Susan Rudd

A generic knowledge-based approach to the analysis of partial discharge data

Partial discharge (PD) diagnosis is a recognized technique to detect defects within high voltage insulation in power system equipment. A variety of methods exist to capture the signals that are emitted during PD, and this paper focuses on the ultra high frequency (UHF) and IEC 60270 techniques. Phase-resolved patterns can be constructed from the PD data captured using either of these techniques and due to the individual signatures that different defects generate, experts can examine the phase-resolved pattern to classify the defect that created it. In recent years, knowledge regarding PD phenomena and phase-resolved patterns has increased, providing an opportunity to employ a knowledge-based system (KBS) to automate defect classification. Due to the consistent physical nature of PD across different high voltage apparatus and the ability to construct phase-resolved patterns from various sensors, the KBS offers a generic approach to the analysis of PD by taking the phase-resolved pattern as its input and identifying the physical PD processes associate with the pattern. This paper describes the advances of this KBS, highlighting its generic application through the use of several case studies, which present the diagnosis of defects captured through both the IEC 60270 and UHF techniques. This paper also demonstrates, in one of the case studies, how a limitation of previous pattern recognition techniques can be overcome by mimicking the approach of a PD expert when the pulses occur over the zero crossings of the voltage waveform of the phase-resolved pattern.

On-line Transformer Condition Monitoring through Diagnostics and Anomaly Detection

This paper describes the end-to-end components of an on-line system for diagnostics and anomaly detection. The system provides condition monitoring capabilities for two in- service transmission transformers in the UK. These transformers are nearing the end of their design life, and it is hoped that intensive monitoring will enable them to stay in service for longer. The paper discusses the requirements on a system for interpreting data from the sensors installed on site, as well as describing the operation of specific diagnostic and anomaly detection techniques employed. The system is deployed on a substation computer, collecting and interpreting site data on-line.

Circuit breaker prognostics using SF 6 data

Control decisions within future energy networks may take account of the health and condition of network assets, pushing condition monitoring within the smart grid remit. In order to support maintenance decisions, this paper proposes a circuit breaker prognostic system, which ranks circuit breakers in order of maintenance priority. By monitoring the SF6 density within a breaker, the system not only predicts the number of days to a critical level, but also incorporates uncertainty by giving upper and lower bounds on the prediction. This prognostic model, which performs linear regression, will be described in this paper, along with case studies demonstrating ranking breakers based on maintenance priority and prognosis of a leaking breaker. Providing an asset manager with this type of information could allow improved management of his/her assets, potentially deferring maintenance to a time when an outage is already scheduled.

Identifying Harmonic Attributes From Online Partial Discharge Data

Partial discharge (PD) monitoring is a key method of tracking fault progression and degradation of insulation systems. Recent research discovered that the harmonic regime experienced by the plant also affects the PD pattern, questioning the conclusions about equipment health drawn from PD data. This paper presents the design and creation of an online system for harmonic circumstance monitoring of distribution cables, using only PD data. Based on machine learning techniques, the system can assess the prevalence of the 5th and 7th harmonic orders over the monitoring period. This information is key for asset managers to draw correct conclusions about the remaining life of polymeric cable insulation, and prevent overestimation of the degradation trend.

Interpretation of partial discharge activity in the presence of harmonics

Recent work has identified that circumstances of equipment operation can radically change condition monitoring data. This contribution investigates the significance of considering circumstance monitoring on the diagnostic interpretation of such condition monitoring data. Electrical treeing partial discharge data have been subjected to a data mining investigation, providing a platform for classification of harmonic influenced partial discharge patterns. The Total Harmonic Distortion (THD) index was varied to a maximum of 40%. The results show progressive development for interpretation of condition monitoring data, improving the asset managers holistic view of an assets health.

Intelligent monitoring of the health and performance of distribution automation

With a move to ‘smarter’ distribution networks through an increase in distribution automation and active network management, the volume of monitoring data available to engineers also increases. It can be onerous to interpret such data to produce meaningful information about the health and performance of automation and control equipment. Moreover, indicators of incipient failure may have to be tracked over several hours or days. This paper discusses some of the data analysis challenges inherent in assessing the health and performance of distribution automation based on available monitoring data. A rule-based expert system approach is proposed to provide decision support for engineers regarding the condition of these components. Implementation of such a system using a complex event processing system shell, to remove the manual task of tracking alarms over a number of days, is discussed.

The role of circumstance monitoring on the diagnostic interpretation of condition monitoring data

Circumstance monitoring, a recently coined termed defines the collection of data reflecting the real network working environment of in-service equipment. This ideally complete data set should reflect the elements of the electrical, mechanical, thermal, chemical and environmental stress factors present on the network. This must be distinguished from condition monitoring, which is the collection of data reflecting the status of in-service equipment. This contribution investigates the significance of considering circumstance monitoring on diagnostic interpretation of condition monitoring data. Electrical treeing partial discharge activity from various harmonic polluted waveforms have been recorded and subjected to a series of machine learning techniques. The outcome provides a platform for improved interpretation of the harmonic influenced partial discharge patterns. The main conclusion of this exercise suggests that any diagnostic interpretation is dependent on the immunity of condition monitoring measurements to the stress factors influencing the operational conditions. This enables the asset manager to have an improved holistic view of an assets health.

Identifying harmonic attributes from on-line partial discharge data

Summary form only given. Partial discharge (PD) monitoring is a key method of tracking fault progression and degradation of insulation systems. Recent research discovered that the harmonic regime experienced by the plant also affects the partial discharge pattern, questioning the conclusions about equipment health drawn from PD data. This paper presents the design and creation of an on-line system for harmonic circumstance monitoring of distribution cables, using only PD data. Based on machine learning techniques, the system can assess the prevalence of the 5th and 7th harmonic orders over the monitoring period. This information is key for asset managers to draw correct conclusions about the remaining life of polymeric cable insulation, and prevent overestimation of the degradation trend.