Improved evolutionary programming maximum power point tracking algorithm for photovoltaic systems
Loading...
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Teknologi Malaysia
Abstract
The most cost-effective approach to improving the overall efficiency of a photovoltaic (PV) system is to enhance its maximum power point tracking (MPPT) algorithm. Under uniform irradiance, conventional MPPTs such as perturb and observe (P&O), hill climbing (HC), and incremental conductance (InCon) are preferable due to their simplicity and ease of implementation. However, under partial shading conditions (PSC), the performance of these MPPTs deteriorates significantly, owing to their inability to distinguish between the global maximum power point (GMPP) and the local maximum power point (LMPP) peaks. Although soft computing (SC) MPPT algorithms, e.g., genetic algorithm (GA), particle swarm optimization (PSO), differential evolution (DE) and evolutionary programming (EP) have been proposed to address this limitation, they require optimal control parameter values, which need to be tuned by an experienced designer. Furthermore, the random nature of these algorithms—particularly during initialization, results in unpredictable GMPP tracking performance. Most importantly, since SC algorithms are iterative by nature, their dynamic is not sufficient to cope with fast irradiance changes, such as the ones required by the EN 50530 standard. To address these problems, an enhanced MPPT algorithm, known as the improved evolutionary programming (IEP) is proposed. It incorporates two new features that do not exist in its predecessor, i.e., conventional evolutionary programming (CEP). The first is the deterministic initialization method (DIM), and the second is the intelligent mechanism for detection of various environmental changes known as real-time change detection (RTCD). The function of DIM is to minimize the search unpredictability due to the random initialization, while the RTCD improves the tracking performance under PSC and gradual changes in irradiance. To assess the feasibility of the proposed MPPT, a stand-alone PV system with a DC–DC boost converter was simulated in MATLAB/Simulink. For 1000 independent optimization runs, the IEP (with DIM) achieved perfect (100%) GMPP tracking; this is 23.43% higher compared to the CEP, which used the conventional pseudo-random number generator for its initialization. Additionally, the IEP outperformed the CEP by 1.5 seconds when tracking the GMPP under PSC. Since the RTCD is based on dynamic perturbation step-sizing, the IEP can closely track the GMPP contour during the EN 50350 test. Hence, power loss (due to inaccurate tracking) is minimized. For the tests involving the lowest and highest slopes of the EN 50530 ramps (Case B), the MPPT efficiency of the IEP was 2.71% and 33.51% higher than the conventional HC, respectively. To validate the findings from the simulation, the algorithm was programmed on the dSPACE DS1104 DSP controller board. A 300 W experimental set-up that comprises a PV array simulator (PVAS), a DC–DC boost converter, sensors, and load banks was used as a testbed for the MPPT algorithms. To ensure consistency, the system configurations and test conditions were maintained as in simulation. For the EN50530, the measurements showed that the system that employs IEP exhibits MPPT efficiency of 99.64% (or 3.24% higher) than the HC algorithm and 98.64% (which was 32.99% higher than HC) for the lowest and highest ramp, respectively. Overall, the experimental findings closely match the simulated results, validating the efficacy and practicality of the proposed IEP MPPT algorithm for stand-alone PV systems.
Description
Thesis (PhD. (Electrical Engineering))
Keywords
Photovoltaic power systems—Research, Evolutionary programming (Computer science), Power resources—Research