US-12618888-B2 - Electrical grid edge event detection and mitigation

Abstract

Embodiments detect one or more events on an electrical grid. Embodiments use a sensor installed at an edge of the electrical grid to generate a sensor waveform at a first sampling rate corresponding to current and/or voltage signals. Embodiments transform the sensor waveform into multiple frequency bands and digitize the multiple frequency bands at a second sampling rate that is lower than the first sampling rate. Embodiments receive, by a pattern recognition machine learning algorithm at the edge, the digitized multiple frequency bands for events and predict, using the ML algorithm, an occurrence of the one or more events.

Inventors

• Bradley Roy WILLIAMS
• Cornell Thomas Eyford, III
• Apirux Bantukul
• Ruixian LIU
• Kenny Gross

Assignees

• ORACLE INTERNATIONAL CORPORATION

Dates

Publication Date

20260505

Application Date

20230720

Claims (20)

1. 1 . A method of detecting one or more events on an electrical grid, the method comprising: using a sensor installed at an edge of the electrical grid, generating a sensor waveform at a first sampling rate corresponding to current and/or voltage signals; transforming the sensor waveform into multiple distinct independent frequency bands; digitizing each of the multiple distinct independent frequency bands at a second sampling rate that is lower than the first sampling rate to generate a plurality of different time-series bands corresponding to each distinct independent frequency band; receiving, by a pattern recognition machine learning (ML) algorithm at the edge, the plurality of different time-series bands corresponding to each distinct independent frequency band in real time or near real time as the time-series bands are produced; and predicting, using the ML algorithm, an occurrence of the one or more events in response to receiving the plurality of different time-series bands, the predicting comprising anomaly detection.
2. 2 . The method of claim 1 , wherein the sensor is an optical accelerometer sensor having a sampling rate of approximately 60 KHz or greater.
3. 3 . The method of claim 1 , wherein the ML algorithm comprises a nonlinear nonparametric regression algorithm.
4. 4 . The method of claim 1 , further comprising: sending a notification of the one or more events to a line controller which is adapted, in response to the notification, to provide mitigation control.
5. 5 . The method claim 1 , wherein the pattern recognition machine learning (ML) algorithm is in remote communication with a Digital Asset Cloud Service (DACS) and a Network Management System (NMS).
6. 6 . The method of claim 1 , wherein the multiple distinct independent frequency bands have a corresponding matrix comprising a plurality of columns, further comprising determining an average magnitude of the columns.
7. 7 . The method of claim 6 , further comprising generating the time-series bands from the columns and inputting the time-series bands to the ML algorithm.
8. 8 . The method of claim 5 , wherein the DACS is remotely located in a cloud infrastructure, and the NMS is located on-premise at a utility that is managing the electrical grid.
9. 9 . The method of claim 1 , wherein the one or more events comprises a tree branch contacting a conductor on the electrical grid.
10. 10 . An electrical grid event detection system comprising: a sensor installed at an edge of the electrical grid, the sensor adapted to generate a sensor waveform at a first sampling rate corresponding to current and/or voltage signals; and one or more processors executing instructions and configured to: transform the sensor waveform into multiple distinct independent frequency bands; digitize each of the multiple distinct independent frequency bands at a second sampling rate that is lower than the first sampling rate to generate a plurality of different time-series bands corresponding to each distinct independent frequency band; receive, by a pattern recognition machine learning (ML) algorithm at the edge, the plurality of different time-series bands corresponding to each distinct independent frequency band in real time or near real time as the time-series bands are produced; and predict, using the ML algorithm, an occurrence of the one or more events in response to receiving the plurality of different time-series bands, the predicting comprising anomaly detection.
11. 11 . The system of claim 10 , wherein the sensor comprises an optical accelerometer sensor having a sampling rate of approximately 60 KHz or greater.
12. 12 . The system of claim 10 , wherein the ML algorithm comprises a nonlinear nonparametric regression algorithm.
13. 13 . The system of claim 10 , the processors further configured to: send a notification of the one or more events to a line controller which is adapted, in response to the notification, to provide mitigation control.
14. 14 . The system claim 10 , further comprising: a Digital Asset Cloud Service (DACS) in remote communication with the ML algorithm; and a Network Management System (NMS) in remote communication with the ML algorithm.
15. 15 . The system of claim 10 , wherein the multiple distinct independent frequency bands have a corresponding matrix comprising a plurality of columns, further comprising determining an average magnitude of the columns.
16. 16 . The system of claim 15 , the processors further configured to generate the time-series bands from the columns and inputting the time-series bands to the ML algorithm.
17. 17 . The system of claim 14 , wherein the DACS is remotely located in a cloud infrastructure, and the NMS is located on-premise at a utility that is managing the electrical grid.
18. 18 . The system of claim 10 , wherein the one or more events comprises a tree branch contacting a conductor on the electrical grid.
19. 19 . A non-transitory computer readable medium having instructions stored thereon that, when executed by one or more processors, cause the processors to detect one or more events on an electrical grid, the detecting comprising: using a sensor installed at an edge of the electrical grid, generating a sensor waveform at a first sampling rate corresponding to current and/or voltage signals; transforming the sensor waveform into multiple distinct independent frequency bands; digitizing each of the multiple distinct independent frequency bands at a second sampling rate that is lower than the first sampling rate to generate a plurality of different time-series bands corresponding to each distinct independent frequency band; receiving, by a pattern recognition machine learning (ML) algorithm at the edge, the plurality of different time-series bands corresponding to each distinct independent frequency band in real time or near real time as the time-series bands are produced; and predicting, using the ML algorithm, an occurrence of the one or more events in response to receiving the plurality of different time-series bands, the predicting comprising anomaly detection.
20. 20 . The computer readable medium of claim 19 , the detecting further comprising: sending a notification of the one or more events to a line controller which is adapted, in response to the notification, to provide mitigation control.

