Accurate PEM fuel cells parameters estimation using hybrid artificial bee colony differential evolution shuffled complex optimizer

Abstract

Summary To get efficient current/voltage (I/V) polarization curves, the parameter identification of the proton exchange membrane (PEM) fuel cells (FCs) model based on experimental datasets and meta‐heuristic algorithms remains an active research field during the past few years. Meanwhile, estimating those parameters accurately is still a challenge. In this work, a new hybridized approach is presented to identify the PEMFC model parameters denominated the artificial bee colony differential evolution shuffled complex (ABCDESC) optimizer. In the proposed algorithm, the double execution of the probabilities evaluation and the selection strategy of ABC optimizer enables to have better exploitation phase without getting stuck into the local optimum. The sum of squared errors (SSE)‐based objective function is used to perform the optimization as it is the well used in the literature. In order to assess this new hybridized approach, a comparative study with the latest published techniques is carried out using six typical test benchmarking PEMFCs modules extensively utilized in the literature. In this context, the reached SSE values and the standard deviations among other challenging methodologies are very competitive with the best convergence speed and reduced number of function evaluations. The ABCDESC algorithm reaches a standard deviation/CPU run time (STD/CRT) of 8.4690e−16/0.298 second, 1.3275e−16/0.290 second, 1.9721e−14/0.356 second, 4.3889e−15/0.343 second, 1.4388e−12/0.49 second, and 1.6398e−17/0.326 second for 250 W, BCS 500 W, NedStack PS6, Ballard Mark V, Horizon H‐12, and Modular SR‐12 stacks; respectively. The comparison results indicate the successful use of the proposed ABCDESC optimizer to characterize the PEMFC model accurately.

Control‐oriented estimation of the exchange current density in PEM fuel cells via stochastic filtering

“…Then N samples are drawn from Gaussian distributions of cell temperature and oxygen partial pressure. Next, N exchange current density particles are created from the temperature and oxygen partial pressures samples through static equation Equation (6). Then, at t = 1 s, every i 0 particle is updated through the state transition Equation ( 9) and for each of these particles a corresponding output voltage is computed.…”

“…Among the most recent algorithms used for fuel cell parameter estimation are bio-inspired optimization algorithms such as hybrid artificial bee colony differential optimizers, 6 genetic algorithms, 7 manta ray foraging optimizers, 8 improved chaotic MayFly optimization, 9 hybrid interior search algorithm, 10 modified gorilla troop optimizer, 11 bi-subgroup optimization, 12 adaptive sparrow search, 13 chaos embedded particle swarm optimization, 14 improved monarch butterfly optimizer, 15 water strider algorithm. 16 These bio-inspired optimization algorithms can find the global minimum in the parameter space and are robust in dealing with the nonlinearities in the fuel cell model.…”

Optimal identification of model parameters for PEMFCs using neoteric metaheuristic methods

“…Then updating the best position is considered according to the following equation: (33) where gbest score denotes the best score of the objective function corresponding position gbest pos . The updating of the coots and leaders' positions can take one of four ways.…”

“…−gbest pos ifR 4 > 0.5 (40) where gbest pos denotes the best position ever found, R 3 and R 4 denote random numbers among [0, 1], and B = 2 − L Iter . The value of the objective function after each updating relating to the coots and leaders' positions can be calculated as Equations ( 30)- (33).…”

“…Another optimization algorithm of Equilibrium optimizer had been applied for solving the same optimization problem of the PEMFC [32]. The results of estimation process have been improved by applying a hybrid optimization algorithm considering artificial bee colony and differential evolution [33]. Coyote-based optimization technique has been applied to enhance the estimation process of the PEMFC parameters.…”