Condition monitoring and faults diagnosis of solar photovoltaic arrays using artificial intelligence technique based on bees algorithm

Suliman, Fouad 2024. Condition monitoring and faults diagnosis of solar photovoltaic arrays using artificial intelligence technique based on bees algorithm. PhD Thesis, Cardiff University. Item availability restricted.

	PDF - Accepted Post-Print Version Restricted to Repository staff only until 28 June 2025 due to copyright restrictions. Download (29MB)
	PDF (Cardiff University Electronic Publication Form) - Supplemental Material Restricted to Repository staff only Download (291kB)

Abstract

The global surge in photovoltaic (PV) capacity, especially in grid-connected photovoltaic (GCPV) plants, emphasizes the need for dependable and efficient PV systems. A crucial component of this reliability hinges on the precise detection and analysis of faults within solar PV arrays. The study commenced by employing an optimisation algorithm to determine the ideal unknown parameters for solar cells and modules. This exploration utilised a new optimisation approach called the Bees Algorithm (BA), aiming to minimise discrepancies between simulated and experimental data. The success of this method is demonstrated by a very low error rate root-mean-square error (RMSE). In conjunction with the corresponding BA-based simulated outcomes, an excellent agreement is evident between the experimental and simulation data, with a standard deviation less than 1%, indicating a strong match between model predictions and experimental results. Furthermore, a small-scale PV system was built to enhance the accuracy of machine learning classifiers for PV system. This system, designed to replicate real-world solar complexities, simulated actual challenges by introducing specific faults. The goal was to produce a high-quality, realistic dataset for training machine learning algorithms, ensuring they’re well-prepared for genuine applications. This research also introduces a comprehensive Fault Detection and Diagnosis (FDD) system that employs Support Vector Machines (SVM) and Extreme Gradient Boosting (XGBoost) for solar array fault detection and classification. To enhance its efficacy, the system integrates nature-inspired algorithms, particularly the Bees Algorithm (BA) and Particle Swarm Optimization (PSO), aiming to improve the model’s precision and overall performance. The faults addressed include short circuits line to line (L-L) and open circuits (OC) faults, and standard test conditions have been considered. The evaluation of these fault detection and classification methods utilizes metrics such as accuracy and confusion matrices. Significantly, the output results for BA-XGBoost yielded the highest overall fault detection accuracy, reaching 87.56%. In contrast, BA-SVM registered just 70.79%. This is in comparison to the standard classifiers, XGBoost and SVM, which achieved accuracies of 65.88% and 63.45%, respectively. In conclusion, the Bees Algorithm (BA) had a significant impact on improving the efficiency of the XGBoost classifier, which indicates the possibility of using this technique not only in fault detection within solar energy but also across other domains.

Item Type:	Thesis (PhD)
Date Type:	Completion
Status:	Unpublished
Schools:	Engineering
Uncontrolled Keywords:	1) Photovoltaic systems 2) PV string 3) I-V curve analysis 4) Faults Diagnosis of Solar Photovoltaic Arrays 5) Support Vector Machine (SVM) 6) Extreme Grading Boosting (XGBoost) 7) Bees Algorithm (BA) 8) Particle Swarm Optimization (PSO)
Date of First Compliant Deposit:	28 June 2024
Last Modified:	28 Jun 2024 15:11
URI:	https://orca.cardiff.ac.uk/id/eprint/170159

Actions (repository staff only)

Edit Item