A MODIFIED FIBONACCI SEARCH ALGORITHM FOR MAXIMUM POWER POINT TRACKING OF PHOTOVOLTAIC SYSTEM UNDER PARTIAL SHADING CONDITIONS Akshaya K. Pati and Nirod C. Sahoo School of Electrical Sciences Indian Institute of Technology, Bhubaneswar-751013, Odisha, India akshaya198310@gmail.com, ncsahoo@iitbbs.ac.in ABSTRACT The performance of Photovoltaic (PV) power system mainly depends on solar insolation and temperature. However, sometimes the performances of the PV systems are affected by partial shading of solar module due to cloud, dust, trees and some other nearby shading objects. During partial shading conditions, the PV characteristics exhibit local and global power maxima. Conventional maximum power point tracking (MPPT) algorithms fail to track the global maximum power point under above operating scenario. A new MPPT algorithm based on two important factors, i.e., study of power versus voltage characteristics of the PV array during partial shading condition and modified Fibonacci search algorithm, is proposed in this paper. The proposed algorithm is better than the conventional Fibonacci search algorithm in terms of power tracking capability of the PV system. Thus, the present proposition is more suitable for both uniform insolation and partial shading conditions. In this study, the PV system is connected with a DC motor and a battery through a bidirectional power converter. Modelling and simulation of the proposed algorithm are implemented on MATLAB/SIMULINK platform and the results are analyzed for validity of the proposed algorithm. KEY WORDS Fibonacci Search Algorithm, MPPT, Partial Shading, Photovoltaic rray A 1. Introduction The global energy demand is increasing day-by-day due to increase in population and urbanization. To meet the increasing energy demand, renewable energy resources are the only alternatives to the conventional resources such as coal, petroleum, natural gas etc. Among the available renewable resources, photovoltaic (PV) system is the most acceptable one, because it is pollution free, requires less maintenance and eco-friendly. The basic component of a PV system is a PV cell. A number of PV cells are connected to form a PV module. The PV cell is a semiconductor diode consisting of a p n  junction which is exposed to solar insolation. Hence, PV cells exhibit nonlinear characteristics with respect to voltage and current output. The major drawbacks of PV system are its nonlinear characteristics and low energy conversion efficiency. The power voltage ( P V  ) characteristics curve of a PV system under uniform insolation has a single peak, at which maximum power can be drawn from the PV system. To operate the PV system at that particular point, maximum power point tracking (MPPT) controller is required. Solar insolation varies with the time of a day, which changes the corresponding generation of power. Maximum power transfer occurs when PV source impedance matches with the load impedance irrespective of solar insolation. To attain maximum power, many algorithms have been proposed in the literature. The detailed overview of the MPPT algorithms has already been discussed and the comparisons are presented in [1]. Among available algorithms, the Perturb & Observation (P&O) and Incremental Conductance (IC) algorithms are most commonly used. The P&O algorithm is based on the principle that perturbation in the operating voltage of a PV array corresponds to change in power generation. If the given perturbation leads to an increase (decrease), the subsequent perturbation is generated in same (opposite) direction of operating voltage. In IC algorithm, the ratio of change in power to change in voltage is zero at maximum power point (MPP). These algorithms have difficulties to find MPP, under partial shading conditions (PSC). A modified hill climbing algorithm and the properties of P V  characteristics have been tested for MPPT under partial shading conditions [2]. Recently, many algorithms based on evolutionary computational techniques such as Genetic Algorithm (GA) [3], Differential Evolution (DE) [4], Particle Swarm Optimization (PSO) [5], and Ant Colony Optimization (ACO) [6] have been proposed. These methods have been tested for both partial shading and uniform insolation conditions. As these algorithms are complex in nature, the application of such algorithms in real time leads to a higher system complexity. Some algorithms are also based on intelligent techniques such as Fuzzy Logic (FL) [7] and Artificial Neural Network (ANN) [8]. These algorithms work under both partial shading and uniform insolation conditions. For the implementation of FL, many fuzzy rules are required; whereas in case of ANN, many training data are necessary. So both the methods are practically not very much suitable for the implementation of MPPT. Similarly Fibonacci search algorithm also does not give guarantee to track the global maximum power for all possible shading patterns [9], [10]. Proceedings of the IASTED International Conference Power and Energy (PE 2013) November 11 - 13, 2013 Marina del Rey, USA DOI: 10.2316/P.2013.806-030 62