Proposing a Novel Method in Diagnosing Power Transformer Failures Based on the Analysis of Morphological Components
محورهای موضوعی : Majlesi Journal of Telecommunication Devices
Amir ZamanVaziri
1
,
Mehran Emadi
2
1 - Department of Electrical Engineering, Mobarkeh Branch, Islamic Azad University, Mobarkeh, Isfahan, Iran
2 - Department of Electrical Engineering, Mobarkeh Branch, Islamic Azad University, Mobarkeh, Isfahan, Iran
کلید واژه: Transformer, Failure, Analysis of Morphological Components, Image Processing,
چکیده مقاله :
Power transformers are very important in electrical energy distribution systems. In establishing a power plant and even in distribution networks, the cost of power transformers is significant. Any damage or failure in transformers can cause irreparable and heavy losses. Therefore, this equipment is always important in periodic visits. In periodic visits, various parameters of this equipment are checked. One of the ways used in periodic visits is to use infrared images to evaluate the health of this equipment. However, the analysis of these images by human experts is always error-prone and expensive. In this article, a new method is proposed in machine vision to diagnose transformer faults. In the proposed method, infrared images are pre-processed and then features based on multi-resolution transformations such as morphological component analysis, wavelet transform and bandlet transform are extracted and dimension reduced with the help of independent component analysis. The given one-dimensional features are classified using random forest classification, support vector machine and k-nearest neighbor. The classification accuracy obtained in the random forest bin class as the best classifier in fault detection is equal to 98.81%, as well as 91.51% sensitivity and 84.54% negative news rate, as well as 91.86% negative news rate compared to other the numbers showed their superiority. In the F criterion, this value has reached 0.99, which shows the efficiency of the proposed method.
Power transformers are very important in electrical energy distribution systems. In establishing a power plant and even in distribution networks, the cost of power transformers is significant. Any damage or failure in transformers can cause irreparable and heavy losses. Therefore, this equipment is always important in periodic visits. In periodic visits, various parameters of this equipment are checked. One of the ways used in periodic visits is to use infrared images to evaluate the health of this equipment. However, the analysis of these images by human experts is always error-prone and expensive. In this article, a new method is proposed in machine vision to diagnose transformer faults. In the proposed method, infrared images are pre-processed and then features based on multi-resolution transformations such as morphological component analysis, wavelet transform and bandlet transform are extracted and dimension reduced with the help of independent component analysis. The given one-dimensional features are classified using random forest classification, support vector machine and k-nearest neighbor. The classification accuracy obtained in the random forest bin class as the best classifier in fault detection is equal to 98.81%, as well as 91.51% sensitivity and 84.54% negative news rate, as well as 91.86% negative news rate compared to other the numbers showed their superiority. In the F criterion, this value has reached 0.99, which shows the efficiency of the proposed method.
[1] X. Ouyang, Q. Zhou, H. Shang, Y. Zheng, S. Pan, and J. Luo, “Towards a Comprehensive Evaluation on the Online Methods for Monitoring Transformer Turn-to-Turn Faults,” IEEE Transactions on Industrial Electronics, 2022.
[2] L. Raeisian, H. Niazmand, E. Ebrahimnia-Bajestan, and P. Werle, “Thermal management of a distribution transformer: An optimization study of the cooling system using CFD and response surface methodology,” International Journal of Electrical Power & Energy Systems, vol. 104, pp. 443-455, 2019.
[3] E. Karimi, and A. Ebrahimi, “Considering risk of cascading line outages in transmission expansion planning by benefit/cost analysis,” International Journal of Electrical Power & Energy Systems, vol. 78, pp. 480-488, 2016.
[4] E. Karimi, A. Ebrahimi, and M. R. Tavakoli, “How optimal PMU placement can mitigate cascading outages blackouts?,” International Transactions on Electrical Energy Systems, vol. 29, no. 6, pp. e12015, 2019.
[5] E. Karimi, and A. Ebrahimi, “Probabilistic transmission expansion planning considering risk of cascading transmission line failures,” International Transactions on Electrical Energy Systems, vol. 25, no. 10, pp. 2547-2561, 2015.
[6] M. Emadi, Z. Jafarian Dehkordi, and M. Iranpour Mobarakeh, “Improving the Accuracy of Brain Tumor Identification in Magnetic Resonanceaging using Super-pixel and Fast Primal Dual Algorithm,” International Journal of Engineering, vol. 36, no. 3, pp. 505-512, 2023.
[7] F. B. Zade, and M. Emadi, “The Improvement of Breast Cancer Diagnosis Rate in Magnetic Resonance Imaging (MRI) using Fusion of Super Pixels and Fuzzy Connectedness,” Majlesi Journal of Telecommunication Devices, vol. 10, no. 4, pp. 137-145, 2021.
[8] S. A. M. Mobarakeh, and M. Emadi, “Improvement of the Identification Rate using Finger Veins based on the Enhanced Maximum Curvature Method using Morphological Operators,” Majlesi Journal of Telecommunication Devices, vol. 11, no. 1, pp. 1-8, 2022.
[9] M. Soleimani, and J. Coykendall, "Divisors and Factorization Type."
[10] R. S. Marques, A. Conci, M. G. Perez, V. H. Andaluz, and T. M. Mejia, “An approach for automatic segmentation of thermal imaging in Computer Aided Diagnosis,” IEEE Latin America Transactions, vol. 14, no. 4, pp. 1856-1865, 2016.
[11] N. Liu, J. Zhang, and X. Wu, “Asset analysis of risk assessment for iec 61850-based power control systems—part i: methodology,” IEEE Transactions on Power Delivery, vol. 26, no. 2, pp. 869-875, 2010.
[12] A.-V. Vo, L. Truong-Hong, D. F. Laefer, and M. Bertolotto, “Octree-based region growing for point cloud segmentation,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 104, pp. 88-100, 2015.
[13] A. Aslam, E. Khan, and M. S. Beg, “Improved edge detection algorithm for brain tumor segmentation,” Procedia Computer Science, vol. 58, pp. 430-437, 2015.
[14] D. Zhou, and Y. Shao, “Region growing for image segmentation using an extended PCNN model,” IET Image Processing, vol. 12, no. 5, pp. 729-737, 2018.
[15] A. Mahami, T. Bettahar, C. Rahmoune, F. Amrane, M. Touati, and D. Benazzouz, "A New Transformer Condition Monitoring Based on Infrared Thermography Imaging and Machine Learning," Advanced Computational Techniques for Renewable Energy Systems, pp. 408-418: Springer, 2023.
[16] G. Sakallı, and H. Koyuncu, “Identification of asynchronous motor and transformer situations in thermal images by utilizing transfer learning-based deep learning architectures,” Measurement, vol. 207, pp. 112380, 2023.
[17] Z. Li, Y. He, Z. Xing, and J. Duan, “Transformer fault diagnosis based on improved deep coupled dense convolutional neural network,” Electric Power Systems Research, vol. 209, pp. 107969, 2022.
[18] J. Jiang, Y. Bie, J. Li, X. Yang, G. Ma, Y. Lu, and C. Zhang, “Fault diagnosis of the bushing infrared images based on mask R‐CNN and improved PCNN joint algorithm,” High voltage, vol. 6, no. 1, pp. 116-124, 2021.
[19] L. Zhou, J. Jiang, X. Zhou, Z. Wu, T. Lin, and D. Wang, “Detection of transformer winding faults using FRA and image features,” IET Electric Power Applications, vol. 14, no. 6, pp. 972-980, 2020.
[20] L. Zhu-Mao, L. Qing, J. Tao, L. Yong-Xin, H. Yu, and B. Yang, "Research on thermal fault detection technology of power equipment based on infrared image analysis." pp. 2567-2571.
[21] S. Anoop, K. Ilango, A. Dhieep, C. Thomas, J. Jose, A. Kumar, and P. Ajith, "Thermal stress monitoring and pre-fault detection system in power transformers using fibre optic technology." pp. 886-891.
[22] H. Shen, L. Zhu, X. Hong, and W. Chang, "ROI extraction method of infrared thermal image based on GLCM characteristic imitate gradient." pp. 192-205.
[23] O. Aljohani, and A. Abu-Siada, “Application of digital image processing to detect transformer bushing faults and oil degradation using FRA polar plot signature,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 24, no. 1, pp. 428-436, 2017.
[24] A. Abu-Siada, and S. Islam, "Image processing-based on-line technique to detect power transformer winding faults." pp. 5549-5554.
[25] M. Soltani, M. Bahadori, and M. Soleimani, “Optimal Predictive Maintenance and Spare Part Inventory Policies for a Degrading System Subjected to Imperfect Actions,” Available at SSRN 4558570, 2023.
[26] A. Mahami, C. Rahmoune, M. Zair, T. Bettahar, and D. Benazzouz, “Automated Transformer fault diagnosis using infrared thermography imaging, GIST and machine learning technique,” Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering, vol. 236, no. 4, pp. 1747-1757, 2022.
[27] S. Aryanmehr, M. Karimi, and F. Z. Boroujeni, "CVBL IRIS Gender Classification Database Image Processing and Biometric Research, Computer Vision and Biometric Laboratory (CVBL)." pp. 433-438.
[28] M. Emadi, M. Karimi, and F. Davoudi, “A Review on Examination Methods of Types of Working Memory and Cerebral Cortex in EEG Signals,” Majlesi Journal of Telecommunication Devices, vol. 12, no. 3, 2023.
[29] M. Harouni, M. Karimi, A. Nasr, H. Mahmoudi, and Z. Arab Najafabadi, "Health Monitoring Methods in Heart Diseases Based on Data Mining Approach: A Directional Review," Prognostic Models in Healthcare: AI and Statistical Approaches, pp. 115-159: Springer, 2022.
[30] M. Aqil, A. Jbari, and A. Bourouhou, “ECG signal denoising by discrete wavelet transform,” International Journal of Online and Biomedical Engineering (iJOE), vol. 13, no. 09, pp. 51-68, 2017.
