Abstract
Recent research has focused on photovoltaic (PV) systems due to their important properties. The efficiency of the PV system can be enhanced by many Maximum Power Point Tracking (MPPT) algorithms proposals. MPPT algorithms are used to achieve maximum PV output power by optimizing the duty cycle of the DC-DC buck/boost converter. This paper introduces an RNA algorithm as an efficient MPPT algorithm for the photovoltaic system. This proposed RNA algorithm consists of two main segments. The first segment is an artificial neural network for generating reference power. The second segment is a proposed Recursive Bit Assignment (RBA) network to allow variable step size of the boost converter duty cycle. The instant PV power adopts the RBA network to produce the variable duty cycle increment value. Additionally, the neural network is implemented in such a way to obtain the best performance. Many simulation results using MATLAB to test the system performance are presented. The performance characteristics of the photovoltaic system with variable irradiance and variable temperature are simulated. From results, the proposed RNA algorithm achieves fast tracking time, high energy efficiency, true maximum power point and acceptable ripple. Additionally, comparisons between the RNA algorithm and other related algorithms such as Perturb and Observe, the Neural Network and the Adaptive Neural Inference System Algorithms are executed. The proposed RNA algorithm achieves the best performance in all case studies such as; irradiance profile variation, severe temperature and irradiance diversions, and partial shading conditions. Besides, the experimental circuit of the PV system is also presented.