Search

It is a great challenge to differentiate partial discharge (PD) induced by different types of insulation defects in high-voltage cables. Some types of PD signals have very similar characteristics and are specifically difficult to differentiate, even for the most experienced specialists. To overcome the challenge, a convolutional neural network (CNN)-based deep learning methodology for PD pattern recognition is presented in this paper. First, PD testing for five types of artificial defects in ethylene-propylene-rubber cables is carried out in high voltage laboratory to generate signals containing PD data. Second, 3500 sets of PD transient pulses are extracted, and then 33 kinds of PD features are established. The third stage applies a CNN to the data; typical CNN architecture and the key factors which affect the CNN-based pattern recognition accuracy are described. Factors discussed include the number of the network layers, convolutional kernel size, activation function, and pooling method. This paper presents a flowchart of the CNN-based PD pattern recognition method and an evaluation with 3500 sets of PD samples. Finally, the CNN-based pattern recognition results are shown and the proposed method is compared with two more traditional analysis methods, i.e., support vector machine (SVM) and back propagation neural network (BPNN). The results show that the proposed CNN method has higher pattern recognition accuracy than SVM and BPNN, and that the novel method is especially effective for PD type recognition in cases of signals of high similarity, which is applicable for industrial applications.

Optimal selection of features of Partial Discharge (PD) signals recorded from defects in High Voltage (HV) cables will contribute not only to the improvement of PD pattern recognition accuracy and efficiency but also to PD parameter visualization in HV cable condition monitoring and diagnostics. This paper presents a novel Random Forest (RF)-based feature selection algorithm for PD pattern recognition of HV cables. The algorithm is applied to feature selection of both PD signals and interference signals with the aim of obtaining the optimal features for data processing. Firstly, the experimental data acquisition and feature extraction processes are introduced. PD signals were captured from faults created in a cable to obtain the raw PD data, then a set of 3500 transient PD pulses and a set of 3500 typical interference pulses were extracted, based on which 34 PD features were extracted for further processing. Furthermore, 119 two-dimensional features and 1082 three-dimensional features were generated. The paper then discusses the basic principle of the RF algorithm. Finally, RF-based feature selection was implemented to determine the optimal features for PD pattern recognition. The results were obtained and evaluated with the Back Propagation Neural Network (BPNN) and Support Vector Machine (SVM). Results show that the proposed RF-based method is effective for PD feature selection of HV cables with the potential for application to additional HV power apparatus.

This paper presents a case study on a 275 kV oil-filled cable. The condition assessment and diagnosis are based on analysis of cable surface temperature in relation to its current load and insulation dielectric loss. The work was initiated by a local abnormal temperature rise of 5.2 °C in cable surface temperature, which was observed during a routine inspection. The temperature rise occurred at bend area with a length of approximately one metre in the Blue phase. No PD activity was identified using on-line PD measurement. The relation between cable surface temperature, cable core temperature and cable insulation condition was then simulated based on the thermal model of power cables. According to simulation analysis, poor condition of cable insulation or oil from an oil duct penetrating a region under the cable surface were identified as possible reasons for the problem observed. An in service X-ray scanning technique was employed for further investigation and to aid diagnosis. The X-ray images revealed a slight distortion of the PVC sheath and the presence of multiple voids between cable insulation paper and the lead sheath. It was concluded that an oil leakage from the oil duct to the voids under the cable lead sheath was responsible for the local cable surface temperature rise. The result removed the concern of incipient cable breakdown, and a potential unplanned outage.

Cross-bonded three-phase single-core Cross-linked Polyethylene (XLPE) cables are widely used in power transmission and distribution systems. A great deal has been published on on-line monitoring of HV cable systems to prevent failure. Relative Dielectric Losses (DL) monitoring reveals the general condition of the cable insulation. Generally, DL measurement relies on measuring operating voltage and insulation leakage current. The main challenge to on-line monitoring of DL in Cross-bonded High voltage (CB HV) cables is the interconnectedness of metal sheath. Measured sheath currents are a combination of the leakage currents of each section and the imbalanced circulating current. Differentiating the leakage current of each interconnected section from the measured data is complex. The second challenge is that co-axial cables are used to connect cable sheaths to cable link boxes, for ease of installation and protection against moisture. As a result, any measured value contains currents from the two sheaths connecting at a joint. The third challenge is that it is hard to obtain the reference voltage synchronously and with high accuracy. Consequently, this paper investigates separation of leakage current components and analysis of DL based on leakage currents without reference voltage. A circuit model is proposed to describe the leakage current separation method for CB HV cables to allow on-line monitoring of DL, and the criterion required for assessing relative change of DL among different phases or sections of the cable for insulation deterioration are presented. Operational measurements from a cable system are analyzed and discussed. Results show that there is no significant deterioration in this monitored major section of a CB HV cable system.

Cross-bonded metal sheath connection is applied in sectioned single-core power cables to reduce or eliminate the voltages that are induced in the sheath over long distances. However, cross-bonded cables present an opportunity as well as a challenge for online measurement and diagnosis of cable conditions. In this paper, a methodology to identify cable sheath faults through analysis of the sheath system currents in a cross-bonded cable system is presented. First, a numerical model is established to simulate the sheath currents in cross-bonded cable systems. Second, analyses of several faults, which happen frequently with serious consequences, are presented on the basis of current measurement at the link cable. Simulations of normal and fault conditions are given to determine the feasibility of fault diagnosis. A case study using field data from a cable tunnel in China considering the normal condition is presented to verify the numerical model. Results in normal condition show good consistency with field data with error less than 5%. Simulation results of fault conditions show that analysis of readings from six current sensors can distinguish different fault types and fault positions using the method proposed. Based on the analyses, criteria are established for sheath loop fault type diagnosis.

