Condition Assessment of Power Transformers through DGA Measurements Evaluation Using Adaptive Algorithms and Deep Learning

This paper presents the developed procedure for condition assessment of power transformer fleet in a utility responsible for supplying electricity in Thailand metropolitan area. The condition assessment is performed by evaluating the quickly-updated test results. The assessment is classified into seven main categories as dissolved gas analysis, oil quality in main tank, oil quality in on-load tap changer, bushing, lightning arrester, visual inspection and thermal scan. The scoring and weighting techniques are applied for condition assessment of each category. The procedure for condition assessment of power transformer components have been further developed as web application software, which displays the results as traffic light indicators. Finally, the developed web application software is now used as a decision support tool for power transformer maintenance in a utility in order to recommend the proper maintenance action.

In an electric power system, a power transformer is one of the most critical equipment and cannot be separated from possible abnormal conditions due to fault. Dissolved Gas Analysis (DGA) is a reliable technique for detecting the presence of a fault condition that just occurred in oil immersed transformer. Basically DGA is a process to calculate the levels or values of hydrocarbon gases that are formed due to abnormalities. The gas inside the transformer can function as a marker for various types of faults. In this paper, for DGA testing and evaluation of the type of fault in the power transformer using the interpretation of IEEE std C57.104 and IEC 60599. The method used for DGA testing according to the IEEE interpretation is total dissolved combustible gas (TDCG), key gas, doernenburg ratio, and roger ratio. While the method used for DGA testing at IEC 60599 is the duval triangle, the basic gas ratio and CO2/CO ratio. From the results of DGA tests that have been carried out, all of these methods will be used to ascertain the type of fault that occurs in the power transformer.

DGA is more sensitive and reliable technique for detecting and evaluating the incipient fault conditions in oil‐immersed transformers. The most frequently interpretation technique that used in dissolved gas analysis is Duval triangle due to its simplicity and accuracy. Unfortunately the impacts of Ethane (C 2 H 6 ) and Hydrogen (H 2 ) were not included in Duval triangle in spite of their importance in diagnosing certain faults types. In addition, the problem of no decision associated with some cases that lie out of the specified codes appeared in Doernenburg, Roger's and IEC techniques. Accordingly, this paper presented a new proposed graphical triangle technique using new three gas concentration ratios. These ratios were converted to new three ratio percentages of DGA based on the consideration of all five combustible gases. The aim of that was to overcome the conflict takes place in Duval triangle and other traditional techniques. The accuracy of the proposed triangle technique was evaluated using practical DGA database obtained from different transformers of different ratings life spans reported by Egyptian Electricity Holding Company, IEC TC 10 and related databases surveyed from actual incipient cases. The comparison with Duval and other traditional techniques showed good diagnostic accuracy of the new proposed triangle technique.

Condition assessment for power transformer requires not only integrating the known artificial intelligence (AI) technology, but also exploiting the interrelation of the measured data. According to the association rule of information data and the variable weight synthesizing theory of factor spaces, an assessment method of transformer condition was proposed in this paper. Via analyzing the interrelation of the independent status parameters (ISP) and transformer fault types, the set of synthetic status parameters (SSP) can be built up. For avoiding interference from subjective experience, association rule theory was used to calculate the constant weight coefficients (CWC) of the ISPs. Since the true transformer condition may not always be accurately reflected under the condition of CWCs of a few SSPs, the method of variable weight synthesizing was used for computing the variable weight coefficients (VWC) of the SSPs. Then, combining with the existing maintenance procedures, a preferable condition assessing system of power transformer was proposed. Operational example proved the condition assessing system may reflect the real operation condition of power transformer.

The application of advanced off-line methods and procedures for condition assessment of new and service-aged power transformers are discussed. The selection of the diagnostic methods is based on the main operational stresses (electrical, mechanical, thermal) and on typical failure modes: (1) detection of polar components in the paper/oil insulating system using frequency domain spectroscopy (FDS) or using measurements of polarisation/depolarisation currents (PDC) in the time domain, (2) localization of inhomogeneities in the insulating system using sensitive partial discharge detection (PD) and (3) detection of deformations and movement of windings in the active part using measurement of the transfer function (FRA).

