Apparatus and method for DC-component-based fault classification of three-phase distribution power cables with magnetic sensing

Abstract

A DC-component-based fault classification apparatus and method for a three-phase power distribution cable utilizes the reconstructed three-phase currents by measuring the magnetic field around the cable with an array of magnetic sensors arranged around the cable surface. A magnetic shield houses the magnetic sensors and blocks background magnetic fields. A data acquisition system acquires analog signals from the sensors and a processing system extracts DC components in the analog signals for the phases during the transient period after a fault. The potential DC components are extracted by mathematical morphology. These DC components arise in the faulted phases when a fault occurs since there is a large current change in the inductive power network.

Claims

1. An apparatus for short-circuit fault classification of a three-phase power distribution cable, comprising: at least three magnetic sensors, wherein the magnetic sensors are arranged in a circular way to form an array about the cable; a magnetic shield to house the magnetic sensors and to block external magnetic fields; a data acquisition system for acquiring analog signals from the magnetic sensors; and a processing and display system to extract DC components in the analog signals for phases during a transient period after a short-circuit fault and to display those DC components as an indication of the short-circuit fault classification.

2. The apparatus according to claim 1, wherein the magnetic sensors can be Hall-effect sensors, anisotropic magneto-resistance (AMR) sensors, tunnel magneto-resistance (TMR) sensors, giant magneto-resistance (GMR) sensors, or other compact magnetic sensors.

3. The apparatus according to claim 1, wherein the magnetic shield is multi-layered and made of high-permeability material.

4. A method of classifying short-circuit faults in a three-phase distribution power cable of a power system, comprising the steps of: sensing a magnetic field around the cable surface with a plurality of magnetic sensors and producing signals related thereto; applying three-phase current extraction (TCE) to reconstruct three-phase currents from the magnetic signals by a stochastic optimization method; extracting a DC component from the reconstructed three-phase currents using mathematical morphology (MM); and classifying a short-circuit fault based on the extracted DC component.

5. The method according to claim 4, wherein the stochastic optimization method is comprised of inverse current program (ICP), magnetic field evaluation (MFE) and source position optimization (SPO).

6. The method according to claim 4 wherein the stochastic optimization method terminates when a Euclidean distance between measured and calculated magnetic fields is smaller than a pre-set threshold.

7. The method according to claim 5, wherein the ICP optimizes the three-phase currents through a least square method based on preset three-phase conductor positions and measured magnetic fields.

8. The algorithm according to claim 5, wherein the MFE calculates magnetic fields at sensing points of the plurality of magnetic sensors with preset three-phase conductor positions and optimized current.

9. The algorithm according to claim 5, wherein the SPO optimizes a three-phase conductor position by using a genetic algorithm.

10. The algorithm according to claim 4, wherein the step of classifying a short-circuit fault involves referencing a logic table of the possibilities of the existence of DC components in the sensed currents of the phases, and using the logic table to identify phase-to-ground short-circuit fault, phase-phase-to-ground short-circuit fault, phase-to-ground short-circuit fault, and three-phase short-circuit fault.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee.

(2) The foregoing and other objects and advantages of the present invention will become more apparent upon reference to the following detailed description and annexed drawings in which like designations denote like elements in the various views, and wherein:

(3) FIG. 1A is a graphical representation of apparatus for measuring magnetic field around a three-phase power distribution cable according to the present invention, FIG. 1B is a circuit schematic for the apparatus of FIG. 1A and FIG. 1C is a cross-sectional view of a three-phase distribution cable showing the placement of a magnetic sensor array;

(4) FIG. 2 is a flowchart of a method for fault classification utilizing the apparatus of FIG. 1 according to the present invention;

(5) FIG. 3 is a flowchart of a three-phase current extraction (TCE) process used in the method of FIG. 2;

(6) FIGS. 4A-4C are exemplary illustration of DC current component extraction during the transient period after the acquisition of the three-phase currents, wherein FIG. 4A shows the extracted and real DC components in the currents for phase-to-ground short-circuit fault (A-G), FIG. 4B shows the phase-to-phase short-circuit fault (B-C), FIG. 4C shows the phase-phase-to-ground short-circuit fault (B-C-G), and FIG. 4D shows the three-phase short-circuit fault (A-B-C); and

(7) FIG. 5 is a logic table to identify the short-circuit fault types (i.e., phase-to-ground short-circuit fault, phase-phase-to-ground short-circuit fault, phase-to-ground short-circuit fault, and three-phase short-circuit fault).

DETAILED DESCRIPTION OF THE INVENTION

(8) The invention relates to an apparatus and method for fault classification of three-phase distribution power cables. It relies on magnetic sensing to identify the existence of DC components in power phases during the transient period after the fault, thus avoiding the problem of waveform distortion which could be incurred by the DC bias of current transformers if the DC component were detected using existing transducers in the power system, which most commonly are current transformers. Current transformers cannot be used since the DC current inversely results in DC bias in the transformers, distorting the secondary output current waveform. While optical current transformers (OCS) based on magneto-optic effect (Faraday effect) could be used, they are very expensive.

(9) The apparatus 10 for measuring the magnetic field around a three-phase power distribution cable 20 is shown in FIG. 1 separated into upper and lower portions 11, 11. The apparatus includes a magnetic sensor array 12, magnetic shielding 14, a data acquisition unit 16 (FIG. 1B), and a data processing & display unit 18 (FIG. 1B). The magnetic sensors 13 (S1-S8 shown in FIG. 1C) of the sensor array 12 are installed in a circular arrangement for measuring the magnetic fields at various azimuths around the cable 20 surface. They are located with respect to the phase conductors A, B, C of FIG. 1C. In order to measure the magnetic fields generated from the three-phase power cable as accurate as possible, a triple-layered magnetic shield 14 made of high-permeability material is located so as to surround the sensors. The shield 14 protects the sensors from background magnetic interference, which may affect the accuracy of the measured magnetic fields generated from the target cable.

(10) The analog outputs of the magnetic sensors 13 are collected by the data acquisition unit 16. The collected signals are then processed by data processing unit 18 to extract DC components in the phases during the transient period after a fault. The computer also displays the components.

(11) The program to extract and identify the existence of DC components in the phases is shown in FIG. 2. The program starts at step S01 with the measured magnetic fields around the cable surface, which are then processed by Three-phase Current Extraction (TCE) sub-program S02, which is shown in more detail in FIG. 3. The essence of TCE is a stochastic optimization method, in which the three-phase currents and the position of the cable conductors are determined by optimization. In particular, the final phase currents are determined by optimizing minimal Euclidean distance between the measured and calculated magnetic fields. By reconstructing the DC components from magnetic sensing, the problem of waveform distortion incurred by the DC bias of current transformers is avoided.

(12) The reconstructed three-phase currents are shown in FIG. 2 as steps S10 (Phase A), S20 (Phase B) and S30 (Phase C). Then each of the reconstructed three-phase currents are processed by the mathematical morphology (MM) method at steps S12, S22, S32 to analyze the DC components per cycle. The extracted DC component of each phase (i.e., |DC.sub.A|, |DC.sub.B| and (|DC.sub.C|) at steps S14, S24, S34 are compared to a pre-set threshold () in steps S16, S26, S36. The threshold () is set because there can be some small DC currents flowing in the system from the renewable energy system and high-voltage-direct-current (HVDC) transmission line system via the grounding points in the AC system. If the condition is satisfied, the indicator in the respective phase would show 1 at steps S18, S28, S38; otherwise the comparison would show 0 at steps S19, S29, S39.

(13) FIG. 3 describes the details of the TCE program according to step S02 in FIG. 2. The TCE starts at step S40 with a default setting of current (I.sub.0) and positions (P.sub.o) of the three-phase conductors. In order to expedite the optimization and find the most feasible solution, the preliminary currents would be set as the rating currents of the power distribution cable. Meanwhile, the position of the three-phase conductors would be set at the angular positions where the three peaks are located in the magnetic-field-distribution pattern measured in a whole circle around the cable surface. In the inverse current program (ICP), the current is optimized by the least square method (I.sub.0) in step S42. The calculated magnetic fields (B.sub.cal) are then updated by the magnetic field evaluation (MFE) program with the optimized current (I.sub.0) and pre-set position (P.sub.0) at step S44. The end conditions at step S46 are as follows: (a) the Euclidian distance between the calculated (B.sub.cal) magnetic field and the measured (B.sub.mea) magnetic field is smaller than the pre-set threshold (.sub.1); (b) the current imbalance degree (K.sub.I) is smaller than the pre-set value (.sub.2, e.g. 10%); and (c) the angle imbalance degree is smaller than the pre-set value (.sub.3, e.g. 20%). The algorithm terminates and outputs the reconstructed three-phase conductor position (P.sub.f) and current (I.sub.p) at step S48.