Dielectric loss has long been recognized as one of the most important indicators of cable insulation health. Although huge efforts have been made in the past to measure the dielectric loss of cable circuits, there has been no report of any successful on-line techniques for the purpose. This paper proposes a new on-line insulation dielectric loss monitoring method, based on separation of currents collected from the co-axial cables connecting the cable sheathes and the cable link boxes. The principle and theoretical foundation of the method are demonstrated. MATLAB simulations, based on the real cable parameters in a 110 kV cable tunnel in China, indicate the error of the proposed method is less than 1 × 10 −3 %. The criteria for determining the relative dielectric loss in cable segments, based on the leakage current separation results, are demonstrated at the end. Initial results and analysis of implementation of the proposed method in the real world 110 kV cable tunnel are presented.

The fast growing wind industry requires a more sophisticated fault detection approach in pitch-regulated wind turbine generators (WTG), particularly in the pitch system that has led to the highest failure frequency and downtime. Improved analysis of data from Supervisory Control and Data Acquisition (SCADA) systems can be used to generate alarms and signals that could provide earlier indication of WTG faults and allow operators to more effectively plan Operation and Maintenance (O&M) strategies prior to WTG failures. Several data-mining approaches, e.g. Artificial Neural Network (ANN), and Normal Behaviour Models (NBM) have been used for that purpose. However, practical applications are limited because of the SCADA data complexity and the lack of accuracy due to the use of SCADA data averaged over a period of 10 min for ANN training. This paper aims to propose a new pitch fault detection procedure using performance curve (PC) based NBMs. An advantage of the proposed approach is that the system consisting of NBMs and criteria, can be developed using technical specifications of studied WTGs. A second advantage is that training data is unnecessary prior to application of the system. In order to construct the proposed system, details of WTG operational states and PCs are studied. Power-generator speed (P-N) and pitch angle-generator speed (PA-N) curves are selected to set up NBMs due to the better fit between the measured data and theoretical PCs. Six case studies have been carried out to show the prognosis of WTG fault and to demonstrate the feasibility of the proposed method. The results illustrate that polluted slip rings and the pitch controller malfunctions could be detected by the proposed method 20 h and 13 h earlier than by the AI approaches investigated and the existing alarm system. In addition, the proposed approach is able to explain and visualize abnormal behaviour of WTGs during the fault conditions.

This paper presents an investigation of effects of high voltage direct current (HVDC) superimposed voltage harmonics on partial discharge (PD) behavior. Firstly, a single air filled void embedded in a slab of polymeric insulation was modeled using finite element methods (FEM). PD phenomena activated under the influence of rippled DC voltages were simulated using an interlinked environment of MATLAB and multiphysics simulation package COMSOL. Secondly, the results of the modeling were backed up by experimental tests, where HVDC superimposed harmonics were applied to artificial void samples in the lab and trends in the resultant PD characterized with varying harmonic orders and amplitudes. The results of both modeling and experiment are generally in agreement showing that as the harmonic orders and levels increase, the repetition rate of PD pulses rises in comparison to a purely DC applied voltage having the same peak amplitude. The findings of this study will assist in understanding of PD behavior under HVDC conditions where harmonic rippling occurs on the voltage waveform.

Dissolved Gas Analysis (DGA) of liquid insulation is an effective means for diagnosing power transformers. The gas contents in insulating oil can be gathered on-line and off-line to indicate the health condition of the transformers, thereafter there are many interpretations of the gas contents. In this work, Seven-fuzzy interpretation modules are individually established, tested and lately combined to monitor power transformers' health. The developed method incorporates trending of the concentration of the dissolved gases over the operating life. The approach processes current and/or historical DGA data, using the 7-developed logic modules, to determine the current state of a transformer, provide information regarding the fault type, fault probability, fault severity and recommended future sampling interval in addition to operating procedure, consistent with industry standards. The developed diagnosis system has been validated using 1290 samples from fresh and previously tested mineral oil filled transformers. The proposed system is proved, based on field data, to be 99% accurate in identifying transformers being in normal or abnormal operation. For the cases where a transformer fault was known, the proposed technique has less than 2% inaccuracy in recognizing the fault's type in comparison to other approaches discussed in literature.

Domestic electricity load profile is essential for energy planning and renewable energy system design. This paper presents analysis of domestic electric load characteristics and a method to model domestic and regional load profile. Multiple Gaussian functions are used to express the load characteristics in the proposed model. This is done by associating the Gaussian function parameters with the peak load changes, e.g. changing height parameters to reflect the peak magnitude. The result of the load curve represented with multiple Gaussian functions allows the model to generate a regional load profile using the number of homes, the number of bedrooms (Nr) and the number of occupants (Np). The proposed model simulates domestic load profile by its load demand change characteristics instead of its appliance ownership and use pattern, etc. Data requirement for the proposed method is significantly lower than the previous top-down and bottom-up approaches. Seasonal change is not included in the present paper, but the method is capable of including seasonal changes if each season’s load demand changes in relation to Np and Nr is available. A demonstration of modelling England and Wales’s national hourly load profile in 2001 and 2011 is presented in this paper. Comparison is made of the proposed method with two other published domestic load profile models. Results show that the proposed method improves the mean percentage errors by at least 5.7% on average hourly load profile.

