In photovoltaic (PV) systems, PV modules are usually cascaded in series to generate higher output power, and the P-V characteristic under partial shading condition (PSC) presents multiple peaks with nonlinearity. Several traditional tracking methods such as Perturbation and Observation (PO) and Incremental Conductance (InC) cannot fulfill the global maximum power point (GMPP) tracing under PSC. Moreover, even the global maximum power point tracking (GMPPT) methods proposed recently such as Particle Swarm Optimization (PSO) may cost lots of time to locate the GMPP of the PV array. As a result, to ensure the PV array under PSC can still generate maximum power efficiently, this paper proposed the Adaptive Neuro Fuzzy Inference System (ANFIS)-based MPPT method, which apples ANFIS and the concept of particles to the MPPT without using expensive irradiance sensors. Instead of searching for the GMPP online, the proposed ANFIS-based MPPT use the training data pairs generated offline to economize the searching time and finish the GMPP tracing simultaneously with zero oscillation under simulation. In addition, it is shown that the tracking time and efficiency of the proposed MPPT method can reach average 6.04 steps and 99.85%, respectively, under simulation, and verifies the fast tracking speed under implementation. Compared to other methods, the proposed MPPT have faster tracking with higher efficiency and zero oscillation around the GMPP.