A method for monitoring the operating status of circuit breakers

Abstract

This invention relates to the field of circuit breaker monitoring and provides a method for monitoring the operating status of circuit breakers, comprising: collecting circuit operating status parameters of the corresponding circuit of the target circuit breaker and simultaneously collecting current data of the control circuit at the same station; constructing a control benchmark at the same station using the control circuit; calculating the current change intensity of the target circuit and the control circuit respectively, calculating relative anomaly indicators, segmenting the operating data into event segments according to event triggering conditions, extracting event feature information, and generating event fingerprint vectors; constructing a normal event library, calculating the maximum similarity between the new event fingerprint and the normal event library, and calculating the risk value by combining the maximum similarity, relative anomaly indicators, and evidence of abnormal temperature rise; determining the anomaly category according to combination rules, and outputting the equipment location information, event time, and location information of evidence indicators; reporting the event fingerprint and risk value summary when the network is restricted, and reporting the event window data and summary when the network is good.

Description

Technical Field

[0001] This invention relates to the field of circuit breaker monitoring, specifically a method for monitoring the operating status of circuit breakers. Background Technology

[0002] Circuit breakers are a critical electrical safety device. Their core function is to automatically cut off the current when abnormal conditions such as overload, short circuit, or undervoltage occur in the circuit, so as to protect electrical equipment and lines from damage. In the power distribution room of the park, common monitoring only covers current, voltage, and temperature, without simultaneously collecting process quantities such as opening and closing coil current, mechanism stroke, and vibration. This makes it impossible to reconstruct the entire evidence chain of a single opening and closing action, and early deterioration cannot be identified.

[0003] Chinese patent application number CN202411230796.X discloses a method and system for monitoring the operating status of circuit breakers. The method includes: acquiring a training set of circuit breaker operating status data; obtaining the probability that any two status data points constitute an open state interval; filtering open state intervals based on the probability of constituting an open state interval; obtaining the probability that each status data point belongs to normal data; clustering all status data points to obtain several feature groups; obtaining the operating status of each feature group based on the probability of belonging to normal data; constructing an optimized decision tree based on the operating status of the feature groups; and classifying the operating status of the circuit breaker using the optimized decision tree, thereby improving the accuracy of the monitoring results for the circuit breaker operating status.

[0004] In the scenario of monitoring the power distribution room in the park, due to the large current fluctuations, frequent operation switching, scattered sites, and few maintenance personnel, as well as the fact that power conversion equipment and charging equipment are prone to introducing irregular fluctuations, circuit breakers may frequently operate, overheat, and accelerate the aging of contact points. Existing methods for monitoring the operating status of circuit breakers often only look at whether they have tripped, making it difficult to form actionable early intervention measures.

[0005] To address the aforementioned issues, this invention proposes a method for monitoring the operating status of circuit breakers. This method involves collecting and aligning data to construct a local benchmark, calculating relative anomaly indicators based on the degree of deviation, generating fingerprints for segmented events where the relative anomaly indicators exceed a threshold, establishing a normal event database, calculating similarity and risk values, determining the anomaly category according to the risk value and discrimination rules, and locating and reporting the anomaly. Summary of the Invention

[0006] In view of the existing problems mentioned above, a method for monitoring the operating status of circuit breakers is proposed.

[0007] The technical solution adopted by this invention to solve the above-mentioned technical problems is: a method for monitoring the operating status of a circuit breaker, comprising:

[0008] Collect the circuit operation status parameters of the circuit corresponding to the target circuit breaker, and simultaneously collect the current data of the reference circuit at the same station; write the data into a unified timestamp and perform time alignment to form unified time axis data, and construct a reference benchmark at the same station based on the reference circuit.

[0009] The current change intensity of the target circuit and the control circuit are calculated separately, and the relative anomaly index is calculated. When the relative anomaly index meets the preset relative anomaly condition, the running data is divided into event segments according to the event triggering condition, and event feature information is extracted from each event segment to generate an event fingerprint vector.

[0010] During the normal operation confirmation phase of the target circuit breaker, event fingerprints are collected to construct a normal event database. The maximum similarity between the new event fingerprints and the normal event database is calculated, and the risk value is calculated by combining the maximum similarity, relative abnormality indicators, and evidence of abnormal temperature rise.

[0011] Based on the combination rules of risk value and post-event stabilization time, the anomaly category is determined, and the device location information, event time, and location information of evidence indicators are output. When the network is restricted, the event fingerprint and risk value summary are reported, and when the network is good, the event window data and summary are reported.

[0012] As a preferred embodiment, the specific steps for collecting the circuit operating status parameters of the circuit corresponding to the target circuit breaker are as follows:

[0013] On the edge computing node, the system topology and asset ledger are read to obtain the circuit breaker list. The circuit breakers to be included in the monitoring are selected from the list, and the selected circuit breakers are bound to the collection points of their respective circuits and sent to the edge computing node to determine the target circuit breaker. The circuit operation status parameters of the target circuit are collected at a preset sampling period. The edge computing node synchronously collects current data from multiple reference circuits as a reference quantity. The target circuit data and the reference circuit data are written to a unified timestamp using the same clock source, and resampling and alignment are performed according to the unified sampling period. When there is sampling jitter, interpolation is performed to fill in missing points and deduplication and merging are performed on duplicate points based on the timestamp to form a unified time axis data sequence covering the target circuit and the reference circuit.