Transformers are an important part of any electrical power system network, warning of fault before failure is vital. The application of on-line fault detection reduces the cost of outages and the possibility of unpredicted failures. This paper outlines an electrically based on-line condition monitoring and short circuit detection method for power transformers. This method uses measurements of voltage and current in both sides of a transformer to classify its health and to identify fault conditions while in operation. Simulation results, relating to a number of short circuit winding faults, are analyzed and discussed. The results confirm that it is possible to identify a variety of short circuit conditions in primary and secondary windings from externally measurable parameters, i.e. measurement of voltages and currents in a transformer’s windings can, based on available winding data, identify changes inside the transformer. The effect of load variation on the magnitude of the measured voltages and currents is also considered in this analysis.

Due to the increasing use of renewable, non-controllable energy generation systems energy storage systems (ESS) are seen as a necessary part of future power delivery systems. ESS have gained research interest and practical implementation over the past decade and this is expected to continue into the future. This is due to the economic and operational benefits for both network operators and customers, battery energy storage system (BESS) is used as the main focus of this research paper. This paper presents an analytical study of the benefits of deploying distributed BESS in an electrical distribution network (DN). The work explores the optimum location of installing BESS and its impact on the DN performance and possible future investment. This study provides a comparison between bulk energy storage installed at three different locations, medium voltage (MV) side and low voltage (LV) side of the distribution transformer (DT) and distributed energy storage at customers&rsquo, feeders. The performance of a typical UK DN is examined under different penetration levels of wind energy generation units and BESS. The results show that the minimum storage size is obtained when BESS is installed next to the DT. However, the power loss is reduced to its minimum when BESS and wind energy are both distributed at load busbars. The study demonstrates that BESS installation has improved the loss of life factor of the distribution transformer.

This paper presents a novel Partial Discharge (PD) localisation method, based on PD pulse Rise-Time and Transfer Function (RTTF), for use in on-line cable PD monitoring systems. This is based on the change in rise time as a function of cable length along which a PD pulse has propagated. A comparison with four other well-known location techniques, i.e. Time-Domain Reflectometry (TDR), Phase difference, Arrival Time Analysis (ATA) and Amplitude-Frequency (AF) mapping has been made and presented. Analysis shows that the TDR and Phase difference methods, although requiring a single measurement point, are only applicable to short length cable systems as both incident pulse and reflected pulse should be observable when it reaches the detection point. On the other hand, the ATA and AF mapping, which can be applied to longer cable systems, need to install detectors at multiple points and detect synchronously. In comparison, PD localisation by the proposed RTTF method, which requires detection of the incident pulse at only one measurement point, can be used to localise a PD source which is further away from the PD detection point than other methods and can be used in on-line PD monitoring systems. As on-line PD monitoring sensors commonly comprise of a high frequency current transformer (HFCT) and a high-pass filter, the rise time of detected PD pulses depends on the attenuation of the cable, the HFCT used and the filter applied. Simulations of pulse propagation in a cable and PD monitoring system are performed, based on analyses in the frequency domain using the concept of transfer functions. Results of RTTF from the simulations have been verified by laboratory experiments and on-site PD measurements, which provide evidence that the RTTF method is a very valuable and useful PD localisation technique.

With global specialization of work units within organizations, interdisciplinary work practices comprised of collaborative efforts between technical and business teams are increasingly common in today's workplace. While higher education has responded by creating opportunities for remote teams to learn from collaborative work, occasions for interaction between remotely situated significantly diverse teams such as business and engineering are few. This study describes a pedagogical offering between U.S.-based business teams and U.K.-based engineering teams that delivers such an experience around mutually dependent, socially responsible projects. Results from student surveys and reflections confirm that the pedagogical design is able to generate perceived interdependence between these diverse teams. Student responses to surveys also suggest that the collaboration is effective in raising student confidence in interdisciplinary and remote team work among students. The study concludes with recommendations for transferability and actualization of interdisciplinary collaboration.

Concerns over global climate changes coupled with growing demand for energy are leading to increased penetration of distributed generation from intermittent sources into low voltage networks. In such cases distribution network (DN) operation will be affected. Consequently, there have been serious concerns over reliability and satisfactory operation of these power systems which contain distributed generation (DG) equipment. Distributed power generated from renewable sources is variable particularly in the case of wind generation or solar energy. The variability affects the stability of the system between supply and consumers. In DN, the losses and voltage drop across the network are significant matters and the DG location has a critical impact on the network operation. So, there is a clear need to optimise the DG size and location in the DN; for example, optimising the number of DG's and co-ordinating their operation can improve voltage drop and network losses. In this paper, an optimisation technique based on the genetic algorithm (GA) in conjunction with the power flow (PF) method is used to improve the DN performance and to identify the best location and size of the DG's. The main goal of the optimisation function is to reduce both the network losses and regulate the voltage level under different loading conditions.

Partial Discharge (PD) in cable systems causes deterioration and failure, identifying the presence of PD is crucial to Asset Management. This paper presents methods for interference signals rejection and for PD interpretation for five types of artificial defect in 11 kV ethylene-propylene rubber (EPR) cable. Firstly, the physical parameters of the artificial defects used for PD signal generation are introduced. Thereafter, the sample stress regime, PD testing and detection systems, including IEC 60270 measurement system and High Frequency Current Transformer (HFCT), are outlined. Following on, a novel Synchronous Detection and Multi-information Fusion (SDMF) based signal identification method is developed, to separate PD and interference signals within raw data. Finally, a comparative PD analysis of two detection systems is carried out and several characteristics of insulation related PD signals of EPR cables are reported. The SDMF based data pre-processing and the comparative PD activity analysis contribute to improvement of PD pattern recognition and assist in on-line PD monitoring systems.

