A Monte-Carlo approach for prognostics of clogging process in HVAC filters using a hybrid strategy a real case study in railway systems

Abstract

This paper discusses an innovative prognostics-oriented method to non-invasively monitor and predict the health of air filters for a HVAC (Heating, Ventilation and Air Conditioning) unit in railway systems. The proposed method permits not only to assess the actual condition of the filter but also to perform prognostics, with predictions about the remaining useful life of the component. The approach described in the present paper consists of a combination of physics-based digital models and operational data acquired during commercial service from the system in the field. First, an appropriate health indicator is calculated based on raw measurements retrieved from the field. The health indicator is computed by taking into account the influence of context factors on the operational behavior of the HVAC unit. Subsequently, learning algorithms are employed to calibrate a carefully designed degradation model of the filter. By using Monte Carlo-based simulation techniques, the remaining useful life (RUL) of the filter is estimated in a probabilistic framework. The proposed approach has been tested on real HVAC unit mounted on a representative Alstom tramway in revenue service. The results of the field study reveal that thanks to the health indicator-based maintenance strategy, the useful life of the filters can be significantly extended. Further, application of the novel physics-based hybrid prognostics model allows to accurately estimate the probabilistic characteristics of the remaining useful life of the target component.