Traffic flow threshold crossing control method and system based on frequency prediction

Abstract

The application provides a frequency pre-judgment-based alternating current flow over-limit control method and system, and belongs to the technical field of power system protection and control, which comprises the following steps: continuously collecting the cycle frequency of the alternating current flow through a frequency measurement circuit, and adaptively sampling a target power source; performing multi-level mutation detection, filtering non-fault transient events in combination with the results, and outputting filtered suspected fault transient events; performing hierarchical over-limit judgment in combination with dynamic judgment criteria and amplitude-duration joint conditions, and counting the number of over-limit judgments that are true, when a preset multi-channel cooperative over-limit condition is met, determining to issue a trip command for action state monitoring of the target power source and the protection object, and performing feedback compensation and recording based on the action state monitoring result. The application solves the problem that the traditional alternating current flow over-limit control method can only respond passively after the frequency fluctuation scene, resulting in insufficient control speed and adaptive ability.

Description

Technical Field

[0001] This invention relates to the field of power system protection and control technology, specifically to a method and system for AC quantity over-limit control based on frequency prediction. Background Technology

[0002] With the continuous expansion of power system scale and the high proportion of renewable energy generation, system frequency fluctuations are becoming increasingly frequent and severe, leading to increasingly prominent issues related to AC quantities such as voltage and current exceeding limits. In scenarios such as industrial power distribution and renewable energy power plant grid connection, frequency deviations can directly cause the operating point of power equipment to deviate from its design value, triggering chain reactions such as overcurrent and overvoltage, and in severe cases, potentially causing equipment damage, protection malfunctions, or even system disconnection. Therefore, timely and accurate detection and control of AC quantities exceeding limits is one of the key technologies for ensuring the safe and stable operation of the power system.

[0003] Currently, traditional AC quantity over-limit control methods mainly rely on real-time monitoring and threshold comparison of voltage and current amplitudes. Specifically, existing solutions typically set a fixed upper or lower limit for amplitude over-limit. When the collected effective or instantaneous voltage or current value exceeds this set threshold and persists for a certain period, it is determined to be an over-limit, triggering corresponding protection or control actions. Some improved solutions introduce frequency tracking technology, such as using a phase-locked loop to achieve synchronous sampling to improve measurement accuracy, but their core judgment logic is still mainly based on amplitude comparison under steady-state or slowly changing conditions. However, the above methods can only initiate control after the actual physical quantity has already exceeded the danger threshold. Faced with a rapid increase in AC quantity caused by rapid frequency shifts, the optimal control window is often missed due to response lag.

[0004] In summary, there is an urgent need in the existing technology for a technical solution that can overcome the passive response after the fact and improve the speed and adaptability of AC quantity over-limit control in frequency fluctuation scenarios. Summary of the Invention

[0005] This invention provides a frequency-prediction-based AC quantity over-limit control method and system to solve the problem that traditional AC quantity over-limit control methods can only respond passively after the fact in frequency fluctuation scenarios, resulting in insufficient control speed and adaptive capability.

[0006] In view of the above problems, the present invention provides a method and system for AC quantity over-limit control based on frequency prediction.

[0007] In a first aspect, the present invention provides an AC quantity over-limit control method based on frequency prediction, comprising: The frequency of the AC quantity is continuously collected through the frequency measurement circuit, and adaptive sampling of the target power supply is performed based on the frequency to obtain the sampling data sequence. Based on the sampled data sequence, multi-level mutation detection is performed, and non-fault transient events are filtered in combination with the multi-level mutation detection results. The filtered mutation events are output as suspected fault transient events. The suspected fault transient events are traversed, and a graded over-limit judgment is performed by combining dynamic judgment benchmarks and amplitude-duration joint conditions. The number of over-limit judgments that are true is counted. When the preset multi-channel coordinated over-limit conditions are met, a trip command is issued. In conjunction with the trip command, the action status monitoring of the target power supply and the protected object is performed, and feedback supplementary actions and records are made based on the action status monitoring results.

