Failure Prognosis Based on Relevant Measurements Identification and Data-Driven Trend-Modeling: Application to a Fuel Cell System

Abstract

Fuel cells are key elements in the transition to clean energy thanks to their neutral carbon footprint, as well as their great capacity for the generation of electrical energy by oxidizing hydrogen. However, these cells operate under straining conditions of temperature and humidity that favor degradation processes. Furthermore, the presence of hydrogen—a highly flammable gas—renders the assessment of their degradations and failures crucial to the safety of their use. This paper deals with the combination of physical knowledge and data analysis for the identification of health indices (HIs) that carry information on the degradation process of fuel cells. Then, a failure prognosis method is achieved through the trend modeling of the identified HI using a data-driven and updatable state model. Finally, the remaining useful life is predicted through the calculation of the times of crossing of the predicted HI and the failure threshold. The trend model is updated when the estimation error between the predicted and measured values of the HI surpasses a predefined threshold to guarantee the adaptation of the prediction to changes in the operating conditions of the system. The effectiveness of the proposed approach is demonstrated by evaluating the obtained experimental results with prognosis performance analysis techniques.

Fractional Order PID Design for a Proton Exchange Membrane Fuel Cell System Using an Extended Grey Wolf Optimizer

“…The extended GWO is the same as the original, where the difference is adding three parameters (α E , β E and δ E ) called the emphasis coefficients to the updated position of Equation (32). Therefore, the extended, updated position can be expressed as Equation (33) [48,49]:…”

“…R is the universal gas constant (8.31451 J•kg −1 •K −1 ). For every hydrogen mole, two electrons pass by the external electrical circuit, and the electrical work is equal to the change in Gibbs free energy if the system has no lossless, the electrical work performed is given in Equation ( 7) [32]:…”

A Study of the Impact of Predictive Maintenance Parameters on the Improvment of System Monitoring

“…System condition monitoring, fault diagnosis, and fault prognosis are the pillars of predictive maintenance. Several approaches for fault diagnosis and prognosis are described in the literature [8][9][10]. These approaches can be data-driven approaches or physical-based model approaches.…”

“…These approaches can be data-driven approaches or physical-based model approaches. Data-driven approaches use real data on the health state of the system to predict the failure of the system while model-based approaches use physical models in the form of physical equations when enough knowledge on the real physical dynamics of a system are available [8,9]. In the existing literature, the health state of the system can be evaluated using different measures [5,[11][12][13][14][15].…”

A Multi-Stage Fault Diagnosis Method for Proton Exchange Membrane Fuel Cell Based on Support Vector Machine with Binary Tree

“…Inspired by the characteristics of the model-based method and data-driven method, these two methods can be combined, namely the hybrid method [22]. Recently, Djeziri et al [32] proposed a hybrid method that combines a prior physical model and data-driven updated kernel for fuel cell failure diagnostics, where the updated kernel is enabled when the estimation error between the predicted and measured values of stack voltage surpasses a predefined threshold. Similarly, Pan et al [33] combined a model-based adaptive Kalman filter and data-driven NARX neural network to realize fuel cell failure diagnostics.…”