This paper presents an effective, Rough Set (RS) based, pattern recognition method for rejecting interference signals and recognising Partial Discharge (PD) signals from different sources. Firstly, RS theory is presented in terms of Information System, Lower and Upper Approximation, Signal Discretisation, Attribute Reduction and a flowchart of the RS based pattern recognition method. Secondly, PD testing of five types of artificial defect in ethylene-propylene rubber (EPR) cable is carried out and data pre-processing and feature extraction are employed to separate PD and interference signals. Thirdly, the RS based PD signal recognition method is applied to 4000 samples and is proven to have 99% accuracy. Fourthly, the RS based PD recognition method is applied to signals from five different sources and an accuracy of more than 93% is attained when a combination of signal discretisation and attribute reduction methods are applied. Finally, Back-propagation Neural Network (BPNN) and Support Vector Machine (SVM) methods are studied and compared with the developed method. The proposed RS method is proven to have higher accuracy than SVM and BPNN and can be applied for on-line PD monitoring of cable systems after training with valid sample data.

This paper addresses the detection and localization of partial discharge (PD) in crossbonded (CB) high voltage (HV) cables. A great deal has been published in recent years on PD based cable insulation condition monitoring, diagnostics and localization in medium voltage (MV) and high voltage (HV) cables. The topic of pulse propagation and PD source localization in CB HV cable systems has yet to be significantly investigated. The main challenge to PD monitoring of CB HV cables is as a result of the interconnectedness of the sheaths of the three single phase cables. The cross-bonding of the sheaths makes it difficult to localize which of the three phases a PD signal has emanated from. Co-axial cables are used to connect cable sheaths to cable link boxes, for ease of installation and protection against moisture. A second challenge is, therefore, the coupling effect when a PD pulse propagates in HV cable joints and the co-axial cables, making PD detection and localization more complex. The paper presents experimental investigations into PD pulse coupling between the cable center conductor and the sheath and the behavior of PD pulse propagation in CB HV cables. It proposes a model to describe PD pulse propagation in a CB HV cable system to allow monitoring and localization, and also presents the knowledge rules required for PD localization in CB HV cable systems.

Partial discharge (PD) based condition monitoring has been widely applied to power cables. However, difficulties in interpretation of measurement results (location and criticality) remain to be tackled. This paper aims to develop further knowledge in PD signal propagation in power cables and attenuation by the PD monitoring system devices to address the localization and criticality issues. As on-line or in-service PD monitoring sensors commonly comprise of a high frequency current transformer (HFCT) and a high-pass filter, the characteristics of detected PD pulses depend on the attenuation of the cable, the HFCT used and the filter applied. Simulation of pulse propagation in a cable and PD monitoring system are performed, based on analyses in the frequency domain using the concept of transfer functions. Results have been verified by laboratory experiments and using on-site PD measurements. The knowledge gained from the research on the change in pulse characteristics propagating in a cable and through a PD detection system can be very useful to PD denoising and for development of a PD localization technique.

Voltage unbalance (VU) in residential distribution networks (RDNs) is mainly caused by load unbalance in three phases, resulting from network configuration and load-variations. The increasing penetration of distributed generation devices, such as small wind turbines (SWTs), and their uneven distribution over the three phases have introduced difficulties in evaluating possible VU. This paper aims to provide a three-phase probabilistic power flow method, point estimate method to evaluate the VU. This method, considering the randomness of load switching in customers’ homes and time-variation in wind speed, is shown to be capable of providing a global picture of a network’s VU degree so that it can be used for fast evaluation. Applying the 2m + 1 scheme of the proposed method to a generic UK distribution network shows that a balanced SWT penetration over three phases reduces the VU of a RDN. Greater unbalance in SWT penetration results in higher voltage unbalance factor (VUF), and cause VUF in excess of the UK statutory limit of 1.3%.

The probabilistic power flow method has previously been used for voltage violation quantification in residential distribution networks (RDNs) with small wind turbines (SWTs) and battery electric vehicles (BEVs). An improved statistical time series approach is developed in this paper to account for factors which impact on system conformity. In addition to consideration of variations in wind speed over time and BEV charging, using a statistical time series approach, the novel method takes into account variations in system load caused by the randomness of load switching by customers. The method has been applied to a generic UK distribution network. Results show that the proposed method provides a closer indication of daily probability distribution of voltage violations in a RDN, based on physical data, than the previous statistical time series approach.

Concerns over changes to the global environment and the growing need for energy have increased the penetration of renewable energy (RE) generation into low voltage distribution networks. The introduction of Energy Storage Systems (ESS) into distribution networks has been proposed to improve the reliability and performance of power systems. Energy storage systems will also be important in helping to regulate the supply from intermittent RE sources that feed into variable load demand. The focus of this paper is to provide an overview of the state of the art of ESS, concentrating on the distinct characteristics and applications of the different types. The analysis includes comparison and evaluation of different storage preferences with regard to integration of these technologies with an electric grid at the distribution network level that may lead to benefits such as distribution upgrade deferral and improvements in power quality and reliability.

Voltage violation tends to be one of the dominant constraints or critical factors that limit the penetration of small wind turbines (SWTs) and battery electric vehicles (BEVs) in distribution networks. To quantify the voltage violation, a statistical time series based approach is developed in this paper. The daily time varying characteristics of electric loads, wind speeds and BEV charging loads have been taken into consideration in a statistical manner. The method has been compared with the conventional time series approach. It is proven that the proposed methodology, while providing a global picture on voltage violations, can potentially be used for fast evaluation of the probability of a voltage violation occurring in a network. The methodology has been applied to an example distribution network. Results show that the proposed methodology can reduce the computational requirement by over 98%, while the difference between the results obtained using the proposed method and applying the conventional time series approach is within 12.5%.

