Spectral Fault Receptive Fields: A MATLAB Toolbox for Health Monitoring of Rotating Machinery

Automatic health assessment of rotating machinery benefits from conceptual and computational models that discern degradation state and trend of components and systems from operational data. Spectral Fault Receptive Fields (SFRFs) have shown promise in capturing degradation trends in rolling bearing vibration data when tuned to optimize condition monitoring and remaining useful life (RUL) estimation tasks. In this study, we refine and extend the SFRF conceptual model by introducing super-Gaussian spectral masks, generalizing response descriptors to Receptive Field Response Functions (RFRFs), and extending the difference-of-Gaussians (DoG) model used for condition indicator (CI) computation to arbitrary contrast mappings. An open-source MATLAB toolbox is engineered around the perception model, providing a computational engine for research and development of prognostics and health management (PHM) pipelines. We showcase the flexibility of the proposed generalizations by computing different CIs, including fractional-order kinetic integrals and entropy channels, that capture degradation trends while exhibiting distinctive, sensitivity and smoothness properties. Furthermore, the scalability of the toolbox is benchmarked and confirmed on the run-to-failure XJTU-SY bearing dataset.