Skip to main content

Techniques of infrared thermography for condition monitoring of electrical power equipment


The application of computer vision continues to widen with advancement in technology. Imaging systems which provide necessary inputs to the computer-vision-based models can come in various ways. Such as X-ray images, Computed Tomography (CT) scan images, and Infrared (IR) images. This paper is a review of different application areas of infrared thermography (IRT) for monitoring the status of electrical power equipment. It summarizes in tabular form recent research and relevant works within the field of condition monitoring of power assets. A general review of the application of IRT in power devices was undertaken before a specific review of selected works based on IRT for important electrical power equipment with a tabular review of possible causes of hotspots using photovoltaic installation as a reference. Results of previous works were presented with highlights on performance metrics used and accuracies achieved. Emphasis where made on the future potential of IRT and some associated techniques. The work saw that heat production within systems during operation is an important characteristic that enables IRT to become applicable for monitoring diverse physical systems, most importantly power systems. The high cost of high-definition, and long-range IR cameras limits the wide adoption of the technology for its potential applications for monitoring power installations. The work recommends future research in the development of affordable IR imaging systems with advanced features for condition monitoring of physical systems such as power installations.


According to Dhimish, et al. [1] as well as in another work by Solo´rzano and Egido [2], hotspots are abnormal high temperatures at connections between power equipment or in power system components themselves caused by faults and improper operation of equipment within the network. Normally, the temperature of objects has been used as a reliable index for gauging the health of biological and inanimate systems. In biomedical fields, Najimi et al. [3] opined that temperature measurements give easy insight into the existence of diseases within the body. The story is not different in physical industrial systems made of interconnected systems that usually convey energy from one point to another. Different types of thermal sensing devices have been developed for various temperature measuring applications; ranging from conventional thermocouples, thermostats, thermistors, thermopiles, thermometers, and state-of-the-art infrared (IR) thermal pointing devices, IR imaging cameras, etc. Hence, Bach et al. [4] saw that detection of hotspots in equipment or installations could become quite a challenge using the conventional methods of temperature measurements because they could become time-consuming, costly, and unsafe for personnel and equipment. However, Infrared imaging or thermography would be suited for applications where safety, cost, and time must be optimized. Moreover, Madding et al. [5] and Usamentiaga et al. [6] inferred that the need for good resolution, optimal temperature range, stability, and accuracy of the temperature measuring device encourages the adoption of the IRT application as the method of choice for condition monitoring of industrial systems.

In order to perform noncontact temperature recording, the camera will be appropriately set for optimal thermal imaging temperature. Some of these settings are not limited to emissivity, resolution, etc. Others include the transmission ability of the transmitting medium (usually air) and the temperature of that transmitting medium. All these settings will affect the ultimate output for the temperature of the object being viewed. These parameters and others ensure that the thermal imaging camera has become an excellent tool for condition monitoring of electrical and mechanical systems in the power industry.

For instance, Alvarado-Hernandez et al [7] pointed out that monitoring of power equipment would inadvertently aid the efficiency of the equipment and improve users’ experience and give leverage for enhancing the operation of the device using advanced intelligent optimization techniques.

This work was organized to give insight into the importance of using IR imaging techniques for monitoring the status of equipment in electrical installation with tabular briefs of diverse works done by different authors using the concept of infrared thermography (IRT). Then, under specific application areas, critical reviews of research done using IRT were presented by showing observed drawbacks of the highlighted methodologies and concluded with recommendations for possible future research.

Review of IRT methodology in various works

Between 2002 till date, many research works have been published in reputable journals. More compelling is that the researches have grown over the years and some of the work are presented in Table 1 which shows the various methods and focus of the works.

Table 1 Recent works in infrared thermography for monitoring electrical equipment

One important advantage of IRT for power equipment monitoring lies in the fact that the effectiveness and efficiency of the technique to produce good results does not depend on the operating voltage of the equipment, hence it has found wide use in many areas of equipment monitoring.

Selected papers where infra-red thermography were applied in different areas or critical devices in the power installations was reviewed in the succeeding sections.

Review of infra-red thermographs in critical power equipment

Power transformers

Utami et al. [38] presented a transformer monitoring scheme with IRT results of the tank, tap regulating device, cooling system, and external insulators. The work involves analyzing the top, bottom, sides, and center of two transformers (normal and abnormal) of the same rating and loading, whereas Asiegbu et al. [39] presented an RLC thermal network model that could reflect the basic equivalent circuit of the transformer which would be applied to develop a comparable analogous thermal model in terms of electrical parameters. Changes in the thermal capacitance of power equipment indicate the condition of the insulation in inductive loads like transformers and cause the working temperature to increase. The analogous model was then used to perform the thermal gradient evaluation of the system. On their part, Fang et al. [40] proposed a method of fault diagnosis of electric transformers using semi-supervised learning to train infrared image processed data. The work adopted a support vector machine for fault classification, whereas the infrared images were clustered utilizing the K-means technique. The authors used feature extraction and generative adversarial networks to get artificial data of the labeled images before applying a semi-supervised graph model to train both the labeled and unlabeled images. However, the SVM does not perform well for large datasets, requires a long period of training, and adversely affected by noisy datasets. Moreover, the SVM technique finds it difficult to compute local optimal condition and could be quite complex to implement. The major issue with K-means is linked to its vulnerability to outliers.

