The paper introduces the task of parameter extraction of photovoltaic (PV) modules as a nonlinear optimization problem. The concerned parameters are the series resistance, shunt resistance, diode ideality factor, and diode reverse saturation current for both the single- and double-diode models. An error function representing the mismatch between computed and targeted performance is minimized using different versions of the bacterial foraging (BF) algorithm of global search and heuristic optimization. The targeted performance is obtained from the nameplate data of the PV module. Five distinct variations of the BF algorithm are used to solve the problem independently for the single- and double-diode models. The best optimization results are obtained when swarming is eliminated, chemotactic step size is dynamically varied, and global best is preserved, all acting together. Under such conditions, the best global minimum of 0.0028 is reached in an average best time of 94.4 sec for the single-diode model. However, it takes an average of 153 sec to reach the best global minimum of 0.0021 in case of double-diode model. An experimental verification study involves the comparison of computed performance to measurements on an Eclipsall PV module. It is shown that all variants of the BF algorithm could reach equivalent-circuit parameters with accepted accuracy by solving the optimization problem. The good matching between analytical and experimental results indicates the effectiveness of the proposed method and validates research findings. (C) 2016 Elsevier Ltd. All rights reserved.