The classical algorithms for maximum power point tracking ensure proper operation under uniform irradiance conditions. However, when photovoltaic (PV) array is subject to partial shading conditions (PSC), several local maxima appear on the P-V characteristics curve of the PV array which are due to the use of the bypass diodes to avoid hot spots effect. The appearance of these multiple peaks on the characteristics of PV array makes the tracking more difficult under these conditions and requires the integration of a more efficient power control system which is able to discriminate between local and global maxima to harvest the maximum possible energy and therefore increase the efficiency of overall system. In addition to implementing a global maximum power point tracking strategies, the mismatch losses associated to the shading effect can further be reduced by using alternative PV arrays' configurations such as Total-Cross-Tied (TCT), Bridge Linked (BL) and Honey-Comb (HC). For this purpose, the main aim of this paper is to design an intelligent MPPT controller that allows predicting and extracting the global maximum power point (GMPP) from PV array under partial shading conditions (PSC) whatever is the used configuration or its size. This intelligent MPPT controller is based on adaptive neuro-fuzzy inference system (ANFIS). The adopted ANFIS network has two inputs and one output. The two inputs of the proposed ANFIS consist of voltage and current while, the output is the output power of each configuration. The ANFIS network is trained using the data derived from performances analysis of different PV array configurations. Furthermore, the ANFIS network uses a hybrid learning algorithm that combines the least squares estimator and the gradient method. The Bishop model of a PV module which describes best the solar cell behavior at negative voltages is considered in this paper for modeling the PV arrays, and it is implemented by using the Simulink and SimPower software. The effectiveness of the proposed method is investigated for TCT configuration under partial shading conditions for various shading scenarios and sudden irradiance change. The results show that the proposed algorithm can track the global MPP effectively and is robust to various shading patterns. Simulation results show high tracking performances in terms of efficiency, tracking speed and system stability. The results are also presented for different configurations such as HC, BL and Series-Parallel (SP) to show the ability of the proposed technique to detect the right peak regardless of the used configuration.