A New Real-Time Reconfiguration Approach Based on Neural Network in Partial Shading for PV Arrays

Abstract

Reconfiguration process in photovoltaic (PV) arrays is very important to improve power-voltage characteristics of the system. In this paper, a new reconfiguration method based on neural network is proposed for PV arrays under partial shadow conditions. A new connection control algorithm based on artificial neural network is presented by the proposed method. This method includes fixed part and adaptive part and uses short circuit currents of PV panel group in every rows of adaptive and fixed part in array. A neural network used for reconfiguration strategy finds new configuration scheme of PV array. Then, adaptive parts are connected to rows of fixed part according to this configuration with switching matrix. Proposed approach has been verified with experimental results obtained using Beagle Board XM microprocessor board in real time for 3x4 array. As shown in results, many contributions such as an improvement in the output power of the PV array, an efficient reconfiguration strategy, real-time applicability, easy measurable parameters, and independence from panel types have been obtained with proposed method.