[0008] Secondly, this invention provides an AC quantity over-limit control system based on frequency prediction, comprising: The adaptive sampling and data acquisition module is used to continuously collect the frequency of AC quantities through a frequency measurement circuit, and to perform adaptive sampling of the target power supply based on the frequency to acquire a sampling data sequence. The non-fault transient filtering and event screening module is used to perform multi-level mutation detection based on the sampled data sequence, and to filter non-fault transient events in combination with the multi-level mutation detection results, and output the filtered mutation events as suspected fault transient events. The graded over-limit judgment and trip decision module is used to traverse the suspected fault transient events, perform graded over-limit judgment combining dynamic judgment benchmark and amplitude-duration joint conditions, count the number of over-limit judgments that are true, and determine to issue a trip command when the preset multi-channel coordinated over-limit conditions are met. The action monitoring and feedback supplementary action module is used to combine the trip command to monitor the action status of the target power supply and the protected object, and to perform feedback supplementary action and record based on the action status monitoring results.

[0009] One or more technical solutions provided in this invention have at least the following technical effects or advantages: The technical solution of this invention first acquires the cycle frequency of the AC quantity in real time through a frequency measurement circuit and calculates the frequency change rate. By combining the current frequency and the frequency change rate, the duration of the next cycle is predicted, and the sampling interval is dynamically adjusted accordingly. Adaptive synchronous sampling is then performed on the target power supply, and after analog-to-digital conversion, a sampling data sequence locked to the system frequency is obtained. This eliminates asynchronous sampling errors caused by frequency offset at the data source, ensuring the complete number of sampling points and accurate sampling time within each power frequency cycle, providing a high-precision data foundation for all subsequent analysis stages.

[0010] Furthermore, based on the sampled data sequence, a half-wave Fourier algorithm is used to calculate the effective values ​​of three time-continuous sampling point groups. By comparing the ratio of adjacent effective values ​​with a preset multiple, a two-layer detection system is achieved for preliminary mutation capture and continuous confirmation. Then, three waveform characteristic parameters—the rise rate, the second harmonic content ratio, and the instantaneous attenuation ratio—are extracted from the mutation event. These parameters are then combined with graded thresholds to filter typical non-fault transient events such as inrush current, motor starting, and capacitor switching, outputting a set of suspected fault transient events. This approach balances sensitive capture of rapid short-circuit faults with reliable identification of slowly developing faults, while effectively eliminating normal operational disturbances and reducing input noise and malfunction risks in subsequent stages.

[0011] Furthermore, a dynamic judgment benchmark based on exponentially weighted moving average is constructed for each sampling channel, combined with an adaptive update of the forgetting factor and a deviation over-limit suspension mechanism. Using the over-limit multiple as a unified metric, a three-level amplitude-duration joint condition is applied to classify over-limit judgments: the highest level is judged without delay, while the intermediate and initial levels are confirmed with different delays. The number of over-limit channels is counted, and a trip command is issued after confirming a fault by combining the over-limit level combination with the multi-channel collaborative condition of directional consistency verification. The dynamic judgment benchmark solves the problem of poor adaptability of fixed thresholds, the three-level graded delay mechanism ensures rapid response in severe faults and prevents erroneous tripping due to instantaneous interference in minor over-limit situations, and the multi-channel collaborative verification completely eliminates false tripping caused by single-channel anomalies.

[0012] Finally, after the trip command is issued, a multi-layered supplementary operation process is executed sequentially, consisting of two trip commands and one backup protection request. The status feedback of the circuit breaker's auxiliary contacts determines whether the switch has successfully tripped. If the switch has not tripped, a backup protection request message is sent to the higher-level protection device. If the switch completely fails to operate, the trip output is locked in a continuous state, and a failure-to-operate alarm is uploaded. After the switch successfully trips, complete fault information is recorded and uploaded to the monitoring terminal. This effectively addresses abnormal situations such as circuit breaker failure to operate, preventing fault escalation through a dual defense of local supplementary operation and regional backup coordination. Full-process recording ensures event traceability.

[0013] In summary, the technical solution of this invention realizes the transformation from passive post-event response to proactive predictive control. It significantly improves the speed and adaptability of control through dynamic benchmarks and graded delays, and comprehensively ensures the reliability and fault tolerance of tripping through multi-channel coordination and multi-layer compensation mechanisms. Ultimately, it achieves an optimal balance in four dimensions: speed, selectivity, reliability, and safety. This solves the problem that traditional AC quantity over-limit control methods can only respond passively after the event in frequency fluctuation scenarios, resulting in insufficient control speed and adaptability. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the AC quantity over-limit control method based on frequency prediction provided by the present invention.

[0015] Figure 2 This is a logic diagram illustrating the trip command issuance in the frequency-prediction-based AC quantity over-limit control method provided by the present invention.

[0016] Figure 3 This is a schematic diagram of the AC quantity over-limit control system based on frequency prediction provided by the present invention.

[0017] In the attached diagram, the labels representing each component are as follows: Adaptive sampling and data acquisition module 11, non-fault transient filtering and event screening module 12, graded over-limit judgment and trip decision module 13, action monitoring and feedback supplementary action module 14. Detailed Implementation

[0018] This invention provides a frequency-prediction-based AC quantity over-limit control method and system, which solves the problem that traditional AC quantity over-limit control methods can only respond passively after the fact in frequency fluctuation scenarios, resulting in insufficient control speed and adaptive capability.

[0019] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0020] Example 1, as Figure 1 As shown, this invention provides an AC quantity over-limit control method based on frequency prediction, the method comprising: S100: The frequency of the AC quantity is continuously collected through the frequency measurement circuit, and adaptive sampling of the target power supply is performed according to the frequency to obtain the sampling data sequence.

[0021] This step continuously acquires the real-time cycle frequency of the target power AC quantity through a frequency measurement circuit and calculates the current frequency change rate. The duration of the next cycle is predicted by combining the current frequency and the frequency change rate. Based on this, the sampling interval is dynamically adjusted, and synchronous sampling is performed for the next cycle. After analog-to-digital conversion, a sampling data sequence adaptively locked with the system frequency is formed.

[0022] Step S100 in the method provided by the present invention includes: The frequency measurement circuit continuously collects the frequency of each cycle of the AC quantity of the target power supply to obtain the current cycle frequency value. Based on the current cycle frequency value, the current frequency change rate is calculated and obtained; Based on the current cycle frequency value and the current frequency change rate, calculate the predicted duration of the next cycle, and evenly divide the predicted duration into a preset number of sampling intervals; According to the sampling interval, the instantaneous values ​​of each sampling point in the next cycle are continuously collected, and the sampled data sequence is formed after analog-to-digital conversion.

[0023] In this step, a hardware frequency measurement circuit, such as a zero-crossing comparator or Schmitt trigger, is first used to shape the AC sine wave into a square wave signal. By capturing the time interval between two adjacent rising edges or falling edges of the square wave, the real-time frequency of the current cycle is calculated.

[0024] For example, in a power system with a rated frequency of 50Hz, the frequency measurement circuit captures a time interval of 19.6ms between the rising edges of the 100th and 101st cycles of the target power supply phase A voltage. The calculated current cycle frequency is approximately 1 / 0.0196 ≈ 51.02Hz.

[0025] Secondly, the processor compares the currently calculated frequency value with the frequency value of the previous cycle, and calculates the rate of frequency change over time through differential operations. That is, the current frequency change rate = (current cycle frequency value - previous cycle frequency value) / previous cycle duration, usually expressed in Hz / s. The frequency change rate reflects the speed and direction of system frequency changes; a positive value indicates that the frequency is increasing, and a negative value indicates that the frequency is decreasing.

