This paper presents an intelligent maximum power point tracker which combines the conventional perturbation and observation method and an artificial neural network. The proposed controller overcomes the drawback of the conventional perturbation and observation under partial shading conditions. A MATLAB model for a photovoltaic system is developed so as to simulate system’s performance and characteristic under partial shading condition. This model includes models for PV array, DC–DC boost converter as well as the proposed algorithm. Results show that the proposed algorithm is able to track the global maxima accurately regardless the number of surrounding local maxima points with an efficiency of 95%.