A new method to analyse temperature measurements and identify failure location in gear transmissions

Abstract

Temperature is a relevant indicator for gearbox health monitoring. Especially in an industrial context where the use of thermocouples and low frequency acquisition system can make this solution simple and cost effective compared to vibration or acoustic ones. Yet, this approach is less investigated in the scientific literature, compared to the dynamic ones. Consequently, this technique lacks specific indicator that can help the diagnostic of a faulty gearbox. In this study, a new indicator that can help locate failure in gearboxes is proposed. It is based on temperature measurements and model analyses through a novel methodology. It consists in: (i) the development of a thermal model of the system under consideration using the thermal network approach; (ii) the use of this model to simulate the faulted system thermal behaviour; (iii) the estimation of a novel fault signature derived from the simulations. Different fault locations are estimated to generate a database of possible defects; (iv) the estimation of a thermal signature for the faulty mechanical system and (v) the comparison of both the experimental and simulated signatures to identify defect localisation. The proposed indicator appears relevant to identify gear failure location in gearbox. For two type of gear defects, the thermal signature estimated from the measurements matches the theoretical thermal signature from the numerical model. The proposed thermal signature approach can be used as a powerful indicator to be used with classical diagnostic methods to identify defect location in geared transmissions.