[0026] For example, the processor reads a frequency value of 50.80Hz from the previous cycle, and the current cycle frequency is 51.02Hz. The interval between the two samplings is 20ms, which is one cycle. Therefore, the current frequency change rate is calculated as (51.02-50.80)Hz / 0.02s = +11.0Hz / s, indicating that the system frequency is rapidly increasing at a rate of 11Hz per second.

[0027] Next, using the current cycle frequency and the current rate of change of frequency, the duration of the next complete cycle is predicted using the average frequency method. The prediction formula can be expressed as: Predicted cycle duration = 1 / [Current cycle frequency + Current rate of change of frequency × Current cycle duration + 0.5 × Current rate of change of frequency × Predicted cycle duration]. Since the right side of the equation still contains the predicted cycle duration to be determined, the current cycle duration is used as the initial estimate of the predicted cycle duration and substituted into the right side. After one iteration, the predicted cycle duration is obtained. The physical meaning of the above formula is: under the assumption of linear frequency change, the average frequency of the next cycle is equal to the average of the instantaneous frequency at the end of the current cycle (current cycle frequency + current rate of change of frequency × current cycle duration) and the instantaneous frequency at the end of the next cycle. By using the current cycle duration as the initial estimate of the predicted cycle duration for iterative solution, the true average frequency of the next cycle can be accurately approximated. After calculating the predicted cycle duration, it is divided by the preset number of sampling points per cycle to obtain the adaptive sampling interval.

[0028] The preset number of sampling points per cycle refers to a fixed parameter pre-set and stored in the device during the design or initialization phase, based on the data density requirements of the protection algorithm, the processor's computing power, and the analog-to-digital converter's sampling rate. It represents the number of samples to be collected within each power frequency cycle. For example, the preset number of sampling points per cycle can be 64, 128, or 256 points. The preset number of sampling points per cycle determines the calculation basis for the adaptive sampling interval: the sampling interval is obtained by dividing the predicted cycle length by the preset number of points. A higher preset value results in higher accuracy in restoring the AC waveform, but also increases the data throughput and the computational burden of subsequent processing.

[0029] For example, the processor calculates the average frequency of the next cycle based on the current cycle frequency value of 51.02Hz, the current frequency change rate of +11.0Hz / s, and the current cycle duration of 19.6ms. First, using the current cycle duration of 19.6ms as an initial estimate of the predicted cycle duration, substituting into the formula: Next cycle average frequency ≈ 51.02 + 11.0 × 0.0196 + 0.5 × 11.0 × 0.0196 ≈ 51.02 + 0.2156 + 0.1078 ≈ 51.34Hz. The corresponding predicted cycle duration is 1 / 51.34 ≈ 19.48ms. With a preset sampling point count of 128 points per cycle, the adaptive sampling interval is calculated to be 19.48ms / 128 ≈ 152.2μs.

[0030] Finally, the analog-to-digital converter triggers sampling according to the calculated sampling interval, sequentially acquiring the instantaneous value of the current within the next complete cycle and converting it into a digital quantity. All sampling points are arranged in chronological order, forming a sampling data sequence of fixed length that is synchronized with the system frequency.

[0031] For example, during the 102nd cycle, the analog-to-digital converter samples every 152.5 μs, sequentially acquiring 128 instantaneous values ​​of the target power supply's A-phase current. For instance, the first sampling point is 0.12 A, the 32nd sampling point is 6.85 A, and the 64th sampling point is -7.02 A. After analog-to-digital conversion, these 128 ordered binary data constitute the sampling data sequence for that cycle.

[0032] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0033] In summary, this step effectively eliminates asynchronous sampling errors caused by system frequency offset, ensures the integrity of the number of sampling points and the accuracy of sampling time within each power frequency cycle, and improves the accuracy and reliability of all subsequent analysis steps such as amplitude calculation, mutation detection, and limit judgment from the data source.

