Smart Monitoring of Power Transformers in Substation 4.0: Multi-Sensor Integration and Machine Learning Approach

Transformers are one of the most critical and cost-intensive components in the electrical power system. Dissolved gas analysis (DGA) has been proven to be a reliable technique for diagnosis of incipient faults in oil-filled transformers. Traditionally, DGA is usually performed in laboratories using standardized methods. In the past decade, an increasing number of multi-gas online monitors have gradually become commercially available. However, it is still not clear whether the monitor readings are accurate and reliable and whether they agree with laboratory results. This raises additional concerns for the application of vegetable oils in power transformers in recent years. This paper comparatively studied the lab and online DGA results of a mineral oil and a vegetable oil under various faults, including thermal faults, partial discharge faults and sparking faults. The results show that online DGA results can be correlated with laboratory DGA results for hydrocarbon gases within an error of 30%. However, gas leakage or air ingression might occur during the transport of oil samples for laboratory analysis. Therefore, oil samples should be analyzed as soon as sampled for laboratory analysis.

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.

Collecting labeled Dissolved Gas Analysis (DGA) data is difficult because the determination of the transformer fault is time-consuming and expensive in the transformer substation, but DGA data without labels is easier to obtain. In order to make full use of DGA data with few labels and a large number of unlabeled data to improve the transformer fault diagnosis rate, it is an important to study the DGA fault diagnosis method based on semi-supervised learning for solving practical problems in the field. Therefore, the paper proposed a novel filter-based semi-supervised feature selection method for building fault diagnosis model and selecting optimal DGA features. The method is test by using IEC T10 dataset and compared with traditional supervised diagnostic models. The results show that the proposed method works in optimizing DGA features and has strong robustness in solving small sample DGA classification problems.

Power transformers are one of the most widely used important components in the electric power system. Dissolved Gas Analysis (DGA) testing is used to diagnose transformer failures before more severe damage occurs by analyzing gas indicators dissolved in transformer oil through several methods, one of which uses conventional methods. However, based on most of the tests conducted by researchers, the detection accuracy of the conventional method is still quite low. Therefore, this research has the main objective of identifying the weaknesses of one of the conventional methods, namely the Rogers Ratio method. This research method uses modifications to the fault diagnosis flow chart which is then applied in the interpretation of DGA test results on power transformers in the case of the GSUT #1 20/11 kV Transformer of Manokwari Gas Engine Power Plant (GEPP). Based on the results of this research, the previous method cannot diagnose the fault (Undetermined) while after being modified it can diagnose the “Thermal Fault 150-200OC. When compared with other conventional methods that have been tested such as the interpretation of the Duval Triangle method, the results of diagnosing “Thermal Fault < 300OC”, it means that in general can be known that there has been a thermal disturbance in the internal transformer at temperatures below 300OC. Thus the results of the modified interpretation of the Rogers Ratio method are better than before so that it can be applied as an additional technique for interpreting DGA test data

With the advent of smart grids, a significant amount of data has become available about the electric infrastructure. Much of research focus has been on exploiting newly available data sources such as smart meters and phasor measurement units. This paper proposes a new class of predictive analytics that can be built to manage existing infrastructure by combining new and old data sources together. Power transformers, one of the most critical assets in the grid, are perhaps frontrunners among ‘smarter’ set of assets which have significant instrumentation already installed to monitor their operating conditions such as load, voltage, and internal oil temperature. While such advanced instrumentation enables detailed operating condition monitoring, manual measurement of dissolved gas concentration has been the primary fault diagnostic method to identify their fault modes. Dissolved gas analysis (DGA) offers great potential to diagnose fault modes in such oil-immersed transformers. This manual routine DGA, however, is costly and not free from error. Fortunately, it is understood that the loading conditions of transformers are major drivers of fault modes in oil-immersed transformers. In this paper, a predictive model is proposed to predict accumulation of dissolved gas concentration in sealed substation transformers based on its historical loading conditions. A multi-dimensional regression approach is used to predict the concentration level of each gas in real-time. Measurements from historical dissolved gas analyses are used to solve the regression problem with a robust optimization framework. The simulation results show that the forecasting of each dissolved gas based on loading characteristics is possible with high regression accuracy ranging from 84% to 97%. Thus this method can be used to optimize DGA inspection schedules as well as to provide “virtual DGA instrumentation” without the associated high cost.

A healthy condition of the power transformer is critical for the reliable operation of the power system. For this reason, the condition assessment of the oil-paper insulation used in power transformers is immensely vital to avoid catastrophic failures in the power system network. Dissolved Gas Analysis (DGA) is a popular technique for diagnosis of power transformer insulation. It provides vital information about the presence of different faults i.e. partial discharge, arcing, low and high magnitude discharges etc. at their incipient stage itself before it can result in total breakdown of the transformer insulation. This paper introduces a DGA method using gas ratio combinations and Random Forest Classifier. Here five different gas concentrations are used, namely hydrogen, ethane, methane, acetylene and ethylene as collected from several power utilities. Five different gases provide ten combination of gas ratios which are used as feature in random forest classifier. The model is trained with 220 DGA data and tested with 109 data which gives an overall accuracy of 89%. This classifier also proves to be extremely suitable for correctly predicting the PD fault which is known to be very difficult to predict by DGA.