Voids in solid insulation give rise to partial discharge (PD). Factors affecting PD activity in internal voids include the type of gas contained, gas pressure and surface properties of the void. Identifying the presence of PD and understanding the measurable parameters may allow improved management of industrial assets. This paper reports on quantifiable and measurable differences in PD activity and material degradation among voids which have vented and unvented channels. To investigate changes in PD activity with void characteristics experiments are carried out using artificial voids created from layered sections of poly(ethylene terephthalate) (PET) polymer. One set of samples has a confined internal void; a second set of samples has a void with a channel having a closed end and a third set of samples has a vented channel connecting the void to the outer atmosphere. The results show measurable differences in PD activity and subsequent material degradation in the various voids.

On-line Partial Discharge (PD) monitoring is being increasingly adopted in an effort to improve asset management of the vast network of MV and HV power cables. This paper presents a novel method for autonomous recognition of PD patterns recorded under conditions in which a phase-reference voltage waveform from the HV conductors is not available, as is often the case in on-line PD based insulation condition monitoring. The paper begins with an analysis of two significant challenges for automatic PD pattern recognition. A methodology is then proposed for applying the K-Means method to the task of recognizing PD patterns without phase reference information. Results are presented to show that the proposed methodology is capable of recognising patterns of PD activity in on-line monitoring applications for both single-phase and three-phase cables and is also effective technique for rejecting interference signals.

Power curve (PC) monitoring can be applied to evaluate the wind turbine generator (WTG) power output and detect deviations between the expected and the measured value, often a precursor of unexpected faults. In this research, the instantaneous SCADA data is used to show the fault forecast ability of Artificial Intelligence (AI) based PC monitoring of a pitch regulated WTG. The measured PCs illustrate that the instantaneous data is better than averaged data, widely used in the literature, to present the dynamics of WTG operation. The influence of ambient temperature, generator speed and pitch angle on WTG power output is analyzed using measured data. The analysis illustrates that the generator speed and pitch angle have a significant effect on WTG power generation. The performance of the proposed model option is compared against previously published option using the same data sets collected from a 2 MW Pitch Regulated WTG. The comparison is based on the mean absolute error (MAE), the root mean squared error (RMSE) and the correlation coefficient (R2). The result shows that models considering generator speed and pitch angle performs better with lowest MAE and RMSE and highest R2 values. A case study illustrated that the AI models, using wind speed, generator speed and pitch angle inputs, would have successfully detected a pitch fault due to the slip ring malfunction nearly 5 hours earlier than the existing fault detection mechanisms.

On-line Partial Discharge (PD) monitoring is being increasingly adopted in an effort to improve the assessment of MV motors. This paper presents a novel method for autonomous analysis and classification of PD patterns recorded in situations in which a phase-reference voltage waveform is not available, as is often the case in on-line PD based insulation condition monitoring systems. The main contributions of the paper are a Polar PD (PPD) representation and a Fractal Theory based autonomous PD recognition method. PPD is applied to convert the traditional phase-resolved PD pattern into a circular form where raw PD data over a power cycle of 20 ms is shown in a polar form so that important information related to PD pattern is retained irrespective of the initial point when data acquisition starts. The fractal theory is then presented in detail to address the task of discrimination of six PD patterns related to motors, as outlined in IEC60034. The classification of known and unknown defects is calculated by centour score. Validation of the proposed method is demonstrated using data from laboratory experiments on three typical PD geometries. The results show that the proposed method performs effectively in recognizing single-source PDs.

The economics of power transmission requires operators to maximise the power through the network. Increasing knowledge of the impact of environmental factors on the current carrying capabilities of bare wire overhead lines is, therefore, important. Cost savings resulting from enhancing thermal transfer capabilities of transmission lines has motivated many utilities and power systems research centres to investigate techniques for maximising operational parameters. Enhancing power ratings of bare power conductors brings, alongside improved generated power dispatch flexibility, cost savings and economical and technical challenges. The aim of this paper is to examine some of the existing techniques employed to compute the ampacity of bare power lines and discuss practical limitations of these techniques. In addition, the impact of each technique will be considered from an operational prospective. On-line and off-line techniques are summarised and compared based on the nature and quality of power sources connected to the system, e.g. power output of windfarms tends to be of intermittent nature.

Transient overvoltages are inevitable phenomena in power systems. They occur in high voltage direct current systems for much the same as the reasons they happen in high voltage alternating current systems and could be of either internal or external origin. Moreover, the switching operation of the power electronic valves in HVDC converters add to these transient sources. The effects of such transient overvoltages become more critical in the case of non-restoring insulation media like solid dielectrics that form the main insulation media of extruded HVDC power cables. In addition, the intrinsic behavior of solid dielectric under DC fields results in space charge accumulation due to inhomogeneity of the insulation, increasing the likelihood of a fault during overvoltage incidents or even during normal operations like polarity reversal. This paper presents the results of PD measurements that have been conducted on artificial test samples at HVDC superimposed with standard switching overvoltages and operational polarity reversal, which are common in modern converter schemes. The results show that under such transient conditions, PD occurs with higher repetition than the normal voltage level in the test sample. The results of this investigation should be of interest to network operators for asset management on fleets of HVDC cables that may typically be subject to the reported transients and polarity reversal effects.

