1949-3029 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TSTE.2018.2878826, IEEE Transactions on Sustainable Energy 1 Abstract—Maximum power point tracking (MPPT) algorithms continuously change duty cycle of a power converter to extract maximum power from photovoltaic (PV) panels. In all of MPPT methods, two parameters, i.e. perturbation period (Tp) and amplitude (D) have a great effect on speed and accuracy of MPPT. Optimum value of the perturbation period is equal to the system settling time which is the system model-dependent parameter. Since the system model varies according to the change of irradiance level and temperature, the value of Tp has to be determined online. In this paper, the parametric identification method is adopted to identify the online value of Tp. The proposed method is based on the dichotomous coordinate descent-recursive least squares (DCD-RLS) algorithm and uses an infinite impulse response (IIR) adaptive filter as the system model. Computation of this algorithm is based on an efficient, fixed-point, and iterative approach with no explicit division operations; these features are highly suitable for online applications. As a result, the proposed method compared to previous works leads to more accurate and faster identification of the system settling time. In order to test and validate the proposed method, it has been simulated and implemented to be further validated with experimental data. Index Terms—Photovoltaic Systems, Maximum Power Point Tracking (MPPT), Perturbation Period, System Identification, Dichotomous Coordinate Descent (DCD), Recursive Least Squares (RLS). I. INTRODUCTION Solar energy is one kind of renewable energies that is noteworthy because of its availability, lack of greenhouse gas emission, and low cost. One way to harness the solar energy is to directly convert it into electricity by photovoltaic (PV) systems. Maximum power point tracking (MPPT) techniques are used in PV systems to extract the maximum power from the PV panels. So far, different MPPT algorithms have been presented that they can fall into two categories of the uniform irradiance and the partial shading condition [1]. Perturb and Observe (P&O) [2] and Incremental Conductance (INC) [3] are the famous algorithms that work in the uniform irradiance; however, Particle swarm optimization (PSO) [4] and Cuckoo search (CS) [5] are algorithms which are able to find a maximum power point (MPP) in the partial shading condition as well as in the uniform irradiance. In Perturb and Observe- based MPPT algorithms, like P&O and INC, an operating point oscillates around MPP in steady states and never locates at MPP; in the best situation, these oscillations are three points [6]. Efficiency of MPPT algorithms highly depends on parameters like perturbation amplitude (D) which is a change in the duty cycle of a power converter applied by MPPT; and perturbation period (Tp) which is the time interval between two successive perturbations. The optimum value of these parameters greatly improves speed and accuracy of MPPT algorithms. In steady states (when MPP is reached), D has to be chosen as small as possible in order to decrease oscillations around MPP, so the optimum value of D is determined according to (1) [6]. ∆ > 1  0 √   . ℎ .| ̇ |.  .  + 1   (1) Moreover, the optimum value of Tp is equal to the system settling time (T  since in this way the PV voltage and current have enough time to settle in an acceptable range before the MPPT algorithm applies the next perturbation; so, the MPPT algorithm makes true decisions based on the settled values of the PV voltage and current. Therefore, in this way oscillations around MPP will be regular and three points [6]. The PV settling time and Tp are shown in Fig. 1 simulated for the 120 W PV system through P&O algorithm. Generally, optimization methods of Tp can be classified into fixed [7-15] and variable [16-23] methods. In fixed methods, Tp is selected in the design stage and remains constant during the MPPT function. Tp smaller than the system settling time increases the transient tracking speed while it leads to irregular and more than three-point oscillation around MPP in the steady state; and Tp longer than the system settling time reduces the transient tracking speed while it leads to regular and three-point oscillation around MPP in the steady state. Therefore, in fixed optimization methods, there is a tradeoff between speed and accuracy of the MPPT algorithm. This problem is solved by variable optimization methods. In such methods, in the steady state, Tp is selected equal to the system settling time by means of online system identification techniques to maintain the three- point oscillation; in the transient state (when irradiance level is changed), Tp is a function of D to provide sufficient time based on the value of D. So far, most of papers have focused on optimization of D, while a few have paid attention to optimization of variable Tp. For instance, there are several novel techniques that reduce or completely remove the steady state oscillations, but the point is that such methods rely on the J. Dadkhah 1 , M. Niroomand 1 , Member, IEEE 1 Department of Electrical Engineering, University of Isfahan, Isfahan, Iran. mehdi_niroomand@eng.ui.ac.ir Real-Time MPPT Optimization of PV Systems by Means of DCD-RLS Based Identification Fig. 1. Demonstration of Tp, T   and D in the PV system with P&O algorithm.