Power transformer is one of the most important and expensive equipment in a power system. Its reliability directly affects a power system. To ensure the reliable operation of a power transformer, its condition needs to be continuously monitored and evaluated. Over the past two decades, a number of diagnostic techniques have been developed for transformer condition assessment such as dissolved gas analysis (DGA), degree of polymerization (DP) measurement, polarization and depolarization current (PDC) measurement, frequency domain spectroscopy (FDS), frequency response analysis (FRA), and partial discharge (PD) detection. However, the interpretations of measurement results acquired from these diagnostics are usually based upon the empirical models, which are sometimes inaccurate and incomplete especially in abnormal transformer operation scenarios. Therefore, accurate interpreting on the measurement data obtained by the above techniques and subsequently making explicit condition assessment of transformers is still a challenge task. Nowadays, considerable efforts have been made in the field of transformer condition monitoring and assessment. Majority efforts are dedicated in developing accurate transformer models and reliable transformer fault diagnosis systems. After completing a comprehensive literature review on various diagnostic techniques for transformers condition assessment, this thesis focuses on three main aspects of transformer’s health condition, including oil characteristics and dissolved gases in transformers, moisture concentration of oil-cellulose insulation and hot spot temperature of transformer windings. Since there is a lack of common framework for applying pattern recognition algorithms (i.e. data centric approaches) to interpret oil characteristics and DGA data, this thesis firstly provides a critical review on various pattern recognition techniques for power transformer insulation diagnosis using DGA and oil characteristics datasets. A general pattern recognition application framework is then proposed. The important issues for improving the applicability of pattern recognition techniques for transformer insulation diagnosis are also discussed. To improve the data quality of training database and enhance the diagnostic accuracy of pattern recognition algorithms, a hybrid algorithm, SMOTEBoost is proposed. It adopts Synthetic Minority Over-sampling Technique (SMOTE) to handle the class imbalance problem, where samples belonging to different fault types (insulation conditions) are unevenly distributed in the training database. By using the boosting approach for reweighting and grouping data points in the training database, the SMOTEBoost can facilitate pattern recognition algorithms consistently attaining desirable diagnosis accuracies. To solve the intricate difficulties in moisture estimation of transformer oil-cellulose insulation system, this thesis introduces two modelling approaches, i.e., multi-physics finite element modelling (FEM) and particle tracing method, where the temperature dependent moisture dynamics in transformers’ insulation system is taken into account. In multi-physics approach, moisture dynamics is modelled by coupling the electromagnetic, thermal, fluid flow and moisture migration physics simultaneously. In particle tracing method, moisture diffusion is formulated from a microscopic view of water molecules’ motion. Especially, the transmission probability of water molecules (termed as particles in the paper) is employed to correlate the microscopic particles’ motion with the macroscopic moisture distribution. Extensive ageing and moisture diffusion experiments have been conducted on a prototype transformer to verify the proposed modelling approaches for an accurate estimation of moisture in transformers. This thesis also proposes a distributed parameter model to investigate the effect of moisture dynamics on dielectric response of a transformer’s cellulose insulation. The correlation between moisture distribution (under non-equilibrium conditions due to thermal transients) and dielectric response parameters (dielectric losses and permittivity) of transformer cellulose insulation is revealed. The proposed methodology can help the proper interpretation of dielectric response measurement of field transformers under thermal transients. To overcome the inaccuracy in empirical thermal dynamic models, in this thesis a moisture dependent thermal model (MDTM) is developed for estimating transformer’s hot spot temperature. In this model, nonlinear thermal resistance is formulated by considering both oil and cellulose (paper and pressboard) of the transformer. Especially, the effect of moisture concentration and hot spot temperature on the thermal resistance of cellulose is taken into account. The proposed MDTM is verified by using historical data of moisture-in-oil and temperature measurements on an in-service vegetable oil-filled transformer. To integrate every piece of data and information obtained from different transformer diagnostic measurements and subsequently evaluating the overall health condition of a transformer, this thesis proposes a data and information fusion framework based on Bayesian Network (BN). Within the Bayesian Network, Monte Carlo and Bootstrap methods are employed to extract the most informative characteristics regarding transformer condition from different diagnostic measurements. Results of case studies demonstrate that the proposed data and information fusion framework can evaluate the effectiveness of combinations of different diagnostic measurements and subsequently facilitate determining optimal diagnostic strategies involved in transformer condition assessment. It is expected that the data centric diagnostic approaches developed in this thesis can provide an accurate modelling and reliable assessment of transformer’s health condition.

Transformer is very critical expansive component based on cost and inconvenience caused from failure. For continuous power supply a preventive maintenance is very much required. Transformer failure causes extensive damage to various other electrical equipment and supply grid which will affect a major part of electrical power system. Condition monitoring systems can help to decrease the transformer life cycle cost and to increases the reliability and availability at high level. It keeps the record of relevant data of a transformer. It tells probably time when transformer going to fail and what changes need to make to ensure that it going to last an expected number of years. Till now there is no as such system which will continuously collect of parameter data from transformer and the analysis thereof to determine the risk of unexpected incapacity and the availability of capacity. All monitoring systems only monitor the limit values for parameters. Both power transformer and distribution transformer are well maintained using the methods discus in this paper.