[0034] S200: Based on the sampled data sequence, perform multi-level mutation detection, and combine the multi-level mutation detection results to filter non-fault transient events, and output the filtered mutation events as suspected fault transient events.

[0035] This step applies multiple mutation detection algorithms with varying time scales or detection sensitivities to the acquired sampled data sequence simultaneously, generating multi-level mutation detection results. Subsequently, by utilizing the spatiotemporal correlations between these detection results or comparing them with features of known non-fault disturbance modes, non-fault transient events are identified and filtered. Finally, events that, even after filtering, are still determined to be mutations and possess potential fault characteristics are marked as suspected fault transient events and output to the next stage.

[0036] In this step, multi-level mutation detection is performed based on the sampled data sequence, including: Select the first sampling point group, the second sampling point group, and the third sampling point group that are sequentially arranged in time from the sampling data sequence; Based on the half-wave Fourier algorithm, the first effective value corresponding to the first sampling point group, the second effective value corresponding to the second sampling point group, and the third effective value corresponding to the third sampling point group are calculated respectively. When the ratio of the second effective value to the first effective value exceeds a first preset multiple, it is determined that there is an initial abrupt change in the AC quantity; When the initial mutation exists and the ratio of the third effective value to the second effective value exceeds a second preset multiple, it is confirmed that there is a continuous mutation in the AC quantity; When the ratio of the second effective value to the first effective value does not exceed the first preset multiple, it is determined that the AC quantity has not undergone a continuous sudden change, and the monitoring status is returned; When the initial mutation occurs and the ratio of the third effective value to the second effective value does not exceed the second preset multiple, it is determined that the AC quantity has not undergone a continuous mutation, and the monitoring status is returned.

[0037] Specifically, firstly, in the acquired continuous-cycle fixed-point sampling data sequence, a cross-cycle sliding method is used to sequentially extract three sampling point groups, each with a data window length of half a power frequency cycle, denoted as the first sampling point group, the second sampling point group, and the third sampling point group, respectively. The three sampling point groups are respectively taken from the second half of the previous cycle, the first half of the current cycle, and the second half of the current cycle, forming a cross-cycle temporal continuous coverage.

[0038] The data window length is chosen to be half a power frequency cycle, or half a cycle, corresponding to 64 points out of 128 samples per cycle. The reason for choosing a half-cycle data window is that it strictly matches the data window requirements of the half-wave Fourier algorithm, ensuring the mathematical orthogonality and accuracy of the algorithm. By using a cross-cycle sliding method, when the third sampling point group (i.e., the data of the second half of the current cycle) is ready, the data of the first sampling point group (i.e., the data of the second half of the previous cycle) and the second sampling point group (i.e., the data of the first half of the current cycle) are already ready. The lag of the entire mutation detection is controlled within half a cycle, balancing detection accuracy and speed.

[0039] For example, with 128 sampling points per cycle, the processor selects points 65 to 128 from the previous cycle (101st cycle) as the first sampling point group, points 1 to 64 from the current cycle (102nd cycle) as the second sampling point group, and points 65 to 128 from the current cycle (102nd cycle) as the third sampling point group. When sampling point 128 of the 102nd cycle is completed, the data in the third sampling point group is just ready, while the data in the first and second sampling point groups have already been calculated, enabling real-time comparison.

[0040] Secondly, a half-wave Fourier transform algorithm is applied to the three selected sampling point groups to extract the fundamental sine and cosine components within each data window, thereby calculating the effective value of the fundamental wave, i.e., the root mean square value. The data window length of the half-wave Fourier transform algorithm is only half a power frequency cycle, strictly matching the data window length of the three sampling point groups, which mathematically guarantees the orthogonality of the sine and cosine basis functions, thus accurately extracting the fundamental wave amplitude information.

[0041] For example, the processor performs a half-wave Fourier transform on the first sampling point group, i.e., points 65 to 128 of the 101st cycle, and obtains a first effective value of 5.02A; performs the calculation on the second sampling point group, i.e. points 1 to 64 of the 102nd cycle, and obtains a second effective value of 6.15A; performs the calculation on the third sampling point group, i.e. points 65 to 128 of the 102nd cycle, and obtains a third effective value of 7.48A.

[0042] Next, the ratio of the second effective value to the first effective value is calculated and compared with a preset first multiple. If the ratio exceeds the first multiple, it indicates that the current has increased significantly within a short time window, and it is preliminarily determined that there is a sudden change in the AC quantity; if the ratio does not exceed the first multiple, it is determined that no continuous sudden change has occurred, and the system returns to normal cycle monitoring status.

[0043] The first preset multiple is usually set to a value between 1.15 and 1.30 to detect whether the current has changed abruptly beyond the normal fluctuation range. For example, a value of 1.2 can be used to sensitively capture moderate current rises.

[0044] For example, the ratio of the second effective value to the first effective value is calculated as 6.15A / 5.02A ≈ 1.225. The preset first multiple is 1.2. Since 1.225 > 1.2, it is determined that there is a preliminary abrupt change in the AC quantity, and the process proceeds to the next confirmation step. If, in another scenario, the ratio of the second effective value to the first effective value is only 1.05, it is directly determined that no persistent abrupt change has occurred, and the process returns to continue monitoring.

[0045] Finally, after determining that a preliminary mutation has occurred, the ratio of the third effective value to the second effective value is calculated and compared with a preset second multiple. If the ratio exceeds the second preset multiple, it indicates that the upward trend of the current is still continuing, rather than a brief single impact, and at this time, it is confirmed that there is a continuous mutation in the AC quantity; if the ratio does not exceed the preset multiple, it indicates that the previous mutation was only a transient disturbance, and it is determined that no continuous mutation has occurred, and the system returns to the monitoring state.

[0046] The second preset multiple is typically set to a value slightly lower than the first preset multiple, such as between 1.10 and 1.25, to confirm whether the upward trend of the current is continuous. For example, a value of 1.15, slightly lower than the first preset multiple, reflects the logic for confirming continuous changes. Both the second and first preset multiples can be adjusted according to the current fluctuation characteristics and protection sensitivity requirements of the specific application scenario.

[0047] For example, after confirming the initial abrupt change, the ratio of the third effective value to the second effective value is calculated as 7.48A / 6.15A ≈ 1.216. The preset second multiple is 1.15. Since 1.216 > 1.15, it indicates that the current is still rising, confirming a continuous abrupt change in the AC quantity.

[0048] In this step, non-fault transient events are filtered based on the multi-level mutation detection results. The filtered mutation events are output as suspected fault transient events, including: Extract the sampled data sequence that confirms the existence of a continuous mutation in the AC quantity from the multi-level mutation detection results, and calculate the waveform characteristic parameters, wherein the waveform characteristic parameters include at least the mutation rise rate, the second harmonic content ratio, and the instantaneous attenuation ratio; Filtering non-fault transient events based on the waveform feature parameters includes: When the mutation rate exceeds a preset high threshold and the instantaneous decay ratio is lower than a preset low threshold, the corresponding mutation event is determined to be a fault-type transient event. When the second harmonic content ratio exceeds the preset harmonic content threshold, the corresponding mutation event is determined to be a non-fault type transient event. When the instantaneous attenuation ratio exceeds the preset high threshold of attenuation ratio, the corresponding mutation event is determined to be a non-fault type transient event. When a mutation event does not meet all classification conditions at the same time, it is merged with the fault type transient event and output as the suspected fault transient event, while non-fault type transient events are filtered out.

[0049] Specifically, after confirming the existence of a continuous mutation in a certain sampled data sequence in the multi-level mutation detection stage, the above data is extracted, and three waveform characteristic parameters are calculated. Among them, the waveform characteristic parameters include at least: mutation rise rate, that is, the increase in effective value per unit time, reflecting the steepness of the fault current rise; second harmonic content ratio, that is, the ratio of the effective value of the second harmonic component to the effective value of the fundamental component, used to identify nonlinear transient characteristics such as transformer inrush current; instantaneous decay ratio, that is, within the current mutation detection window, the peak value of the impulse is located by the peak finding algorithm, the stable effective value at the end of the window is calculated, and the ratio obtained by dividing the difference between the peak value of the impulse and the stable effective value at the end of the window by the peak value of the impulse is used to instantly determine whether the current is showing a rapid decline trend within the detection window.

[0050] For example, after multi-level mutation detection confirms a persistent mutation in the 102nd cycle, the processor extracts 128 sampling data points for that segment and calculates: mutation rise rate = (7.48A - 5.02A) / (0.5 × 20ms) ≈ 246A / s, where the unit time is half a cycle; the second harmonic content ratio is 8.5% after full-wave Fourier analysis; the instantaneous attenuation ratio is calculated by finding the peak value of the impact within the current detection window as 7.48A, and the stable effective value corresponding to the end of the window, i.e., points 65 to 128, is 7.12A, so the instantaneous attenuation ratio ≈ (7.48 - 7.12) / 7.48 ≈ 4.8%.

[0051] Secondly, a high threshold for the rate of rise and a low threshold for the attenuation ratio are set. When the sudden rate of rise exceeds the high threshold and the instantaneous attenuation ratio is lower than the low threshold, it indicates that the current not only rises extremely rapidly but also does not show a significant drop within the detection window, remaining at a high level continuously, which is a characteristic of typical fault short-circuit currents. Events that meet this condition are judged as fault-type transient events and directly enter the suspected fault event set.

[0052] The high threshold for the sudden rise rate is used to identify transient events of the fault type. Taking a 5A rated current system as an example, it is usually set in the range of 100A / s to 500A / s, with a typical value of 200A / s. It needs to be determined in conjunction with the system's short-circuit capacity and 80% of the minimum rise rate in the typical fault waveform, and must be higher than the maximum rise rate of non-fault events such as motor starting. The low threshold for the attenuation ratio is used to identify the characteristic that the fault current basically does not attenuate within the window. It is usually set between 3% and 10%, with a typical value of 5%. It is taken as 2% to 3% higher than the maximum value of the instantaneous attenuation ratio of the short-circuit current in the fault waveform within the detection window.

[0053] For example, the preset high threshold for the rise rate is 200 A / s, and the low threshold for the decay ratio is 5%. If the current event suddenly rises at a rate of 246 A / s > 200 A / s and the instantaneous decay ratio is 4.8% < 5%, meeting the fault determination criteria, then the event is determined to be a transient event of fault type.

[0054] Secondly, a harmonic content threshold is set. When the second harmonic content ratio exceeds the threshold, it indicates that the current is rich in second harmonics, a typical characteristic of transformer inrush current, and belongs to non-fault disturbances. The current rise rate of inrush current at the moment of closing is extremely steep, comparable to the initial rise rate of a short-circuit fault. The reliable characteristic distinguishing the two lies in the second harmonic content rather than the rise rate; therefore, only the second harmonic content ratio is used as the criterion for inrush current. Events that meet this condition are judged as non-fault transient events and filtered out.

[0055] The threshold for second harmonic content is usually set between 12% and 20%, with a typical value of 15%. It needs to be adjusted according to the characteristics of the transformer core material. For amorphous alloy transformers, it can be reduced to 10% to 12%, and for systems with severe harmonic pollution, it needs to be appropriately increased.

[0056] For example, in another sudden event, the second harmonic content near the 200th cycle is 18.5%, and the sudden rise rate is 350 A / s. The preset harmonic content threshold is 15%. Since 18.5% > 15%, despite the extremely high rise rate, this event is still judged as a non-fault type transient event caused by transformer inrush current and is filtered out from the sudden event set.

