A method for determining the type of power failure event for a station

Abstract

This invention provides a method for determining the type of power outage event in a station. After continuously collecting characteristic data of key electrical quantities in the station's power system, a dynamic operating baseline is constructed. The real-time collected key electrical quantities are then compared with the dynamic operating baseline, and a pre-defined health deterioration trend combination logic rule is used to determine whether the system has entered a pre-event state. When a pre-event state is entered, the data cache is locked, historical cached data is saved, and an enhanced recording mode is activated to continuously capture subsequent evolution data, ultimately forming a full-process panoramic data package. This package can then be used for tracing the source, determining the type and root cause of the power outage event, achieving a fundamental shift from passive post-event classification to proactive full-process tracing. This method can accurately locate the root cause and dynamic evolution path of the power outage event, providing a direct decision-making basis for preventative maintenance and system optimization.

Description

Technical Field

[0001] This invention relates to the field of power system operation and maintenance technology, and in particular to a method for determining the type of power outage event at a substation. Background Technology

[0002] With the continuous advancement of smart grid construction and the increasing demands for power supply reliability, the safe and stable operation of substation power systems has become crucial. As the "heart" of a substation, the power system provides power for critical loads such as monitoring, protection, communication, and operation. Once it loses power, it may trigger a chain of faults, leading to serious consequences. Therefore, quickly and accurately determining the type and root cause of power outages is of great significance for achieving rapid power restoration, preventing the escalation of accidents, and optimizing operation and maintenance strategies.

[0003] Existing power outage event analysis systems typically monitor electrical quantities such as voltage and current at key nodes. When a clear fault signal such as voltage loss or protection action is detected, the system collects information such as switch changes and protection outputs within a short time window before and after the event. The system then uses preset logical rules to reverse-engineer and compare the data to classify and determine the event type.

[0004] Existing technologies are based on analytical methods that take snapshots of the state immediately after a fault occurs. Their logical starting point is the occurrence of the fault, which has significant limitations: First, this method cannot effectively capture and characterize the gradual process and abnormal precursors of the system before complete collapse, resulting in insufficient analytical capabilities for power loss events caused by dynamic processes such as slow insulation degradation and gradual load increases. Second, due to the lack of full-process information, it is easy to misjudge the root cause of complex faults accompanied by chain reactions or to sever the causal relationship between events. Third, conclusions drawn from instantaneous phenomena often only guide post-event handling such as equipment replacement, failing to reveal deeper systemic operational risks and unable to provide effective decision support for early warning and preventative maintenance. Summary of the Invention

[0005] In view of this, the present invention proposes a method for determining the type of power outage event in a power station, realizing a fundamental shift from passive post-event classification to proactive full-process tracing, which can accurately locate the root cause and dynamic evolution path of the power outage event, and provide a direct decision basis for preventive maintenance and system optimization.

[0006] The technical solution of this invention is implemented as follows:

[0007] A method for determining the type of power outage event at a power station includes the following steps:

[0008] Step S1: Continuously collect characteristic data of key electrical quantities of the power system of the station, and store the characteristic data sequence in the data buffer area;

[0009] Step S2: Based on the feature data sequence in the data buffer, a dynamic operating baseline for key electrical quantities to be automatically adjusted over time and under operating conditions is established using a sliding time window algorithm;

[0010] Step S3: Collect the characteristic data of real-time key electrical quantities of the station's power system, and compare them with the dynamic operating baseline based on the preset health deterioration trend combination logic rules to determine whether the station's power system has entered the pre-event state.

[0011] Step S4: If it is determined that the event is in the pre-event state, lock the data cache area and save the historical cache data from the data cache area back a preset time from the current moment.

[0012] Step S5: Continuously capture subsequent evolution data until the station power system status is stable, and output the historical cached data and subsequent evolution data as a full-process panoramic data package;

[0013] Step S6: Call the full-process panoramic data package, and through the complete evolution process of the key electrical quantities of the power system of the replay station, locate the first abnormal point where the anomaly first appeared and the abnormal evolution path, and determine the type and root cause of the power failure event.