[0014] As a preferred embodiment, the specific steps for constructing a co-station reference standard using a reference loop are as follows:

[0015] Edge computing nodes statistically aggregate the current sequences of the control loop on a unified time axis using a sliding time window to generate a benchmark quantity characterizing the intensity of load fluctuations in the same station group. The formula is as follows:

[0016] ,

[0017] in This represents the co-station reference quantity within the sliding time window T, where K represents the number of reference loops, k represents the index of the reference loop, and t represents the current time. represents the sequence of current sampled values ​​of the k-th control loop on a uniform time axis, where var represents the variance operator;

[0018] The intensity of change of the reference quantity and the target circuit within the same time window is stored in the same window to serve as a comparison benchmark for subsequent relative anomaly identification and event discrimination.

[0019] As a preferred embodiment, the specific steps for calculating the current change intensity of the target circuit and the control circuit, and calculating the relative anomaly index, are as follows:

[0020] Based on the target loop current sequence and the control loop current sequence at the same station, the current change intensity is calculated within the same sliding time window. A control benchmark is constructed using the change intensity statistics of the control loop. The change intensity of the target loop is then compared with the control benchmark to obtain a relative anomaly index. The formula is as follows:

[0021] ,

[0022] in Indicates a relative abnormality index. Represents the target loop current sequence. It represents a fixed positive number, used to avoid the denominator being zero;

[0023] As a preferred embodiment, the specific steps of segmenting the runtime data into event segments according to event triggering conditions, extracting event feature information from each event segment, and generating an event fingerprint vector are as follows:

[0024] When the relative anomaly index reaches the preset relative anomaly condition, it is determined that there is evidence of abnormal fluctuation in the target loop that is independent of the common mode fluctuation of the same station, and the event-based processing flow is triggered accordingly. The edge computing node segments the running data according to the preset event triggering conditions.

[0025] When the relative anomaly index is not less than 1.80 and meets the condition for three consecutive adjacent time windows, it is determined that a relative anomaly has been established; when the relative anomaly index is between 1.50 and 1.80 and meets the condition for five consecutive adjacent time windows, it is determined that a relative anomaly has been established; when the relative anomaly index is less than 1.50, it is determined that a relative anomaly has not been established and event segmentation is not triggered.

[0026] The current and temperature sequences within the event segment are denoised and outlier removed. A baseline value is calculated within a preset baseline interval before the triggering time. This baseline value characterizes the steady-state level before the event. Using the baseline value as a reference, the start and end positions of the current change process within the event segment are determined. The current change amplitude is determined by the difference between the average current in the steady-state interval after the event and the average current in the baseline interval. The current change process is positioned according to a fixed threshold ratio, with a high threshold of 90% and a low threshold of 10%. The time span for the current to rise from the low threshold to the high threshold is determined as the rise duration, and the time span for the current to fall back from the high threshold to the low threshold is determined as the fall duration. The circuit breaker state sequence within the event segment is read. If a state reversal is detected within the event segment, the reversal direction is encoded as an action flag and written into the event feature information. If no reversal occurs, the action flag is set to no action. When temperature acquisition is present, the temperature rise amplitude is determined by the difference between the average temperature at the end of the event segment and the average temperature at the beginning. The temperature rise amplitude is written into the event feature information as a thermal feature to form the event feature information.

[0027] Edge computing nodes extract a fixed-length data window before and after the triggering time as an event segment, and perform feature extraction on the event segment to form event feature information. The event feature information is encoded into an event fingerprint vector according to a fixed dimension and a fixed order. The event fingerprint vector is bound to the triggering type, triggering time, circuit to which it belongs, and device identification information for storage and reporting.

[0028] As a preferred implementation, the specific steps for calculating the maximum similarity between the new event fingerprint and the normal event database are as follows:

[0029] By continuously collecting loop operation status data and extracting event feature information, a normal event database is constructed. The event fingerprint vectors in the normal event database represent typical event characteristics of the target loop under normal operation and serve as a benchmark for anomaly detection. Whenever a new event occurs, the edge computing node first extracts event feature information from the new event and compares it with the event fingerprint vectors in the normal event database to calculate the maximum similarity. The similarity calculation is based on the relative distance metric between event fingerprints. By calculating the similarity between the new event fingerprint and all fingerprints in the normal event database, the normal event fingerprint that is closest to the new event fingerprint is selected as a reference. The formula is as follows:

[0030] ,

[0031] in The fingerprint vector representing a new event. This represents a fingerprint vector in the normal event database. This indicates the normal event library. Let represent the j-th event fingerprint in the normal event database, and n represent the dimension of the event fingerprint. This represents the i-th feature value in the fingerprint of a new event. Let represent the i-th feature value in the fingerprint of the j-th normal event, and max represent the maximum value function;

[0032] As a preferred implementation method, the specific steps for calculating the risk value by combining maximum similarity, relative anomaly index, and evidence of abnormal temperature rise are as follows:

[0033] By combining the maximum similarity and relative anomaly indicators, the evidence of abnormal temperature rise in an event is analyzed. The evidence of abnormal temperature rise is obtained by calculating the temperature change within the event segment and the difference between the temperature at the end of the event and the temperature at the beginning of the event. The risk value is calculated by combining the maximum similarity, relative anomaly indicators and evidence of abnormal temperature rise according to preset weights. This risk value is used to characterize whether the new event exceeds the normal operating range and the severity of the anomaly.

[0034] As a preferred implementation, the specific steps for determining the anomaly category based on the combination rule of risk value and post-event stabilization duration are as follows:

[0035] The event category is determined based on the risk value. If the risk value exceeds the preset threshold, the device location information, event time, and location information of evidence indicators will be output.

[0036] When the risk value is below 0.5, it indicates that the event is within the normal operating range and no intervention is required. Such events will not trigger alarms and require no further maintenance. When the risk value is between 0.5 and 0.75, it indicates that the event has potential anomalies. Such events will trigger alarms, and regular checks are recommended, but immediate remedial measures are not required. When the risk value is greater than 0.75, it indicates that the anomaly of the event is severe. Such events will trigger alarms and require intervention measures.

[0037] As a preferred implementation, the specific steps for reporting event fingerprints and risk value summaries when the network is limited, and reporting event window data and summaries when the network is good, are as follows:

[0038] The communication link quality parameters, including packet loss rate and round-trip time, are obtained by the edge computing node. When the packet loss rate is not less than the packet loss threshold and the round-trip time is not less than the delay threshold, the network is determined to be restricted and the event fingerprint and risk value summary are reported. When the packet loss rate is less than the packet loss threshold and the round-trip time is less than the delay threshold, the network is determined to be good and the event window data and the summary are reported.

[0039] Beneficial effects

[0040] Compared with the prior art, the present invention has the following advantages:

[0041] 1. By constructing a benchmark for comparison at the same station and calculating relative anomaly indicators, and by dividing the operation process into event segments and generating event fingerprints, the system can identify early degradation that does not exceed the limit but deviates in form by combining the maximum similarity of the normal event database, thereby improving the ability to detect latent faults.

[0042] 2. It not only outputs location information including device location information, event time, and evidence indicators, but also uses event fingerprints and key features as evidence chains to facilitate operation and maintenance review and repair decisions, thereby improving the accuracy and verifiability of location. Attached Figure Description

[0043] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0044] Figure 1 This is a flowchart illustrating the present invention;

[0045] Figure 2 This is a comparison diagram of the effects of the present invention and the prior art, where gray bars represent the prior art and black bars represent the present invention. Detailed Implementation

[0046] To make the technical means, creative features, objectives, and effects of this invention easier to understand, the invention is further described below with reference to specific embodiments. However, the following embodiments are merely preferred embodiments of this invention and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments described herein without creative effort are all within the protection scope of this invention.

[0047] Example 1:

[0048] To achieve the above objectives, please refer to Figures 1 to 2 This invention provides a method for monitoring the operating status of a circuit breaker, the method comprising the following steps:

[0049] Step S1: Collect and align data to construct a baseline for comparison at the same station;

[0050] Step S2: Calculate relative anomaly indicators and segment events to generate fingerprints;

[0051] Step S3: Establish a normal event database and calculate similarity and risk value;

[0052] Step S4: Determine the anomaly type, locate the output, and report it.

[0053] This method is implemented in the order of S1–S4, and its overall process is as follows:

[0054] Collect the circuit operation status parameters of the circuit corresponding to the target circuit breaker, and simultaneously collect the current data of the reference circuit at the same station; write the data into a unified timestamp and perform time alignment to form unified time axis data, and construct a reference benchmark at the same station based on the reference circuit.

[0055] The current change intensity of the target circuit and the control circuit are calculated separately, and the relative anomaly index is calculated. When the relative anomaly index meets the preset relative anomaly condition, the running data is divided into event segments according to the event triggering condition, and event feature information is extracted from each event segment to generate an event fingerprint vector.

[0056] During the normal operation confirmation phase of the target circuit breaker, event fingerprints are collected to construct a normal event database. The maximum similarity between the new event fingerprints and the normal event database is calculated, and the risk value is calculated by combining the maximum similarity, relative abnormality indicators, and evidence of abnormal temperature rise.

[0057] Based on the combination rules of risk value and post-event stabilization time, the anomaly category is determined, and the device location information, event time, and location information of evidence indicators are output. When the network is restricted, the event fingerprint and risk value summary are reported, and when the network is good, the event window data and summary are reported.

[0058] The specific steps for collecting and aligning data to construct a peer-to-peer benchmark are as follows:

[0059] At the edge computing node in the power distribution room of the park, such as an industrial IoT gateway, the system topology and asset ledger of the power distribution room are read to obtain a list of circuit breakers. Maintenance personnel then select the circuit breakers to be included in the monitoring from the list, bind the selected circuit breakers to the collection points of their respective circuits, and send the data to the edge computing node to determine the target circuit breaker. The node then electrically connects to the metering and acquisition devices, switch quantity acquisition terminals, and temperature measurement acquisition terminals of the target circuit breaker's circuit. The circuit operating status parameters of the target circuit are collected at a preset sampling period. For example, a layered sampling period is used, setting the electrical parameter sampling period to 1 second and the circuit breaker open / closed state sampling period to 1 second. The sampling period is 100ms, and the action statistics sampling period is 60s. The circuit operation status parameters include electrical parameter information, switch status information, action statistics information, and thermal status information. The edge computing node synchronously collects current data from multiple reference circuits in the same power supply zone as the target circuit as the reference quantity. The target circuit data and the reference circuit data are written to a unified timestamp using the same clock source, and resampling and alignment are performed according to the unified sampling period. When there is sampling jitter, the missing points are interpolated and the duplicate points are deduplicated and merged based on the timestamp to form a unified time axis data sequence covering the target circuit and the reference circuit.

[0060] Furthermore, the edge computing nodes statistically aggregate the current sequences of the control loops on a unified time axis using a sliding time window to generate a benchmark quantity characterizing the intensity of load fluctuations in the same station group. The formula is as follows:

[0061] ,

[0062] in This represents the co-station reference quantity within the sliding time window T, where K represents the number of reference loops, k represents the index of the reference loop, and t represents the current time. This represents the sequence of current sampled values ​​of the k-th control loop on a unified time axis. var represents the variance operator. Specifically, the variance operator refers to taking all current sampled values ​​within a time window of a preset length as a set, calculating the degree of dispersion of the set relative to the mean, and obtaining the change intensity index of the time window.

[0063] Specifically, the reference quantity characterizing the load fluctuation intensity of the same station group is taken as an example of the same 10kV distribution cabinet in a power distribution room of a certain park. The outgoing circuit where the target circuit breaker is located is selected as the target circuit, and the other 6 outgoing circuits in the same cabinet are selected as the reference circuits. The edge computing node collects the current with a sampling period of 1 second and aligns the timestamp. The variance of the current of each circuit is calculated within a 60-second sliding time window. Assuming that the variances of the 6 reference circuits are 0.80, 0.95, 1.05, 0.90, 1.10 and 0.88 respectively, the reference quantity of the same station is the average value of the above variances, 0.95.

[0064] The intensity of change of the reference quantity and the target circuit within the same time window is stored in the same window to serve as a comparison benchmark for subsequent relative anomaly identification and event discrimination.

[0065] By combining the collection of all state parameters of the target loop with the construction of a reference benchmark through time window statistics, this approach differs from existing technologies that only apply fixed threshold alarms to a single loop and a single variable. It reduces false alarms caused by fluctuations in operating conditions and provides a repeatable and traceable alignment data foundation and reference benchmark for subsequent event segmentation, fingerprint generation, and risk assessment.

[0066] The specific steps for calculating relative anomaly indicators and segmenting events to generate fingerprints are as follows:

[0067] Based on the target loop current sequence and the control loop current sequence at the same station, the current change intensity is calculated within the same sliding time window. The current change intensity is the degree of dispersion relative to the mean calculated by the variance operator. A control benchmark is constructed using the change intensity statistics of the control loop. The change intensity of the target loop is then compared with the control benchmark to obtain a relative anomaly index. The formula is as follows:

[0068] ,

[0069] in Indicates a relative abnormality index. Represents the target loop current sequence. This indicates a fixed positive number, used to avoid the denominator being zero; for example, it can be 0.0001.

[0070] When the relative anomaly index reaches the preset relative anomaly condition, it is determined that there is evidence of abnormal fluctuation in the target circuit that is independent of the common mode fluctuation of the same station, and the event-based processing flow is triggered accordingly. The edge computing node segments the running data according to the preset event triggering conditions, including circuit breaker state change triggering, electrical parameter mutation triggering, and thermal state trend anomaly triggering.

[0071] Specifically, the preset relative anomaly conditions are determined as follows: when the relative anomaly index is not less than 1.80 and is satisfied for three consecutive adjacent time windows, a relative anomaly is determined to be established; when the relative anomaly index is between 1.50 and 1.80 and is satisfied for five consecutive adjacent time windows, a relative anomaly is determined to be established; when the relative anomaly index is less than 1.50, a relative anomaly is determined to be not established and event segmentation is not triggered.

[0072] Edge computing nodes extract fixed-length data windows before and after the trigger moment as event segments, and perform feature extraction on these event segments to form event feature information. This event feature information includes the current change amplitude, the duration of the current change process, the circuit breaker operation flag, the temperature change within the event window, and the stabilization time after the event. The event feature information is encoded into an event fingerprint vector according to a fixed dimension and a fixed order, for example, defined as 6 dimensions, arranged in the following order: the first position is the current change amplitude, the second position is the duration of the current rise, the third position is the duration of the current fall, the fourth position is the circuit breaker operation flag, the fifth position is the temperature rise within the event window, and the sixth position is the stabilization time after the event. The event fingerprint vector is bound to the trigger type, trigger time, circuit, and equipment identification information for storage and reporting.

