| 1. |
- Prasshanth, C.V., et al.
(författare)
-
Enhancing photovoltaic module fault diagnosis: Leveraging unmanned aerial vehicles and autoencoders in machine learning
- 2024
-
Ingår i: Sustainable Energy Technologies and Assessments. - : Elsevier. - 2213-1388 .- 2213-1396. ; 64
-
Tidskriftsartikel (refereegranskat)abstract
- Photovoltaic (PV) modules play a pivotal role in renewable energy systems, underscoring the critical need for their fault diagnosis to ensure sustained energy production. This study introduces a novel approach that combines the power of deep neural networks and machine learning for comprehensive PV module fault diagnosis. Specifically, a fusion methodology that incorporates autoencoders (a deep neural network architecture) and support vector machines (SVM) (a machine learning algorithm) is proposed in the present study. To generate high-quality image datasets for training, unmanned aerial vehicles (UAVs) equipped with RGB cameras were employed to capture detailed images of PV modules. Burn marks, snail trails, discoloration, delamination, glass breakage and good panel were the conditions considered in the study. The experimental results demonstrate remarkable accuracy of 98.57% in diagnosing faults, marking a significant advancement in enhancing the reliability and performance of PV modules. This research contributes to the sustainability and efficiency of renewable energy systems, underlining its importance in the quest for a cleaner, greener future.
|
|
| 2. |
|
|
| 3. |
|
|
| 4. |
- Srivatsan, B, et al.
(författare)
-
Fault diagnosis of air compressors using transfer learning: A comparative study of pre-trained networks and hyperparameter optimization
- 2024
-
Ingår i: Journal of Low Frequency Noise, Vibration and Active Control. - 1461-3484 .- 2048-4046.
-
Tidskriftsartikel (refereegranskat)abstract
- Air compressors are critical components in many industries whose catastrophic failure results in huge financial losses anddowntime leading to accidents. Hence, real time fault diagnosis of air compressor is essential to predict the health conditionof air compressor and plan scheduled maintenance thereby reducing financial losses and accidents. Fault diagnosis usingtransfer learning aids in real time fault detection. In the present study, five air compressor conditions were considerednamely, check valve fault, inlet and outlet reed valve fluttering fault, inlet reed valve fluttering fault, outlet reed valvefluttering fault, and good condition. The raw vibration data was converted to radar plot images that were pre-processed andclassified using four pre-trained networks (ResNet-50, GoogLeNet, AlexNet, and VGG-16). The hyperparameters likeepochs, batch size, optimizer, train-test split ratio, and learning rate were varied to find out the best network for aircompressor fault diagnosis. ResNet-50 among all other pre-trained networks produced the maximum classificationaccuracy (average of five trials) of 98.72%.
|
|
| 5. |
- Venkatesh, Naveen, et al.
(författare)
-
A comparative study on bayes classifier for detecting photovoltaic module visual faults using deep learning features
- 2024
-
Ingår i: Sustainable Energy Technologies and Assessments. - 2213-1388 .- 2213-1396. ; 64
-
Tidskriftsartikel (refereegranskat)abstract
- Renewable energy is found to be an effective alternative in the field of power production owing to the recent energy crises. Among the available renewable energy sources, solar energy is considered the front runner due to its ability to deliver clean energy, free availability and reduced cost. Photovoltaic (PV) modules are placed over large geographical regions for efficient solar energy harvesting, making it difficult to carry out maintenance and restoration works. Thermal stresses inherited by photovoltaic modules (PVM) under varying environmental conditions can lead to failure of internal components. Such failures when left undetected impart a number of complications in the system that will lead to unsafe operation and seizure. To avoid the aforementioned uncertainties, frequent monitoring of PVM is found necessary. The fault identification in PVM using essential features taken from aerial images is presented in this study. The feature extraction procedure was carried out using convolutional neural networks (CNN), while the feature selection process was carried out by the J48 decision tree method. Six test conditions were considered such as delamination, glass breakage, discoloration, burn marks, snail trail, and good panel. Bayes Net (BN) and Naïve Bayes (NB) classifiers were utilized as primary classifiers for all the test conditions. Results obtained from the classifiers were compared and the best classifier for fault detection in PVM is suggested.
|
|
| 6. |
- Venkatesh S, Naveen, et al.
(författare)
-
Detection of visual faults in photovoltaic modules using a stacking ensemble approach
- 2024
-
Ingår i: Heliyon. - : Elsevier. - 2405-8440. ; 10:6
-
Tidskriftsartikel (refereegranskat)abstract
- Faults in photovoltaic (PV) modules may occur due to various environmental and physical factors. To prevent faults and minimize investment losses, fault diagnosis is crucial to ensure uninterrupted power production, extended operational lifespan, and a high level of safety in PV modules. Recent advancements in inspection techniques and instrumentation have significantly reduced the cost and time required for inspections. A novel stacking-based ensemble approach was performed in the present study for the accurate classification of PV module visible faults. The present study utilizes AlexNet (a pre-trained network) to extract image features from the aerial images of PV modules with the aid of MATLAB software. Furthermore, J48 algorithm was applied to perform the feature selection task to determine the most relevant features. The features derived as output from the J48 algorithm were passed onto train eight base classifiers namely, Naïve Bayes, logistic regression (LR), J48, random forest (RF), multilayer perceptron (MLP), logistic model tree (LMT), support vector machines (SVM) and k-nearest neighbors (kNN). The best performing five classifiers on the front run with higher classification accuracies were selected to formulate three categories of stacking ensemble groups as follows: (i) three-class ensemble (SVM, kNN, and LMT), (ii) four-class ensemble (SVM, kNN, LMT, and RF), and (iii) five-class ensemble (SVM, kNN, LMT, RF, and MLP). A comparison in the performance of the aforementioned stacked ensembles was evaluated with different meta classifiers. The obtained results infer that the four-class stacking ensemble model (SVM, kNN, LMT, and RF) with RF as the predictor achieved the highest possible classification accuracy of 99.04%.
|
|
| 7. |
- Vivek, Joseph, et al.
(författare)
-
Wear particle image analysis: feature extraction, selection and classification by deep and machine learning
- 2024
-
Ingår i: Industrial Lubrication and Tribology. - : Emerald Group Publishing Limited. - 0036-8792 .- 1758-5775. ; 76:5, s. 599-607
-
Tidskriftsartikel (refereegranskat)abstract
- PurposeThis study aims to explore the integration of machine learning (ML) in tribology to optimize lubrication interval decisions, aiming to enhance equipment lifespan and operational efficiency through wear image analysis.Design/methodology/approachUsing a data set of scanning electron microscopy images from an internal combustion engine, the authors used AlexNet as the feature extraction algorithm and the J48 decision tree algorithm for feature selection and compared 15 ML classifiers from the lazy-, Bayes and tree-based families.FindingsFrom the analyzed ML classifiers, instance-based k-nearest neighbor emerged as the optimal algorithm with a 95% classification accuracy against testing data. This surpassed individually trained convolutional neural networks’ (CNNs) and closely approached ensemble deep learning (DL) techniques’ accuracy.Originality/valueThe proposed approach simplifies the process, enhances efficiency and improves interpretability compared to more complex CNNs and ensemble DL techniques.
|
|