[31] A. R. Khan, F. Doosti, M. Karimi, M. Harouni, U. Tariq, S. M. Fati, and S. Ali Bahaj, “Authentication through gender classification from iris images using support vector machine,” Microscopy research and technique, vol. 84, no. 11, pp. 2666-2676, 2021.
[32] M. Karimi, M. Harouni, and S. Rafieipour, "Automated medical image analysis in digital mammography," Artificial intelligence and internet of things, pp. 85-116: CRC Press, 2021.
[33] M. Harouni, M. Karimi, and S. Rafieipour, “Precise segmentation techniques in various medical images,” Artificial Intelligence and Internet of Things, pp. 117-166, 2021.
[34] M. Soleimani, F. Mahmudi, and M. H. Naderi, “On the Maximal Graph of a Commutative Ring,” Mathematics Interdisciplinary Research, 2021.
[35] M. Soleimani, and A. S. Mirshahzadeh, “Multi-class Classification of Imbalanced Intelligent Data using Deep Neural Network,” EAI Endorsed Transactions on AI and Robotics, vol. 2, 2023.
[36] M. Soleimani, F. Mahmudi, and M. Naderi, “Some results on the maximal graph of commutative rings,” Advanced Studies: Euro-Tbilisi Mathematical Journal, vol. 16, no. supp1, pp. 21-26, 2023.
[37] A. Rehman, M. Harouni, F. Zogh, T. Saba, M. Karimi, and G. Jeon, “Detection of Lung Tumors in CT Scan Images using Convolutional Neural Networks,” IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2023.
[38] M. Karimi, M. Harouni, A. Nasr, and N. Tavakoli, "Automatic lung infection segmentation of covid-19 in CT scan images," Intelligent Computing Applications for COVID-19, pp. 235-253: CRC Press, 2021.
[39] A. Raftarai, R. R. Mahounaki, M. Harouni, M. Karimi, and S. K. Olghoran, "Predictive models of hospital readmission rate using the improved AdaBoost in COVID-19," Intelligent Computing Applications for COVID-19, pp. 67-86: CRC Press, 2021.
[40] V. Seena, and J. Yomas, "A review on feature extraction and denoising of ECG signal using wavelet transform." pp. 1-6.
[41] A. Rehman, M. Harouni, M. Karimi, T. Saba, S. A. Bahaj, and M. J. Awan, “Microscopic retinal blood vessels detection and segmentation using support vector machine and K‐nearest neighbors,” Microscopy research and technique, vol. 85, no. 5, pp. 1899-1914, 2022.
[42] L. Xie, Z. Li, Y. Zhou, Y. He, and J. Zhu, “Computational diagnostic techniques for electrocardiogram signal analysis,” Sensors, vol. 20, no. 21, pp. 6318, 2020.
[43] K. Zhang12, Y. Huang, C. Song, H. Wu, L. Wang, and S. M. Intelligence, “Kinship verification with deep convolutional neural networks,” 2015.
[44] T. G. Jan, and S. M. Khan, “A Systematic Review of Research Dimensions Towards Dyslexia Screening Using Machine Learning,” Journal of The Institution of Engineers (India): Series B, pp. 1-12, 2023.
[45] M. Soleimani, A. Harooni, N. Erfani, A. R. Khan, T. Saba, and S. A. Bahaj, “Classification of cancer types based on microRNA expression using a hybrid radial basis function and particle swarm optimization algorithm,” Microscopy Research and Technique, 2024.
Amir ZamanVaziri1, Mehran Emadi2
1- Department of Electrical Engineering, Mobarkeh Branch, Islamic Azad University, Mobarkeh, Isfahan, Iran.
Email: amir_zamnvaziri@yahoo.com
2- Department of Electrical Engineering, Mobarkeh Branch, Islamic Azad University, Mobarkeh, Isfahan, Iran.
Email: emadi.mehran49@gmail.com (Corresponding author)
ABSTRACT: Power transformers are very important in electrical energy distribution systems. In establishing a power plant and even in distribution networks, the cost of power transformers is significant. Any damage or failure in transformers can cause irreparable and heavy losses. Therefore, this equipment is always important in periodic visits. In periodic visits, various parameters of this equipment are checked. One of the ways used in periodic visits is to use infrared images to evaluate the health of this equipment. However, the analysis of these images by human experts is always error-prone and expensive. In this article, a new method is proposed in machine vision to diagnose transformer faults. In the proposed method, infrared images are pre-processed and then features based on multi-resolution transformations such as morphological component analysis, wavelet transform and bandlet transform are extracted and dimension reduced with the help of independent component analysis. The given one-dimensional features are classified using random forest classification, support vector machine and k-nearest neighbor. The classification accuracy obtained in the random forest bin class as the best classifier in fault detection is equal to 98.81%, as well as 91.51% sensitivity and 84.54% negative news rate, as well as 91.86% negative news rate compared to other the numbers showed their superiority. In the F criterion, this value has reached 0.99, which shows the efficiency of the proposed method.