Transformer failure in electricity supply grids has a high financial impact, due to failure to meet commercial contract and possibility of transformer replacement cost. Therefore, detecting fault inception is very important in order to keep the transformer operating with recommended efficiency, hence ensuring stability of the electric network. Dissolved gas analysis (DGA) of a transformer can provide clear indication of thermal and electrical stresses on power transformer insulation and is considered as one of the most effective tools for oil-filled power transformer diagnostics. DGA is used to detect incipient faults in order to manage the fault severity. Both on-line and offline condition monitoring methods can be applied to obtain gas content, thereafter there are many interpretation techniques for DGA results. The accuracy of these techniques is dependent on the operator's experience and knowledge of the materials and equipment involved. In this work, a combined fuzzy logic analysis technique for monitoring of power transformers based on DGA analysis is proposed, the system uses the 7 key gases to diagnose the health of the transformer and, where applicable, fault type. Initially, gas levels are considered using the IEEE standard as a basis to indicate the health of the transformer. A combined fuzzified analytical tool, based on Duval Triangle, Doernenburg ratio and Key gas method, are analysed to identify the fault type, improving on the accuracy of the individual interpretation techniques. The analytical tool has been applied to 444 sample faults reported in the literature to assess the accuracy of the proposed system. Results presented show the system's overall decision has improved capability of identifying the transformer condition over individual methods. The proposed system is proved to have 99 % accuracy in identifying the transformer normality. For cases where transformers were faulty, the approach has 98.76% accuracy in recognising the actual fault, superior to individual approaches.

This paper presents an investigation into the influence of different operating conditions of HVDC converters on partial discharge behavior occurring within solid dielectrics. Cables at the DC side undergo different stresses due to transients and harmonics that may result in an intensified level of partial discharge (PD) activity and could affect their insulation integrity. Most DC converter schemes, nowadays, are bidirectional with the valve firing angles at the heart of power conversion and power flow control. Depending on the mode of converter operation and the properties of the interconnected AC systems, varying levels of harmonics, either characteristic or non-characteristic, appear on the DC side. By reproducing down-scaled equivalent DC outputs of conventional converters under controlled laboratory conditions, PD pulses occurring under characteristic HVDC voltages have be measured. PD was produced using a range of insulation samples containing known defects and measured using an HFCT sensor. Characteristics of the acquired PD pulses have been analyzed with the aim of finding a promising condition monitoring tool to identify the influence of the superimposed harmonics on PD activity. The results will aid HVDC network operators in identifying incipient cable faults by contributing to diagnostic knowledge rules for interpretation of PD parameters under known operating conditions.

Insulation is a critical component of any electrical power system, which, when subjected to excessive stress, may lead to premature ageing and eventual failure. Partial discharge (PD) measurement is a well-developed insulation diagnostic method, particularly under alternating voltage (AC), for quality assessment and condition monitoring of power apparatus. However, this technique is still immature when applied to DC operating equipment, especially high voltage direct current (HVDC) systems where rippled and distorted voltage waveforms appear at the output terminal of the converters. This paper presents the results of experimental PD measurements and analysis for commonly encountered PD sources: corona, surface and void discharging defects under the application of DC and DC superimposed with harmonics. The results show that PD activity under rippled voltage is different than that under pure DC. PD occurs with higher repetition rate, and is - in varying degrees - influenced by the relative magnitudes of the harmonics, depending on the PD source. Since it appears that PD is more severe under the influence of harmonic voltage ripple, the results should be of interest to network operators and will contribute to understanding of the phenomena in HVDC insulation diagnostics.

Partial discharge (PD) measurement as a non-destructive diagnostic tool is a proven technique that has served in both design improvement and condition monitoring of insulation systems. An understanding and interpretation of the physical phenomena and influencing parameters of PD in AC systems have been established through experience over the past decades, however, flexible alternating current transmission system (FACTS) and high-voltage direct current (HVDC) systems present new challenges. These are due to the differences in physical space charge phenomena and the complexities introduced by power electronic switching. More meaningful information on the physical condition of the dielectric insulation could be gained given a better understanding of the physical discharge phenomena occurring in the region of a defect. Numerical modelling of these complex phenomena can go some way to achieving this. In this paper, PD behavior in HVDC systems has been investigated with the employment of Finite Element Methods (FEM). An insulation defect, consisting of a single void within a polymeric insulation layer, has been modeled to investigate the idealized behavior of such a defect in, for example, a sub-sea HVDC shielded cable. The multi-physics simulation package COMSOL interlinked with MATLAB were used to study the electric field distribution imposed by the HVDC waveform and the PD process. It was found that additional harmonic content results in an increased PD repetition rate to be proportional to the dominant harmonic amplitude.

This paper presents a methodology of optimizing power systems demand due to electric vehicle (EV) charging load. Following a brief introduction to the charging characteristics of EV batteries, a statistical model is presented for predicting the EV charging load. The optimization problem is then described, and the solution is provided based on the model. An example study is carried out with error and sensitivity analysis to validate the proposed method. Four scenarios of various combinations of EV penetration levels and charging modes are considered in the study. A series of numerical solutions to the optimization problem in these scenarios are obtained by serial quadratic programming. The results show that EV charging load has significant potential to improve the daily load profile of power systems if the charging loads are optimally distributed. It is demonstrated that flattened load profiles may be achieved at all EV penetration levels if the EVs are charged through a fast charging mode. In addition, the implementation of the proposed optimization is discussed with analyses on the impact of travel pattern and the willingness of customers.

Partial discharge (PD) measurement is an established condition monitoring technique used to facilitate the detection of incipient faults in high voltage electrical insulation systems such as gas-insulated switchgear and power transformers. By simultaneously recording partial discharge using both the IEC60270 'apparent charge' measurement technique and the more recent Radio Frequency method, which measures the energy radiated from the discharge, it is thought that more information can be gained about the nature of the PD source. Combined measurement has been carried out on various common PD source topologies recreated under laboratory conditions using PD test cells in a range of insulating media including SF6, oil and epoxy resin. Test cell geometries include floating electrodes, free particles and protrusions. By plotting apparent charge against the energy of the RF signal for a large number of PD pulses, the relationship between the two techniques can be quantified for a given test configuration. It has been found that the correlation between the two techniques produces characteristic patterns specific to each defect type. It is envisioned that combined RF/IEC measurements will contribute to a more widespread acceptance of the RF technique in terms of its ability to quantify PD severity by relating the RF energy to the more widely accepted IEC60270 pC level.

