Failure rate estimation of overhead electric distribution lines considering data deficiency and population variability

Abstract

Summary Estimating a precise data‐driven failure rate of electrical components in the distribution networks is a prominent task in asset management of the network. To estimate the failure rates in the overhead distribution lines, there are two main challenges: data deficiency and population variability. In order to overcome difficulties, this paper proposes an applicable method based on hierarchical Bayesian Poisson regression (HBPR). The proposed method is applied to the real distribution system with 34 feeders. The deviance information criterion and model checking procedure are used to compare the goodness of fit between HBPR and exchangeable hierarchical Bayesian model (EHBM). Finally, to show the functionality of the HBPR model, the failure rates obtained from HPBR and EHBM are used to calculate reliability indices, and the results are compared with the actual value of indices. Copyright © 2014 John Wiley & Sons, Ltd.

Bayesian Estimates of Transmission Line Outage Rates That Consider Line Dependencies

“…There is previous research predicting outage rates using Bayesian methods. Li [1], Iešmantis [4], and Moradkhani [5] present three Bayesian hierarchical models. All three hierarchical models have a Poisson distribution for outage counts, but how the outages are counted and lower levels of the model are different.…”

“…The grid cells are introduced to model spatial dependence by constructing a correlation matrix in the Gamma field. The hierarchical model in Moradkhani [5] estimates failure rates of individual overhead distribution feeders, which are assumed to be independent of each other. To have an analytical form for the posterior distribution, conjugate priors are used, which results in a Gamma posterior distribution.…”

“…Transmission line outage rates are often supposed to be proportional to line length, and they are often quoted as rates per unit length [4], [5]. However, a line's outage rate is not strictly proportional to the line length because of substation and other effects, making the dependence on line length only a partial dependence.…”

Do we have enough data? Robust reliability via uncertainty quantification

“…Different endogenous and operational-environmental factors will most likely influence the ageing of the components and produce a very different failure behaviour even for identical lines. For more details on the problem, the reader is referred to [23].…”

Improving distribution system reliability calculation efficiency using multilevel Monte Carlo method

“…Distribution system reliability is generally assessed with the useful life period of the component. The failure rate of a component is assumed as constant, and exponential probability distribution is usually used for modelling the uncertainty of TTF . Similarly, in this study, repair time is also assumed as constant, and exponential distribution could be used for modelling TTR randomness.…”