KEYWORDS: Transformer, Failure, Analysis of Morphological Components, Image Processing.
|
1. Intoraduction
As the heart of a substation, the power transformer acts as an important link for voltage conversion and energy delivery. Survey results show that most of the power transformers around the world were installed in the 1980s and many of them are reaching the end of their life. The failure rate of these transformers has been continuously increasing in recent years, especially in power companies with poor maintenance and asset management methods, causing the loss of this valuable equipment [1]. Any change and failure in the transformer causes unwanted heat in this equipment. Hot spots in electrical equipment, especially power transformers, often occur as a result of looseness, oxidation and corrosion of connections, asymmetry of phases, and insulation failure of coils. All these things can be done by using thermal or infrared IR imaging equipment to measure the temperature of equipment such as energy distribution overhead lines, cables, transformers and fuses, cables, electrical panel equipment and all electrical distribution network equipment, including substations. Determined ground [2].Diagnosing transformer failure is done online with the help of infrared image. These images are sensitive to the heat caused by various errors in the transformer. The performance of these cameras is limited by the level of image processing used. In other words, in infrared imaging in thermal cameras, the proposed method is important in detecting failures in image processing[3-6]. With the development and spread of substation inspection robots and smart substations in distribution networks, a large number of infrared fault images need to be analyzed. The infrared image transformer fault detection method can automatically identify and analyze the collected infrared images, which can reduce the maintenance costs of transformers and the dependence on technicians. and reduce manpower. Therefore, it is important to develop an infrared image recognition method to help deal with large amounts of infrared image data. In the early stage, the detection based on infrared or infrared thermal was very limited by thermography and the level of computer calculations. Researchers focused on removing image noise with filter algorithms including Gaussian filter, adaptive median filter, median filter, etc. Also, the image segmentation technique, including object segmentation or fault region segmentation, is performed simultaneously, represented by threshold segmentation by the OTSU method[7-9].Gray histogram methods [10], K-means [10,11], similarity of adjacent regions, growth algorithm based on area and morphological opening and closing operations [12], edge detection [13] and PCNN (pulse coupled neural network) pulse) [14] is presented. In [15] , a new non-contact and non-intrusive method for monitoring electrical transformers and detecting their defects based on infrared thermography imaging techniques (IRT) imaging techniques is adopted. In [16], a method based on deep learning is proposed in transformer failure detection. In this paper, in addition to accuracy-based analysis, an in-depth evaluation is presented to show the most suitable architecture in thermal image classification. [17]proposes a fault detection model that includes adaptive synthetic oversampling (ADASYN), reconstructed data method, and an improved Mask Region convolutional neural network (CDCN). In [18], by building an object recognition system with a Convolutional Neural Network (CNN) frame, the Bush frame can be accurately extracted. To detect the fault area of bushings and the background, a pulse connected neural network based on simple linear iterative clustering is proposed to improve the fault area segmentation performance. In [19] a new interpretation of transformer failure analysis is proposed including new image features based on image processing technique. In [20], it has been proposed to detect faults in the power network using Gabor filters. In this method, the investigated image passes through a filter bank, then the output of the filter is thresholded. By combining the output of different filters, a suitable pattern of the defect can be obtained. In [21], in order to solve the key problem of automatic fault detection technology of power equipment, projected transformation has been introduced to extract the features of the thermal image in the system, using the characteristics of the equipment in this design. In [22]. they have proposed GLCM based equipment fault detection. In [23], a new transformer failure interpretation approach has been proposed to detect winding short circuit fault in transformer, radial deformation and bushing faults using polar diagram and have introduced digital image processing in which various unique image features of a polar plot are extracted using geometric dimensions, invariant moments. In [24], an online technique is introduced to detect internal faults in a power transformer by constructing a voltage-current (V-I) location diagram to provide the current status of the transformer's health status. However, the above methods are only applicable for images with a relatively simple background. For those infrared images with complex background, the main image segmentation methods have strong limitations. In order to overcome these complications, multi-resolution transformation methods are proposed[25]. Wavelet transform is one of the important mathematical multi-resolution transforms, which obtains a time-scale representation of the digital signal using digital filtering techniques [26]. In this transformation, it will be calculated by changing the scale of the analysis window, changing the location of the window in time, multiplying it by the signal and integrating it in all times. In the discrete case, filters of different cutoff frequencies are used to decompose the signal at different scales[27-29]. The signal is passed through a series of high-pass filters to analyze high frequencies and through a series of low-pass filters to analyze low frequencies. The degree of signal resolution, which is a measure of the amount of detail in the signal, is varied by filtering operations and scaled by upsampling and Subsampling (downsampling)[30].Bandelet are one of the multi-resolution transforms that have been developed to overcome the challenges and shortcomings of the wavelet transform. Bandelet calculates a geometric flow of an image to better extract smooth edge information. Bandelet decomposition is applied to the orthogonal wavelet coefficients or wavelet filter bank of an image and is calculated by geometric orthogonal transformation through orthogonal filters. As a result of