Machine Learning-Based Condition Monitoring for PV Systems: State of the Art and Future Prospects

Tarek Berghout,Mohamed Benbouzid,Toufik Bentrcia,Leïla-Hayet Mouss,Siniša Djurović,Xiandong Ma

doi:10.3390/en14196316

Abstract

To ensure the continuity of electric power generation for photovoltaic systems, condition monitoring frameworks are subject to major enhancements. The continuous uniform delivery of electric power depends entirely on a well-designed condition maintenance program. A just-in-time task to deal with several naturally occurring faults can be correctly undertaken via the cooperation of effective detection, diagnosis, and prognostic analyses. Therefore, the present review first outlines different failure modes to which all photovoltaic systems are subjected, in addition to the essential integrated detection methods and technologies. Then, data-driven paradigms, and their contribution to solving this prediction problem, are also explored. Accordingly, this review primarily investigates the different learning architectures used (i.e., ordinary, hybrid, and ensemble) in relation to their learning frameworks (i.e., traditional and deep learning). It also discusses the extension of machine learning to knowledge-driven approaches, including generative models such as adversarial networks and transfer learning. Finally, this review provides insights into different works to highlight various operating conditions and different numbers and types of failures, and provides links to some publicly available datasets in the field. The clear organization of the abundant information on this subject may result in rigorous guidelines for the trends adopted in the future.

Highlights

We focus on failures common to the varying aforementioned PV systems types
By interpreting the findings provided by this literature review, it can be deduced that the three main machine learning (ML) categories, namely, traditional ML tools, deep learning, and knowledgedriven learning with several learning paradigms, have numerous important characteristics, as indicated below:
All of the ML models are subject to the MPPT conditions, excluding the work of Bakdi et al [18], which addressed both MPPT and Intermediate Power Point Tracking (IPPT); Most of the ML models depend on data generated from simulation models; A limited number of fault classes are considered, with the exception of a number of works, such as Momeni et al [60], Akram et al [102], Liu et al [76], and Bakdi et al [18], in which 10 or more faults are considered; Traditional ML models usually have I-V/P-V signals as inputs; Deep Learning (DL) and knowledge-driven models are generally used to manipulate all kinds of images

Summary

Introduction

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. The consumption of renewable energy has received increased acceptance in a wide range of sectors due to the clear advantages it offers. The inherent environmentally friendly power generation process has stimulated global interest in the development of renewable energies as the only solution for a cleaner environment and the satisfaction of increased energy demands [1,2]. In 2020, the statistical studies of the “World Energy Data” reports stated that, in regard to the global consumption of energy, renewable energies account for

Objectives

Findings

Discussion

Conclusion