A modern electric power system is a complex organizational structure that coordinates its intelligent components through the definition of roles, communication channels and powers. The management system of intelligent components of the electric power system should ensure the consistency of their work at technological stages of generation, transport, distribution and consumption of electric energy, while achieving the targets and reducing the value of resource consumption. The disadvantage of the process management system that is currently used in electric power systems is that the hierarchical management structure is applied to the network topology. Thus, there is a conflict of resources and processes of generation, transport, distribution and consumption of electricity. The authors propose a concept of a distributed resource and process management system in electric power systems using digital twin technology. The electrical power system is modeled as a polystructural one. The concepts of the system of polystructure indicators, the metric system of polystructure, the body of polystructure are used. Representation of electric power system components and technological processes of generation, transport, distribution and consumption by means of digital twin technology makes it possible to exclude conflicts of resources and processes in the electric power system while maintaining the requirements for reliability and safety of the system. Digital twin technology, as applied to polystructured systems, provides the developers of distributed management systems with a methodology for creating a modern management system, where the production of management decisions does not lead to conflicts between the components of the power system. The proposed distributed management system is built as a polystructure, the body of which ensures the consistency of technological processes, equipment resources and electricity consumption.

Power transformer is one of the most important equipments in electrical power system. The detection of certain gases generated in transformer is the first indication of a malfunction that may lead to failure if not detected. Dissolved gas analysis (DGA) of transformer oil has been one of the most reliable techniques to detect the incipient faults. Many conventional DGA methods have been developed to interpret DGA results obtained from gas chromatography. Although these methods are widely used in the world, they sometimes fail to diagnose, especially when DGA results falls outside conventional method codes or when more than one fault exist in transformer. To overcome these limitations, fuzzy inference system (FIS) is proposed. 250 different cases are used to test the accuracy of various DGA methods in interpreting the transformer condition.

A novel hybrid machine learning model for prediction of CO2 using socio-economic and energy attributes for climate change monitoring and mitigation policies

Among the most critical components in the electric power system is the power transformer. As such, a significant body of research has been put forward to attempt to anticipate the needs for maintenance to be performed. Traditional health assessment has required sampling of oil for submission to a laboratory for analysis, but this has been deemed undesirable in light of budgetary constraints on maintenance staffing, and new predictive maintenance philosophies for substation equipment. A number of processes have been developed in recent years for online health assessment of transformers, most of which have focused on dissolved gas analysis. This paper describes a novel optical methodology for on-line transformer health assessment that utilizes an ultraviolet absorption measurement to identify the degradation of the transformer oil. An optical system was selected because of its immunity to the electromagnetic noise typical of substations, and because of the minimal impact that non-conducting materials have on the insulation system design of the transformer. The system is designed to identify deterioration and premature aging resulting from overheating, low level arcing or excessive exposure to atmospheric air. The system consists of a light source, filter, guide and detection components, and a very simple computational requirement. The measurements performed with the prototype system are validated with a high precision spectrophotometry measurement and an independent oil-testing laboratory.

As a major component in electric power systems, power transformers are one of the most expensive and important pieces of electrical equipment. The trouble-free operation of power transformers is an important criterion for safety and stability in a power system. Technical diagnostics of electrical equipment are a mandatory part of preventing accidents and ensuring the continuity of the power supply. In this study, a complex diagnostic methodology was proposed and applied for special power transformers’ risk estimation. Twenty special power transformers were scored with the proposed risk estimation methodology. For each transformer, dissolved gas analysis (DGA) tests, transformer oil quality analysis, visual inspections of all current equipment on-site and historical data for the operation of each electrical research were conducted. All data were collected and analyzed under historical records of malfunctioning events. Statistical data for expected fault risk, based on long-term records, with such types of transformers were used to make more precise estimations of the current state of each machine and expected operational resource. The calculated degree of insulation polymerization was made via an ANN-assisted predictive method. Assessment of the collected data was applied to allow detailed information of the state of the power transformer to be rated. A method for risk assessment and reliability estimation was proposed and applied, based on the health index (HI) for each transformer.

Active power losses have the potential to affect the distribution of power flows along transmission lines as well as the mix of energy used throughout power networks. Grey wolf optimization algorithms (GWOs) are used in electrical power systems to reduce active power losses. GWOs are straightforward algorithms to implement because of their simple structure, low storage and computing needs, and quicker convergence from the constant decrease in search space. The electrical power system may be separated into three primary components: generation, transmission, and distribution. Each component of the power system is critical in the process of distributing electricity from where it is produced to where it is used by customers. By using the GWO, it is possible to regulate the active power delivered by a high-voltage direct current network based on a multi-terminal voltage-source converter. This review focuses on the role of GWO in reducing the amount of active power lost in power systems by considering the three major components of electrical power systems. Additionally, this work discusses the significance of GWO in minimizing active power losses in all components of the electrical power system. Results show that GWO plays a key role in reducing active power losses and consequently reducing the impact of power losses on the performance of electrical components by different percentages. Depending on how the power system is set up, the amount of reduction can be anywhere from 12% to 65.5%.