Description

FIELD One embodiment is directed generally to an electrical grid, and in particular to the detection and mitigation of events on the electrical grid. BACKGROUND INFORMATION An electrical grid is an interconnected network for electricity delivery from producers to consumers. Electrical grids vary in size and can cover whole countries or continents. An electrical grid includes: (1) power stations that are often located near energy and away from heavily populated areas; (2) electrical substations to step voltage up or down; (3) electric power transmission to carry power long distances; and (4) electric power distribution to individual customers, where voltage is stepped down again to the required service voltage. One challenge for utilities that manage and operate distributed electrical grids is the fairly common scenario where overhead conductors come into contact with a falling branch, creating a high impedance fault. In response, protection systems need to de-energize the respective conductors, but this can take a long time (e.g., minutes). The falling branch on conductors can cause electrocutions and can start wildfires and cause huge potential liability to the electric utility. In generally, for the utility to de-energize the falling conductors when utility protection equipment does not sense the faults, a human will make an emergency call to police/fire dispatchers, and the dispatchers will contact the utility emergency response center with the reported location of the fallen tree branch in the line and/or the actual fallen conductor. However, it can take 30 minutes or longer before a utility crew can respond and de-energize the impacted conductor, or longer if, for example, a tree branch falls into a line and/or takes down a line in the middle of the night with no witnesses. A similar challenge for utilities is to support grid stability in view of the rapid growth of inverter based distributed energy resources (e.g., rooftop solar photovoltaic (“PV”) panels, battery energy storage, most wind turbine DC-AC connections, and EV charging systems) and the decommissioning of large central power plants with their synchronous generators that have traditionally held grids together through disturbances with their spinning inertia of the generator turbine. To support decarbonization and grid resiliency, new approaches are needed to provide what is referred to as “synthetic inertia” through autonomous real-time operations at the edge devices. Another challenge for utilities in supporting grid edge processing is manageability of equipment deployed at the grid edge. Due to the distributed nature, there will be numerous distributed equipment spread across wide geographical areas. The status, such as configuration, software version, computing resource utilization, etc., must be managed to ensure the proper functionalities and compatibility across the power grid. Manual management of the equipment, such as sending service personnel to each site to update equipment's software, is too slow due to the geographically distributed nature of these equipment as well as highly error prone. Remote management of the equipment, such as managing software versioning, managing configurations, etc., is needed to ensure effective power grid performance. SUMMARY Embodiments detect one or more events on an electrical grid. Embodiments use a sensor installed at an edge of the electrical grid to generate a sensor waveform at a first sampling rate corresponding to current and/or voltage signals. Embodiments transform the sensor waveform into multiple frequency bands and digitize the multiple frequency bands at a second sampling rate that is lower than the first sampling rate. Embodiments receive, by a pattern recognition machine learning algorithm at the edge, the digitized multiple frequency bands for events and predict, using the ML algorithm, an occurrence of the one or more events. BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate various systems, methods, and other embodiments of the disclosure. It will be appreciated that the illustrated element boundaries (e.g., boxes, groups of boxes, or other shapes) in the figures represent one embodiment of the boundaries. In some embodiments one element may be designed as multiple elements or that multiple elements may be designed as one element. In some embodiments, an element shown as an internal component of another element may be implemented as an external component and vice versa. Furthermore, elements may not be drawn to scale. FIG. 1 is an overview diagram of a distributed electrical grid system that can implement embodiments of the invention. FIG. 2 is a block diagram of a computer server/system in accordance with an embodiment of the present invention that can be used to implement any of the functionality disclosed herein. FIG. 3 is a flow/block diagram of functionality of the grid edge detection