A novel MPPT controller using Neural Network and Gain-Scheduled PI for Solar PV system under rapidly varying environmental condition

Article type: Research Article

Authors: Ulaganathan, M.S.^{; *} | Devaraj, D.

Affiliations: Department of Electrical and Electronics Engineering, Kalasalingam Academy of Research and Education, Krishankoil Tamilnadu, India

Correspondence: [*] Corresponding author. Ulaganathan Meenakshi Sundaram, Department of Electrical and Electronics Engineering, Kalasalingam Academy of Research and Education, Krishankoil-626126 Tamilnadu, India. E-mail: ulagu.er@gmail.com

Abstract: The Maximum Power Point Tracking (MPPT) controller plays a vital role in maximizing power output from the solar Photovoltaic (PV) sources. The tracking efficiency of the MPPT controller is affected by a rapidly varying environmental condition. This paper presents a novel MPPT controller for standalone PV system based on a Neural Network (NN) and Gain-scheduled Proportional Integral (GS-PI) controller to track the fast-changing Maximum Power Point (MPP).The NN model is trained to predict the operating parameters of the PV array at which maximum power is generated. The gain scheduled PI controller parameters are optimally tuned with Real-coded Genetic Algorithm (RGA) to improve the controller performance. The developed MPPT controller is used to control the power converter in the solar PV system. The PV array along with the control scheme is developed using LabVIEW and Multisim environment. Further, the performance of the developed control strategy is experimentally validated with solar PV emulator and DC-DC boost converter under the varying irradiation conditions. The tracking performance of the developed MPPT controller is compared with the modified Perturb and Observe and NN+PI controller based MPPT controller. The experimental results reveal that the tracking performance of the developed MPPT technique is much improved and more accurate in MPP tracking.

Keywords: Neural networks, Real Coded Genetic Algorithm, gain scheduled PI controller, P&O algorithm

DOI: 10.3233/JIFS-182556

Journal: Journal of Intelligent & Fuzzy Systems, vol. 37, no. 1, pp. 1085-1098, 2019