A sparse FP-VAR model for representing multichannel non-stationary baseline vibration signals from a gearbox

Abstract

Time series model-based approaches are widely employed for condition monitoring of gearboxes. These approaches rely on accurate time series modeling of the baseline vibration signals. This paper proposes a sparse functional pooled-vector auto-regression (FP-VAR) model for representing multichannel non-stationary baseline vibration signals from a gearbox. A model-based fault detection scheme is also presented. The proposed sparse FP-VAR model combines the advantages of the sparse time series modeling and utilization of multichannel vibration signals, where the former has shown to have higher modeling accuracy than conventional non-sparse time series modeling, and the latter enables the removal of the correlated random noise between channels. Simulation results have demonstrated that the sparse FP-VAR model has higher modeling accuracy than the sparse FP-AR and conventional non-sparse FP-VAR models. Subsequently, the model-based fault detection scheme has a higher fault detection rate than using the other two models.