this process, a different transformation is obtained from each geometric direction and they can be processed to find the optimal set of filters with the best basic algorithm. The characteristics of wavelets can be manipulated by selecting low-pass and high-pass square mirror filters [31].Morphological component analysis (MCA) can decompose an image into two components[32]. The advantages of MCA include completeness and effectiveness. (1) Completeness: In multiscale transformation, transformation bases are built to reveal salient features of an image. (2) Effectiveness: The sparse representation uses a complete dictionary with more columns than rows and models the image of the 1-D representation as a linear combination of columns (atoms) [33].The stability and reliability of a power system depends in many ways on the condition and health of power transformers[29]. As one of the most expensive and important elements, the power transformer is a very necessary element whose failure and damage may cause the power system to be interrupted. Therefore, it seems that transformers should be constantly inspected and reviewed in order to apply preventive repairs. In addition to the cost and the need for experienced manpower, visual inspection for power transformers is not very accurate and is always associated with errors. For this purpose, using methods based on image processing can be helpful[34]. Thermal or infrared IR cameras that work with infrared technology are an efficient way to provide preventive maintenance in power transformers and failure detection along with efficient image processing methods. It seems that it is necessary to propose a new and efficient method in diagnosing the failure of power transformers[35]. Therefore, in this research, a method based on the morphological component analysis will be presented in order to detect faults in power transformers. In the proposed method, taking advantage of the morphological diversity of images and the advantages of MCA, the component (carton, texture, and edge) is considered as a feature and will be entered into a classification for failure detection. The innovations of this research can be stated as follows:
• Improving the accuracy of transformer failure detection based on the analysis of morphological components
In the following, this article is divided as follows. In the second part, the principles of image processing in detecting defects in infrared images are presented. In the third part, the suggestion method will be presented. In the fourth part, the evaluation of the proposed method will be done. Finally, the conclusion of the article will be presented in the fifth section.
2. material and method
The passage of electric current through various circuits and components in a device, devices and electrical equipment is always associated with heat production, it seems that by measuring the temperature and preparing thermal photographs of electrical components and equipment, it is a reliable guide in determining the weak points that may be in The future will lead to major connections. Hot spots in electrical equipment are often caused by not being strong, oxidation or corrosion of connections, as well as asymmetry of phases or insulation failure of coils. By using IR infrared image recording equipment in measuring the temperature of equipment such as power transmission lines, high voltage substations, transformers, switches, fuses, cables and all control equipment and electrical panels, it reveals them before they lead to destructive events. It was solved in the electrical system [36].Unique information is extracted from infrared images related to electric energy distribution network equipment and used to show the performance status of this equipment in electric energy distribution smart networks. In addition to morphological features that can be observed, features such as wavelet features and statistical features are also effective in diagnosis. The method of analyzing infrared images in electrical energy distribution network equipment based on machine vision and image processing, data pre-processing, dimension extraction and reduction, classification and application are discussed[37].
2.1. Wavelet Transform
Wavelet transformation is one of the mathematical transformations in multi-precision domains. This conversion is a desirable strategy for establishing an optimal balance between time accuracy and frequency accuracy [38]. At higher frequencies, the wavelet transform gains time-domain information at the cost of losing frequency-related information. While at lower frequencies, it gains frequency information at the expense of temporal information loss. As the Fourier transform is defined based on an integral, the wavelet transform can also be defined based on an integral as follows:
| (1) |
(5) |
|
(6) |
|
Fig. 2. Block diagram of the proposed method.
The first step: improving the quality of the infrared image
Infrared images consist of a gray color spectrum that lies between white and black. Sometimes, due to the low contrast in the image, it is very difficult to distinguish the difference between adjacent pixels that have very close colors. Since the defect in an equipment discussed in this research may be similar to the rest of the parts, it is necessary to improve the quality of the image in such a way that it becomes easier to detect this defect in the transformer [42].Bidirectional filtering is used as an edge preservation tool in image enhancement applications [4]. Along with a low-pass spatial kernel (which helps with smoothing), it uses a kernel to prevent smoothing near the edges. As a result, the filter is able to smooth homogeneous areas and preserve sharp edges at the same time. It was shown that spatial and domain kernels are typically Gaussian, which can be improved by adjusting the width and center of the Gaussian domain kernel at each pixel, and the enhancement capacity of the two-way filter.
|
|
Fig. 3. (a) The original image and (b) the improved rendering based on filtering in the proposed method.
The second step: applying MCA principal components analysis
Morphological component analysis (MCA) is a new method that allows us to separate features in an image when these features present different morphological aspects. We show that MCA can be very useful for decomposing images into textures, edges and piecewise smooth (cartoon)or for feature extraction applications. To classify the defects in the infrared images using the neural network method or any other classifier in machine learning, first the features based on the morphological component analysis (MCA) are extracted from the infrared image and by them the neural network or another classifier. It is taught in machine learning.