[0073] Specifically, the process involves extracting features from event segments to form event feature information, denoising and removing outliers from the current and temperature sequences within the event segments, and calculating a benchmark value within a preset benchmark interval before the triggering time. This benchmark value characterizes the steady-state level before the event occurs. Using this benchmark value as a reference, the start and end positions of the current change process within the event segment are determined. The amplitude of the current change is determined by the difference between the average current in the steady-state interval after the event and the average current in the benchmark interval. The current change process is then positioned according to a fixed percentage threshold. Specifically, the low threshold is 10%, and the high threshold is 90%. The time span for the current to reach the high threshold is determined accordingly. The rising duration is then determined by the time span during which the current falls from a high threshold to a low threshold. The circuit breaker state sequence within the event segment is read. If a state reversal is detected within the event segment, the reversal direction is encoded as an action flag and written into the event feature information. If no reversal occurs, the action flag is set to inactive. When temperature acquisition is present, the temperature rise amplitude is determined by the difference between the average temperature at the end of the event segment and the average temperature at the beginning. The temperature rise amplitude is written into the event feature information as a thermal feature, forming event feature information that includes the current change amplitude, circuit change duration, action flag, temperature rise amplitude, and the stabilization time after the event. This information is used for subsequent event fingerprint encoding and similarity comparison.

[0074] Through the above-mentioned processing link of relative anomaly screening using the same-station comparison benchmark set, common-mode fluctuations are suppressed by using the same-station comparison benchmark. Then, event-level evidence solidification is performed on the target loop that meets the relative anomaly conditions. This is different from the existing technology that only performs fixed threshold judgment on the instantaneous quantity of a single loop. This reduces false alarms caused by operating condition fluctuations and makes the subsequent anomaly category judgment have verifiable event evidence and comparable fingerprint features.

[0075] The specific steps for establishing a normal event database and calculating similarity and risk values ​​are as follows:

[0076] By continuously collecting loop operation status data and extracting event feature information, a normal event database is constructed. The event fingerprint vectors in the normal event database represent typical event characteristics of the target loop under normal operation and serve as a benchmark for anomaly detection. Whenever a new event occurs, the edge computing node first extracts event feature information from the new event and compares it with the event fingerprint vectors in the normal event database to calculate the maximum similarity. The similarity calculation is based on the relative distance metric between event fingerprints. Specifically, it calculates the similarity between the new event fingerprint and all fingerprints in the normal event database, and selects the normal event fingerprint that is closest to the new event fingerprint as a reference. The formula is as follows:

[0077] ,

[0078] in The fingerprint vector representing a new event. This represents a fingerprint vector in the normal event database. This indicates the normal event library. Let represent the j-th event fingerprint in the normal event database, and n represent the dimension of the event fingerprint. This represents the i-th feature value in the fingerprint of a new event. Let represent the i-th feature value in the fingerprint of the j-th normal event, and max represent the maximum value function;

[0079] By combining maximum similarity and relative anomaly indicators, edge computing nodes also analyze evidence of abnormal temperature rise in events. Specifically, evidence of abnormal temperature rise is obtained by calculating the temperature change within an event segment, i.e., the difference between the temperature at the end of the event and the temperature at the beginning of the event. Based on preset weights, the risk value is calculated by combining maximum similarity, relative anomaly indicators, and evidence of abnormal temperature rise. This risk value is used to characterize whether a new event exceeds the normal operating range and the severity of the anomaly. The preset weights are determined by analyzing the impact of different fault types on the equipment, such as electrical faults, mechanical faults, and faults caused by temperature rise, to determine which factors have a greater impact on fault risk.

[0080] The dynamic anomaly detection mechanism based on a normal event database and similarity calculation improves the early identification capability of new event anomalies and effectively distinguishes between minor anomalies caused by factors such as load fluctuations and equipment aging and real faults, avoiding the waste of resources due to false alarms in traditional methods. Unlike existing technologies, this invention utilizes a combination of event fingerprint similarity, relative anomaly indicators, and evidence of abnormal temperature rise to comprehensively assess the degree of anomaly of events, thereby providing a more accurate and actionable risk assessment.

[0081] The specific steps for determining the anomaly category, locating the output, and reporting are as follows:

[0082] The event category is determined based on the risk value. If the risk value exceeds the preset threshold, such as higher than 0.75, the location information of the equipment, the event time and the location information of the evidence indicators will be output, including the circuit number, circuit breaker number, distribution cabinet number, the time of the event and the corresponding abnormal evidence.

[0083] Specifically, the event category determination based on risk value is as follows: when the risk value is below 0.5, it indicates that the event is within the normal operating range and no intervention is required. This type of event does not trigger an alarm and requires no further maintenance. When the risk value is between 0.5 and 0.75, it indicates that the event has potential anomalies but does not significantly affect the normal operation of the equipment. This type of event triggers an alarm, and regular inspection is recommended, but immediate remedial measures are not required. When the risk value is greater than 0.75, it indicates that the anomaly of the event is severe and may affect the equipment. This type of event triggers an alarm and requires immediate intervention measures, such as arranging maintenance or shutdown.

