A Fault Detection Scheme Utilizing Convolutional Neural Network for PV Solar Panels with High Accuracy Mary Pa Amin Kazemi

2025-04-30 0 0 490.23KB 5 页 10玖币

侵权投诉

A Fault Detection Scheme Utilizing Convolutional Neural

Network for PV Solar Panels with High Accuracy

Mary Pa, Amin Kazemi

Department of Electrical Engineering, Lakehead University

Department of Mechanical and Industrial Engineering University of Toronto

Email: mpaparim@lakeheadu.ca, amin.kazemi@utoronto.ca

Abstract

Solar energy is one of the most dependable renewable energy

technologies, as it is feasible almost everywhere globally. However,

improving the efficiency of a solar PV system remains a significant

challenge. To enhance the robustness of the solar system, this paper

proposes a trained convolutional neural network (CNN) based fault

detection scheme to divide the images of photovoltaic modules. For

binary classification, the algorithm classifies the input images of PV

cells into two categories (i.e. faulty or normal). To further assess the

network's capability, the defective PV cells are organized into

shadowy, cracked, or dusty cells, and the model is utilized for multiple

classifications. The success rate for the proposed CNN model is

91.1% for binary classification and 88.6% for multi-classification.

Thus, the proposed trained CNN model remarkably outperforms the

CNN model presented in a previous study which used the same

datasets. The proposed CNN-based fault detection model is

straightforward, simple and effective and could be applied in the fault

detection of solar panel.

I- Introduction

Traditional fossil fuel-based power generations are shifting

towards renewable energy to achieve net-zero greenhouse gas

emissions by 2050. Renewable energy may be cheaper as well

as be friendly to the environment. Examples of the most

promising renewable energy sources are hydroelectric power,

solar, and wind. Photovoltaic energy is one of the cleanest and

most available renewable resources[1], [2], which has attracted

much attention in recent decades [3]. Solar energy utilization

is expected to increase more globally in the coming years. It is

a promising alternative to fossil fuels and has a low adverse

environmental impact. The use of solar energy can be

downscaled to individual homes by using solar panels. These

panels absorb the energy from the sun and provide power for

a particular use, which makes the power system independent

of larger electrical grids. Solar panels are usually designed to

generate electricity in recent decades. However, they may face

issues during their operation, which can reduce their efficiency

or cause complete failure. Like any other electrical energy

production system, photovoltaic power plants require

monitoring and supervision to detect defects or abnormalities

that may develop during operation and ensure their appropriate

functioning and longevity while minimizing energy losses

[4].The faults seen in a PV system can be grouped into several

categories, such as a line-to-line defect[5] . The most common

solar panel defects are the generation of a hot spot that causes

degradation of the cells, microcracks due to thin construction,

broken glass, and dust accumulation under the glass. All these

defects may severely diminish the performance of the solar

modules. The monitoring can be done on-site (e.g.[6]) or

remotely (e.g.[7]). The author applied two CNN strategies to

recognize issues in PV frameworks with a normal exactness of

73.5%, which isn't palatable and needs more improvement. A

PV imperfection forecast approach was proposed by

combining the fuzzy hypothesis and ANN and using voltage

and power proportions as input factors to distinguish different

PV issues. The author connected neural network methods for

fault localization and classification of PV frameworks and

reached good results even with noisy data [8].

Despite a lot of research in the intelligent algorithm-based

fault detection of PV panels, determination of the best

performing classifiers remains a challenge since their

performances depend on various parameters such as the type

of the problem, quality of the input signals or images, the

number of inputs, number of layers, and the adjusting

parameters in the networks. The current study provides a

feature extraction and classification method based on a deep

two- dimensional (2-D) CNN. An overview of the CNN-based

fault detection algorithm is illustrated in Figure-1. Initially, the

algorithm classifies the input images of PV cells into two

simple categories, faulty or normal, called binary classification.

Thereafter, the defective PV cells are further classified into

shadowy, cracked, or dusty cells, known as multiple

classifications. The approach used in this work is relatively

simple while providing satisfactory outcomes. Moreover, the

algorithm can be used to analyze several pictures of grid-

connected solar PV panels and locate the faulty cells, which

improves the durability and reliability of the PV systems.

Figure-1 A general perspective of the CNN-based fault detection

algorithm.

The convolutional neural network is a type of deep learning

commonly utilized in image recognition and classification in

remote sensing. The CNN transforms input information into

numbers using several layers through its different topologies.

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

AFaultDetectionSchemeUtilizingConvolutionalNeuralNetworkforPVSolarPanelswithHighAccuracyMaryPa,AminKazemiDepartmentofElectricalEngineering,LakeheadUniversityDepartmentofMechanicalandIndustrialEngineeringUniversityofTorontoEmail:mpaparim@lakeheadu.ca,amin.kazemi@utoronto.caAbstractSolarenergyisoneoft...

展开>> 收起<<

A Fault Detection Scheme Utilizing Convolutional Neural Network for PV Solar Panels with High Accuracy Mary Pa Amin Kazemi.pdf

共5页,预览1页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

A Fault Detection Scheme Utilizing Convolutional Neural Network for PV Solar Panels with High Accuracy Mary Pa Amin Kazemi

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: