Fault classiﬁcation of distribution power cables by detecting decaying DC components in fault currents with magnetic sensing: a robust non-invasive technique without pre-calibration
In order to trip the relays with proper protection schemes, a very critical step is to classify the fault types. Though traditional fault classiﬁcation methods working in the post-fault steady-state conditions have been adopted, they exhibit the following drawbacks: ﬁrstly, pre-calibration is required to set the thresholds for each power distribution cable. Secondly, these methods can be unreliable since the fault conditions (e.g., fault location, short-circuit resistance, etc.) can be very complex and the threshold may not be reached. Thirdly, the classiﬁcation process is delayed since the signals are extracted in the post-fault steady-state conditions. Last but not least, the wiring conﬁguration is complicated because both current and voltage signals are needed. Recently, a number of transient methods in the post-fault conditions have been recently developed, while there still lacks a fault classiﬁcation method that does not require pre-calibration, invulnerable to EMI, and can reliably identify three-phase short-circuit fault.
This project aims to develop a DC-component-based fault classiﬁcation for power distribution cables. When a fault occurs, there is a large change of current in the faulted phases. Since the power network is highly inductive (generators, transformers, etc.), the DC components arises in the faulted phases. This method is promising because: (1) the faulted and healthy phases can be reliably distinguished even for identifying three-phase short circuit fault; (2) no pre-calibration is needed; (3) background EMI does not affect DC component. We will develop a robust algorithm for extracting three-phase DC component in fault current from the magnetic ﬁeld around a multi-core power cable. This method will be veriﬁed by simulation ﬁrst. A prototype platform measurement platform will be established for implementation. Finally, this method will be tested and veriﬁed by experiment on a laboratory testbed and also on a scaled power grid established by our industrial partner.
This proposed method will provide a more reliable fault classiﬁcation to ensure the proper function of relays, boosting self-healing ability in distribution systems to realize smart grid, saving the manpower of pre-calibration and controlling the adverse effects of faults to facilitate the development of smart city. Our research team with both academic and power industry investigators has been working on extracting and analyzing the current signals from the magnetic ﬁelds in power grid for more than eight years. We thus are capable to take up this research challenge.
|Effective start/end date||1/1/17 → 10/31/20|
- University Grants Committee: $76,996.00