Development of entropy based searching algorithms for reconfiguration

Photovoltaic (PV) panels are used for the conversion of solar energy which is the most important energy source into electrical energy. Nowadays, PV arrays and even PV power plants are created by connecting these panels in various ways to utilize energy from solar energy. However, the efficiency of PV systems is based directly on irradiance level, temperature and connection structure of panels. Especially, the maximum power point obtained from PV arrays is greatly affected when partial shading conditions occur on PV system. Various tracking algorithms are used to run the PV arrays with the maximum power point. On the other hand, the maximum power point can be obtained from the PV array can be increased by changing the connection structure of panels. For this process, reconfiguration algorithms are used. The reconfiguration algorithms in PV systems is aimed to change panel connections in real time according to partial shading conditions to obtain more power from system. In other words, reconfiguration is searching and finding the most appropriate connection structure between possible connections of panels. This process is performed by changing all or some of these panel connections with certain principles. The methods in literature, are often used algorithms which are difficult to apply such as bubble sort. For an efficient reconfiguration algorithm, the connection structure which can obtain the maximum power point should be found in real time. In this study, a new reconfiguration approach is proposed to obtain maksimum power from PV arrays with partial shading. Entropy based genetic algorithm is used to find the optimal connection structure among millions even and billions possibility. This new method is presented an efficient and high accuracy searching method. A PV system that consist of adaptive panel and fixed panel will be had a different configuration with same PV panels by connecting modules of adaptive panel to modules of fixed panel. A switching matrix circuit is used for this. Thus, the structure that obtain more power under the same conditions is provided. Two different measurement information have been used as input parameters for developed algorithms: short circuit currents and shading ratio values obtaing by image processing. For experiments, a PV system which consist of three adaptive PV panels and nine fixed PV panels is created and the method of any size to be used in real time in the PV array is illustrated by the effective results. Obtained comparative power-voltage graphs are demonstrated the performance of the proposed method. As a result, the scope of this study, a new reconfiguration method which used entropy based genetic algorithm and can be applied in reat time is proposed to obtain more power under partial shading for every size of PV arrays. Also, this method is verified by simulations and experiments. The techniques developed in this study have been supported research TUBITAK 1001 project with No: 112E214.