Mlakic et al. [41] applied modern machine learning tools for power asset monitoring inspired to present a fault identification technique in transformers using deep learning tools for the analysis of IR images of 10/0.4 kV distribution transformers. The authors applied the AlexNet CNN-based learning algorithm in Matlab for processing the raw image datasets. The work restated that as a diagnostic tool, the IR imaging tool can yield important insight into the heat intensity and its distribution within power equipment as well as the rate of energy flow from the hotspots in the device. These data can be applied to isolate the level of disorder within the unit. The authors emphasized that unequal distribution of heat radiated from the cooling fins could be a result of the presence of air pouches or low oil levels therein. The research described the CNN architecture with the human visual system and how its layers effect the process of image recognition. The case study involves image acquisition during normal and abnormal conditions of the equipment. The datasets are labeled based on the brightness of identified hotspots. However, large convolution filters used in the AlexNet is not quite optimal because of issue of overfitting, longer training time, etc. because it would increase the number of parameters, thereby raising the amount of unrelated features that can be extracted which limits the learning ability of the algorithm with respect to features common to different situations and therefore generalize poorly. And the depth of the AlexNet is not sufficient in comparison with other deep models like ResNet, VGGNet, etc. Again, the adoption of the normal distribution instead of the Xavier Glorot (XG) method for weight initialization, hinders the ability of the algorithm to overcome issues of vanishing gradient. The XG function can initialize weights of neural networks in a manner to limit the variance of the activations in each layer, thereby solving the problem of vanishing or exploding gradients, hence the XG function has been applied to later versions of AlexNet.

Similarly, Shiravan et al. [42] IRT and computational fluid dynamics (CFD) of 3 transformers of different ratings; 630 kVA, 400 kVA, and 50 kVA. And predicted faults in the transformers’ cooling mechanism by a combination of both IRT and CFD techniques. The authors developed a thermal model of the transformers using nonlinear thermal resistive components which were developed further into differential equations considering design-dependent parameters (DDP) and empirical factors (EFs). The CFD technique used was finite volume based and simulated in ANSYS FLUENT version 19. Validation of the proposed model was done by comparing the maximum error and the root mean square error (RMSE) of the CFD model with that of the thermal model. Decision criteria were based on the difference in temperature between both methods was more than 4 °C for the radiator or top oil, then the unit is not okay; if it is 6 °C or more, then the unit is not only faulty but its cooling system is not good. The challenge of using the RMSE of evaluation metric lies in the fact that it is affected by the scale of data used such that as the error rises its value also increases. The RMSE is also dependent on the distribution of outliers within the dataset and would normally increase with the extent of the dataset.

But Jiang et al. [43] proposed a mask R-CNN and modified Pulse Coupled neural network joint method for determining faults in bushings of electrical equipment using IRT images. Noting that problems in bushings make up 5–50% of faults in transformers, the authors used relative position and coverage area of fault as parameters of concern for feature extraction. The work noted that when PCNN method was used for segmenting the target pixels around the feature-mapped areas and extraction of the faulty regions, the result was an image with lots of difficulties to remove noises. This encouraged them to apply a simple linear iterative clustering (SLIC) to produce definite frames, thereby limiting undesirable effects of borders on the PCNN by finding the mean of nearby colors. By this method, the metrics of the PCNN were enhanced. Different bushing fault conditions were evaluated as dielectric loss, connection fault, oil leakage, and partial discharge with regard to coverage areas. With over 2000 images from 51 stations in China, a batch size of 2, number of epochs and iterations per epoch set to 30 and 100, respectively; the work used validation steps of 50 and simulated with an XEON-W3 processor running Nvidia GTX 1080 graphics on Ubuntu 16.04 LTS operating system installed on 64 GB memory. Nevertheless, the use of the PCNN could become cumbersome with respect to the complexity and number of parameters required to be set for optimum performance according to Huang, et al. [44]. In the same way, the author’s use of SLIC though commendable based on its preeminence over other methods could come with more challenges at the grouping stage of the K-means process, any improperly classified pixels may be transferred to yield undesirable superpixels and little regions are joined adjoining neighbor irrespective of the semblance in terms of the color Kim et al. [45].

Rotating machines

Electric machines constitute the main engine driving the power sector. Whether AC or DC-operated, they can be deployed as power generators or motors. Hence, they can function as sources of power or loads. The work by Phuc et al. [46] developed a thermal model independent of exact motor identification. The authors presented a Lumped-Parameter Thermal Network (LPTN) and Dual Kalman Filtering (DKF)-based technique to monitor the temperature profile of Rotors in Induction Machines. And stated that the accuracy of the system depends on the thermal parameters, low computational effort, and rotor temperature observed with IRT and applied to a dual Kalman filter for reviewing the evolution of thermal model parameters.

However, Zarghani et al. [47] revealed that the LPTN requires lots of human expertise to model the circuit especially when a large number of parameters are present, thereby making it more challenging to debug errors. Also, there could be a lack of clarity on the uncertainty concerning the requisite model of the power loss. Therefore, it would be very difficult to model the temperature using deep learning-based algorithms in line with the little number of model parameters that can fit easily with the LPTN to give the same estimation accuracy Kirchgässuer et al. [48].

Resendiz-Ochoa et al. [49] proposed automatic Infrared thermography for analyzing faults in induction motors. The proposed technique detects the concerned area with an automatic image segmentation based on Otsu thresholding process. The goal is to accomplish features extraction of temperatures for thermal analysis of the defective induction motor. The technique has the potential for automatic fault classification in pattern recognition to be applied in image segmentation applications. Anurag Choudhary et al. [50] diagnosed defects in the bearings of rotating machines by observing their operational thermal images and developed a Convolutional Neural Network better than the associated ANN. Six conditions of bearings including a good state were reviewed and compared, employing ANN and LeNet-5-based CNN structure. The suggested method consists of bearing thermal image data collection, extraction of features, and training of the ANN and CNN model, each used as the classifier to classify the different bearing conditions. The authors tested the CNN method on large datasets with up to 99.80% classification accuracy which significantly outperforms the ANN. The technique was not designed for a particular fault, hence Padmanabhan et al. [51] presented a method for identifying inter-turn faults in the status of operational electric machines. The authors applied Thermal and Magnetic Sensor Arrays to the Stator Ends of the winding, thereby obtaining the distribution of the heat and magnetic flux throughout the region encompassed by the end-winding. During abnormal conditions, it would be easy to observe the asymmetry in the thermal and magnetic signature caused by inter-turn short circuits. The presented HESA technique offered better versatility and early fault identification than the comparable IRSA method when viewed within a 1.5-kW induction motor test set. The potential of the techniques lies in their reliability, fast fault detection, and scalability to different types of machines. Moreover, there is an opportunity for future research into an estimation of the extent or expected severity of defects for different operating conditions.

Power electronics

As the power system becomes more sophisticated, demand for improved performance grows, and the need for automatic and fast-switching electrical devices becomes critical. Moreover, the need to ensure optimal performance of the power system makes power electronics systems vital for optimum system operation. Weifei Li et al. [52] presented a method of predicting in real-time the temperature at the PN junctions of an IGBT-based inverter with a concise specified model of the power loss. The model combines dual-impedance temperature model of the inverter with considerations for the results of computation fluid dynamics (CFD) analysis. CFD-based techniques are usually complex models that take more computing resources that impact on cost and simulation time. Also, they are generally built on approximations of practical models thereby limiting the accuracy of the models. They can be prone to errors due to factors like boundary conditions, mesh size, etc.

Leppänen et al. [53] proposed how to mitigate failure in power diodes used in converters by investigating humidity-induced failure in power diodes. Bearing in mind the adverse conditions such as high humidity, extreme temperature, and elevated reverse voltage profiles most power converters are regularly exposed during operation. Boost stage power semiconductor modules from two vendors were investigated. The leakage current measurement of the modules used for the study was monitored on-site during the test for two types of passivated modules, viz., glass passivation and polyimide passivation. And water treeing was found to dissolve lead in particular areas, and glass passivation could lead to multi-modal failures when exposed to the H3TRB-HVDC test conditions and appear as hotspots in the panels. The hotspots could stem from voltage-blocking degradation that finally causes short circuits. According to the authors, a “lock-in” thermography above the glass passivation enclosing the high voltage edge termination elements is susceptible to challenging humidity conditions, this is where the hotspots were observed. Further analysis using a scanning electron microscope (SEM) confirmed that the edge termination contained several tree-like structures growing from both ends of the edge termination on top of the passivation film. The use of SEM could be costly in many ways based on the price paid for the power electronic unit.

Surge arresters

Arup Kumar Das et al. [54], were interested in using IRT for the assessment of surge arresters (SA) using a transfer learning approach to gauge the extent and presence of surface contaminants such as dust, salt, etc. on metal oxide surge arrester (MOSA) to monitor the condition of the device in power installation and counteract the tendency of the SAs to fail before its lifespan. The authors applied IRT and studied the third harmonic leakage current drain by the device. Thereafter analyzed the relationship between the third harmonic leakage current and the temperature of the arrester using a neural network. The neural network was trained with three input parameters of arrester temperature, ambient temperature, and relative humidity of the environment, whereas its output is the third harmonic resistive current. IR thermal images of metal oxide surge arrester at different levels of pollution were observed, preprocessed, and its features of interest were extracted when applied to the ResNet50-based CNN. And, produced up to 98% testing accuracy on simulation with an 11 kV arrester. The authors applied the extracted features to classifiers like k-nearest neighbor, support vector machine, naïve Bayes, and random forest observing that the random forest showed the best performance. In terms of monitoring different pollution levels, the proposed technique shows the capacity to identify the severity of contamination of the surface of the surge arrester with good accuracy. The techniques give automatic, fast, reliable, and remote observation. As the SA ages, its leakage current increases with the deterioration of performance. The aging factor can also be monitored when the leakage current is observed with the temperature at the surface of the device. According to He et al. [55] the concept of ResNet for feature extraction was a game-changer in many computer vision applications as it has enabled the training of datasets with deeper neural networks without compromising the training error and addresses the issues of vanishing gradient, improved model accuracy and relatively fast training time through identity mapping or skip connections.

Andrade et al. [56] investigated the issue of heat transfer in ceramic surge arresters using thermography and computer simulations based on finite element analysis (FEA). Normally, the air gap within arresters could lead to more thermal resistance between the varistor’s column and the ambient. So, techniques for proper analysis of the phenomenon are necessary and should include consideration for heat loss via conduction, convection, and radiation. The authors proposed the use of computational simulations in combination with thermography as a tool for a temperature-based estimate of the varistor’s state.

For the analysis, thermography measurements were performed in a 69-kV ceramic-housed arrester subject to a thermal cycle and the results were compared with finite element simulations to picture the relationship between varistor’s temperature and the temperature of the outer part of the arrester to aid field assessment of the polymeric and porcelain-housed surge arrester. The work highlighted better results for varistors enclosed with polymeric than ceramic materials. And showing the potential of the technique to be applied for optimization of parameters of heat transfer mechanisms that most define the thermal behavior of the arrester. The authors saw the prospect of developing an equivalent mathematical expression that can relate external temperature and that of the internal components for predicting the temperature of the varistor column. However, this method would not be generic because issues of heat loss are related to the physical shape and geometry of the device, hence different arrester shapes would need specific simulations to be effective.

Solar photovoltaic modules

Condition monitoring of Solar Photovoltaic systems is one area that has attracted the interest of researchers in power system analysts in recent years. The growing interest could be a result of a worldwide policy shift toward renewable sources of energy by the year 2030. There is a compelling need to optimize available sources of renewable energy by preventing or reducing energy loss. And, as solar PV systems become the easy focus of researchers, with hotspots known to contribute up to 49% of faults in PV modules. The concept of IRT has become a veritable choice technique for monitoring their performance. Pramana and Dalimi [57] were interested to identify hotspot faults in PV Modules with the hope of classifying them appropriately. Table 2 depicts the main cause of hotspot faults in PV panels. The hotspot faults can manifest due to internal or external factors in the PV modules. The external hotspots are caused by the prevalence of adverse environmental conditions within which the modules are operated, whereas the internal issues are usually linked to power diode failure.

Table 2 Root cause analysis of hotspots in pv panels

Simons and Meyer [67], presented a method for Detecting and analyzing the occurrence of hotspots in PV solar cells by scanning the face of the modules to observe the thermal profile of the cells in reverse bias. Heterogeneous thermal distribution across the modules indicates hotspots. Thereafter, the authors used a scanning electron microscope (SEM), to further observe microscopic images of areas where the hotspots manifested and identified irreparable damage to the cells due to the heating effect. The work highlighted the relationship between contaminated portions of cells, especially by transition elements along with oxygen, carbon, iron, and platinum, and the hotspot phenomenon.

Pallavi et al. [68], highlighted the effect of hotspots on the energy output from solar PV modules. Localized heating within a solar cell gives rise to hotspot formation, which further leads to module damage and system degradation. A detailed model encompassing non-uniform temperature distribution across a series of connected PV cells is presented and is seen to give high accuracy with respect to the experimental measurements as compared to the average temperature and the standard conditions-based output prediction.

In this work, the output performance of such PV modules is estimated based on non-uniform temperature distribution and is shown to match well with the experimental results with 98% accuracy. The proposed method performs better than the average temperature-based approach with more than 5% improvement in maximum power point prediction accuracy. The proposed method would be inefficient for monitoring large-scale PV installation. And, there is a need for the acquisition and auto-processing of numerous IR images. And, the need to apply modern deep learning methods would entail the acquisition of numerous IR images which would be quite challenging. So, artificial intelligent solar panel monitoring systems would be appropriate as proposed by Wang et al. [69] whose work deployed a combination of U-Net neural network and a supervised machine learning model-like decision tree to achieve 99.8% fault diagnosis ability of PV panel faults. However, U-Net’s use of a large number of parameters as a result of more layers and skip connections could cause overfitting when used on limited images.

Power lines

Transmission and distribution lines, generally referred to as power lines are to the power system what arteries and veins are to the human body. They are the physical channels through which electrical power is conveyed from one node to the other. They are the most vulnerable part of the power system and can reach thousands of kilometers in length, passing through very challenging weather and terrain; and must be in good condition always. Hence, the need to monitor their operating state is paramount to linesmen and system operators. Jalil et al. [70] proposed a technique for monitoring power lines using a FLIR A65sc IR camera mounted on an unmanned aerial vehicle (UAV) to acquire infrared and visible images of power lines, which they subjected to a fusion-based images processing algorithm. And passed through a canny edge detection model, before detecting the linear features in the images using the Hough transformation technique. The power lines were segregated from other portions of the pictures and a thresholding method based on histogram technique was used to profile and identify the hotspots in the lines. But Vozikis and Jansab [71] opined that one major issue with the Hough stems from the fact that it is highly susceptible to giving incorrect results when encountering complex geometric images as well as dark shadows in images.


The variety of methods applied and accompanying results in different works concerning infrared thermography are presented in Table 3 shows the predisposition of researchers to adopt intelligent techniques over classical methods when developing their models.

Table 3 Results of techniques used for infrared thermography


Infrared Thermography for the past two decades has continued to attract attention as a useful tool for monitoring electrical power equipment and installed system especially when in operation. There is no limit to areas they can be applied, whether biological or physical systems. Once heat is radiated from such objects above absolute temperature, then IR cameras can provide instantaneous images of the temperature distribution within that equipment. The focus of most research in IRT-based condition monitoring lies in the development of intelligent systems and improving the accuracy of neural network-based models. There is a need for more work in comparative assessment of the response times for most of the models. And, the importance of more research in developing models that would not require lots of IR images to train, yet give very good results, because data acquisition in this research area takes a lot of time, energy, human expertise, and money. There is a need to develop more affordable IR cameras with good video recording and storage features which would greatly aid data acquisition of infrared images of equipment and installations.

Availability of data and materials

Not Applicable.



Average precision


Area under the receiver operating characteristics


Bag of visual word


Computational fluid dynamics


Condition monitoring


Convolutional neural network


Computed tomography


Design-dependent parameters


Dissolved gas analysis


Dual Kalman filtering


Discrete wavelet transforms


Empirical factors


Extremely randomized tree


Furan current analysis


Finite element analysis


Frequency response analysis


Generative adversarial network


High humidity, high temperature and high voltage reverse bias test


Hall effect sensor array


Health index


Hue, saturation, intensity


Hue, saturation, lightness


Hue, saturation, value


Joint Photographic Experts Group


Recovery voltage method


High voltage


Insulated gate bipolar transistor


Induction motor/machine


Infrared thermopile sensor array




Infrared thermography


Joint systems engineering group algorithm


Lumped parameter thermal network


Mean absolute deviation


Motor current signature analysis


Multi-layer perceptron


Metal oxide surge arrester


Mean square error


Normalized cross-correlation


Oil natural, air natural


Oil forced, air forced


Pulse coupled neural network


Portable network graphics


Peak signal-to-noise ratio




Residual network


Random forest


Resistive–inductive and capacitive


Root mean square error


Surge arrester


Scanning electron microscope


Scale invariant feature transform


Simple linear iterative cluster


Signal-to-noise ratio


Speeded up robust feature


Support vector machine


Unmanned aerial vehicle


Variational autoencoder


Wasserstein autoencoder reconstruction-based differential image classification


Wavelet transform


  1. Dhimish M, Holmes V, Mather P, Sibley M (2018) Novel hot spot mitigation technique to enhance photovoltaic solar panels output power performance. Sol Energy Mater Sol Cells 179:72–79.

    Article  Google Scholar 

  2. Solor’zano J, Egido MA (2014) Hot-spot mitigation in PV arrays with distributed MPPT (DMPPT). Sol Energy 101:131–137.

    Article  Google Scholar 

  3. Najmi A, Kaore S, Ray A, Sadasivam B (2020) Use of handheld infrared thermometers in COVID-19 pandemic for mass screening: understanding its implications through a case report. J Fam Med Prim Care 9(10):5421–5422.;PMCID:PMC7773056

    Article  Google Scholar 

  4. Bach AJ, Stewart IB, Disher AE, Costello JT (2015) A comparison between conductive and infrared devices for measuring mean skin temperature at rest, during exercise in the heat, and recovery. PLoS ONE.

    Article  Google Scholar 

  5. Madding R, Orlove GL, Lyon BR (2007) The Importance of spatial resolution in infrared thermography temperature measurement: three brief case studies. Proc SPIE: Int Soc Opt Eng.

    Article  Google Scholar 

  6. Usamentiaga R, Venegas P, Guerediaga J, Vega L, Molleda J, Bulnes FG (2014) Infrared thermography for temperature measurement and non- destructive testing. Sensors.

    Article  Google Scholar 

  7. Hernandez A, Ivan A, Ramirez IZ, Cuellar AYJ, Osornio-Rios RA, Donderis-Quiles V, Antonino-Daviu JA (2022) Infrared thermography smart sensor for the condition monitoring of gearbox and bearings faults in induction motors. Sensors.

    Article  Google Scholar 

  8. Feng JQ, Sun P, Tang WH, Buse DP, Wu QH, Richards Z, Fitch J (2022) Implementation of a power transformer temperature monitoring system. In: IEEE proceedings, international conference on power system technology. pp 1980–1983.

  9. Younus AM, Yang BS (2010) Wavelet co-efficient of thermal image analysis for machine fault diagnosis. In: Prognostics and system health management conference (PHM2010 Macao).

  10. Manana M, Arroyo A, Ortiz A, Renedo CJ, Perez S, Delgado F (2011) Field winding fault diagnosis in DC motors during manufacturing using thermal monitoring. Appl Therm Eng 31(5):978–983.

    Article  Google Scholar 

  11. Eftekhari M, Moallem M (2013) A novel indicator of stator winding interturn fault in induction motor using infrared thermal imaging. Infrared Phys Technol 61:330–336.

    Article  Google Scholar 

  12. Cui H, Xu Y, Zeng J, Tang Z (2013) The methods in infrared thermal imaging diagnosis technology of power equipment. In: IEEE 4th international conference on electronics information and emergency communication, Beijing.

  13. Jadin MS, Ghazali KH, Taib S (2013) Thermal condition monitoring of electrical installations based on infrared image analysis. In: Saudi international electronics, communications and photonics conference, Riyadh. pp 1–6.

  14. Huda ASN, Taib S (2013) Suitable features selection for monitoring thermal condition of electrical equipment using infrared thermography. Infrared Phys Technol 61:184–191.

    Article  Google Scholar 

  15. Huda ASN, Taib S (2013) Application of infrared thermography for predictive/preventive maintenance of thermal defect in electrical equipment. Appl Therm Eng 61:220–227.

