Mismatching losses reduction of photovoltaic (PV) array has been intensively discussed through the increasing penetration of residential and commercial PV systems. Many causes of mismatching losses have been identified and plenty of proposed methods to solve this problem have been recently proposed. This paper deals with reducing method of mismatching losses due to the non-uniform irradiance conditions. It is well-known that a certain number of multiple peaks occur on the power-voltage curve as the number of PV modules in one-string increases under non-uniform operating conditions. Since the conventional control method only drives the operating points of PV system to the local maxima close to open circuit voltage, only small portion of power can be extracted from the PV system. In this study, a radial basis function neural network (RBF-ANN) based intelligent control method is utilized to map the global operating voltage and non-irradiance operating condition in string and central based MPPT systems. The proposed method has been tested on 10 x 3 (2.2 kW), 15 x 3 (2.5 kW) and 20 x 3 (3.3 kW) of series-parallel PV array configuration under random-shaded and continuous-shaded patterns. The proposed method is compared with the ideal case and conventional method through a simple power-voltage curve of PV arrays. The simulation results show that there are significant increases of about 30-60% of the extracted power in one operating condition when the proposed method is able to shift the operating voltage of modules to their optimum voltages. (C) 2012 Elsevier Ltd. All rights reserved.