In the framework of thin representation, a dictionary 
is viewed as an N×T matrix. When T > N or even T >>N, the dictionary is too complete and is built by merging several dictionaries. An image x∈RN (an image with N pixels can be expressed as a lexically ordered 1-D vector) as a linear combination of elementary atoms M(M < T) of the dictionary, according to the model equation has been made
(7) |
| ||
(8) |
|
(9) |
|
(10) |
| ||
(11) |
| ||
(12) |
|
Random forest (%) | Support vector machine (%) | K nearest neighbor (%) | Evaluation criteria |
71.76 | 68.13 | 75.26 | Precision |
72.72 | 66.75 | 71.38 | Sensitivity |
70.27 | 63.75 | 72.33 | Accuracy |
69.43 | 71.89 | 67.21 | Recall |
74.00 | 80.01 | 87.01 | Criterion F |
Table 2. The results of the simulation with dimension reduction with the help of PCA.
Random forest (%) | Support vector machine (%) | K nearest neighbor (%) | Evaluation criteria |
75.64 | 79.13 | 86.26 | Precision |
66.72 | 80.57 | 81.38 | Sensitivity |
76.26 | 83.75 | 89.33 | Accuracy |
75.43 | 85.89 | 77.21 | Recall |
74.02 | 80.80 | 87.25 | Criterion F |
Table 3. Simulation results along with dimension reduction in the proposed method.
Random forest (%) | Support vector machine (%) | K nearest neighbor (%) | Evaluation criteria |
98.18 | 94.10 | 95.81 | Precision |
86.25 | 85.37 | 91.51 | Sensitivity |
63.38 | 81.14 | 84.54 | Accuracy |
97.87 | 90.87 | 91.25 | Recall |
99.25 | 91.23 | 94.28 | Criterion F |
As it is clear from the results of Table 4, the RF classifier with K nearest neighbor has been able to get better results. It should be noted that the accuracy and sensitivity values obtained and the recall rate are also acceptable results. As it can be seen from the graph, the combination of the extracted features has led to better results, and this proves the important sub-hypothesis of this research, including the improvement of results in classifications by reducing the dimension with the help of the proposed method. This superiority is still perceptible in all the criteria used. The F criterion is also included in these evaluations. In this case, the random forest classifier has been able to obtain the best result with 0.99 in the accuracy criterion as well as the F criterion with one hundred features.
4.4. Comparison with other studies
The proposed method in this article is compared with other related researches in fault diagnosis in transformer equipment. These methods include RNN recurrent neural networks [32] and deep learning [33]. In recent years, methods based on deep learning are among the most common methods and of course with good efficiency. Various researches have used methods based on deep learning. Although these methods are acceptable and highly accurate, they require a large database for simulation. The desired methods for comparison have been simulated on the database of this research. Therefore, a correct comparison and correct evaluation has been made. The results of the comparison of the proposed method with the desired researches are shown in table (5). As can be seen from the results of this table, the proposed method has better accuracy than other methods based on deep learning. The database used in all these researches is the database collected in this research by Saman Niro Sepahan Company.
Table 4. Comparison of the proposed method with other articles.
Average accuracy | Research | The presented model |
94.10 | [32] | 2 D-CNN |
95.91 | [33] | RNN |
99.00 | - | Proposed |
5. Conclusion
In this research, the relationship between In this article, the proposed method for detecting defects in transformer equipment was evaluated in infrared images. In the proposed method, several features in the time domain including morphological features (edge and cartoon), multi-resolution features based on wavelet transform with Dabich's filter bank were also extracted. A method based on PCA principal component analysis was introduced. The feature vector matrix was classified by support vector machine, random forest and k nearest neighbor classifiers. The parameters of recall rate, precision, accuracy, specificity and f-criterion are evaluated in three categories: SVM, KNN and RF. The RF classification has the best result with numerical values higher than 99% in the proposed method in feature selection based on the PCA algorithm.
REFERENCES
[1] X. Ouyang, Q. Zhou, H. Shang, Y. Zheng, S. Pan, and J. Luo, “Towards a Comprehensive Evaluation on the Online Methods for Monitoring Transformer Turn-to-Turn Faults,” IEEE Transactions on Industrial Electronics, 2022.
[2] L. Raeisian, H. Niazmand, E. Ebrahimnia-Bajestan, and P. Werle, “Thermal management of a distribution transformer: An optimization study of the cooling system using CFD and response surface methodology,” International Journal of Electrical Power & Energy Systems, vol. 104, pp. 443-455, 2019.
[3] E. Karimi, and A. Ebrahimi, “Considering risk of cascading line outages in transmission expansion planning by benefit/cost analysis,” International Journal of Electrical Power & Energy Systems, vol. 78, pp. 480-488, 2016.
[4] E. Karimi, A. Ebrahimi, and M. R. Tavakoli, “How optimal PMU placement can mitigate cascading outages blackouts?,” International Transactions on Electrical Energy Systems, vol. 29, no. 6, pp. e12015, 2019.