Transformer condition monitoring and assessment is crucial as it can be applied to evaluate its maintenance needs. The current trend is to leave critical power transformers in service uninterrupted unless the monitored parameters indicate otherwise. This ensures optimal use of the facility and uninterrupted power supply to consumers. This paper presents the evaluation of the health status of a transformer using the dissolved gas analysis (DGA), oil testing and evidential reasoning criterion. In the analysis, the dissolved gases and oil testing parameters are first normalized and then transformed into fuzzy variables using trapezoid membership function. Then from fuzzy variables, the condition assessment is transformed into a multiple-attribute decision making (MADM) problem under an evidential reasoning (ER) framework. Two case studies are evaluated using actual field data to verify the methodology. Both demonstrate the applicability of the proposed methodology. This emerges as a good logical and systematic way of assessing the maintenance needs of a power transformer using operational data.

A Review of Dissolved Gas Analysis in Power Transformers

Power transformers and shunt reactors are an integral part in ascertaining the reliability of any power utility. Utilities are facing considerable pressure to ensure their critical asset's reliability due to deregulation, increased demand, economic pressures, and profit constrains. Most of the power transformers and shunt reactors develop faults and get critical over a period of time. A transformer's and reactor's condition is normally assessed based on DGA supported by electrical tests. In this paper three Case studies on 315 MVA transformers 63 MVAr are investigated. These jobs have seen high hydrogen and methane, showing faults in these jobs. However it was not supported by electrical tests such as isolation test and low voltage tests in any of the case. These jobs were physically inspected and investigated for further details. DGA was monitored throughout their period. The fault analysis is covered in this paper.

Transformer condition should be maintained because of its importance to electricity network. Each utility, in this case PLN P3B Jawa-Bali, has to assure their reliability in order to maintain electrical power system stability by assessing transformer conditions. Once transformer condition attained, appropriate maintenance and operation strategy can be defined. The most important things influence transformer condition is its insulation. Since transformer insulation is also influenced by thermal stress, another stresses also influence insulation condition i.e. chemical, electrical and mechanical stress. We calculate FAA (factor accelerated aging) of the transformer by estimating hot spot from top oil temperature. Calculated FAA is correlated to dissolved gasses analysis (DGA) and oil quality analysis (OQA). Data used were taken from 150/20 kV 60 MVA power transformer from the year of 2007 until 2008.

The most common fault diagnosis method for oil-immersed power transformers is dissolved gas analysis (DGA). Doernenburg ratios, Rogers ratios, IEC (International Electrotechnical Commission) ratios, and Duval's triangle are conventional DGA techniques for insulating oil in power transformers. In this study, Scikit-learn known as a popular open-source free machine learning tool for Python programming language has been used to develop different machine learning (ML) classifiers to effectively detect defects in oil-immersed power transformers. These classifiers are Decision Trees, Support Vector Machines, Gaussian Naive Bayes, k-Nearest Neighbours, Random Forests, and Multi-Layer Perceptron. The input vector of each classifier has been formed by Doernenburg ratios, Rogers ratios, IEC ratios, and CSUS (California State University Sacramento) method. After these classifiers are completely trained, unseen DGA data sets are then used to evaluate their performances. Based on a statistical analysis, the study can indicate the most effective type of the input vector and ML classifier for precisely detecting faults in power transformers.

Condition assessment of power transformers are important to avoid power system failures because it usually cause significant economic losses. In order to detect the mechanical integrity of power transformer windings, Frequency response measurements (FRA) has been developed. It is an effective tool to detect the internal mechanical integrity. However, interpretation of FRA results can sometimes be tedious when comparing transformers with different structures. This paper presents a FRA measurements based AI approach to recognize different faulty and non-faulty transformers irrespectively of their design structures. FRA data sets were collected from 10, Field aged power transformers with different capacities and each data set consisted of 100 features i.e., 4 FRA frequency bands, 5 FRA measuring methods and 5 statistical analyzing parameters. It was found that with two Principal Components, faulty and non faulty transformers could be easily separated. Accordingly, the 100 features were reduced to 64 features by optimizing FRA measuring methods from 5 to 3. Then different ML techniques and NN algorithms were used to further classify the faulty transformers into three different fault types. It was found that fine kNN was the best method for transformer classification with a 100% accuracy of the tested samples.

Condition Assessment of Power Transformers based upon Weighted Criteria OWA Operators and Fuzzy Model

The high cost and the critical nature of power transformers in power substations makes it is very important to prevent failures and detect problems as early as possible. This presentation reviews the diagnostic and monitoring tests and equipment available to assess the condition of transformers and provide an early warning of potential failure.