The application of artificial intelligence algorithms for transformer incipient fault classification using dissolved gas analysis (DGA) is an interesting engineering approach. However, there are various factors that affect the performance of artificial intelligence algorithms. This article presents the influence of the data balancing approach on transformer DGA fault classification with the machine learning (ML) approach. In this work, a total of 4580 DGA samples from in-service transformers are considered for training various ML models. The main challenge for the DGA problem lies in the availability of the normal degradation transformer data and its uniformity corresponding to different faults is almost impossible. This is because DGA is not an exact science, but an empirical approach subjects to variability. Thus, it is a usual practice to apply data sampling techniques that largely influence the efficiency of the algorithms. The present work reports the impact of the data balancing schemes on the performance of the fault classification and demonstrates that a careful choice of the data sampling method and ML algorithm is essential for DGA problems. To demonstrate the global scale ability of the propose model, the model is tested on the IEC TC ten data (field inspection data), while these data are not exposed to the machine during the learning stage. The ability of the ADASYN method in significant enhancement of the global scale capability of AI-based transformer DGA fault classification is reported. This approach will be helpful for condition monitoring engineers in transformer insulation diagnosis in implementing the monitoring modules for large transformer fleets and understanding the insulation oil behavior over years.

Power transformer is vital equipment in any electrical power system. So any fault in the power transformer may lead to the interruption of power supply and accordingly the financial losses will also be great. So it is important to detect the incipient faults of transformer as early as possible. Among the existing methods for identifying the incipient faults, dissolved gas analysis (DGA) is the most popular and successful method. Any kind of fault inside transformer gives rise to overheating and will produce characteristics amount of gases in transformer oil. In this paper classical methods of DGA such as Key Gas Method, Rogers Ratio Method and Duval Triangle Method are reviewed first and the need to integrate with the artificial intelligence (AI) methods for improving the performance of diagnosis is justified. Reported work presents a new and efficient artificial intelligence technique that is support vector machine (SVM) for transformer fault diagnosis using dissolved gas analysis. The proposed method i.e. Support Vector Machine is a classification tool based on statistical learning theory. Here 3 types of multiclass SVM method that is One - against-One, One-against-All and binary decision tree have been used for the fault diagnosis. Each SVM method has been trained and tested with many practical fault data of power transformers.

Transformer is the key electric installation in power system. Dissolved gas Analysis （ DGA ） on-line monitoring technique can discover early faults of transformer by analyzing the concentration of dissolved gas in transformer oil [1][2]. However, the on-line system can't monitoring the status of transformer accuracy in time because there are many invalid sensors make the quality of the on-line data is relatively low in field[3]. Although there are some studies in improving the reliability of DGA sensor by designing the new sensor and using distributed sensor technology, but there are few studies in validity assessment, since the DGA sensor was installed and running in special environment. Therefore this paper proposed an assess method of validity for DGA sensor based on multiple criterion. This method is based on large amounts on-line DGA data in field, and use the sliding window to acquire characteristic set form the sensor data stream. The characteristic set is used to assess the validity of sensor, and the length of the sliding window can be changed according the actual. Four kinds of criterion are applied to characteristic set to assess the validity of sensor. The first criterion is the number and distribution of the abnormal value, the abnormal value often reflect the invalid of sensor directly; the second one is the distribution of the continual same value, there are many continual same value in the data in field usually, the continual same value means that the sensor is abnormal; the next criterion is the changes of Coefficient of Variation (C.V), the coefficient of variation is a common and important indicator which is used to reflect the fluctuation of one set, and the statistical fluctuations can be found through it; The change of gas increase rate also be used as an criterion since it is an important indicator of DGA on-line monitoring system, and this criterion can reflect the abnormal condition of trend. Applying those three criteria to the characteristic set, five discriminant values can be gotten. Considering there are many sensors with different property and many sensors with same property but was installed in different environments, every discriminant value was assigned a weight value. Then the state value of sensor can be calculated by summing the discriminant values with weight value. Comparing the state value with tolerance value which was set in advance, the validity of sensor can be acquired. The assessment program based on multiple criterion assess method is designed by using MATLAB because its advantages in data processing and interface design. Using the on-line DGA data and the sensor identification reports from a reginal grid to verify the accuracy of this assess method. The verification result show that the correct recognition rate of invalid sensor can be reached to 100% and erroneous recognition rate of valid sensor is 3.1%. It can be concluded from the paper that the assess method based on multiple criterion is efficient and accurate. Applying the assess method into field, the DGA sensor which is invalid can be found accurately and timely.