Grouping electrical disturbance events

Abstract

Automatically grouping electrical disturbance events and generating reports facilitates prioritization of maintenance and repair in an electrical power delivery system. A grouping technique entails automatically obtaining electrical disturbance event and outage data that is based on multiple waveform measurement units (WMUs). It then forms groups of related events based on predefined criteria, including timing, asset matching, event type, phases, and reactance. Finally, it generates reports, including visualizations, based on the grouped event data, which are used for resource allocation, including maintenance and repair activities. This automated process enhances or expedites the efficiency and effectiveness of maintaining and repairing electrical power delivery systems by providing prioritized and actionable insights.

Description

Docket No. 503461 / 70031GROUPING ELECTRICAL DISTURBANCE EVENTSRELATED APPLICATION

[0001] This application claims priority benefit of U.S. Provisional Patent Application No. 63 / 686,773 filed August 24, 2024, which is hereby incorporated by reference in its entirety.TECHNICAL FIELD

[0002] This disclosure relates generally to power system diagnostics and, more particularly, to fault localization and detection in electrical networks.BACKGROUND INFORMATION

[0003] In the IEEE Power & Energy Magazine, September / October 2023 edition (Vol. 21, No. 5), an article starting on page 68 by Hamed Mohsenian-Rad, Ph.D., and Wilsun Xu, Ph.D., titled “Synchro-Waveforms,” describes a power system measurement technology approach. This technology, referred to as a waveform measurement unit (WMU), obtains time-synchronized waveform measurements — i.e., synchro-waveforms — from different locations within a power system. WMUs can capture inconspicuous disturbances that are often overlooked by other types of time-synchronized sensors, such as phasor measurement units (PMUs). WMUs also monitor system dynamics at much higher frequencies, as well as much lower frequencies, than the fundamental components of voltage and current that are commonly monitored by PMUs.

[0004] Another paper by V. Barrera Nunez et al., titled “Feature Analysis and Classification Methodology for Overhead Distribution Fault Events,” presented in the IEEE / PES Transmission and Distribution Committee on July 25-29, 2010, describes the use of waveform data from power quality monitors (PQM) installed in substations. The paper presents an analysis of unique features and a classification method for identifying the root cause of overhead distribution fault events, which are multi-cycle events. In particular, it focuses on faults caused by animal contacts, tree contacts, and lightning-induced disturbances.

[0005] Additionally, there are a number of patents describing the calculation of fault locations. FIG. 1, for example, shows an example set of electrical schematics and waveform diagrams arranged in a sequence for representing a SLG (single line-to-ground) incipient fault location estimation. The estimation technique is described in U.S. Patent No. 8,941,387 of Kim. The ’387 patent describes a time-domain analysis for determining net fault electrical parameters, including inductive reactance.Docket No. 503461 / 70031SUMMARY OF THE DISCLOSURE

[0006] This disclosure relates to automatically grouping electrical disturbance events and generating reports to prioritize maintenance and repair activities in an electrical power delivery system. For instance, a method, performed by a server, involves automatically obtaining electrical disturbance event and outage data from multiple WMUs. It then forms groups of related events based on predefined criteria, which include timing, asset matching, fault type, phases, and reactance.

[0007] The grouping of events begins by identifying and grouping them according to the substation within the electrical power delivery system where the events occurred. The method further refines the grouping by selecting events that occur on the same electrical circuit connected to the identified substation and categorizing those that affect the same phase of the circuit, such as phase A, phase B, or phase C.

[0008] Additionally, the method includes grouping events based on an impedance component value (e.g., cumulative reactance to a fault). This involves calculating a cumulative reactance value for each event and grouping those where the cumulative reactance values fall within a predefined tolerance range. To improve the accuracy of event grouping, the method may also involve adjusting the predefined tolerance range for cumulative reactance based on historical data associated with the specific substation or circuit. As used herein, the term impedance component refers to any impedance or reactance value that is suitable for use in estimating the location of an electrical disturbance or classifying an event type. An impedance component may include, for example, a complex impedance value (Z = R + jX), the magnitude of impedance (|Z|), or a derived reactance value (X), depending on the nature of the fault analysis or the structure of the electrical model being used. In some embodiments, a reactance value may be extracted directly by analyzing the phase difference between voltage and current waveforms, without computing the full complex impedance. This flexibility allows the system to match the form of impedance-related data found in utility network models, which often express line characteristics in terms of per-segment reactance values for use in location estimation.

[0009] Finally, events grouped based on substation, circuit, phase, and cumulative reactance are further analyzed to determine if they share a common root cause. This analysis enables more targeted maintenance and repair actions. As used herein, a “common root cause” refers to the underlying physical, environmental, or operational condition responsible for multiple related electrical disturbance events. Examples include, without limitation, a defective or degraded electrical component (such as a connector, cable, transformer, orDocket No. 503461 / 70031 protection device), an environmental interference (such as vegetation contact, lightning, or animal intrusion), or an operational factor (such as incorrect switching or configuration). In some embodiments, the determination of a common root cause is based at least in part on outage metadata, such as cause codes, device types, and overhead / underground designations, combined with patterns in precursor event data and, in some cases, results from field inspections.

[0010] The method concludes by generating reports that include visualizations of the grouped event data, which are used for resource allocation, including maintenance and repair activities.

[0011] Grouping electrical disturbance events (referred to as the grouper process, or simply, grouper) has several uses. First, it provides the capability to link electrical disturbance events, including both precursors and outages, that share the same underlying cause. This linking can be achieved through rules-based methods, but machine learning approaches can also be utilized. The grouper process aids in automating the reporting of successful cases and enables the tracking of prediction evolution over time. Additionally, it supports the generation of alerts for groups of precursors that have not yet led to an outage. The processed datasets created by the grouper are useful for further research and development by the data science team, as they allow outage log data to be used as labels for precursor events. When used in conjunction with an outage log analysis, the grouper can help assess weaknesses in the existing chain of capabilities, providing actionable predictions for clients. This assessment can guide the prioritization of improvements in areas such as event detection, event classification, and location, and clarify the reasons for such prioritizations.

[0012] In the context of the document, “artificial Intelligence (Al)” and “machine learning (ML)” are terms that relate to the application of computer algorithms and systems to perform tasks that typically require human intelligence. Al is a broader concept that encompasses machine learning as well as other techniques that enable machines to mimic human behavior and decision-making processes. ML, a subset of Al, specifically refers to algorithms that allow computers to learn from and make predictions or decisions based on data. In this document, ML is primarily mentioned in relation to the analysis of waveform data and fault detection in electrical power delivery systems. It involves training models on historical data to identify and predict fault types and causes, such as vegetation-induced faults. The term Al might also appear to broadly reference the intelligent processes involved, including the automation of complex data analysis tasks that go beyond mere programmed instructions to include adaptation and learning from new data, enhancing theDocket No. 503461 / 70031 system’s predictive accuracy and operational efficiency. This distinction is helpful for understanding the capabilities and functions described, recognizing that while all ML is Al, not all Al involves ML.

[0013] Additional aspects and advantages will be apparent from the following detailed description of embodiments, which proceeds with reference to the accompanying drawings.BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0014] To easily identify the discussion of any particular element or act, the most significant digit or digits in a reference number refer to the figure number in which that element is first introduced.

[0015] FIG. l is a set of electrical schematic and waveform diagrams arranged in a sequence for representing a SLG (single line-to-ground) incipient fault location estimation technique in accordance with the prior art.

[0016] FIG. 2 is a block diagram of a WMU network architecture in accordance with one embodiment.

[0017] FIG. 3 is a block diagram showing an electrical power delivery system and WMU synchro-waveform event analysis in accordance with one embodiment.

[0018] FIG. 4 is a set of sample electrical disturbance event waveforms in accordance with one embodiment.

[0019] FIG. 5 is a screenshot of an event overview timeline user interface in accordance with one embodiment.

[0020] FIG. 6 is a screenshot showing a detail view of the events user interface in accordance with one embodiment.

[0021] FIG. 7 is a table showing grouped electrical disturbance events in accordance with one embodiment.

[0022] FIG. 8-FIG. 14 are a sequence of screenshots of an original signal user interface showing different precursor events in accordance with one embodiment.

[0023] FIG. 15-FIG. 17 are screenshots of reports provided based on the grouper process in accordance with one embodiment.

[0024] FIG. 18 is a data report user interface in accordance with one embodiment.

[0025] FIG. 19 is an annotated flow diagram showing an example of the grouper process in accordance with one embodiment.Docket No. 503461 / 70031

[0026] FIG. 20 is a flowchart of a process representing a computer-implemented method performed by a WMU data analysis system for automatically grouping electrical disturbance events detected in an electrical power delivery system.

[0027] FIG. 21 is a block diagram showing components of a WMU or server in accordance with one embodiment.DETAILED DESCRIPTION OF EMBODIMENTS

[0028] In this disclosure, the term electrical disturbance event is used to generically refer to any of a precursor event, incipient fault, or incipient anomaly. A precursor event and an incipient fault (or incipient anomaly) are closely related terms in the context of system monitoring and fault diagnosis, but they can carry slightly different implications.

[0029] A precursor event in the context of power systems refers to an abnormality or a small disturbance that precedes a larger system event, such as a fault or failure. It is a warning sign or early indication that something is not operating as it should within the system. Detecting these precursor events is important because it can allow for preventive actions to be taken before a more significant and potentially damaging incident occurs. For example, a small fluctuation in voltage or a transient spike might be a precursor to a larger fault in a transformer or other piece of electrical equipment. Examples include unusual perturbation, temperature changes, or electrical noise that precedes a mechanical or electrical failure. By analyzing such precursors, engineers can identify and rectify potential issues before they lead to power outages, equipment damage, or safety hazards. Precursor event analysis is part of predictive maintenance strategies in modern power systems, which entail not just the detection of precursor events using advanced monitoring equipment and analytical tools but also the use of data analysis, ML algorithms, and historical performance data to predict when a fault is likely to occur, thereby allowing for maintenance to be scheduled at the optimal time to enhance reliability and reduce downtime.

[0030] An incipient fault is a small or developing fault that has not yet resulted in a complete failure or system interruption. It indicates the initial stage of a fault, where the effects may be subtle and not yet causing significant performance issues. Incipient faults, if undetected and unaddressed, can develop into more serious faults. Examples include a small insulation breakdown in a cable.

[0031] An incipient anomaly is a term that is often used interchangeably with incipient fault, but it can also refer to irregularities that are not strictly faults but could lead to one. An anomaly might be a deviation from the expected behavior, which could be due to a fault, an external influence, or a non-ideal condition in the system.Docket No. 503461 / 70031

[0032] Electrical disturbance events refer to instances where the electrical power’s voltage, current, or frequency deviates from the norm, potentially leading to malfunction or failure of the equipment connected to the power system. These events can be brief and transient or can last for longer periods, impacting the performance and lifespan of electrical devices and infrastructure. Electrical disturbance is a broad term encompassing various types of disturbances, including, but not limited to:

[0033] Voltage / current sags (dips): these are reductions in voltage / current levels for a short duration, often caused by fault conditions or sudden large load connections within the transmission or distribution network. They can lead to sensitive equipment tripping or malfunctioning in industries and commercial facilities connected to the network.

[0034] Voltage / current swells: temporary increases in voltage / current occur due to sudden load drops or equipment failures in the network. Swells can damage equipment and appliances, leading to costly repairs and replacements for both utilities and consumers.

[0035] Transients: high-frequency, short-duration surges in voltage or current are usually triggered by lightning strikes, switching operations, or fault clearing in the transmission and distribution system. These events can cause insulation failure, equipment damage, and operational disruptions.

[0036] Harmonics: distortions in the electrical current or voltage waveform are primarily generated by non-linear loads but can be exacerbated by the transmission and distribution system itself, especially when it has a large number of such loads. Harmonics can reduce the efficiency of power delivery, cause overheating in electrical components, and lead to premature equipment failure.

[0037] Frequency variations: changes from the standard power frequency, caused by imbalances in power generation and consumption. In transmission and distribution systems, maintaining frequency stability is crucial for the synchronization and operation of the entire grid. Frequency variations can disrupt the operation of sensitive equipment and, in severe cases, lead to widespread power outages.

[0038] FIG. 2 shows a WMU network architecture 200 for monitoring and analyzing faults within a power system, such as an electrical power delivery system 300 (FIG. 3). According to some embodiments, WMU network architecture 200 integrates subcomponents provided by a power transmission and distribution (T&D) company with infrastructure and software available from the applicant for this disclosure, Toumetis, Inc. of Boise, Idaho. This integrated system allows for comprehensive monitoring and analysis of the power grid, aiming to quickly and efficiently detect and respond to transitory electrical anomalies.Docket No. 503461 / 70031

[0039] The main components shown in the example of FIG. 2 are a substation 202 and a WMU data analysis system 204, which are communicatively coupled via an internet connection 206. Internet connection 206 is available at substation 202 using cellular connection equipment 208 (e.g., LTE or 5G), fiber, or other communication devices. In some embodiments, WMU data analysis system 204 is a remotely located server (e.g., AWS server) such that the analysis is performed via cloud computing. In other embodiments, WMU data analysis system 204 may be located at an edge, such as within substation 202.

[0040] In this example, substation 202 serves as the physical location where power system monitoring occurs. It contains equipment for controlling and monitoring the electricity flow in the power grid. Substation 202 also includes an internal LAN 210, which is communicatively coupled to cellular connection equipment 208 via wireless or ethernet connection 212 with a router 214.

[0041] LAN 210 interconnects all the monitoring and computing devices located in LAN 210. For instance, a substation server 216 includes local PC 218 with having an optional storage interface 220 for optional removeable storage 222 and a wireless or wired router 214 connected via router interface 224 to local PC 218. Local PC 218 is configured with software processes 226 including a process 228 to get FTP data from devices and a process 230 push data to storage 232 in connection with cloud storage and processing system 234.

[0042] Cloud storage and processing system 234 is configured to read instructions from a machine-readable or computer-readable medium (e.g., a non-transitory machine-readable storage medium), including storage 232, and perform any one or more of the methods discussed herein (or portions thereof), such as discussed for process 1900 (FIG. 19), or any portions of these embodiments. Additional details are described later with reference to FIG. 21.

[0043] Router 214 also connects to multiple WMU devices 236 via server connection 238. Multiple WMU devices 236 include monitoring device 240, which samples synchrowaveform data and provides it via an FTP interface 242 for processes 226.

[0044] FIG. 3 shows electrical power delivery system 300 and WMU synchro-waveform event analysis 302, representing the capability to use waveform data to detect or predict, classify, and locate faults within electrical power delivery system 300, which is useful for quick response and maintenance to ensure reliability and safety in the electrical grid. Example classifications of faults include a tee connector fault, load break elbow fault, cable (to ground or other line) fault, capacitor fault, transformer fault, or other device faults.Other types of events are also possible.Docket No. 503461 / 70031

[0045] Electrical power delivery system 300 includes a power generation plant 304 (which could be a hydroelectric dam, wind generation, or various types of thermal power plants), a step-up transmission substation 306, high-voltage transmission lines 308, a step-down transmission substation 310, a step-down distribution substation 312, distribution network poles 314, and residential or commercial users 316.

[0046] In electrical power delivery system 300, WMU synchro-waveform event analysis 302 is used for fault location estimation 318. Specifically, FIG. 3 shows a time-domain waveform 320 exhibiting an electrical disturbance event 322. Electrical disturbance event 322 is then analyzed to estimate fault location estimation 318, which in this example is estimated at a distribution line 324 between step-down distribution substation 312 and distribution network poles 314 or feeder lines 326.

[0047] Time-domain waveform 320 shows three overlapping waveforms, representing differing views of the impacted phase of an electrical power system. Voltage spikes on all three waveforms indicate an example of a precise, sub-cycle moment when electrical disturbance event 322 appeared in electrical power delivery system 300. The present applicant, Toumetis, Inc., has developed other techniques for location estimation, which are a subject of techniques described in International Patent Application No. PCT / US2025 / 030681, titled “Phasor-Based Fault Location Estimation in Electrical Power Delivery System,” which is hereby incorporated by reference in its entirety. The techniques described in the ’681 application may be used for multi- or sub-cycle faults (with the underlying analysis being performed per half cycle or less).

[0048] FIG. 4 shows a set of waveform samples 402, each exhibiting electrical disturbance events in order to provide an example of criteria for grouping events together. In the example of FIG. 4 a first event 404 is on a first phase whereas a second event 406 and a third event 408 occur on a second phase. For separate events to belong to same group (i.e., due to same cause), they would be detected (1) on the same circuit, (2) on the same (single or multiple) faulty phases (3) at approximate the same location estimates (e.g., as described in the ’681 application), (4) as occurring before the outage, and (5) any other optional criteria.

[0049] When a sustained outage (or simply, an outage) occurs, there are systems or processes that a utility company has in place to assess the details around the fault, e.g., where the physical damage was, what device types were involved, what was the likely cause of the damage, and other metadata for outage events.Docket No. 503461 / 70031

[0050] Some precursors result in system interruption, an outage or momentary outage with an unknown cause. For instance, when the protection system operates, the fault is cleared and the outage is momentary, e.g., five minutes or less, as specified in IEEE 1366. The cause may then temporarily disappear, allowing the system to resume operations, but the unresolved cause may return, causing more / larger outages. If the underlying cause of a system interruption has not be addressed / removed, this may cause further system interruptions later.

[0051] Precursor events detected by the monitoring system typically have little or no such metadata available from the utility company. But if the precursors are grouped to an outage event, then the metadata assigned to the outage event can then be associated with the grouped precursor events as well. This provides for further understanding of the precursor events, and the metadata may also serve as labels for the precursor events that are used for training and assessing the Toumetis physics-based / rules-based / ML models. Thus, one embodiment of the disclosed grouper process is to be able to group together all these events with the same underlying cause.

[0052] To monitor and view electrical disturbance events and automated inferences, FIG.5, FIG. 6, and FIG. 8-FIG. 14 show a set of screenshots for a power event monitoring software application, referred to as Cascadence. Cascadence uses ML to derive analytic insights and predictions from power quality meters, other data sources, outage management system (OMS), and weather data from across the grid. Those predictions inform operations that specific equipment is failing so proactive maintenance / replacement can be carried out and customer outage time is reduced and satisfaction improved. Cascadence feeds that data into its ML algorithms (see, e.g., FIG. 17 of the ’681 application) in order to automatically perform the following tasks: event type classification (i.e., correlate electrical disturbance events to device types), and physics-based incipient fault location (i.e., determine the position of the failing equipment).

[0053] In this example, the Cascadence application presents to a user an event overview timeline user interface 500 (FIG. 5), an original signal user interface 800 (FIG. 8-FIG. 14). With reference to FIG. 5, event overview timeline user interface 500 is accessible via an overview tab 502 and shows several timelines on which events are depicted. The timelines include a precursor timeline 504 and an outage timeline 506 for each day. An individual day can be selected and shown in a detailed view 512. A time slot on the timelines is selectable to view in greater detail particular events occurring in that time slot. In this example, outages 508 are selected so that specific information about an outage event 510 can be viewed.Docket No. 503461 / 70031

[0054] FIG. 6 shows in greater detail how precursor timeline 504, outage timeline 506, and detailed view 512 provide a visualization of separate events that may have the same root cause, in which case these events could be grouped together to prioritize maintenance. For example, precursor event 606, precursor event 608, and precursor event 610 all occurred on the same day before an outage event 612, which has been manually labeled (“M”). This disclosure describes techniques for determining whether any of these precursors (or earlier precursor events 604) are directly related to outage event 612 (or other outage).

[0055] FIG. 7 shows a table of partly grouped results in an example dataset. In the table, “resolved” means group has ended with an outage, “multiple precursors” may be more valuable than “single precursor” groups in terms of their prioritization, and “none” means event does not belong to any group. All events in a “resolved” group are assigned a dw ticket key indicator 702 of the related outage. For “not resolved,” the group is assigned the event id indicator 704 of the earliest precursor in the group. Customer minutes of interruption (CMI) is a measure of the duration of interruptions experienced by customers.

[0056] FIG. 8 shows original signal user interface 800 for viewing the WMU data of a particular March 28, 2024 precursor event 802. In this example, event 802 is selected to show its original three-phase voltage signal waveforms 804, an original three-phase current signal waveforms 806, a reactance to fault 808 (i.e., calculated as described in the ’681 application), and inferences 810 that were determined using machine learning or Al processes. For instance, inferences 810 include an event persistence type 812 (shown as a precursor event) and a fault type 814 that is shown as a line-to-ground (LG) event.

[0057] Event persistence type 812 includes precursor events, outages, and operational issues (e.g., schedule maintenance). This term conveys the various ongoing or transient states of events in the context of a power system, i.e., whether the events are persistent (ongoing or long-lasting in the case of planned outages) or transient (temporary or shortlived), which can be crucial for understanding and managing their impact on the power system.

[0058] Fault type 814 includes single line-to-ground (LG) faults, line-to-line (LL) faults, line-to-line-to-ground (LLG) faults, and three-line faults (LLL).

[0059] In some embodiments, events can also be manually or auto-labeled based on additional inferences. For example, labelling 816 includes a cause 818, a device 820 (not shown), and line placement 822 (not shown, overhead (OH) or underground (UG)).

[0060] Causes 818 indicate the cause of the event and can include any of the following: cap switch, cap switch re-strike, cap switch stuck contact, customer problem, deenergizedDocket No. 503461 / 70031 for safety, dig in, equipment failed / damage, faulty meter, fire, foreign object in lines, improper installation, lightning, loose connection, meter power supply, non-standard construction, other, overloaded normal conditions, planned outage, power electronics switching, restoration switching event, severe weather, snapshot, switch (load off), switch (load on), switching error, switching event, sympathetic event, timing pulse, transformer inrush current, transmission event, tree / limb preventable, tree / limb unpreventable, under investigation, unknown, and vehicle contact.

[0061] Device 820 indicates an electrical component that is determined to be involved with the event. Example electrical components include: conductor (OH), connector (UG), cross arm, customer equipment, cut out, disconnect, disconnect switch, down guy, elbow (L.B.), elbow (N.L.B.), electronic sectionalizer, feed through, fuse, fuse cabinet, fuse switch, ground bank, handhole, hotline clamp, insulator / pin, jumper, meter, miscellaneous substation, other equipment, pole, polymer arrester, pothead, probes, RA switch, RC-off (distribution only), recloser, regulator, relay, splice, step down transformer, switch, and T- connector.

[0062] In the example of FIG. 8, event 802 is captured by a power quality meter (or equivalent device) and classified by Cascadence and its underlying ML algorithms: event type precursor, fault type LG, phases A, and then the location algorithm provide a cumulative reactance to the fault of 10.104 ohms.

[0063] Approximately four minutes later, FIG. 9 shows that another event 902 is captured by a power quality meter, with identical Cascadence classifications, but with a cumulative reactance to the fault of 10.129 ohms. Given the similarity of events 802 and 902, when the grouper process is performed on all events (e.g., later that day), these events 802 and 902 are grouped on the basis of substation, circuit, phase and cumulative reactance to the fault. They are grouped as unresolved multiple events because in this example there is no outage log associated with the momentary outages.

[0064] FIG. 10 shows that, on April 3, 2024, an event is captured by a power quality meter and classified by Cascadence with its underlying ML algorithms: event type precursor, fault type LG, phases A and then the location algorithm provide a cumulative reactance to the fault of 10.233 ohms. When the grouper process is run later that day, this event is added to the two March 28thevents and there are now three unresolved events in this group. While the location values are slightly different, they are within a predefined tolerance value.

[0065] FIG. 11 shows that on May 29, 2024, an event is captured by a power quality meter and classified by Cascadence and its underlying ML algorithms: event type precursor, faultDocket No. 503461 / 70031 type LG, phases A, and then the location algorithm provides a cumulative reactance to the fault of 9.925 ohms. When the grouper process is run later that day, this event is now added to the previous three events and there are now four unresolved events in this group. While the location values are slightly different, they are all within a predefined tolerance value.

[0066] FIG. 12 shows that on July 13, 2024, an event is captured by a power quality meter and classified by Cascadence and its underlying ML algorithms: event type precursor, fault type LG, phases A, and then the location algorithm provides a cumulative reactance to the fault of 10.209 ohms. When the grouper process is run later that day, this event is now added to the previous four events and there are now five unresolved events in this group. While the location values are slightly different, they are all within a predefined tolerance value.

[0067] FIG. 13 shows that on August 2, 2024, an event is captured by a power quality meter and classified by Cascadence and its underlying ML algorithms: event type precursor, fault type LG, phases A, and then the location algorithm provide a cumulative reactance to the fault of 9.995 ohms. When the grouper process is run later that day, this event is now added to the previous five events and there are now six unresolved events in this group. While the location values are slightly different, they are within a predefined tolerance value. The grouper process continues to update the grouping as long as the events have the same substation, circuit, phase and cumulative reactance to the fault.

[0068] FIG. 14 shows that on August 4, 2024, an outage event is captured by a power quality meter and classified by Cascadence and its underlying ML algorithms. Although the “Information” panel is not shown, the event type is an outage, fault type LG, phases A, and then the location algorithm provide a cumulative reactance to the fault of 10.873 ohms (which is shown in the “Associations” panel). When the grouper process is run later that day, this event transitions the group from a multiple unresolved group to a multiple resolved group because a sustained outage event occurs. Additionally, as shown in FIG. 14, clicking on the Associations icon in Cascadence lists all the events in the group.

[0069] Electric utility company personnel, which act as users or clients of Cascadence, may prioritize what type of events are important to them, considering the potential impact to the customers due to a sustained outage. For instance, FIG. 15-FIG. 17 shows an example of reports that are generated and provided to a client, based on the grouper process. In these embodiments, the reports describe a successful grouping, provides additional underlying information, a GIS map with the highlighted search area and some of the waveforms in the grouping. As explained below, example items in the report may include an interactive mapDocket No. 503461 / 70031 based on averaged location, with optional photographic evidence. Photos can be attached to the report, along with additional reports related to the patrol, find, or outage in question. The waveform preview displays the selected member of the group, chosen from the location-days plot, with a clickable link that directs the user to original signal user interface 800 for further analysis of this waveform. The location-days plot is also featured, where red indicates an outage event and a red cross marks the ground truth location from Synergi or other WMU.

[0070] FIG. 15 illustrates an example client-facing “Prediction & Outage Overview” report generated by the grouper process. This report consolidates the details of a successfully matched group of precursor events and a corresponding outage. On the left panel, “Prediction Details” summarize the statistical and analytical data derived from WMU inputs, including the count of related precursor events, classification (e.g., overhead or underground), and both base and corrected cumulative reactance values for the event group. The base reactance value represents the raw cumulative reactance calculated directly from the WMU waveform data. The corrected reactance value is obtained by applying a “reactance adjustment” factor to the base reactance; this adjustment is location-specific, determined based on the substation or feeder from which the measurement originates, and compensates for known impedance characteristics of that network segment. For example, in the illustrated case, the system applies a -3.34 Q adjustment to the base reactance to yield the corrected reactance used for grouping logic and location estimation. Outage-specific metadata, such as the date and time of occurrence, utility outage ticket reference, type of interruption, CMI, identified cause, and affected equipment type, are also included. A “Prediction Validation” section indicates the forecast lead time before failure, classification accuracy, and location accuracy metrics.

[0071] On the right, a geographic information system (GIS) map visualizes the estimated fault search area, overlaid with key network elements including substations, isolating devices, triggered / untriggered fault circuit indicators (FCIs), and fuses. The map legend differentiates device types and structural line status (solid for overhead, dashed for underground). As described in U.S. Provisional Patent Application No. 63 / 777,617 filed March 25, 2025, which is hereby incorporated by reference in its entirety, red lines and purple lines outline the calculated search zone based on fault location algorithms and alarm data, enabling rapid field crew dispatch and targeted inspections.

[0072] FIG. 16 provides a waveform-centric “Prediction & Outage Overview” view for the same grouped events, emphasizing time-domain characteristics and matching logic. The top left panel lists the extracted prediction waveform attributes, such as the detected fault phase,Docket No. 503461 / 70031 fault type (e.g., single line-to-ground), fault duration (in cycles), and grouping methodology (by phase, fault type, estimated location, and calculated fault current). Outage characteristics for the resolved event (e.g., phase, reclosing attempts) are also documented. The right panel presents representative waveform snapshots for selected precursor events and the eventual outage. Each snapshot displays synchronized three-phase voltage and current traces, annotated with the computed cumulative reactance to the fault at that moment. For example, precursor events on March 28 (10.129 Q), May 29 (9.925 Q), and August 2 (9.955 Q) closely match the outage reactance observed on August 4 (10.449 Q), confirming spatial correlation within the system. The “Matching Logic” section outlines how the grouper process confirms linkage, correlating outage phase / location with predictions, validating FCI alarm patterns, and using waveform features (e.g., wave duration, point-on-wave, and energy signatures) in combination with ML algorithms to establish causal relationships between precursor events and the outage.

[0073] FIG. 17 depicts a grouped-event “Summary and Verification” report user interface 1700 integrating selectable success criteria 1702, location mapping 1704, waveform previews 1706, and optional field inspection imagery.

[0074] For instance, the upper left panel provides a concise status for an example event group 1710, including whether it is resolved (e.g., by association with an unplanned or planned outage to fix the issue), the site, affected equipment type, cause of failure, number of related precursors, average computed location reactance (XI), and time-to-failure interval. Green-highlighted badges 1712 allow a user to confirm whether grouping 1710 met the defined precursor, location, and lead-time criteria.

[0075] The upper right panel shows a GIS map with network topology. Similar to FIG. 16, this panel includes the calculated search area and precise outage location (red pin), overlaid with the utility’s asset layout.

[0076] The lower left panel displays a waveform preview for a selected precursor event 1714 in the group, including its voltage and current traces, phase / fault type classification, and reactance measurement, with a link to view the full original waveform in the analysis interface. Selecting a different precursor will then show its waveforms.

[0077] The lower right panel presents an example thermal image captured during a field patrol, showing elevated surface temperatures at the suspected failure point, providing corroborative physical evidence that links the WMU-based prediction to an observable defect in the field.Docket No. 503461 / 70031

[0078] FIG. 18 shows a data report summarizing performance metrics for grouped electrical disturbance events and associated outages. For instance, the upper left panel shows successful predictions averaged over the last 12 months, with the target shown for comparison. A “successful prediction” in this context is one made at least four days prior to the outage and located within the correct region. In some embodiments, success is defined as correctly classifying the event type (e.g., Tee connector) within an agreed-upon location tolerance on the correct circuit. This lead time is client-specific and is selected to enable proactive patrolling and defect remediation before failure. The adjacent bar chart distinguishes between successful predictions realized (“saved”) and those classified as missed opportunities.

[0079] The upper right panel presents reliability metrics widely used in the electric utility industry: CMI, system average interruption duration index (SAIDI), and system average interruption frequency index (SAIFI). CMI is calculated by multiplying the number of affected customers by the outage duration in minutes. SAIDI is calculated by dividing the total duration of all customer interruptions by the total number of customers served, and SAIFI is calculated by dividing the total number of customer interruptions by the total number of customers served. The bar chart differentiates between estimated savings in CMI and missed opportunities.

[0080] The bottom left panel provides a hierarchical breakdown of outage statistics for the selected date range, showing total outages, those with PQ meters present, potentially preventable outages, outages with a corresponding PQ event, grouped precursors, and those classified as successful predictions. Percentages and counts are indicated at each level.

[0081] The bottom right panel lists recent precursor groups, with columns for substation, circuit, number of precursor events, and patrol status. This tabular view allows users to quickly identify active groups requiring field verification or follow-up. In some embodiments, the patrol status can be enhanced with additional information provided by utility personnel, such as uploaded images from field inspections (see, e.g., FIG. 17) or annotations describing observed conditions. Supplemental data, including success criteria for confirming or dismissing a suspected fault, may be associated with an event group and stored for historical reference.

[0082] By allowing utility personnel to add such images and supplemental data to an event group, a historical dataset can be compiled to guide prioritization of precursor investigations. Statistical information derived from the dataset (for example, the number of precursors in a group and the intervals between them before resolution) may be used toDocket No. 503461 / 70031 improve the accuracy of predicted time to failure. In general, as the number of precursors increases, as current magnitudes rise, and as events occur more frequently in a group, the expected time to failure decreases. Thus, aggregating these data provides a richer dataset for training and validation, improving prediction accuracy.

[0083] Predicted location can also be used in conjunction with patrol status to prioritize investigations. For instance, potential faults on a feeder with high customer impact (CI) may be investigated before faults on laterals with lower CI. Location estimates can further help utilities isolate which section of a circuit is most likely to contain the fault, enabling crews to target inspections and repairs more efficiently and restore service more quickly.

[0084] The patrol status panel may also integrate configurable thresholds for user notification. A utility operator can specify conditions that trigger alerts, such as a minimum number of precursors in a group within a given time window, a current magnitude of a group above a set threshold, or a predicted fault location on a critical feeder for a group. When these conditions are met, the system can automatically update patrol status and generate alerts through channels such as SMS, email, or push notification. These proactive alerts enable the operator to dispatch crews or schedule inspections before an outage occurs.

[0085] FIG. 19 shows a precursor event grouping process 1900. The aim of process 1900 is to group precursor events that might be related to each other and any related outage. Process 1900 involves several steps, starting with arranging existing group data 1902.

[0086] First, authentication 1904 with Cognito OAuth2 token endpoint 1906 is performed to obtain an access token for storage requests using a back-end Cognito app client.

[0087] Next, 120 days of groups 1908 are retrieved 1910 from storage. Each group 1908 is deemed to be within the last 120 days if an outage start time or the latest precursor event in that group occurred within this period. Metadata for outages and precursor events in groups 1908, including location and alarms, is then obtained 1912 from storage, stored in grouped 1914 and formatted as format groups 1916.

[0088] Following this, ungrouped events 1918 and ungrouped outages 1920 from the last 60 days are retrieved 1922 and 1924, where the outage or event start time is within this timeframe. This period allows for any issues with the ingest process and captures manually labeled events, such as when a user changes an event’s type from “Operational” to “Precursor” and it is in the last 60 days.

[0089] When ungrouped outages 1920 are being retrieved 1924, client-specific search criteria are applied, and that the outage ID is populated. An example of client-specific search criteria includes an interruption type code that is one of “FDR [feeder],” “LATDocket No. 503461 / 70031[lateral],” “OCR [oil circuit recloser],” “SYS [system, not a local circuit],” or “TX [transformer].” More generally, ungrouped outages 1920 are those electrical disturbance events having a manual event type of “Outage” or “NULL.” Other client-specific search criteria may apply.

[0090] When ungrouped events 1918 are being retrieved 1922, search criteria may specify location criteria, taking into account manual, grouper, and model label preference order. This includes an event type of “Outage” (e.g., for a momentary outage) and fault type of NULL / LG / LL, or an event type of “Precursor,” fault type of LG / LL, with a cycle length of Transient - 0.25 (when fault type is LG).

[0091] Next, process 1900 entails determining 1926 whether there are possible outage or restoration events from existing not resolved * groups 1908. Possible outage / restoration events have a manual event type set to “Outage” or set to “NULL.” Similarly, process 1900 entails determining 1928 whether there are possible outage or restoration events from ungrouped events 1918. This ensures manually labeled precursor events are not used for the outage / restoration event matching 1930 or 1932.

[0092] Process 1900 then entails matching 1930 events to existing outage no fault groups, resulting in unchanged outage no fault groups and new outage no precursor groups. An “outage no fault” refers to the case where it is known that there is an outage due to the outage log, but there has been no observed corresponding waveform event / disturbance. There might be different explanations for this: the waveform data was somehow absent at the time of the outage, or it was a planned outage by the utility and therefore a fault waveform disturbance would not have been expected, etc.

[0093] An ungrouped event is an outage event if it occurs within X seconds of the outage start time (X is client-specific). A match occurs when the event asset matches the outage asset. When an outage is present, the outage asset can be used to group to. The criteria are the same whether finding outage events or precursor events for the outage: the event is on the same circuit as the outage; the event has a corresponding alarm that matches the outage circuit; or the event is on a parent asset of the outage circuit. Similarly, matching 1930 for restoration events occurs when the event takes place during the outage with a client-specific tolerance, and the event asset matches the outage asset.

[0094] Given the criteria, an event can match against multiple outages if they are close together in terms of time. A different group is created for each match, which introduces a problem of what the grouper labels should be if an event matches multiple outages. In this case, the oldest outage is used to populate the labels. Grouper labels are discussed at the endDocket No. 503461 / 70031 of process 1900. When modifying a group, the previous group ID (or IDs) is stored to create a group history in the database.

[0095] Ungrouped events, together with any possible outage / restoration events found in existing not resolved * groups, are matched 1932 against ungrouped outages using the same criteria mentioned above. This leads to new outage no fault groups and new outage no precursor groups.

[0096] Finally, as shown in matching 1930, ungrouped events and possible outage / restoration events are checked against existing outage no fault groups. This step is for when a client has been found to adjust outage start times in subsequent outage logs. This means that an event that may not have been deemed an outage event when the outage was first ingested may now be one. The outcome includes unchanged outage no fault groups and new outage no precursor groups.

[0097] In determining 1926, using events from existing not resolved * groups in the latter two steps may reveal that one or more events in the group are outage / restoration events. This leads to modified or sometimes empty not re solved * groups. The outcome is modified not resolved * groups.

[0098] Next, process 1900 entails determining 1934 possible precursor events from ungrouped events. Possible precursor events are those having a manual event type set to “Precursor” or the following conditions: no manual event type (NULL) and no restoration event grouper labels (grouper event type is “Outage,” grouper cause is “Restoration Switching Event”).

[0099] Ungrouped precursor events are matched 1936 to resolved * groups. The outcome includes unchanged resolved * groups and updated resolved * groups. A resolved single precursor group’s status may be unchanged if the new precursor added to the group is deemed to be a duplicate event. This step involves taking all resolved * groups and checking ungrouped events to see if they can be assigned to the groups as precursors. It is unlikely that group changes occur here, but it may happen if an older event has been manually labeled. This is the first step where the following concepts are introduced: unpredictable outages, group event / outage ID blacklist, and outage / precursor event grouping criteria. Unpredictable outages are those with causes that are unpredictable in nature, such as animal contact, severe weather, and planned outages. The group event / outage ID blacklist involves storing any event removed from a group in the database, preventing the grouper from putting the event back in the same group. The opposite also applies, preventing an outage from being grouped to the event. Outage / precursor event grouping criteria determineDocket No. 503461 / 70031 if an outage event can be grouped to a precursor event. The precursor must occur before the oldest outage event in the group.

[0100] In some embodiments, matching means the asset must match (see outage asset matching mentioned in matching 1930). The group must contain at least one event with a fault type of LG or LL or other fault type with location data. If the group consists of only LL events and null / LLG / LLL, the event must have a fault type of LL to be matched with this group. If the group has any LG events, the event must have a fault type of LG to be matched. This is so that the location data can be compared.

[0101] Some faults evolve over time to become more complex, as described in U.S. Provisional Patent Application No. 63 / 849,279 filed July 23, 2025, which is hereby incorporated by reference in its entirety. For that reason, other embodiments may group earlier, less complex events with later, more complex events as they evolve, provided they are relatively close in terms of timing and the calculated location. An example of this is phenomena is when a wind gust causes a tree to initially produce an LL or LG fault, which then collapses an overhead distribution pole causing an LLL fault, and eventually grounds causing an LLG fault. This evolving event can be detected and grouped even though there are different fault types or phases involved at different times.

[0102] The phases of all events in the group are collected, and all of the event’s phases must intersect with the group’s phases. Reactance matching depends on the group’s fault type. If the group consists of only LL events and null / LLG / LLL, the event’s LL reactance value is used. Otherwise, the LG reactance value is used. The lowest and highest reactance values in the group are determined, and the event’s reactance value must lie within these bounds, including a client-specific tolerance on both sides.

[0103] In some embodiments, the tolerance range for the impedance component value is dynamically adjusted based on historical event data for the same substation or feeder. Such historical data can include prior precursor and outage events for which the impedance component values and corresponding field-confirmed fault locations are known. By analyzing the typical variation in impedance component values for related events on that part of the network, the system can calibrate the tolerance range to account for local network characteristics, such as impedance modeling discrepancies or asset configuration differences. This adaptive tolerance adjustment improves the accuracy of grouping by reducing false positives and false negatives when matching events.

[0104] Ungrouped precursor events are then matched 1938 to outage no precursor groups using similar criteria as in the previous step. This step is largely the same as the previousDocket No. 503461 / 70031 matching 1936 except that the remaining precursor events are matched to outage no precursor groups. Any events added to resolved * groups are no longer available for this step, as precursors can only be in one group. One other difference from the previous step is the use of both existing outage no precursor groups and newly formed outage no precursor groups. It is unlikely that new resolved * groups are formed in this step because of the timing of data ingestion. A precursor event ingested on the same day as an outage, occurring a few hours before the outage, may be an example. The outcome includes new resolved * groups, which could arise from precursor events being added to existing or new outage no precursor groups, and unchanged outage no precursor groups.

[0105] Following this, ungrouped precursor events are matched 1940 to not resolved groups. This involves matching existing precursor groups to new precursors, including updated precursor groups from matching 1930 and 1932. The outcome includes updated not resolved * groups and unchanged not resolved * groups.

[0106] Other than the asset criterion, the other criteria are the same as steps 1936 and 1938. Without an outage asset to match against, the asset criterion differs slightly in these steps. For some embodiments, this criterion is the events must be on the same feeder. For other embodiments, alarm and hierarchy data are used to group similar events together. If an event has alarms, all events in the group must share at least one of those alarms. If the event is on a circuit, it must be on a parent bus of the event’s circuit, on a bank sharing a parent bus with the event’ s circuit, or on a circuit sharing a parent bus with the event’ s circuit. If the event is on a bank, it must be on a child circuit of the event’s bank, on a bank sharing a parent bus with the event’s bank, or on a bus that is a parent of the event’s bank. If the event is on a bus, it must be on a child circuit of the event’s bus, on a child bank of the event’s bus, or on a bus sharing one or more child circuits with the event’s bus. As used herein, “hierarchical relationships” refer to the parent-child relationships between different asset types represented in the utility’s network model. For example, a bus is a junction in the substation or network where circuits and banks connect. A bank is a group of transformers or other equipment connected to a bus. A circuit (or feeder) is the distribution path that runs from a substation through the network to serve downstream loads. The grouping process uses these hierarchical relationships to determine whether two events are electrically related and occur along the same path in the network, even if the events are not associated with the exact same physical device.

[0107] Remaining ungrouped precursor events are then grouped 1942 into new not resolved * groups using the same criteria as the previous step, resulting in new not resolved * groups.Docket No. 503461 / 70031

[0108] Not resolved * groups are then matched 1944 to resolved * groups, attempting to combine them using the same criteria as in step 1936, resulting in updated resolved * groups or unchanged resolved * groups.

[0109] Not resolved * groups are also matched 1946 to outage no precursor groups, attempting to combine them using similar criteria, resulting in new resolved * groups or unchanged outage no precursor groups.

[0110] Next, process 1900 entails obtaining 1948 all new or updated groups and storing 1950 the groups. These are sent to storage 1952 (POST request to / event / group).

[0111] Finally, process 1900 entails obtaining 1954 grouper label changes for storing 1956. As noted previously with reference to matching 1930, there might have events that could be outage or restoration events for multiple outages. The workaround for this was to associate the event with the oldest outage it matches to. There is a mapping of events to their matched outages used to create grouper labels. This covers all outage and restoration events. The cause, device, and OH / UG properties on the outage are assigned to the events as labels. For precursor events, the cause, device, and OH / UG labels can also be taken if the event is part of a resolved * group. Otherwise, they are left blank. The final step is checking if the label has changed since the last run of the grouper. If it has not changed, it does not need to be stored again. All the labels are then sent to storage 1958 (POST request to / event / grouper_label).

[0112] FIG. 20 shows a process 2000 representing a computer-implemented method performed by a WMU data analysis system (e.g., system 204, FIG. 2) for automatically grouping electrical disturbance events detected in an electrical power delivery system. In block 2002, process 2000 receives, over a network, synchronized multi-phase voltage and current waveform measurements generated by a WMU deployed in the electrical power delivery system. In block 2004, process 2000 detects, from the waveform measurements, electrical disturbance events. In block 2006, process 2000 for each detected event, determines an event type, a disturbed phase, an impedance component value, and a location within the electrical power delivery system based on the waveform measurements and a network model of the electrical power delivery system. In block 2008, process 2000 groups the events by determining, from the network model, whether the events occur within a common feeder or substation, confirming that the events share at least one common disturbed phase, confirming that the events have compatible event types, and retaining those events whose impedance component values are within a predefined tolerance for the group. In block 2010, process 2000 generates, for a grouped set of events, an event-location reportDocket No. 503461 / 70031 comprising a geographic search map overlaid on a topology of the electrical power delivery system, and waveform previews of events in the group.

[0113] Additional embodiments of the method include the following refinements. The method may further include associating precursor events, which lack complete outage metadata, with outage events that have corresponding outage metadata, and assigning at least a portion of the outage metadata to the grouped precursor events. The method may also include determining that events occur on the same feeder or substation by converting the impedance component value for each event to a location in the network model of the electrical power delivery system and mapping that location to the corresponding feeder or substation. In another embodiment, the determination that events occur on the same feeder or substation may be based on analyzing hierarchical relationships among assets in the network model, including buses, banks, and circuits, to establish whether events trace to the same portion of the network.

[0114] The method may further include retaining an event in a group only when its set of disturbed phases is identical to the set of disturbed phases of other events in the group. In some embodiments, the impedance component value may comprise a cumulative reactance value, and grouping includes determining minimum and maximum cumulative reactance values for events in the group and retaining only those events whose values fall within a client-specific tolerance. The tolerance itself may be adjusted based on historical event data associated with the feeder or substation, thereby improving grouping accuracy.

[0115] In another embodiment, the method may further include analyzing the grouped events that have been matched by feeder or substation, disturbed phase, event type, and impedance tolerance to determine whether the events share a common root cause. Grouping may also include matching events based on a common alarm identifier or outage management system (OMS) record, or associating events with an outage when they occur within a client-specific time interval of an outage start time and share an asset match with the outage. In some cases, associating precursor events with outage events further comprises assigning outage cause, device type, and overhead / underground designation from the outage record to the precursor events.

[0116] Finally, generating the event-location report may further include displaying a calculated search area overlaid on the electrical network topology and highlighting network devices located within that search area.

[0117] FIG. 21 is a block diagram illustrating components 2100, according to some example embodiments, configured to read instructions from a machine-readable orDocket No. 503461 / 70031 computer-readable medium (e.g., a non-transitory machine-readable storage medium) and perform any one or more of the methods discussed herein (or portions thereof) in connection with WMUs, the grouper event-grouping process, and associated data analysis and reporting functions of WMU data analysis system 204 (FIG. 2). In particular, these components can execute the grouping logic described in process 1900 or process 2000, store and retrieve historical event / outage data, and generate visualizations and reports, such as those shown in FIG. 15-FIG. 18, for presentation in the Cascadence user interface.

[0118] In this example, hardware resources 2102 include one or more processors 2104 (or processor cores), one or more memory / storage devices 2106, and one or more communication resources 2108, each communicatively coupled via a bus 2110. In some embodiments, a hypervisor 2112 may be executed to provide a virtualized execution environment for one or more instances of the grouper process or other analysis services.

[0119] Processors 2104 may include any suitable general-purpose or specialized processors capable of executing software instructions for tasks such as WMU data ingestion, waveform analysis, cumulative reactance calculation, ML-based fault classification, group matching, and report generation.

[0120] Memory / storage devices 2106 may include volatile or non-volatile memory for storing instructions 2122 and datasets used by the grouper process, including historical outage and precursor event records, trained ML model parameters, and intermediate grouping results for active event sets.

[0121] Communication resources 2108 provide network connectivity for exchanging data with peripheral devices 2118 (e.g., WMUs, operator workstations, or field crew tablets) and with databases 2124 via network 2120. Such communication resources may include wired or wireless interfaces appropriate for server-to-database communication, secure web service APIs, or field device data synchronization.

[0122] Instructions 2122 may include executable code for the grouper’s back-end logic, WMU data processing and storage routines, location estimation algorithms, ML model execution, and front-end report rendering for the Cascadence interface. Instructions 2122 may reside, in whole or in part, within processors 2104, memory / storage devices 2106, peripheral devices 2118, or databases 2124, and may be transferred between these components during operation.

[0123] In light of this disclosure, skilled persons will appreciate that many changes may be made to the details of the above-described embodiments without departing from theDocket No. 503461 / 70031 underlying principles of the invention. The scope of the present invention should, therefore, be determined only by claims and equivalents.

Claims

Docket No. 503461 / 70031CLAIMSWhat is claimed is:

1. A computer-implemented method performed by a waveform measurement unit (WMU) data analysis system for automatically grouping electrical disturbance events detected in an electrical power delivery system, the method comprising: receiving, over a network, synchronized multi-phase voltage and current waveform measurements generated by a WMU deployed in the electrical power delivery system; detecting, from the waveform measurements, electrical disturbance events; for each detected event, determining an event type, a disturbed phase, an impedance component value, and a location within the electrical power delivery system based on the waveform measurements and a network model of the electrical power delivery system; grouping the events by determining, from the network model, whether the events occur within a common feeder or substation, confirming that the events share at least one common disturbed phase, confirming that the events have compatible event types, and retaining those events whose impedance component values are within a predefined tolerance for the group; and generating, for a grouped set of events, an event-location report comprising a geographic search map overlaid on a topology of the electrical power delivery system, and waveform previews of events in the group.

2. The method of claim 1, wherein the grouping includes associating precursor events, which lack complete outage metadata, with outage events that have corresponding outage metadata, and assigning at least a portion of the corresponding outage metadata to the grouped precursor events.

3. The method of claim 1, in which determining that events occur on the common feeder or substation comprises converting the impedance component value for each event to a location in the network model of the electrical power delivery system and determining a corresponding substation or feeder for that location.

4. The method of claim 1, in which determining that events occur on the common feeder or substation comprises analyzing hierarchical relationships between assets in the network model, including bus, bank, and circuit relationships.Docket No. 503461 / 700315. The method of claim 1, in which grouping the events further comprises retaining an event in the group only when its set of affected phases is identical to the set of affected phases of the other events in the group.

6. The method of claim 1, in which the impedance component value comprises a cumulative reactance value, and retaining events in the group includes determining minimum and maximum cumulative reactance values for events in the group and retaining only those events whose cumulative reactance values are within a client-specific tolerance of the minimum and maximum values.

7. The method of claim 6, further comprising adjusting the client-specific tolerance for the impedance component value based on historical data associated with a corresponding substation or feeder to improve the accuracy of event grouping.

8. The method of claim 1, further comprising analyzing the events in a group that has been matched by feeder or substation, disturbed phase, event type, and impedance component value tolerance to determine whether the grouped events share a common root cause.

9. The method of claim 1, in which grouping the events further comprises matching events based on a common alarm identifier or outage management system (OMS) record associated with each event.

10. The method of claim 1, in which grouping includes associating events with an outage event when the events occur within a client-specific time interval of an outage start time and share an asset match with the outage event.

11. The method of claim 1, in which associating precursor events with outage events further comprises assigning outage cause, device type, and overhead / underground designation from the outage record to each precursor event in the group.

12. The method of claim 1, in which generating the event -location report further comprises displaying a calculated search area overlaid on an electrical network topology and highlighting network devices within the search area.

13. An apparatus for automatically grouping electrical disturbance events and generating reports for prioritization of maintenance and repair in an electrical power delivery system, the apparatus comprising: a processor; andDocket No. 503461 / 70031 a computer-readable media communicatively coupled to the processor, the computer- readable media storing instructions that, when executed by the processor, cause the apparatus to: receive, over a network, synchronized multi-phase voltage and current waveform measurements generated by a waveform measurement units (WMU) deployed in the electrical power delivery system; detect, from the waveform measurements, electrical disturbance events; for each detected event, determine an event type, a disturbed phase, an impedance component value, and a location within the electrical power delivery system based on the waveform measurements and a network model of the electrical power delivery system; group the events by determining, from the network model, whether the events occur within a common feeder or substation, confirming that the events share at least one common disturbed phase, confirming that the events have compatible event types, and retaining those events whose impedance component values are within a predefined tolerance for the group; and generate, for a grouped set of events, a user interface display configured to assist in locating events and prioritizing maintenance, the report comprising visualizations of the grouped event data.

14. The apparatus of claim 13, in which the computer-readable media further stores instructions that, when executed by the processor, cause the apparatus to identify and group events according to the substation within the electrical power delivery system where the events occurred.

15. The apparatus of claim 14, in which the computer-readable media further stores instructions that, when executed by the processor, cause the apparatus to select events that occur on the same electrical circuit connected to an identified substation.

16. The apparatus of claim 15, in which the computer-readable media further stores instructions that, when executed by the processor, cause the apparatus to categorize events that affect the same phase of the identified circuit.

17. The apparatus of claim 13, in which the computer-readable media further stores instructions that, when executed by the processor, cause the apparatus to calculate a cumulative reactance value for each event and group events whose cumulative reactance values fall within a predefined tolerance range.Docket No. 503461 / 7003118. The apparatus of claim 17, in which the computer-readable media further stores instructions that, when executed by the processor, cause the apparatus to adjust the predefined tolerance range for cumulative reactance based on historical data associated with the specific substation or circuit to improve the accuracy of event grouping.

19. The apparatus of claim 17, in which the computer-readable media further stores instructions that, when executed by the processor, cause the apparatus to analyze the grouped events to determine whether they share a common root cause, thereby enabling more targeted maintenance and repair actions.

20. The apparatus of claim 13, in which the computer-readable media further stores instructions that, when executed by the processor, cause the apparatus to generate reports including the visualizations of the grouped event data for use in resource allocation pertaining to maintenance and repair activities.

Images