Non-intrusive load monitoring (NILM) is a convenient method to determine the electrical energy consumption and operation of individual appliances based on analysis of composite load measured at the entry of a building. It avoids installation of parallel sensors for monitoring individual appliances and could be applied in the smart metering system to obtain useful information for load management. This paper presents an improved NILM method that is capable of recognizing Electric Vehicle Battery (EVB) charging as a load type. Based on the proposed framework, a special pattern recognition algorithm is used to perform load disaggregation. A random switching simulator is developed to examine the performance of the improved NILM under various scenarios. The results demonstrate that the EVB charging load is recognized as well as other traditional appliances. The overall success rate of the disaggregation reaches 94.5% at typical circumstance. Through sensitivity analysis it is also shown that the EVB charging load makes a small impact on the overall performance.

For many years incipient partial discharge (PD) faults in power cables have been identified through off-line investigation techniques. More recently, in an effort to allow pro-active asset management of the medium-voltage (MV) cable network to be carried out, continuous on-line monitoring systems are being installed with the aim of reducing unexpected failures. This study presents work on the analysis and handling of acquired data, from the point of view of asset management and the PD activities observed in an on-line cable monitoring systems. Initially, a review of on-line against off-line cable PD monitoring is presented, in terms of their setups and their respective advantages and disadvantages. The study then presents the authors’ experience of applying wavelet-based denoising techniques [both the discrete wavelet transform (DWT) and the second generation wavelet transform (SGWT)] to PD data denoising. Results of a study on the on-line PD-based monitoring of MV underground cables that are presented in the following section demonstrate that PD activity which is observed in on-line monitoring also has an associated high level of electrical noise that must be removed to allow proper identification of the PD signals and the denoised PD activity is seen to vary with time. Finally, the necessity of developing a means by which knowledge rules can automatically be acquired from on-line condition monitoring data, to reduce reliance on human expertise, is discussed.

Detection and diagnosis of partial discharge (PD) activity has been widely adopted in electrical plant condition monitoring. Analysis and detection of PD in practical applications is often hampered by noise in the signal. Recent research has shown that the discrete wavelet transform (DWT) is effective in extracting PD pulses from severe noise. One disadvantage, however, is that DWT does not reproduce accurate PD pulse magnitude and pulse shape after thresholding in the presence of strong noise. This paper presents the application of the second generation wavelet transform (SGWT), as an improved algorithm, to extraction of PD pulse from electrical noise. The paper begins with the description of the fundamental theory and structure of SGWT analysis and comparisons with DWT. The method is then applied to both simulated and real world PD data. Results prove that SGWT can significantly improve the effectiveness of PD denoising.

Supervisory control and data acquisition (SCADA) data has been widely applied to identify abnormal conditions in wind turbine generators (WTG). One approach was to apply Artificial Intelligence (AI) to SCADA data, comparing the predicted power output of a WTG and its actual output and using the prediction error as an indicator to detect faults. However, complicated training processes limit its application. This paper presents a normal behavior model (NBM), based on power output-generator speed (P-N) curve, to analyze SCADA data from modern pitch regulated WTGs for detecting anomalies. Through analysis of the operational characteristics of the pitch regulated WTG, it is found that inaccuracies in wind speed measurement, the inertia of the rotor, yaw and pitch misalignments, and air density fluctuation may affect the performance of the power curve monitoring algorithms. This paper shows that under normal conditions the P-N curve based NBM performs better when fitting the SCADA data to WTG output under normal conditions than the power curve. Results demonstrate that it can give alarm to forthcoming faults earlier than the existing condition monitoring system (CMS).

Transformers are an important part of the electrical power system network, therefore, its fault detection is vital. Off-line methods are commonly used for their fault detection. These methods have associated costs derived from the necessity of taking the transformer out of service. The application of on-line methods reduces the expected costs and the possibility of unpredicted failures. In the present study, off-line and on-line methods are applied to the detection of short circuits in transformers, demonstrating the possibility of moving from off-line to on-line methods. Short circuits between sections in a transformer winding, between winding and core and between windings have been considered. PSPICE software is used to simulate the transformer for both detection methods. A comparison of the fault indication in both techniques proves the possibility of moving from off-line to on-line method. Sweep Frequency Response Analysis (SFRA) is considered an accurate technique for off-line tests. The changes in the windings inductance and capacitance affect the number of poles in the system response, so the number of poles will indicate the number of healthy sections. For on-line method, measurement of primary/secondary voltage/current is used to determine the measurable values which would result from a range of internal short circuit faults. From the simulated results, it is found that primary current can be used as the main indicator for the primary winding short circuits; a combination of secondary voltage and primary current is found to be useful for detecting secondary winding faults. Secondary winding voltage can be the main indicator for the cross windings short circuits.

Partial Discharges (PDs) within an AC motor may occur when local electrical stress exceeds the inception voltage. The phase of local electrical stress at each position in the motor varies with the three phase power supply, the PD activity at a location, therefore, varies with the voltage phase. In this research, the motor electric field model is calculated by a Finite Element Method (FEM) based simulation. The paper provides a theoretical foundation for the online PD localization methodology in motor stator windings, which forms also a basis for future work for an on-line motor insulation condition monitoring system to facilitate effective warning of imminent failure with added information of defect location.

