Hill Climbing Artificial Electric Field Algorithm for Maximum Power Point Tracking of Photovoltaics

Abstract

In this paper, maximum power point tracking (MPPT) of a photovoltaic (PV) system is performed under partial shading conditions (PSCs) using a hill climbing (HC)–artificial electric field algorithm (AEFA) considering a DC/DC buck converter. The AEFA is inspired by Coulomb’s law of electrostatic force and has a high speed and optimization accuracy. Because the traditional HC method cannot perform global search tracking and instead performs local search tracking, the AEFA is used for a global search in the proposed HC-AEFA. The critical advantage of the HC-AEFA is that it is desirable performing local and global searches. The proposed hybrid method is implemented to derive an MPP by tuning the converter duty cycle, considering the objective function for maximizing the PV system extracted power. Its capability is evaluated and compared with well-known particle swarm optimization (PSO), considering standards, PSCs, and radiation changes conditions. The tracking efficiency for the most challenging shading pattern (third pattern) using the HC-AEFA, HC, AEFA and PSO is obtained at 99.93, 90.35, 98.85, 99.80%, respectively. The analysis of the population-based optimization process for different algorithms proved the HC-AEFA faster convergence at lower iterations than the other methods. So, the superiority of the proposed HC-AEFA subjected to different patterns is confirmed with higher tracking efficiency and global power peak, fewer fluctuations, higher convergence speed, and higher dynamic and Static-efficiency compared to the other methods.

A Comprehensive Investigation into the Application of Convolutional Neural Networks (ConvNet/CNN) in Smart Grids

“…It is essential for developing oil and gas fields [9]. An artificial electric field algorithm that climbs hills can be used to track a photovoltaic system's greatest Power [10]. Powerful models like AlexNet may produce results with high accuracy on even the most challenging datasets [11].…”