[5] E. Karimi, and A. Ebrahimi, “Probabilistic transmission expansion planning considering risk of cascading transmission line failures,” International Transactions on Electrical Energy Systems, vol. 25, no. 10, pp. 2547-2561, 2015.
[6] M. Emadi, Z. Jafarian Dehkordi, and M. Iranpour Mobarakeh, “Improving the Accuracy of Brain Tumor Identification in Magnetic Resonanceaging using Super-pixel and Fast Primal Dual Algorithm,” International Journal of Engineering, vol. 36, no. 3, pp. 505-512, 2023.
[7] F. B. Zade, and M. Emadi, “The Improvement of Breast Cancer Diagnosis Rate in Magnetic Resonance Imaging (MRI) using Fusion of Super Pixels and Fuzzy Connectedness,” Majlesi Journal of Telecommunication Devices, vol. 10, no. 4, pp. 137-145, 2021.
[8] S. A. M. Mobarakeh, and M. Emadi, “Improvement of the Identification Rate using Finger Veins based on the Enhanced Maximum Curvature Method using Morphological Operators,” Majlesi Journal of Telecommunication Devices, vol. 11, no. 1, pp. 1-8, 2022.
[9] M. Soleimani, and J. Coykendall, "Divisors and Factorization Type."
[10] R. S. Marques, A. Conci, M. G. Perez, V. H. Andaluz, and T. M. Mejia, “An approach for automatic segmentation of thermal imaging in Computer Aided Diagnosis,” IEEE Latin America Transactions, vol. 14, no. 4, pp. 1856-1865, 2016.
[11] N. Liu, J. Zhang, and X. Wu, “Asset analysis of risk assessment for iec 61850-based power control systems—part i: methodology,” IEEE Transactions on Power Delivery, vol. 26, no. 2, pp. 869-875, 2010.
[12] A.-V. Vo, L. Truong-Hong, D. F. Laefer, and M. Bertolotto, “Octree-based region growing for point cloud segmentation,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 104, pp. 88-100, 2015.
[13] A. Aslam, E. Khan, and M. S. Beg, “Improved edge detection algorithm for brain tumor segmentation,” Procedia Computer Science, vol. 58, pp. 430-437, 2015.
[14] D. Zhou, and Y. Shao, “Region growing for image segmentation using an extended PCNN model,” IET Image Processing, vol. 12, no. 5, pp. 729-737, 2018.
[15] A. Mahami, T. Bettahar, C. Rahmoune, F. Amrane, M. Touati, and D. Benazzouz, "A New Transformer Condition Monitoring Based on Infrared Thermography Imaging and Machine Learning," Advanced Computational Techniques for Renewable Energy Systems, pp. 408-418: Springer, 2023.
[16] G. Sakallı, and H. Koyuncu, “Identification of asynchronous motor and transformer situations in thermal images by utilizing transfer learning-based deep learning architectures,” Measurement, vol. 207, pp. 112380, 2023.
[17] Z. Li, Y. He, Z. Xing, and J. Duan, “Transformer fault diagnosis based on improved deep coupled dense convolutional neural network,” Electric Power Systems Research, vol. 209, pp. 107969, 2022.
[18] J. Jiang, Y. Bie, J. Li, X. Yang, G. Ma, Y. Lu, and C. Zhang, “Fault diagnosis of the bushing infrared images based on mask R‐CNN and improved PCNN joint algorithm,” High voltage, vol. 6, no. 1, pp. 116-124, 2021.
[19] L. Zhou, J. Jiang, X. Zhou, Z. Wu, T. Lin, and D. Wang, “Detection of transformer winding faults using FRA and image features,” IET Electric Power Applications, vol. 14, no. 6, pp. 972-980, 2020.
[20] L. Zhu-Mao, L. Qing, J. Tao, L. Yong-Xin, H. Yu, and B. Yang, "Research on thermal fault detection technology of power equipment based on infrared image analysis." pp. 2567-2571.
[21] S. Anoop, K. Ilango, A. Dhieep, C. Thomas, J. Jose, A. Kumar, and P. Ajith, "Thermal stress monitoring and pre-fault detection system in power transformers using fibre optic technology." pp. 886-891.
[22] H. Shen, L. Zhu, X. Hong, and W. Chang, "ROI extraction method of infrared thermal image based on GLCM characteristic imitate gradient." pp. 192-205.
[23] O. Aljohani, and A. Abu-Siada, “Application of digital image processing to detect transformer bushing faults and oil degradation using FRA polar plot signature,” IEEE Transactions on Dielectrics and Electrical Insulation, vol. 24, no. 1, pp. 428-436, 2017.
[24] A. Abu-Siada, and S. Islam, "Image processing-based on-line technique to detect power transformer winding faults." pp. 5549-5554.
[25] M. Soltani, M. Bahadori, and M. Soleimani, “Optimal Predictive Maintenance and Spare Part Inventory Policies for a Degrading System Subjected to Imperfect Actions,” Available at SSRN 4558570, 2023.
