Proton exchange membrane fuel cells (PEMFCs) have become the center of attention for energy conversion in many areas such as automotive industry, where they confront a high dynamic behavior resulting in their characteristics variation. In order to ensure appropriate modeling of PEMFCs, accurate parameters estimation is in demand. However, parameter estimation of PEMFC models is highly challenging due to their multivariate, nonlinear, and complex essence. This paper comprehensively reviews PEMFC models parameters estimation methods with a specific view to online identification algorithms, which are considered as the basis of global energy management strategy design, to estimate the linear and nonlinear parameters of a PEMFC model in real time. In this respect, different PEMFC models with different categories and purposes are discussed first. Subsequently, a thorough investigation of PEMFC parameter estimation methods in the literature is conducted in terms of applicability. Three potential algorithms for online applications, Recursive Least Square (RLS), Kalman filter, and extended Kalman filter (EKF), which has escaped the attention in previous works, have been then utilized to identify the parameters of two well-known semi-empirical models in the literature, Squadrito et. al and Amphlett et. al. Ultimately, the achieved results and future challenges are discussed.

No abstract

Cold start of proton exchange membrane fuel cells (PEMFCs) at sub-zero temperatures is perceived as one of the obstacles in their commercialization way in automotive application. This paper proposes a novel internal-based adaptive strategy for the cold start of PEMFC to control its operating current in real time in a way to maximize the generated heat flux and electrical power in a short time span. In this respect, firstly, an online parameter identification method is integrated into a semi-empirical model to cope with the PEMFC performances drifts during cold start. Subsequently, an optimization algorithm is launched to find the best operating points from the updated model. Finally, the determined operating point, which is the current corresponding to the maximum power, is applied to PEMFC to achieve a rapid cold start. It should be noted that the utilization of adaptive filters has escaped the attention of previous PEMFC cold start studies. The ultimate results of the proposed strategy are experimentally validated and compared to the most commonly used cold start strategies based on Potentiostatic and Galvanostatic modes. The experimental outcomes of the comparative study indicate the striking superior performance of the proposed strategy in terms of heating time and energy requirement.

Output power of a fuel cell (FC) stack can be controlled through operating parameters (current, temperature, etc.) and is impacted by ageing and degradation. However, designing a complete FC model which includes the whole physical phenomena is very difficult owing to its multivariate nature. Hence, online identification of a FC model, which serves as a basis for global energy management of a fuel cell vehicle (FCV), is considerably important. In this paper, two well‐known recursive algorithms are compared for online estimation of a multi‐input semi‐empirical FC model parameters. In this respect, firstly, a semi‐empirical FC model is selected to reach a satisfactory compromise between computational time and physical meaning. Subsequently, the algorithms are explained and implemented to identify the parameters of the model. Finally, experimental results achieved by the algorithms are discussed and their robustness is investigated. The ultimate results of this experimental study indicate that the employed algorithms are highly applicable in coping with the problem of FC output power alteration, due to the uncertainties caused by degradation and operation condition variations, and these results can be utilized for designing a global energy management strategy in a FCV.

No abstract