    Article  Google Scholar 

  16. Garcia-Ramirez AG, Morales-Hernandez LA, Osornio-Rios RA, Garcia-Perez A, Romero-Troncoso RJ (2014) Thermographic technique as a complement for MCSA in induction motor fault detection. In: IEEE 2014 International conference on electrical machines (ICEM), Berlin, pp 1940–1945.

  17. Karvelis P, Georgoulas G, Stylios CD, Tsoumas IP, Antonino-Daviu JA, Picazo-Ródenas MJ, Climente-Alarcón V (2014) An automated thermographic image segmentation method for induction motor fault diagnosis. In: IECON 2014: 40th annual conference of the IEEE industrial electronics society, Dallas, pp 3396–3402.

  18. Janssens O, Schulz R, Slavkovikj V, Stockman K, Loccufier M, Van de Walle R, Hoecke SV (2015) Thermal image based fault diagnosis for rotating machinery. Infrared Phys Technol 73:78–87.

    Article  Google Scholar 

  19. Zou H, Huang F (2015) Novel intelligent fault diagnosis method for electrical equipment using infrared thermography. Infrared Phys Technol 73:29–35.

    Article  Google Scholar 

  20. Munoz-Ornelas O, Elvira-Ortiz DA, Osornio-Rios RA, Romero-Troncoso RJ, Morales- Hernandez LA (2016) Methodology for thermal analysis of induction motors with infrared thermography considering camera location. In: IECON 2016: 42nd annual conference of the IEEE industrial electronics society, Florence.

  21. Ramirez-Nunez JA, Morales-Hernandez LA, Osornio-Rios RA, Antonino-Daviu JA, Romero-Troncoso RJ (2016) Self-adjustment Methodology of a thermal camera for detecting faults in industrial machinery. In: IECON 2016: 42nd annual conference of the IEEE industrial electronics Society, Florence.

  22. Singh G, Anil-Kumar TC, Naikan VNA (2016) Fault diagnosis of induction motor cooling system using infrared thermography. In: IEEE 6th international conference on power systems (ICPS), New Delhi.

  23. Singh G, Anil-Kumar TC, Naikan VNA (2016) Induction motor inter turn fault detection using infrared thermographic analysis. Infrared Phys Technol 77:277–282.

    Article  Google Scholar 

  24. Dragomir A, Adam M, Andruşcă M, Munteanu A (2016) Aspects concerning the influence of environmental factors in infrared monitoring of electrical equipment. In: IEEE 2016 international conference and exposition on electrical and power engineering (EPE 2016), Iasi, Romania, pp 133–138.

  25. Khan Q, Khan AA, Ahmad F (2016) Condition monitoring tool for electrical equipment: thermography. In: IEEE international conference on electrical, electronics, and optimization techniques (ICEEOT)—2016, Chennai, pp 2802–2806.

  26. Dutta T, Sil J, Chottopadhyay P (2016) Condition monitoring of electrical equipment using thermal image processing. In: IEEE First international conference on control, measurement and instrumentation (CMI), Kolkata, pp 311–315.

  27. Resendiz-Ochoa E, Osornio-Rios RA, Benitez-Rangel JP, Morales-Hernandez LA (2017) Segmentation in thermography images for bearing defect analysis in induction motors. In: IEEE 2017 IEEE 11th international symposium on diagnostics for electrical machines, power electronics and drives (SDEMPED) Tinos, pp 572–577.

  28. Liu Z, Duan L, Fu Q, Wang J, Shi T (2017) Infrared image combined with CNN based fault diagnosis for rotating machinery. In: International conference on sensing, diagnostics, prognostics, and control, Shanghai, pp 137–142.

  29. Lopez-Perez D, Antonino-Daviu J (2017) Failure detection in industrial electric motors through the use of infrared-based isothermal representation. In: IECON 2017: 43rd annual conference of the IEEE industrial electronics society, Beijing. pp 3822–3827.

  30. Dragomir A, Adam M, Andruçcâ M, Munteanu A, Boghiu E (2017) Considerations regarding infrared thermal stresses monitoring of electrical equipment. In: International conference on electromechanical and power systems (SIELMEN), Iasi, pp 100–103.

  31. Mariprasath T, Kirubakaran V (2018) A Real time study on condition monitoring of distribution transformer using thermal imager. Infrared Phys Technol.

    Article  Google Scholar 

  32. Sangeetha MS, Nandhitha NM, Karthikeyan S, Venkatesh N (2018) Mathematical relationship between hotspot temperature, emissivity and distance in thermographs for condition monitoring of electrical equipment. In: Proceedings of the 2nd international conference on trends in electronics and informatics (ICOEI 2018) IEEE conference record, pp 984–988.

  33. Fambrini F, Iano Y, Caetano DG, Rodriguez AAD, Moya C, Carrara E, Arthur R, Cabello FC, Zubem JV, DeVal Cura LM, Destro-Filho JB, Campos JB, Saito JH (2018) GPU cuda JSEG segmentation algorithm associated with deep learning classifier for electrical network images identification. In: International conference on knowledge based and intelligent information and engineering systems, KES2018, 3–5 September 2018, Belgrade, Serbia, Procedia Computer Science. 126:557–565.

  34. Sahu M, Sharma SR, Singh A, Kumar R, Sood YR (2020) An improved infrared thermography technique for hotspot temperature, per unit life and aging accelerating factor computation in transformers. In: International conference on computing, power and communication technologies (GUCON) galgotias university, India.

  35. Najafi M, Baleghi Y, Gholamian SA, Mirimani SM (2020) Fault diagnosis of electrical equipment through thermal imaging and interpretable machine learning applied on a newly-introduced dataset. In: 2020 6th Iranian Conference on signal processing and intelligent systems (ICSPIS), Mashhad, pp 1–7.

  36. Vidhya R, Ranjan PV, Shanker NR (2021) Transformer breather thermal image decomposition for fault diagnosis. In: 7th International conference on electrical energy systems (ICEES) Chennai, pp 448–453.

  37. Mahami A, Rahmoune C, Bettahar T, Benazzouz D (2021) Induction motor condition monitoring using infrared thermography imaging and ensemble learning techniques. Adv Mech Eng 13(11):1–13.

    Article  Google Scholar 

  38. Utami NY, Tamsir Y, Pharmatrisanti A, Gumilang H, Cahyono B, Siregar R (2009) Evaluation condition of transformer based on infrared thermography results. In: Proceedings of the 9th international conference on properties and applications of dielectric materials, Harbin, pp 9–23

  39. Asiegbu GO, Haidar AMA, Hawari K (2013) Thermal defect analysis on transformer using a RLC network and thermography. Circuits Syst 4:49–57.

    Article  Google Scholar 

  40. Fang J, Yang F, Tong R, Yu Q, Dai X (2021) Fault diagnosis of electric transformers based on infrared image processing and semi-supervised learning. Glob Energy Interconnect 4(6):596–607.

    Article  Google Scholar 

  41. Mlakić D, Nikolovski S, Majdandžić L (2018) Deep learning method and infrared imaging as a tool for transformer faults detection. J Electr Eng 6:98–106.

    Article  Google Scholar 

  42. Shiravand V, Faiz J, Samimi MH, Mehrabi-Kermani M (2021) Prediction of transformer fault in cooling system using combining advanced thermal model and thermography. Gener Transmission Distrib 15:1972–1983.

    Article  Google Scholar 

  43. Jiang J, Bie Y, Li J, Yang X, Ma G, Lu Y, Zhang C (2021) Fault diagnosis of the bushing infrared images based on Mask R-CNN and improved PCNN joint algorithm. High Volt 6:116–124

    Article  Google Scholar 

  44. Huang C, Tian G, Lan Y, Peng Y, Ng EYK, Hao Y, Cheng Y, Che W (2019) A new pulse coupled neural network (PCNN) for brain medical image fusion empowered by shuffled frog leaping algorithm. Front Neuro Sci.

    Article  Google Scholar 

  45. Kim KS, Kang MC, Zhang D, Ko SJ (2013) Improved simple linear iterative clustering superpixels. In: Digest of technical papers IEEE: 17th international symposium on consumer electronics (ISCE), pp 259–260.

  46. Phuc PN, Bozalakov D, Vansompel H, Stockman K, Crevecoeur G (2021) Rotor temperature virtual sensing for induction machines using a lumped-parameter thermal network and dual Kalman filtering. IEEE Trans Energy Convers.

    Article  Google Scholar 

  47. Zarghani A, Torkaman H, Arbab N, Toulabi MS (2022) Lumped parameter thermal network for thermal analysis of a rotor-excited axial flux switching machine with electromagnetic-thermal design. Measurement 193:110971.

    Article  Google Scholar 

  48. Kirchgässner W, Wallscheid O, Böcker J (2023) Thermal neural networks: Lumped-parameter thermal modeling with state-space machine learning. Eng Appl Artif Intell.

    Article  Google Scholar 

  49. Resendiz-Ochoa E, Osornio-Rios RA, Benitez-Rangel JP, Romero-Troncoso RDJ, Morales-Hernandez LA (2017) Induction motor failure analysis: an automatic methodology based on infrared imaging. IEEE Access.

    Article  Google Scholar 

  50. Choudhary A, Mian T, Fatima S (2021) Convolutional neural network based bearing fault diagnosis of rotating machine using thermal images. Measurement.

    Article  Google Scholar 

  51. Kumar PS, Xie L, Halick MSM, Vaiyapuri V (2021) Stator end-winding thermal and magnetic sensor arrays for online stator inter-turn fault detection. IEEE Sens J 21(4):5312–5321.

    Article  Google Scholar 

  52. Li W, Li G, Sun Z, Wang Q (2021) Real-time estimation of junction temperature in IGBT inverter with a simple parameterized power loss model. Microelectron Reliab 127:2021.

    Article  Google Scholar 

  53. Leppänen J, Ross G, Vuorinen V, Ingman J, Jormanainen J, Paulasto-Krӧckel M (2021) A humidity-induced novel failure mechanism in power semiconductor diodes. Microelectron Reliab 123:2021.

    Article  Google Scholar 

  54. Das AK, Dey D, Chatterjee B, Dalai S (2021) A transfer learning approach to sense the degree of surface pollution for metal oxide surge arrester employing infrared thermal imaging. IEEE Explore.

    Article  Google Scholar 

  55. He K, Zhang X, Ren S, Sun J (2016) Deep residual learning for image recognition. In: Proceedings of the IEEE conference on computer vision and pattern recognition, pp 770–778.

  56. Andrade AF, Costa EG, Fernandes JMB, Alves HMM, Cícero RC, Filho A (2019) Thermal behavior analysis in a porcelain-housed ZnO surge arrester by computer simulations and thermography. High Volt 4(3):173–177.