[0014] Preferably, step S1 includes the following steps:

[0015] Simultaneously collect characteristic data of key electrical quantities, including three-phase voltage, three-phase current, zero-sequence voltage, harmonic content, and power, from each section of the busbar and incoming circuit of the power supply system at the station.

[0016] For the characteristic data of the collected key electrical quantities, calculate their effective values ​​within a fixed short period to form a characteristic data sequence;

[0017] Each critical electrical quantity is configured with an independent first-in-first-out data buffer, and the characteristic data sequence is stored in the corresponding data buffer.

[0018] Preferably, the specific steps for calculating the effective value of the collected key electrical quantity characteristic data within a fixed short period are as follows:

[0019] Within a fixed short period, N consecutive sampling points are taken. The characteristic data of the key electrical quantities within each of the N sampling points are squared, summed, divided by the number of sampling points N, and then the square root is taken to obtain the effective value R within the fixed short period. The calculation formula is as follows:

[0020] ;

[0021] in, , The feature data represents the first to Nth sampling points within the fixed short period, where N represents the total number of sampling points within the fixed short period.

[0022] Preferably, step S2 includes the following specific steps:

[0023] The first preset duration is used as the short-term calculation window, and the second preset duration is used as the long-term learning window;

[0024] The moving average and standard deviation of the feature data sequence in the data buffer are calculated in real time, and short-term and long-term baselines are generated respectively.

[0025] Preferably, step S3 includes the following specific steps:

[0026] The characteristic data of key electrical quantities collected in real time are compared item by item with the dynamic operating baseline of the corresponding circuit, and a joint judgment is made according to the pre-set health deterioration trend combination logic rules, wherein the health deterioration trend combination logic rules include:

[0027] When the moving average of the effective value of a certain bus voltage shows a unidirectional continuous downward trend over multiple consecutive sampling periods, and the cumulative decline exceeds the preset proportional threshold of its short-term baseline average value, the trend deviation criterion is triggered.

[0028] When the instantaneous fluctuation amplitude of a feeder circuit current exceeds a preset multiple of its short-term baseline standard deviation multiple times consecutively, the fluctuation anomaly criterion is triggered.

[0029] When the amplitude of the zero-sequence voltage at the neutral point of the system continues to rise and exceeds a preset multiple of its long-term baseline typical value, the parameter imbalance criterion is triggered.

[0030] The health status of the power system is managed by a logic state machine. The state machine includes a normal state, a precursor state, and a fault state. When any combination of one or more logic rules is satisfied, the state machine switches from the normal state to the precursor state.

[0031] Preferably, step S4 includes the following specific steps:

[0032] When the power system of the judgment station enters the pre-event state, a lock command is sent to the data buffer, and all historical buffer data that has been backed up for a preset time from the current moment is transferred to a specific accident data file.

[0033] Preferably, step S5 includes the following specific steps:

[0034] Enabling enhanced recording mode increases the sampling frequency to an integer multiple of the normal frequency and expands the recording range to all logically related circuits and equipment in the station's power system;

[0035] The enhanced recording mode continues to run until the state machine confirms that the power system of the station has entered a new stable state or the state machine has clearly jumped to the fault state. The stable state is defined as the critical electrical quantities continuously recovering to the long-term baseline range and all switch states remaining unchanged for a preset period of time.

[0036] Historical cached data, evolution data captured in enhanced recording mode, and switch change and protection action sequence records are aligned, merged, and encapsulated according to a unified time scale to generate a full-process panoramic data package.

[0037] Preferably, the specific steps in step S6 for locating the first anomaly point and the anomaly evolution path that first showed an anomaly include:

[0038] Parallel traversal scanning is performed on the curves of the characteristic data of all key electrical quantities in the full-process panoramic data package as a function of time. During the traversal scanning process, the characteristic data of each key electrical quantity is compared with its corresponding dynamic operating baseline point by point to identify the direction and degree of deviation of each key electrical quantity from the baseline, and all deviation times are recorded.

[0039] Compare the earliest deviation times of all critical electrical quantities, determine the deviation point with the earliest global timestamp as the first anomaly point, and record the specific electrical quantity type and measurement point location corresponding to the first anomaly point;

[0040] Starting from the first anomaly point, trace forward along the time axis to determine the temporal sequence and logical relationship of subsequent abnormal changes in other key electrical quantities or switch states;

[0041] Based on the electrical topology of the station power system, the propagation direction of the critical electrical quantities that are abnormal in space is analyzed, and the timing of the switch change and protection action information is verified to match the causal logic of the abnormal changes in the critical electrical quantities.

[0042] By combining the chronological relationships in the time series with the transmission paths in the topological space, the complete abnormal evolution path from the first anomaly point to the final power loss is reconstructed.

[0043] Preferably, the specific steps for determining the time sequence and logical association of subsequent abnormal changes in other key electrical quantities or switch states are as follows:

[0044] After identifying the first anomaly, the analysis begins with the timestamp of the first anomaly and proceeds in the positive direction along the time axis.

[0045] The system retrieves all other key electrical quantity characteristic data that broke through their respective dynamic operating baselines after the first anomaly point in the full-process panoramic data package, as well as all circuit breaker opening and closing switch change events and all protection device output action events, and outputs them as abnormal events.

[0046] Each retrieved abnormal event is strictly sorted according to its absolute timestamp to generate a complete event sequence arranged chronologically.

[0047] Based on the main electrical wiring diagram of the station's power system, establish a database of topological connections between all key electrical measurement points and switchgear.

[0048] Based on the aforementioned topology connection database, determine whether each newly emerging critical electrical quantity anomaly event is located downstream of the previously abnormal equipment in the electrical path, or whether it is physically related to the previous anomaly through electromagnetic coupling or control loop;

[0049] Based on the principle of power system fault propagation, this study verifies whether the complete event sequence conforms to the basic logic of fault source propagation along electrical path and triggering the action of protection devices at each level in sequence. A logic diagram of abnormal diffusion and chain reaction with time as the vertical axis and electrical topology as the horizontal axis is constructed.

[0050] Preferably, the specific steps for verifying whether the timing of the switch change and protection action information matches the causal logic of the abnormal changes in key electrical quantities are as follows:

[0051] Extract the key electrical quantity abnormal change sequence and the switch and protection action sequence from the full-process panoramic data package. The key electrical quantity abnormal change sequence includes the start and end times of abnormal voltage drop, abnormal current increase, and abnormal zero-sequence voltage rise. The switch and protection action sequence includes the absolute time of each circuit breaker's opening and closing action, as well as the absolute time of each protection device from startup to output action.

[0052] Perform time alignment and logical comparison of the abnormal change sequence of key electrical quantities and the action sequence of switches and protection devices;

[0053] Retrieve the preset protection setting list and protection action logic, and verify each protection action event and each circuit breaker non-manual operation tripping event;

[0054] For any action event, if no prior critical electrical quantity abnormality that meets the set value conditions can be found as the cause, or if the timing interval does not conform to the inherent characteristics of the equipment, then the action is marked as a timing logic mismatch event.

[0055] Compared with the prior art, the beneficial effects of the present invention are:

[0056] This invention discloses a method for determining the type of power outage event in a station power supply system. During daily operation, the station power supply system continuously collects characteristic data of key electrical quantities, forms a characteristic data sequence, and stores it in a data buffer. By extracting the characteristic data sequence within a preset time period, a dynamic operating baseline can be established. Then, based on the dynamic operating baseline, real-time monitoring of power outages in the station power supply system can be performed. The collected real-time characteristic data of key electrical quantities in the station power supply system can be compared with the dynamic operating baseline, and a judgment can be made based on preset logical rules to determine whether an event precursor state has been entered. When an event precursor state is confirmed, the data buffer is locked, and full-process enhanced recording is activated to construct a full-process panoramic data package. Then, the source analysis is performed through the full-process panoramic data package to determine the type and root cause of the power outage event, so as to facilitate timely maintenance and repair. This solves the problem in the prior art that it cannot effectively analyze the dynamic evolution process and root cause of process and complex faults due to reliance on snapshots of the fault's instantaneous state. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only preferred embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a flowchart of a method for determining the type of power outage event in a station according to the present invention. Detailed Implementation

[0059] To better understand the technical content of this invention, a specific embodiment is provided below, and the invention will be further described in conjunction with the accompanying drawings.

[0060] See Figure 1 The present invention provides a method for determining the type of power outage event at a power station, comprising the following steps:

[0061] Step S1: Continuously collect characteristic data of key electrical quantities of the power system at the station, form a characteristic data sequence, and store it in the data buffer. Specific steps include:

[0062] Simultaneously collect characteristic data of key electrical quantities, including three-phase voltage, three-phase current, zero-sequence voltage, harmonic content, and power, from each section of the busbar and incoming circuit of the power supply system at the station.

[0063] For the collected key electrical quantity characteristic data, calculate their effective values ​​within a fixed short period to form a characteristic data sequence with seconds as the basic unit. The fixed short period is a complete cycle of the power frequency AC current or an integer multiple thereof. The steps for calculating the effective values ​​are as follows:

[0064] The raw AC sampling sequence from the voltage transformer and the current transformer is processed, wherein the raw AC sampling sequence is instantaneous value collected at equal time intervals;

[0065] Take N consecutive sampling points within a fixed short period, square the characteristic data of the key electrical quantities within each of the N sampling points, sum them, divide by the number of sampling points N, and then perform a square root operation to obtain the effective value R within the fixed short period. The calculation formula is as follows:

[0066] ;

[0067] in, , The feature data represents the first to Nth sampling points within the fixed short period, where N represents the total number of sampling points within the fixed short period and is a dimensionless positive integer.

[0068] Each critical electrical quantity is configured with an independent first-in-first-out data buffer, and the characteristic data sequence is stored in the corresponding data buffer. The data buffer continuously saves the characteristic data sequence of the most recent complete cycle in a cyclic overwrite manner.

[0069] Step S2: Based on the feature data sequence in the data buffer, a dynamic operating baseline for key electrical quantities is established using a sliding time window algorithm to automatically adjust over time and under operating conditions. Specific steps include:

[0070] The first preset duration is used as the short-term calculation window, and the second preset duration is used as the long-term learning window;

[0071] The moving average and standard deviation of the characteristic data sequences in the data buffer are calculated in real time, and a short-term baseline describing the recent real-time fluctuation of electrical quantities and a long-term baseline characterizing their historical typical operating range are generated respectively.

[0072] The short-term and long-term baselines are automatically clustered and matched to the typical operating condition category of the current period based on day and night time, weekday and holiday patterns, and seasonal load variation patterns through machine learning algorithms. The baseline parameters are dynamically calibrated and smoothed by calling the long-term statistical model of the corresponding historical period to establish a healthy profile.

[0073] Step S3: Collect real-time key electrical quantity characteristic data of the station's power system, and compare it with the dynamic operating baseline based on the preset health deterioration trend combination logic rules to determine whether the station's power system has entered the pre-event state. Specific steps include:

[0074] The characteristic data of key electrical quantities collected in real time are compared with the dynamic operating baseline of the corresponding circuit item by item, and a joint judgment is made based on the pre-set health deterioration trend combination logic rules.

[0075] The health status of the power system is managed by a logic state machine. The state machine includes a normal state, a precursor state, and a fault state. When any combination of one or more logic rules is satisfied, the state machine switches from the normal state to the precursor state.

[0076] The combined logic rules for the health deterioration trend mentioned above include:

[0077] When the moving average of the effective value of a certain bus voltage shows a unidirectional continuous downward trend over multiple consecutive sampling periods, and the cumulative decline exceeds the preset proportional threshold of its short-term baseline average value, the trend deviation criterion is triggered.

[0078] When the instantaneous fluctuation amplitude of a feeder circuit current exceeds a preset multiple of its short-term baseline standard deviation multiple times consecutively, the fluctuation anomaly criterion is triggered.

[0079] When the amplitude of the zero-sequence voltage at the neutral point of the system continues to rise and exceeds a preset multiple of its long-term baseline typical value, the parameter imbalance criterion is triggered.

[0080] Step S4: If it is determined that the event is in a pre-event state, lock the data cache area and save historical cache data from the data cache area for a preset time backward from the current moment. The specific steps include:

[0081] When the power system of the judgment station enters the pre-event state, a lock command is sent to the data buffer, and all historical buffer data that has been backed up for a preset time from the current moment is transferred to a specific accident data file.

[0082] During normal operation, the background process continuously writes the high-speed sampling raw data and feature calculation results of all critical electrical quantities into a first-in-first-out (FIFO) cyclic data buffer with a fixed capacity. If the state machine determines that the station power system has entered the pre-event state, it immediately sends a lock command to the data buffer and transfers all historical cached data to a specific accident data file in non-volatile memory.

[0083] Step S5: Continuously capture subsequent evolution data until the station's power system status stabilizes. Output the historical cached data and subsequent evolution data as a full-process panoramic data package. Specific steps include:

[0084] While locking historical cached data, an enhanced recording mode is enabled, increasing the sampling frequency to an integer multiple of the normal frequency and expanding the recording range to all logically related circuits and equipment in the station's power system;

[0085] The enhanced recording mode continues to run until the state machine confirms that the power system of the station has entered a new stable state or the state machine has clearly jumped to the fault state. The stable state is defined as the critical electrical quantities continuously recovering to the long-term baseline range and all switch states remaining unchanged for a preset period of time.

[0086] Historical cached data, evolution data captured in enhanced recording mode, and switch change and protection action sequence records are aligned, merged, and encapsulated according to a unified time scale to generate a single continuous full-process panoramic data packet containing the entire time chain of events.

[0087] Step S6: Call the full-process panoramic data package. By replaying the complete evolution process of key electrical quantities in the power system of the replay station, locate the first anomaly point where the anomaly first appeared and the anomaly evolution path. The specific steps are as follows:

[0088] Parallel traversal scanning is performed on the curves of the characteristic data of all key electrical quantities in the full-process panoramic data package as a function of time. During the traversal scanning process, the characteristic data of each key electrical quantity is compared with its corresponding dynamic operating baseline point by point to identify the direction and degree of deviation of each key electrical quantity from the baseline, and all deviation times are recorded.

[0089] Compare the earliest deviation times of all critical electrical quantities, determine the deviation point with the earliest global timestamp as the first anomaly point, and record the specific electrical quantity type and measurement point location corresponding to the first anomaly point;

[0090] Starting from the first anomaly point, trace forward along the time axis to analyze the temporal sequence and logical relationship of subsequent abnormal changes in other key electrical quantities or switch states;

[0091] Based on the electrical topology of the station power system, the propagation direction of the critical electrical quantities that are abnormal in space is analyzed, and the timing of the switch change and protection action information is verified to match the causal logic of the abnormal changes in the critical electrical quantities.

[0092] By combining the chronological relationships in the time series with the transmission paths in the topological space, the complete abnormal evolution path from the first anomaly point to the final power loss is reconstructed.

[0093] Preferably, the specific steps for determining the time sequence and logical association of subsequent abnormal changes in other key electrical quantities or switch states are as follows:

[0094] After identifying the first anomaly, the analysis begins with the timestamp of the first anomaly and proceeds in the positive direction along the time axis.

[0095] The system retrieves all other key electrical quantity characteristic data that broke through their respective dynamic operating baselines after the first anomaly point in the full-process panoramic data package, as well as all circuit breaker opening and closing switch change events and all protection device output action events, and outputs them as abnormal events.

[0096] Each retrieved abnormal event is strictly sorted according to its absolute timestamp to generate a complete event sequence arranged chronologically.

[0097] Based on the main electrical wiring diagram of the station's power system, establish a database of topological connections between all key electrical measurement points and switchgear.

[0098] When analyzing logical connections, based on the topology connection relationship database, it is determined whether each newly emerging critical electrical quantity abnormal event is located downstream of the equipment that previously had an abnormality in the electrical path, or whether it is physically related to the previous abnormality through electromagnetic coupling or control loop.

[0099] Based on the principle of power system fault propagation, this study verifies whether the complete event sequence conforms to the basic logic of fault source propagation along electrical path and triggering the action of protection devices at each level in sequence. A logic diagram of abnormal diffusion and chain reaction with time as the vertical axis and electrical topology as the horizontal axis is constructed.

[0100] Preferably, the specific steps for verifying whether the timing of the switch change and protection action information matches the causal logic of the abnormal changes in key electrical quantities are as follows:

[0101] Extract the key electrical quantity abnormal change sequence and the switch and protection action sequence from the full-process panoramic data package. The key electrical quantity abnormal change sequence includes the start and end times of abnormal voltage drop, abnormal current increase, and abnormal zero-sequence voltage rise. The switch and protection action sequence includes the absolute time of each circuit breaker's opening and closing action, as well as the absolute time of each protection device from startup to output action.

[0102] Perform time alignment and logical comparison of the abnormal change sequence of key electrical quantities and the action sequence of switches and protection devices;

[0103] Retrieve the preset protection setting list and protection action logic. The protection action logic defines the causal relationship between the continuous fulfillment of preset value conditions for specific electrical quantity abnormalities and the corresponding switch action, and verifies each protection action event and each circuit breaker non-manual operation tripping event.

[0104] Protection action event verification: If the electrical quantity corresponding to the action is abnormal, and its occurrence time is earlier than the protection action time, and the abnormal characteristics continuously meet the protection's set value and delay conditions before the action, then the causal logic of the sequence is determined to be matched.

[0105] Circuit breaker non-manual operation tripping event verification: If the upstream protection device of the circuit has taken an output action before the tripping event, and the time difference between the two events is within a reasonable range of the circuit breaker's inherent tripping time and the protection relay transmission delay, then the causal logic of the sequence is determined to be matched.

[0106] For any action event, if no prior critical electrical quantity abnormality that meets the set value conditions can be found as the cause, or if the timing interval does not conform to the inherent characteristics of the equipment, then the action is marked as a timing logic mismatch event.

[0107] This invention establishes a dynamic operational baseline to identify precursory states of events during system operation. This baseline triggers the instantaneous locking of historical cached data and the activation of enhanced full-process recording to construct a comprehensive data package for source tracing analysis. This solves the problem in existing technologies that rely on snapshots of fault states, making it impossible to effectively analyze the dynamic evolution and root causes of process-oriented and complex faults. It also addresses the issue that existing methods based on snapshots of fault states fail to capture process precursors and the complete evolutionary chain, leading to misjudgments of the root causes of progressive and complex faults and limited operational guidance value. This invention represents a fundamental shift from passive post-event classification to proactive full-process source tracing, enabling precise location of the root cause and dynamic evolution path of power outages, providing direct decision-making support for preventative maintenance and system optimization.

[0108] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for determining the type of power outage event at a substation, characterized in that, Includes the following steps: Step S1: Continuously collect characteristic data of key electrical quantities of the power system of the station, and store the characteristic data sequence in the data buffer area; Step S2: Based on the feature data sequence in the data buffer, a dynamic operating baseline for key electrical quantities to be automatically adjusted over time and under operating conditions is established using a sliding time window algorithm; Step S3: Collect the characteristic data of real-time key electrical quantities of the station's power system, and compare them with the dynamic operating baseline based on the preset health deterioration trend combination logic rules to determine whether the station's power system has entered the pre-event state. Step S4: If it is determined that the event is in the pre-event state, lock the data cache area and save the historical cache data from the data cache area back a preset time from the current moment. Step S5: Continuously capture subsequent evolution data until the station power system status is stable, and output the historical cached data and subsequent evolution data as a full-process panoramic data package; Step S6: Call the full-process panoramic data package, and through the complete evolution process of the key electrical quantities of the power system of the replay station, locate the first abnormal point where the anomaly first appeared and the abnormal evolution path, and determine the type and root cause of the power failure event. The specific steps for locating the first anomaly point and the anomaly evolution path in step S6 include: Parallel traversal scanning is performed on the curves of the characteristic data of all key electrical quantities in the full-process panoramic data package as a function of time. During the traversal scanning process, the characteristic data of each key electrical quantity is compared with its corresponding dynamic operating baseline point by point to identify the direction and degree of deviation of each key electrical quantity from the baseline, and all deviation times are recorded. Compare the earliest deviation times of all critical electrical quantities, determine the deviation point with the earliest global timestamp as the first anomaly point, and record the specific electrical quantity type and measurement point location corresponding to the first anomaly point; Starting from the first anomaly point, trace forward along the time axis to determine the temporal sequence and logical relationship of subsequent abnormal changes in other key electrical quantities or switch states; Based on the electrical topology of the station power system, the propagation direction of the critical electrical quantities that are abnormal in space is analyzed, and the timing of the switch change and protection action information is verified to match the causal logic of the abnormal changes in the critical electrical quantities. By combining the chronological relationships in the time series with the transmission paths in the topological space, the complete abnormal evolution path from the first anomaly point to the final power loss is reconstructed.

2. The method for determining the type of power outage event at a station according to claim 1, characterized in that, The specific steps of step S1 include: Simultaneously collect characteristic data of key electrical quantities, including three-phase voltage, three-phase current, zero-sequence voltage, harmonic content, and power, from each section of the busbar and incoming circuit of the power supply system at the station. For the characteristic data of the collected key electrical quantities, calculate their effective values ​​within a fixed short period to form a characteristic data sequence; Each critical electrical quantity is configured with an independent first-in-first-out data buffer, and the characteristic data sequence is stored in the corresponding data buffer.

3. The method for determining the type of power outage event at a station according to claim 2, characterized in that, The specific steps for calculating the effective value of the collected key electrical quantities within a fixed short period are as follows: Within a fixed short period, N consecutive sampling points are taken. The characteristic data of the key electrical quantities within each of the N sampling points are squared, summed, divided by the number of sampling points N, and then the square root is taken to obtain the effective value R within the fixed short period. The calculation formula is as follows: ； in, , The feature data represents the first to Nth sampling points within the fixed short period, where N represents the total number of sampling points within the fixed short period.

4. The method for determining the type of power outage event at a station according to claim 1, characterized in that, The specific steps of step S2 include: The first preset duration is used as the short-term calculation window, and the second preset duration is used as the long-term learning window; The moving average and standard deviation of the feature data sequence in the data buffer are calculated in real time, and short-term and long-term baselines are generated respectively.

5. The method for determining the type of power outage event at a station according to claim 4, characterized in that, The specific steps of step S3 include: The characteristic data of key electrical quantities collected in real time are compared item by item with the dynamic operating baseline of the corresponding circuit, and a joint judgment is made according to the pre-set health deterioration trend combination logic rules, wherein the health deterioration trend combination logic rules include: When the moving average of the effective value of a certain bus voltage shows a unidirectional continuous downward trend over multiple consecutive sampling periods, and the cumulative decline exceeds the preset proportional threshold of its short-term baseline average value, the trend deviation criterion is triggered. When the instantaneous fluctuation amplitude of a feeder circuit current exceeds a preset multiple of its short-term baseline standard deviation multiple times consecutively, the fluctuation anomaly criterion is triggered. When the amplitude of the zero-sequence voltage at the neutral point of the system continues to rise and exceeds a preset multiple of its long-term baseline typical value, the parameter imbalance criterion is triggered. The health status of the power system is managed by a logic state machine. The state machine includes a normal state, a precursor state, and a fault state. When any combination of one or more logic rules is satisfied, the state machine switches from the normal state to the precursor state.

6. The method for determining the type of power outage event at a station according to claim 1, characterized in that, The specific steps of step S4 include: When the power system of the judgment station enters the pre-event state, a lock command is sent to the data buffer, and all historical buffer data that has been backed up for a preset time from the current moment is transferred to a specific accident data file.

7. The method for determining the type of power outage event at a station according to claim 5, characterized in that, The specific steps of step S5 include: Enabling enhanced recording mode increases the sampling frequency to an integer multiple of the normal frequency and expands the recording range to all logically related circuits and equipment in the station's power system; The enhanced recording mode continues to run until the state machine confirms that the power system of the station has entered a new stable state or the state machine has clearly jumped to the fault state. The stable state is defined as the critical electrical quantities continuously recovering to the long-term baseline range and all switch states remaining unchanged for a preset period of time. Historical cached data, evolution data captured in enhanced recording mode, and switch change and protection action sequence records are aligned, merged, and encapsulated according to a unified time scale to generate a full-process panoramic data package.

8. The method for determining the type of power outage event at a station according to claim 1, characterized in that, The specific steps for determining the time sequence and logical association of subsequent abnormal changes in other key electrical quantities or switch states are as follows: After identifying the first anomaly, the analysis begins with the timestamp of the first anomaly and proceeds in the positive direction along the time axis. The system retrieves all other key electrical quantity characteristic data that broke through their respective dynamic operating baselines after the first anomaly point in the full-process panoramic data package, as well as all circuit breaker opening and closing switch change events and all protection device output action events, and outputs them as abnormal events. Each retrieved abnormal event is strictly sorted according to its absolute timestamp to generate a complete event sequence arranged chronologically. Based on the main electrical wiring diagram of the station's power system, establish a database of topological connections between all key electrical measurement points and switchgear. Based on the aforementioned topology connection database, determine whether each newly emerging critical electrical quantity anomaly event is located downstream of the previously abnormal equipment in the electrical path, or whether it is physically related to the previous anomaly through electromagnetic coupling or control loop; Based on the principle of power system fault propagation, this study verifies whether the complete event sequence conforms to the basic logic of fault source propagation along electrical path and triggering the action of protection devices at each level in sequence. A logic diagram of abnormal diffusion and chain reaction with time as the vertical axis and electrical topology as the horizontal axis is constructed.

9. The method for determining the type of power outage event at a station according to claim 1, characterized in that, The specific steps for verifying whether the timing of the switch change and protection action information matches the causal logic of the abnormal changes in key electrical quantities are as follows: Extract the key electrical quantity abnormal change sequence and the switch and protection action sequence from the full-process panoramic data package. The key electrical quantity abnormal change sequence includes the start and end times of abnormal voltage drop, abnormal current increase, and abnormal zero-sequence voltage rise. The switch and protection action sequence includes the absolute time of each circuit breaker's opening and closing action, as well as the absolute time of each protection device from startup to output action. Perform time alignment and logical comparison of the abnormal change sequence of key electrical quantities and the action sequence of switches and protection devices; Retrieve the preset protection setting list and protection action logic, and verify each protection action event and each circuit breaker non-manual operation tripping event; For any action event, if no prior critical electrical quantity abnormality that meets the set value conditions can be found as the cause, or if the timing interval does not conform to the inherent characteristics of the equipment, then the action is marked as a timing logic mismatch event.

Images