Dissertation Announcement for Zeeshan Ahmed — 04/25/2019 at 2:00 PM
April 5, 2019
Dear Faculty, Graduate and Undergraduate Students,
You are cordially invited to my PhD. dissertation defense.
Dissertation Title: Development of autonomous condition monitoring systems based on probabilistic modelling of lifetime partial discharge data
When: Thursday, April 25, 2019, 02:00 pm
Where: Simrall 228
Candidate: Zeeshan Ahmed
Degree: PhD, Electrical and Computer Engineering
Committee:
Dr. Joni Kluss
Assistant Professor of Electrical and Computer Engineering
(Major Professor)
Dr. J. Patrick (Pat) Donohoe
Professor of Electrical and Computer Engineering
(Committee Member)
Dr. Yong Fu
Associate Professor of Electrical and Computer Engineering
(Committee Member)
Dr. Masoud Karimi-Ghartemani
Associate professor of Electrical and Computer Engineering
(Committee Member)
Abstract
Partial discharge (PD) measurements have been widely accepted as an efficient online insulation condition assessment method in high voltage equipment. The detection of PD activity in the earlier stages of their development is an effective solution to avoid catastrophic breakdown failures in power systems networks. Two sets of experimental PD measuring setups were established with the aim to study the variations in the partial discharge characteristics over the insulation degradation in terms of the physical phenomena taking place in PD sources, up to the point of failure. Furthermore, the experimentally observed data was used to for mulate predictive numerical models analyzing long term historical data, minimizing data storage requirements, while maintaining sufficient details to effectively depict temporal characteristics and generalized PD behavior.
Probabilistic lifetime modeling techniques based on classification, regression and multivariate time series analysis were performed for a system of several PD response variables, i.e. average charge, pulse repetition rate, average charge current, and largest repetitive discharge magnitude over the data acquisition period. Experimental lifelong PD data obtained from cable samples subjected to accelerated degradation was used to study the dynamic trends and relationships among those aforementioned response variables. Distinguishable data clusters detected by the T-Stochastics Neighborhood Embedding (tSNE) algorithm allows for the exam ination of the state-of-the-art modeling techniques over PD data. Dendrograms, silhouette plots, and parallel coordinate plots are used to visualize and observe the consistency of these data clusters. The Bagged Tree Ensemble has been recognized as a relatively accurate classifier for estimating the predictive behavior of PD activity. The said classifier was trained and retrained for multi-experimental data sets distinguishing the hidden similarities and patterns among the PD response variables. An alternative approach utilizing a multivariate time series analysis was used to compare the performance of Classification and Regression models. Stochastically formulated cointegrated variables determined by the Johansen and Engle-Granger cointegration and constraint tests can be combined to form new stationary variables to estimate the parameters for the Vector Auto Regression (VAR) and Vector-Error Correction (VEC) models. The validity of both models was evaluated by generating M onte Carlo and Minimum Mean Squared Error (MMSE) simulated forecasts. True observed data and forecasted data mean values lie within the 95th percentile confidence interval responses, which demonstrates the soundness and accuracy of both models. A life-predicting model based on the cointegrated relations between the multiple response variables, correlated with experimentally evaluated time-to-breakdown values, can be used to set an emergent alarming trigger and as a step towards establishing long-term continuous monitoring of partial discharge activity.
Furthermore, this dissertation also proposes an effective PD monitoring system based on wavelet and deflation compression techniques required for an optimal data acquisition as well as an algorithm for high-scale, big data reduction to minimize PD data size and account only for the useful PD information.
Best Regards,
Zeeshan Ahmed