[26] A. Mahami, C. Rahmoune, M. Zair, T. Bettahar, and D. Benazzouz, “Automated Transformer fault diagnosis using infrared thermography imaging, GIST and machine learning technique,” Proceedings of the Institution of Mechanical Engineers, Part E: Journal of Process Mechanical Engineering, vol. 236, no. 4, pp. 1747-1757, 2022.
[27] S. Aryanmehr, M. Karimi, and F. Z. Boroujeni, "CVBL IRIS Gender Classification Database Image Processing and Biometric Research, Computer Vision and Biometric Laboratory (CVBL)." pp. 433-438.
[28] M. Emadi, M. Karimi, and F. Davoudi, “A Review on Examination Methods of Types of Working Memory and Cerebral Cortex in EEG Signals,” Majlesi Journal of Telecommunication Devices, vol. 12, no. 3, 2023.
[29] M. Harouni, M. Karimi, A. Nasr, H. Mahmoudi, and Z. Arab Najafabadi, "Health Monitoring Methods in Heart Diseases Based on Data Mining Approach: A Directional Review," Prognostic Models in Healthcare: AI and Statistical Approaches, pp. 115-159: Springer, 2022.
[30] M. Aqil, A. Jbari, and A. Bourouhou, “ECG signal denoising by discrete wavelet transform,” International Journal of Online and Biomedical Engineering (iJOE), vol. 13, no. 09, pp. 51-68, 2017.
[31] A. R. Khan, F. Doosti, M. Karimi, M. Harouni, U. Tariq, S. M. Fati, and S. Ali Bahaj, “Authentication through gender classification from iris images using support vector machine,” Microscopy research and technique, vol. 84, no. 11, pp. 2666-2676, 2021.
[32] M. Karimi, M. Harouni, and S. Rafieipour, "Automated medical image analysis in digital mammography," Artificial intelligence and internet of things, pp. 85-116: CRC Press, 2021.
[33] M. Harouni, M. Karimi, and S. Rafieipour, “Precise segmentation techniques in various medical images,” Artificial Intelligence and Internet of Things, pp. 117-166, 2021.
[34] M. Soleimani, F. Mahmudi, and M. H. Naderi, “On the Maximal Graph of a Commutative Ring,” Mathematics Interdisciplinary Research, 2021.
[35] M. Soleimani, and A. S. Mirshahzadeh, “Multi-class Classification of Imbalanced Intelligent Data using Deep Neural Network,” EAI Endorsed Transactions on AI and Robotics, vol. 2, 2023.
[36] M. Soleimani, F. Mahmudi, and M. Naderi, “Some results on the maximal graph of commutative rings,” Advanced Studies: Euro-Tbilisi Mathematical Journal, vol. 16, no. supp1, pp. 21-26, 2023.
[37] A. Rehman, M. Harouni, F. Zogh, T. Saba, M. Karimi, and G. Jeon, “Detection of Lung Tumors in CT Scan Images using Convolutional Neural Networks,” IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2023.
[38] M. Karimi, M. Harouni, A. Nasr, and N. Tavakoli, "Automatic lung infection segmentation of covid-19 in CT scan images," Intelligent Computing Applications for COVID-19, pp. 235-253: CRC Press, 2021.
[39] A. Raftarai, R. R. Mahounaki, M. Harouni, M. Karimi, and S. K. Olghoran, "Predictive models of hospital readmission rate using the improved AdaBoost in COVID-19," Intelligent Computing Applications for COVID-19, pp. 67-86: CRC Press, 2021.
[40] V. Seena, and J. Yomas, "A review on feature extraction and denoising of ECG signal using wavelet transform." pp. 1-6.
[41] A. Rehman, M. Harouni, M. Karimi, T. Saba, S. A. Bahaj, and M. J. Awan, “Microscopic retinal blood vessels detection and segmentation using support vector machine and K‐nearest neighbors,” Microscopy research and technique, vol. 85, no. 5, pp. 1899-1914, 2022.
[42] L. Xie, Z. Li, Y. Zhou, Y. He, and J. Zhu, “Computational diagnostic techniques for electrocardiogram signal analysis,” Sensors, vol. 20, no. 21, pp. 6318, 2020.
[43] K. Zhang12, Y. Huang, C. Song, H. Wu, L. Wang, and S. M. Intelligence, “Kinship verification with deep convolutional neural networks,” 2015.
[44] T. G. Jan, and S. M. Khan, “A Systematic Review of Research Dimensions Towards Dyslexia Screening Using Machine Learning,” Journal of The Institution of Engineers (India): Series B, pp. 1-12, 2023.
[45] M. Soleimani, A. Harooni, N. Erfani, A. R. Khan, T. Saba, and S. A. Bahaj, “Classification of cancer types based on microRNA expression using a hybrid radial basis function and particle swarm optimization algorithm,” Microscopy Research and Technique, 2024.
سکوی نشر دانش
پیوندهای سایت
مراکز مرتبط
پشتیبانی
صفحات رسمی
حقوق این وبسایت متعلق به سامانه مدیریت نشریات دانشگاه آزاد اسلامی است.
حق نشر © 1403-1400