[0084] The edge computing node obtains communication link quality parameters, including packet loss rate and round-trip time (RTT). When the packet loss rate is not less than a first threshold and the RTT is not less than a second threshold, the network is deemed restricted and an event fingerprint and risk value summary are reported. When the packet loss rate is less than the first threshold and the RTT is less than the second threshold, the network is deemed good and event window data and the summary are reported. Specifically, the packet loss threshold and RTT threshold are selected by the operations and maintenance personnel based on historical data analysis. For example, the packet loss threshold is 1% and the RTT threshold is 200ms.

[0085] By classifying events into different categories based on risk values, loop anomaly events can be dynamically and accurately assessed and judged, improving the accuracy and response speed of anomaly detection, while reducing maintenance costs caused by false alarms and missed alarms.

[0086] like Figure 2 A comparative chart of the effects of a circuit breaker operation status monitoring method is presented, comparing key performance indicators of the present invention and existing technologies in circuit breaker operation status monitoring in the form of a bar chart. Gray bars represent existing technologies, and black bars represent the present invention. As can be seen from the chart, the present invention has lower false alarm rate and false negative rate than existing technologies, indicating that it is more effective in suppressing common-mode fluctuations at the same station. It has higher fault location accuracy and early warning lead time than existing technologies, indicating that the present invention achieves earlier and more accurate anomaly identification by comparing the similarity between event fingerprints and the normal event database. It has lower data reporting volume than existing technologies, indicating that the network adaptive summary reporting mechanism of the present invention reduces communication and storage overhead.

[0087] Example 2:

[0088] Based on the above embodiment 1, a method for monitoring the operating status of circuit breakers in a power distribution room monitoring scenario in a park is as follows:

[0089] Step 1: In the power distribution room of the park, the edge computing node is set as a power distribution automation data acquisition terminal. The edge computing node establishes a communication connection with the integrated measurement and control device, protection measurement and control device, and energy meter in the power distribution cabinet via fieldbus. It automatically reads the system topology file and equipment identification mapping table, automatically selects the monitoring object and determines the target circuit breaker according to the preset rule of prioritizing the incoming circuit breaker, and automatically binds the target circuit breaker to the electrical parameter acquisition points, switch status acquisition points, action count acquisition points, and temperature measurement acquisition points of its circuit. A unified sampling scheduling strategy is used to collect data for the target circuit, where the electrical parameters are collected according to... The system collects data in 1-second cycles. Switch status is reported based on event triggers and verified by polling every 100 milliseconds. The number of actions is read in 60-second cycles and updated incrementally when the status changes. Edge computing nodes select three reference circuits within the same power supply zone to synchronously collect current as a reference quantity. All sampled data is written to a unified timestamp based on the local clock, and a unified time axis data is formed through time alignment. The fluctuation intensity of the reference circuit current is calculated by sliding time window on the unified time axis, and the average is taken to form a reference quantity for the same station. This reference quantity is then associated with and stored with the fluctuation intensity of the target circuit within the same window for subsequent relative anomaly identification and event discrimination.

[0090] Step 2: Differential energy characterization is used to represent the current change intensity of the target circuit current sequence on a unified time axis. Specifically, the absolute value of the current difference between adjacent sampling points is accumulated within a sliding time window to obtain the target circuit change intensity. The edge computing node compares the target circuit change intensity with the reference standard to obtain a relative anomaly index. The relative anomaly condition is set to exceed a threshold and continuously meet a preset window number to suppress false triggering of instantaneous disturbances. When the relative anomaly condition is met, the edge computing node divides the running data into event segments according to the event triggering conditions. The event triggering conditions include circuit breaker state changes, sudden changes in electrical parameters, and abnormal temperature rise trends. A preset length window is extracted before and after the triggering time. Feature extraction is performed on the event segments to extract the current change amplitude, duration of the change process, action flag, temperature rise amplitude, and stabilization time after the event. These are encoded into a fixed-dimensional event fingerprint vector in a preset order. The event fingerprint is associated with the trigger type, event time, circuit to which it belongs, and equipment identification information, stored, and reported for subsequent similarity comparison and risk assessment.

[0091] Step 3: Taking the period of confirmed normal operation as the data collection phase, event fingerprints are aggregated according to event type to construct a normal event database. Continuous features in the event fingerprints are standardized, and discrete action markers are subject to matching constraints to form a feature index that can be used for retrieval. When a new event occurs, a nearest neighbor search is performed on the new event fingerprint in the normal event database. Candidate sets with consistent action markers are prioritized, and similarity is calculated based on weighted distance, with the maximum similarity selected as a reference. Edge computing nodes simultaneously calculate relative anomaly indicators, and the temperature rise anomaly evidence is formed by the temperature difference between the start and end of the event window. The degree of morphological deviation, relative anomaly, and temperature rise anomaly are fused according to preset weights to obtain a risk value, which is used to characterize the degree and severity of the new event deviating from normal operation.