    Article  Google Scholar 

  57. Pramana PAA, Dalimi R (2020) Hotspot detection method in large capacity photovoltaic (PV) Farm. In: IOP Conf. Series: materials Science and Engineering, ICETIR 2020, Vol. 982, IOP Publishing.

  58. Paggi M, Corrado M, Rodriguez MA (2013) A multi-physics and multi-scale numerical approach to microcracking and power-loss in photovoltaic modules. Compos Struct 95:630–638.

    Article  Google Scholar 

  59. Roth T, Siebert R, Meyer K (2014) From short-term hotspot measurements to long-term module reliability. Energy Proc 55:504–508.

    Article  Google Scholar 

  60. Van-Mölken JI, Yusufoğlu UA, Safiei A, Windgassen H, Khandelwal R, Pletzer TM, Kurza H (2012) Impact of micro-cracks on the degradation of solar cell performance based on two-diode model parameters. Energy Proc 27:167–172

    Article  Google Scholar 

  61. Berardone I, Corrado M, Paggi M (2014) A generalized electric model for mono and polycrystalline silicon in the presence of cracks and random defects. Energy Proc 55:22–29.

    Article  Google Scholar 

  62. El-Gharabawy ASA (2018) Review on corrosion in solar panels. Int J Smart Grid 2(4):218–220

    Google Scholar 

  63. Fairbrother A, Gnocchi L, Ballif C, Virtuani (2022) Corrosion testing of solar cells: Wear-out degradation behavior. Solar Energy Mater Solar Cells 248:111974

    Article  Google Scholar 

  64. Dhakshinamoorthy M, Sundaram K, Murugesan P, David PW (2022) Bypass diode and photovoltaic module failure analysis of 1.5 kW solar PV array. Energy Sourc Part A: Recov Utilization Environ Effects 44(2):4000–4015.

    Article  Google Scholar 

  65. Maghami MR, Hizam H, Gomes C, AmranRadzi M, Rezadad MI, Hajighorbani S (2016) Power loss due to soiling on solar panel: a review. Renew Sustain Energy Rev 59(2016):1307–1316.

    Article  Google Scholar 

  66. Pareek S, Chaturvedi N, Dahiya R (2017) Optimal interconnections to address partial shading losses in solar photovoltaic arrays. Sol Energy 155:537–551.

    Article  Google Scholar 

  67. Simon M, Meyer EL (2010) Detection and analysis of hot-spot formation in solar cells. Sol Energy Mater Sol Cells 94(2):106–113.

    Article  Google Scholar 

  68. Bharadwaj P, Karnataki K, John V (2018) Formation of hotspots on healthy PV modules and their effect on output performance. In: IEEE 7th world conference on photovoltaic energy conversion (WCPEC) (A Joint Conference of 45th IEEE PVSC, 28th PVSEC & 34th EU PVSEC), Waikoloa, pp 0676–0680.

  69. Wang X, Yang W, Qin B, Wei K, Ma Y, Zhang D (2022) Intelligent monitoring of photovoltaic panels based on infrared detection. Energy Rep 8:5005–5015.

    Article  Google Scholar 

  70. Jalil B, Pascali MA, Leone GR, Martinelli M, Moroni D, Salvetti O (2019) To identify hot spots in power lines using infrared and visible sensors. In: Choroś K, Kopel M, Kukla E, Siemiński A (eds) Multimedia and network information systems. MISSI 2018. Advances in intelligent systems and computing, vol 833. Springer, Cham.

    Chapter  Google Scholar 

  71. Vozikis G, Jansab J (2008) Advantages and disadvantages of the hough transformation in the frame of automated building extraction. In: The international archives of the photogrammetry, remote sensing and spatial information sciences. 2008, Vol. XXXVII. Part B3b. Beijing, pp 719–724

  72. Zou H, Huang F (2015) An intelligent fault diagnosis method for electrical equipment using infrared images. In: Proceedings of the 34th Chinese control conference July 28–30, Hangzhou, pp 6372–6376

  73. Fanchiang KH, Huang YC, Kuo CC (2021) Power electric transformer fault diagnosis based on infrared thermal images using wasserstein generative adversarial networks and deep learning classifier. Electronics 10:1161.

    Article  Google Scholar 

  74. Fang J, Yang F, Tong R, Yu Q, Dai X (2021) Fault diagnosis of electric transformers based on infrared image processing and semi-supervised learning. Global Energy Interconnection 4(6):596–607.

    Article  Google Scholar 

  75. Fanchiang KH, Kuo CC (2022) Application of thermography and adversarial reconstruction anomaly detection in power cast-resin transformer. Sensors 22:1565.

    Article  Google Scholar 

Download references


Not applicable.


There has been no significant financial support for this work that could have influenced its outcome.

Author information

Authors and Affiliations



All authors read and approved the final manuscript. Each author made substantial contributions to the conception or design of the work and analysis, conclusion, and subsequent revision.

Corresponding author

Correspondence to Ezechukwu Kalu Ukiwe.

Ethics declarations

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ukiwe, E.K., Adeshina, S.A. & Tsado, J. Techniques of infrared thermography for condition monitoring of electrical power equipment. Journal of Electrical Systems and Inf Technol 10, 49 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: