Condition Evaluation of Large Generator Stator Insulation Based on Partial Discharge Measurement

Abstract

The large generator faults are mostly from the stator insulation, which is aged by exposure to a combination of thermal, electrical, mechanical, and harsh environment stresses. Condition evaluation of stator insulation is an important measure of ensuring the safe operation and extending the remaining life of large generator. In this paper, a diagnosis method based on the partial discharge (PD) measurement was used for the condition evaluation of the stator insulation. The statistical parameters of partial discharge phase resolved distribution were proposed to assess the aging condition of stator insulation. A partial least square (PLS) approach was used to explore and extract relationships between the statistical parameters of partial discharge distribution with physical properties and performance such as electrical breakdown strength. Results of the PD testing and statistical analysis show that the statistical parameter Sk extent of partial discharge distribution changes much more remarkably with the aging time than do the other parameters. The statistical parameter Sk can be considered as a potential indicator of stator insulation aging. The PLS approach can effectively assess the stator insulation aging condition and can provide a predictive capability for the stator insulation diagnosis.

Partial discharge pattern recognition using variable predictive model‐based class discrimination with kernel partial least squares regression

“…The PRPD method is classically used to realise the feature extraction of a PD signal. In the method, discharge magnitude (q), the corresponding phase angle, or discharge epoch (ϕ), at which they occur and their number densities or discharge rates (n) over some chosen time interval are used to quantify each of pulses [16,17]. From those quantities, the following PD distributions which are functions of the phase angle ϕ are computed for each PD signal, and the main distributions are as follows: (a) the maximum discharge magnitude phase resolved distribution H qmax (ϕ) that represents the maximum discharge magnitude of the observed discharges in each phase window as a function of the phase angle; (b) the mean of discharge magnitude phase resolved distribution H qn (ϕ) that represents the average amplitude of the observed discharges in each phase window as a function of the phase angle; and (c) discharge rates distribution (ϕ) that represents the number of the observed discharges in each phase window as a function of the phase angle.…”

“…Due to the PD phenomenon which is the phase-dependent behaviour appears under AC voltages conditions, several quantities as a function of the phase angle can be used to describe the PD phenomena [16]. The shapes of H qmax (ϕ), H qn (ϕ), and H n (ϕ) would vary with the pattern of the defects [18].…”

Generator Insulation-Aging On-Line Monitoring Technique Based on Fiber Optic Detecting Technology

“…They also have different coefficients of expansion. Severe vibration can introduce cracks inside the thick epoxy mica and voids inside the mica facilitating PDs [48][49][50][51][52][53][54][55][56]. Of course, winding temperature would depend on load which will show up in the temperature trending profile in an online monitoring configuration.…”

Statistical Health Reasoning of Water-Cooled Power Generator Stator Bars Against Moisture Absorption

“…Popular tools used for this function include statistical methods [1], artificial intelligence [2], support vector machines [3], kernel estimation [4], and Mahalanobis distance (MD) [5]- [7]. CM has been applied to various engineering applications: machine components [8], [9], machine tools [10], and power transformers [11]- [13] or generators [14]- [20]. 3) Health prognostics function: This function aims at predicting the remaining time before an engineered system no longer performs the required function(s) or the remaining useful life (RUL) in real time.…”

“…Fernando et al found that the frequency-domain spectroscopy and estimated parameters are informative to assess the generator insulation condition with a laboratory generator bar experiment [19]. Li et al used a partial least-square approach and a statistical parameter skewness to diagnose the stator bar insulation [20]. General electric (GE) suggests the standard outage test program [31] to indicate the presence of a water leakage through a water-cooled generator stator bar insulation and possible avert stator bar insulation failure.…”