[0092] Step 4: Map risk values ​​to event categories according to grading rules, and introduce continuity criteria to suppress short-term disturbances. Specifically, when a risk value is below the first threshold twice consecutively, it is judged as a normal event and only archived; when a risk value is between the first and second thresholds twice consecutively, it is judged as a warning event and inspection suggestions are generated; when a risk value reaches or exceeds the second threshold at any time, it is judged as an alarm event and location information is immediately output. The location information includes equipment location information, event time, and evidence indicators. The equipment location information consists of power distribution unit identifiers, circuit identifiers, and switch unit identifiers. The evidence indicators include at least maximum similarity, relative anomaly indicators, abnormal temperature rise evidence, and post-event stabilization time. The reporting content is adaptively selected based on link quality. When the packet loss rate exceeds a preset threshold, summary reporting is used. The summary includes the event fingerprint, risk value, event category, and location information. When the link quality meets the threshold, full reporting is used. Full reporting adds the original sequence data of the event window and feature extraction results to the summary for platform verification and retention.

[0093] The embodiments of the present invention described above are subject to modification and change of method by those skilled in the art without departing from the embodiments and broader aspects of the present invention. The appended claims are intended to include all such modifications and changes of method that do not depart from the present invention.

Claims

1. A method for monitoring the operating status of a circuit breaker, characterized in that, include: Collect the circuit operation status parameters of the circuit corresponding to the target circuit breaker, and simultaneously collect the current data of the reference circuit at the same station; The data is written with a unified timestamp and time-aligned to form a unified timeline data, and a station-wide reference benchmark is constructed using the reference loop. The current change intensity of the target circuit and the control circuit are calculated separately, and the relative anomaly index is calculated. When the relative anomaly index meets the preset relative anomaly condition, the running data is divided into event segments according to the event triggering condition, and event feature information is extracted from each event segment to generate an event fingerprint vector. During the normal operation confirmation phase of the target circuit breaker, event fingerprints are collected to construct a normal event database. The maximum similarity between the new event fingerprints and the normal event database is calculated, and the risk value is calculated by combining the maximum similarity, relative abnormality indicators, and evidence of abnormal temperature rise. Based on the combination rules of risk value and post-event stabilization time, the anomaly category is determined, and the device location information, event time, and location information of evidence indicators are output. When the network is restricted, the event fingerprint and risk value summary are reported, and when the network is good, the event window data and summary are reported.

2. The method for monitoring the operating status of a circuit breaker according to claim 1, characterized in that, The specific steps for collecting the circuit operating status parameters of the circuit corresponding to the target circuit breaker are as follows: On the edge computing node, the system topology and asset ledger are read to obtain the circuit breaker list. The circuit breakers to be included in the monitoring are selected from the list, and the selected circuit breakers are bound to the collection points of their respective circuits and sent to the edge computing node to determine the target circuit breaker. The circuit operation status parameters of the target circuit are collected at a preset sampling period. The edge computing node synchronously collects current data from multiple reference circuits as a reference quantity. The target circuit data and the reference circuit data are written to a unified timestamp using the same clock source, and resampling and alignment are performed according to the unified sampling period. When there is sampling jitter, interpolation is performed to fill in missing points and deduplication and merging are performed on duplicate points based on the timestamp to form a unified time axis data sequence covering the target circuit and the reference circuit.

3. The method for monitoring the operating status of a circuit breaker according to claim 1, characterized in that, The specific steps for constructing a co-station reference standard using a reference loop are as follows: Edge computing nodes statistically aggregate the current sequences of the control loop on a unified time axis using a sliding time window to generate a benchmark quantity characterizing the intensity of load fluctuations in the same station group. The formula is as follows: ， in This represents the co-station reference quantity within the sliding time window T, where K represents the number of reference loops, k represents the index of the reference loop, and t represents the current time. represents the sequence of current sampled values ​​of the k-th control loop on a uniform time axis, where var represents the variance operator; The intensity of change of the reference quantity and the target circuit within the same time window is stored in the same window to serve as a comparison benchmark for subsequent relative anomaly identification and event discrimination.

4. The method for monitoring the operating status of a circuit breaker according to claim 1, characterized in that, The specific steps for calculating the current change intensity of the target circuit and the control circuit respectively, and calculating the relative anomaly index are as follows: Based on the target loop current sequence and the control loop current sequence at the same station, the current change intensity is calculated within the same sliding time window. A control benchmark is constructed using the change intensity statistics of the control loop. The change intensity of the target loop is then compared with the control benchmark to obtain a relative anomaly index. The formula is as follows: ， in Indicates a relative abnormality index. Represents the target loop current sequence. It represents a fixed positive number, used to avoid the denominator being zero.

5. The method for monitoring the operating status of a circuit breaker according to claim 1, characterized in that, The specific steps for dividing the runtime data into event segments according to event triggering conditions, extracting event feature information from each event segment, and generating an event fingerprint vector are as follows: When the relative anomaly index reaches the preset relative anomaly condition, it is determined that there is evidence of abnormal fluctuation in the target loop that is independent of the common mode fluctuation of the same station, and the event-based processing flow is triggered accordingly. The edge computing node segments the running data according to the preset event triggering conditions. A relative anomaly is determined to be established when the relative anomaly index is not less than 1.80 and is satisfied for three consecutive adjacent time windows; a relative anomaly is determined to be established when the relative anomaly index is between 1.50 and 1.80 and is satisfied for five consecutive adjacent time windows; a relative anomaly is determined to be established when the relative anomaly index is less than 1.50, and no relative anomaly is determined to be established and event segmentation is triggered.

6. The method for monitoring the operating status of a circuit breaker according to claim 5, characterized in that, The specific steps of dividing the running data into event segments according to event triggering conditions, extracting event feature information from each event segment, and generating an event fingerprint vector further include: The current and temperature sequences within the event segment are denoised and outlier removed. A baseline value is calculated within a preset baseline interval before the triggering time. This baseline value characterizes the steady-state level before the event. Using the baseline value as a reference, the start and end positions of the current change process within the event segment are determined. The current change amplitude is determined by the difference between the average current in the steady-state interval after the event and the average current in the baseline interval. The current change process is positioned according to a fixed threshold ratio, with a high threshold of 90% and a low threshold of 10%. The time span for the current to rise from the low threshold to the high threshold is determined as the rise duration, and the time span for the current to fall back from the high threshold to the low threshold is determined as the fall duration. The circuit breaker state sequence within the event segment is read. If a state reversal is detected within the event segment, the reversal direction is encoded as an action flag and written into the event feature information. If no reversal occurs, the action flag is set to no action. When temperature acquisition is present, the temperature rise amplitude is determined by the difference between the average temperature at the end of the event segment and the average temperature at the beginning. The temperature rise amplitude is written into the event feature information as a thermal feature to form the event feature information. Edge computing nodes extract a fixed-length data window as an event segment before and after the triggering time, and perform feature extraction on the event segment to form event feature information. The event feature information is encoded into an event fingerprint vector according to a fixed dimension and a fixed order. The event fingerprint vector is bound to the triggering type, triggering time, belonging loop and device identification information for storage and reporting.

7. The method for monitoring the operating status of a circuit breaker according to claim 1, characterized in that, The specific steps for calculating the maximum similarity between the new event fingerprint and the normal event database are as follows: By continuously collecting loop operation status data and extracting event feature information, a normal event database is constructed. The event fingerprint vectors in the normal event database represent typical event characteristics of the target loop under normal operation and serve as a benchmark for anomaly detection. Whenever a new event occurs, the edge computing node first extracts event feature information from the new event and compares it with the event fingerprint vectors in the normal event database to calculate the maximum similarity. The similarity calculation is based on the relative distance metric between event fingerprints. By calculating the similarity between the new event fingerprint and all fingerprints in the normal event database, the normal event fingerprint that is closest to the new event fingerprint is selected as a reference. The formula is as follows: ， in The fingerprint vector representing a new event. This represents a fingerprint vector in the normal event database. This indicates the normal event library. Let represent the j-th event fingerprint in the normal event database, and n represent the dimension of the event fingerprint. This represents the i-th feature value in the fingerprint of a new event. Let represent the i-th feature value in the fingerprint of the j-th normal event, and max represent the maximum value function.

8. The method for monitoring the operating status of a circuit breaker according to claim 1, characterized in that, The specific steps for calculating the risk value by combining maximum similarity, relative anomaly indicators, and evidence of abnormal temperature rise are as follows: By combining the maximum similarity and relative anomaly indicators, the evidence of abnormal temperature rise in an event is analyzed. The evidence of abnormal temperature rise is obtained by calculating the temperature change within the event segment and the difference between the temperature at the end of the event and the temperature at the beginning of the event. The risk value is calculated by combining the maximum similarity, relative anomaly indicators, and evidence of abnormal temperature rise according to preset weights. This risk value is used to characterize whether the new event exceeds the normal operating range and the severity of the anomaly.

9. The method for monitoring the operating status of a circuit breaker according to claim 1, characterized in that, The specific steps for determining the anomaly category based on the combination rule of risk value and post-event stabilization time are as follows: The event category is determined based on the risk value. If the risk value exceeds the preset threshold, the device location information, event time, and location information of evidence indicators will be output. When the risk value is below 0.5, it indicates that the event is within the normal operating range and no intervention is required. Such events do not trigger alarms and do not require further maintenance. When the risk value is between 0.5 and 0.75, it indicates that the event has potential anomalies. Such events trigger alarms and it is recommended to perform regular checks, but no immediate remedial measures are required. When the risk value is greater than 0.75, it indicates that the anomaly of the event is severe. Such events trigger alarms and require intervention measures.

10. A method for monitoring the operating status of a circuit breaker according to claim 1, characterized in that, The specific steps for reporting event fingerprints and risk value summaries when the network is limited, and reporting event window data and summaries when the network is good, are as follows: The communication link quality parameters, including packet loss rate and round-trip time, are obtained by the edge computing node. When the packet loss rate is not less than the packet loss threshold and the round-trip time is not less than the delay threshold, the network is determined to be restricted and the event fingerprint and risk value summary are reported. When the packet loss rate is less than the packet loss threshold and the round-trip time is less than the delay threshold, the network is determined to be good and the event window data and the summary are reported.

Images