Single-core Cross-linked Polyethylene (XLPE) cables are commonly used to transmit 3-phase power in long distance transmission and distribution lines. Cross bonding of metal sheaths across phases limits induced currents. Assessing cable condition allows asset managers to prevent failure. On-line monitoring of Dielectric Loss (DL) indicates the general state of the cable insulation. In practice, DL measurement relies on measurement of a reference voltage and insulation leakage current. Circulating currents, which are generated in unbalanced lengths of interconnected cable sections, are superimposed on the leakage currents of each section. This makes measurement and differentiation of leakage currents in interconnected sections more complex. An on-line method of assessing relative DL in HV cross-bonded XLPE cables, with no requirement of reference voltage, has been developed. The method uses data from power frequency current sensors applied to collect sheath currents. The method, named Leakage Current Separation Method (LCSM), can distinguish the leakage currents under cross-bonded connection. The Relative Tendency of DL in Comparison of Three-phase Leakage Currents (RTDL-CTLC) has been put forward to determine the relative insulation conditions of interconnected sections. The actual data obtained from an 110kV cable tunnel, China, was analyzed to verify the consistency. Error analysis, including the characteristics of current sensors, the precision of data acquisition, synchronization error and the fluctuations of system frequency, was established to improve the on-line measurement.

Worldwide, most PILC MV underground cables in use are approaching the end of their design life; hence, failures are likely to increase. This paper studies the electric field and potential distributions within the PILC insulted cable containing common void-defect. The finite element model of the performance of the belted PILC MV underground cable is presented. The variation of the electric field stress within the cable using the Finite Element Method (FEM) is concentrated. The effects of the void-defect within the insulation are given. Outcomes will lead to deeper understanding of the modeling of Paper Insulated Lead Covered (PILC) and electric field response of belted PILC insulted cable containing void defect.

In order to address the detection and localization of partial discharge (PD) in cross-bonded (CB) high voltage (HV) cables, this paper investigates the characteristics of PD propagation in CB HV cables. On-line PD detection and location in power cables has been a research topic for many years. The issue has not been solved due to surrounding interference and complicated propagation modes in cable joints. It becomes even more complex with CB cables. The main challenge to PD monitoring of CB cables is as a result of the interconnectedness of the sheaths of the three single phase cables. The cross-bonding of the sheaths makes it difficult to localize which of the three phases a PD signal has emanated from. Co-axial cables are used to connect cable sheaths to cable link boxes, for ease of installation and protection against moisture. A second challenge is, therefore, the coupling characteristic when a PD pulse propagates in HV cable joints and the co-axial cables, making PD detection and localization more complex. The paper presents experimental investigations into PD pulse coupling characteristic between the cable center conductor and the sheath and the behavior of PD pulse propagation in CB HV cables, which can be used for PD localization in CB HV cable system.

Epoxy resins are used in the production of high-voltage insulation components. Partial discharges, inherent in high-voltage installations, degenerate the resinous material and cause degradation of the insulation thus reducing the working life of the component, An understanding of the nature of the chemical changes taking place under partial discharge stressing should allow the production of improved insulation systems and/or identification of diagnostic methods for monitoring the state of existing equipment. Attenuated total reflectance Fourier transform infrared (ATR-FTIR) microspectroscopy is used to determine the chemical changes to the surface of a bisphenol-A based epoxy resin resulting from partial discharge stressing in an air environment. Differences in stressed resin surface chemistry are found to be dependent upon the humidity of the stressed air. The variation in degradation products occurring in dry and in moist air are discussed and possible reactions to explain these differences are presented.

A wide range of polymeric materials, including epoxy resins, are moulded to produce components for various uses. Silicon-based release agents are commonly used to ensure that the component can be released from the mould. This work demonstrates that the release agent is absorbed by resin during casting. The reaction of samples to various stress regimes, i.e. chemical, mechanical, radiative and electrical, are considered. The reaction of epoxy resin contaminated with the release agent to electrical discharge stressing is different to that of resin with no contaminant. The differences in reaction to the stress are considered and the possible effects for other conditions are presented.

Epoxy resin is moulded to produce high voltage insulation components. To release the component from the mould, among other techniques, silicon-based release agents are used. The authors demonstrate that release agent is absorbed by resin during casting, and describe the change in reaction to electrical discharge stressing. Partial discharge stressing of epoxy resin contaminated with release agent causes surface crazing and silicon products, not evident prior to stressing, are detectable on the surface. Noncontaminated samples exhibit signs of erosion and chemical change, but show no evidence of cracking nor of silicon products. Chemical, mechanical and radiative stresses applied to contaminated resin failed to produce silicon material on the resin surface. It is proposed that partial discharge stressing of noncontaminated resin causes erosion of the sample surface, producing volatile products and chemical changes: there is no crazing of the surface layer, as the pure resin is homogeneous. Partial discharge stressing of contaminated epoxy resin results in erosion of the organic resinous material, but leaves nonvolatile silicon products on the resin surface. These mask the chemical changes occurring in the resin: the surface crazing is due to a differential stress reaction between the contaminated surface layer and the noncontaminated region below.

This paper investigates the impact of unbalanced loads and SWT penetrations over three phases on voltage violation of residential distribution networks (RDNs). In this research, the point estimate method (PEM) is applied to identify the voltage violation of the network with various SWT penetration and distribution scenarios, taking into account the time varying characteristics of loads and wind speeds. The results quantify the voltage violation that will help distribution network operators to assess and constrain SWT impact on network configuration and operation.