complex systems, over the conventional condition monitoring of components separately.