(14) If the end conditions are not met, the conductor positions are optimized (P.sub.u) by source position optimization (SPO) at step S45 in which a genetic algorithm is used. The iteration continues to optimize the position and currents of the three-phase conductors until the end conditions are satisfied. Then the optimized currents (I.sub.p) would be accepted as the three-phase currents. Because the positions of three-phase conductors A, B, C are fixed, the three-phase position can be reserved for executing the program the next time. The algorithm of FIG. 3 is operated per quarter of a cycle to extract the DC components for achieving a fast response.

(15) The fault classification method of the present invention is critical for tripping the relays for isolating the faulted areas, or to make an overall decision for auto-reclosing on specific phases.

(16) FIGS. 4A-4D are exemplary illustrations of DC component extraction during the transient phase after the acquisition of the three-phase currents. The tested power distribution network consisting of a feeder (voltage: 22 kV; short-circuit capacity, 2.2 MVA; inductance, 0.35 H; resistance, 13.96), a 20-km distribution power cable (resistance r.sub.1=0.024 /km, r.sub.0=0.412 /km; induction I.sub.1=0.4278 mH/km, r.sub.0=1.5338 mH/km; capacitance c.sub.1=0.2811 F/km, c.sub.0=0.1529 F/km) and the loading (1 MW). The fault is assumed to have occurred at 10 km away from the feeder side at 0.10 s for all fault types, i.e., phase-to-ground short-circuit fault (A-G) in FIG. 4A, phase-to-phase short-circuit fault (B-C) in FIG. 4B, phase-phase-to-ground short-circuit fault (B-C-G) in FIG. 4C, and three-phase short-circuit fault (A-B-C) in FIG. 4D. For the illustration shown in FIGS. 4A-4D the MM method was applied to extract the DC component of three-phase current in the post-fault period. The result shows that the DC components only arise in the faulted phases, and the ones extracted by MM method matched with the accurate values in the faulted phases.

(17) The logic table of fault classification of the power distribution cable is shown in FIG. 5. According to the flowchart in FIG. 2, the faulted phase would be reported as a 1, otherwise 0. The single-phase-to-ground short-circuit fault and three-phase short-circuit fault can be easily identified by judging the existence of logic indication 1 in respective phases, namely, A-G (1-0-0), B-G (0-1-0), C-G (0-0-1), and A-B-C (1-1-1). Two logic 1 indications are shown in either phase-to-phase short-circuit fault or phase-phase-to-ground short-circuit fault. However, since the short-circuit current direction is different (i.e., currents would be flowing between phases in phase-to-phase short-circuit fault, and into the ground in the phase-phase-to-ground short-circuit fault), the DC component would also be different. Therefore, the polarity of the DC components can be used to further distinguish the fault types. The opposite polarity of DC components is for the phase-to-phase short-circuit fault, and the identical polarity is for the phase-phase-to-ground short-circuit fault.

(18) The benefits of the present invention are (1) reliability enhancement of fault classification in distributed networks for ensuring the proper function of relays and reducing the time needed for repair; (2) the elimination of manpower costs in terms of a pre-calibration process for installing relays on each underground power cable for achieving a more cost-effective power system upgrade; and (3) the facilitation of Smart Grid construction by improving the self-healing ability in distribution systems. The reliability is enhanced by tripping the proper relays consistent with the protection scheme (e.g., single-phase re-closing) to reduce the adverse effects of faults. The accurate fault classification is critical for calculation of the distance from the relay to the fault point, saving the time to locate and repair the fault by service personnel. Self-healing in a Smart Grid also requires pinpointing the faulted point as accurately as possible. With accurate fault location based on the correct fault classification, the faults can be isolated in a minimal area, or the network can be re-configured in a fast way to sustain the customers with optimized alternative power supplies.

(19) Further, the system of the present invention is cost effective in that its magnetic sensors are less expensive than optical current transformers. The system of the present invention is also fast, establishing classification in one single cycle, while the prior art requires several cycles.

(20) Specific features of the invention are shown in one or more of the drawings for convenience only, as each feature may be combined with other features in accordance with the invention. Alternative embodiments will be recognized by those skilled in the art and are intended to be included within the scope of the claims. Accordingly, the above description should be construed as illustrating and not limiting the scope of the invention. All such obvious changes and modifications are within the patented scope of the appended claims.