Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

As transformers age, their internal condition degrades, which increases the risk of failure. To prevent these failures and to maintain transformers in good operating condition is a very important issue for utilities. Traditionally, routine preventative maintenance programs combined with regular testing were used. The change to condition-based maintenance has resulted in the reduction, or even elimination, of routine time-based maintenance. Instead of doing maintenance at a regular interval, maintenance is only carried out if the condition of the equipment requires it. Hence, there is an increasing need for better nonintrusive diagnostic and monitoring tools to assess the internal condition of the transformers. If there is a problem, the transformer can then be repaired or replaced before it fails. An extensive review is given of diagnostic and monitoring tests, and equipment available that assess the condition of power transformers and provide an early warning of potential failure.

Review of condition assessment of power transformers in service

Pattern recognition

Dissolved gas analysis (DGA) is used to assess the condition of power transformers. It uses the concentrations of various gases dissolved in the transformer oil due to decomposition of the oil and paper insulation. DGA has gained worldwide acceptance as a method for the detection of incipient faults in transformers.

A review of dissolved gas analysis measurement and interpretation techniques

A concise review is given of the progress made in the field of partial discharges (PD) at DC voltage. Although ample reference will be made to work of other authors in this field, the paper will concentrate on the progress that was made at Delft University of Technology over a period of 14 years in three PhD projects. In the first project, a start was made with the analysis of the physics of partial discharges at DC voltage and different types of PD were characterized based on parameters like time interval between PD and PD magnitude. In a second project, PD analysis was applied to HVDC apparatus and different means of classification of PD at DC voltage were proposed. In the third project, PD analysis was applied to HVDC mass-impregnated cables and test specifications were proposed. In this paper the work performed in the above three Ph.D. projects is summarized with ample reference to papers of other workers in this field. Attention is given to the mechanism of PD at DC voltage as compared to AC voltage and techniques for measurement and analysis of DC PD patterns. Examples of practical application of DC PD testing are given. Finally, some thoughts on future work are presented.

/pdf/partial-discharges-at-dc-voltage-their-mechanism-detection-2ka0gekh8r.pdf

Partial discharges at DC voltage: their mechanism, detection and analysis

Electrical treeing is an electrical prebreakdown phenomenon commonly found in solid dielectrics. It is a damaging process due to partial discharge (PD) and progresses through the stressed insulation by chemical degradation whose path resembles the form of a tree root. The characteristics of PDs due to electrical treeing before the insulation breakdown are discussed in this paper. Several point to plane electrode specimens were prepared using epoxy resin as the insulation material, and tested under different voltage levels from 10kV to 12kV AC. While PD data acquisitions were in progress by means of a precision computer-based measuring system, the progress of treeing in the samples was also visually monitored and photos were captured in specific time intervals for each sample. The trends of the phase resolved patterns prior to breakdown and far before breakdown were compared. Also the trends of PD sequences were analysed for critical and non-critical conditions.

Partial discharge characteristics of electrical trees prior to breakdown

The general method for identifying the partial discharge type in a power transformer is based on their fingerprints in the form of phase-resolved discharge patterns. In the case of multiple defects, traditional clustering methods can be applied for separation of active sources. However, such an approach is impractical for online real-time monitoring due to the very large data size. In this paper a new method using stream clustering is introduced. The method separates the active sources by processing the signal once it is captured, then only a synopsis of the discharge data is stored. Two stream clustering algorithms: Density Grids and DenStream are employed. Through measurements obtained from laboratory experimental setups (corona, surface discharge, transformer defect model) performance of the proposed algorithms are evaluated. It is shown that stream clustering method is able to separate the constituent components involved in the stream of a multi-source discharge signal without the need to store a large amount of information. The performance of the Density Grids method depends on a limited number of features that it can accommodate. In comparison, the DenStream method can capture more features which enable better separation of active sources at the expense of longer processing time.

https://ietresearch.onlinelibrary.wiley.com/doi/epdf/10.1049/iet-smt.2018.5389

Online monitoring of transformer through stream clustering of partial discharge signals

This paper presents the characteristics of partial discharge activity in oil-impregnated insulation under different test voltage frequencies and oil contamination. The aim of this experiment is to investigate the effect of the voltage frequency variation in accelerating the deterioration of this particular insulation. The test results indicate that, for the conditions tested, there is not a linear scaling of the current, power and the repetition rate of the discharges, which may require some re-assessment of the aging determination for such configurations.

Characteristics of partial discharge in oil impregnated insulation using different test voltage frequencies

In this paper a method for selecting optimal wavelet is introduced, which is implemented for evaluating electrical measured partial discharges on a 66 kV/25 MVA fully interleaved winding of a power transformer. Applying wavelet transform to a signal produces a wavelet detail coefficient distribution throughout the for generating coefficients with maximal values. The basic idea of this method is described and applications to partial discharge studies are explored. The paper demonstrates that the wavelet based de-noising method proposed in the paper can he employed in extracting the PD pulses from the measured signal successfully to provide enhanced information and further infer the original site of the PD pulse through capacitive ratio method.

http://ieeexplore.ieee.org/iel5/9968/32026/01490078.pdf

An optimal wavelet filtering method for noise suppression of PD measured signal and its location in power transformer winding

Power transformer partial discharge signal measurement using Radio Frequency method has advantages over IEC 60270 ‘apparent charge’ measurement method for online monitoring due to greater immunity against external interference. However, the lack of a well-defined calibration relationship between the two methods is main disadvantage of RF method. Simultaneous measurements, using these two methods, carried out on various transformer PD source models, in laboratory, to investigate main parameters that affect the relationship of results of these two methods. In this paper, factors that influence this relationship are classified into three groups: (i) type of PD sources, (ii) transformer windings as PD signal transmitter, and (iii) the inner structure of the transformer. Firstly, the relationship between these two methods has been investigated for some simple types of PD sources. Subsequently, by adding the winding structure model to the void, which is the PD source, its impact on each measurement method and the relationship between them have been studied. The impact of inner structure on the relation between the two methods is compensated through a formula which is proposed as a means for calibration of the RF partial discharge measurement setup. However to compensate for internal insulating system effects more research is required.

Trevor Blackburn

Papers

Partial discharge characteristics of electrical trees prior to breakdown

Online monitoring of transformer through stream clustering of partial discharge signals

Characteristics of partial discharge in oil impregnated insulation using different test voltage frequencies

An optimal wavelet filtering method for noise suppression of PD measured signal and its location in power transformer winding

Transformer components impact on compatibility of measured PDs: comparison of IEC60270 and RF methods