[0057] Furthermore, a high threshold for the attenuation ratio is set. When the instantaneous attenuation ratio exceeds this high threshold, it indicates that the current rapidly drops from the peak value within the detection window. This is a typical characteristic of instantaneous impacts such as motor starting or capacitor switching, and belongs to non-fault disturbances. Events that meet this condition are judged as non-fault transient events and are filtered out.

[0058] The high threshold for attenuation ratio is used to identify the rapid attenuation characteristics of instantaneous impacts. It is usually set between 30% and 60%, with a typical value of 50%, which is 90% of the lower limit of the attenuation ratio for non-fault impact events.

[0059] For example, in another sudden event, the instantaneous attenuation ratio near the 350th cycle is 45%, meaning the current rapidly decreases by nearly half from its peak. The preset high threshold for the attenuation ratio is 50%. Since 45% < 50%, this non-fault criterion is not met. However, if the instantaneous attenuation ratio of another event is 55% > 50%, then this event is determined to be a non-fault transient event caused by motor starting and is filtered out from the set of sudden events.

[0060] Finally, for mutation events that do not match any of the above classification rules, in accordance with the safety and conservatism principle of protecting against unauthorized actions, they are merged with events clearly identified as fault types and output together as suspected fault transient events to the subsequent classification and limit-breaking judgment stage. All events identified as non-fault types are completely filtered out.

[0061] For example, another sudden event has a rise rate of 150 A / s, a second harmonic content ratio of 5%, and an instantaneous attenuation ratio of 20%. It does not meet the fault criteria (rise rate not reaching the 200 A / s high threshold), the inrush current criterion (harmonic ratio not exceeding 15%), or the instantaneous impact criterion (attenuation ratio not exceeding the 50% high threshold). This indicates that although the event is atypical, it poses a potential danger. Therefore, it is merged with the identified fault type events and output as a suspected transient fault event.

[0062] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0063] In summary, this step balances speed and reliability through multi-level detection, avoiding over-response to single instantaneous disturbances. Furthermore, by using multi-feature joint filtering based on instantaneous rise rate, second harmonic content ratio, and instantaneous attenuation ratio calculated in real time within the same detection window, it effectively eliminates various non-fault transient events common in normal operation without waiting for subsequent cycle data. This satisfies the stringent requirements of instantaneous overcurrent protection for output time while reducing input noise and the risk of protection maloperation in subsequent over-limit judgment stages. It ensures that only events with genuine potential fault characteristics enter the next level of judgment, thereby significantly improving the accuracy and anti-disturbance capability of the entire control process while ensuring that the protection does not fail to operate.

[0064] S300: Traverse the suspected fault transient events, perform graded over-limit judgment combining dynamic judgment benchmark and amplitude-duration joint conditions, and count the number of over-limit judgments that are true. When the preset multi-channel coordinated over-limit conditions are met, determine to issue a trip command.

[0065] like Figure 2 As shown, this step generates a dynamic judgment benchmark in real time for each suspected transient fault event selected, based on the current operating status of the system. Then, based on this dynamic benchmark, the AC amplitude of the event and the duration exceeding the benchmark are jointly judged, and the degree of exceeding the limit is divided into multiple levels. After traversing all suspected events, the cumulative number of true limit exceeding judgments is counted, and it is checked whether the preset multi-channel coordinated limit exceeding conditions are met. Once the conditions are met, a trip command is issued.

[0066] Step S300 in the method provided by the present invention includes: The suspected fault transient events confirmed in each sampling channel are traversed, and the effective value of the AC quantity of each channel is calculated based on the half-wave Fourier algorithm. Based on the dynamic judgment criteria corresponding to each channel, and according to the amplitude-duration joint condition, the effective value of the AC quantity of each channel is judged in a graded manner to obtain the graded over-limit judgment result of each channel. The number of channels with true over-limit determination results is counted. When the number of channels reaches a preset channel number threshold and meets the preset multi-channel collaborative over-limit conditions, the protected object is confirmed to be in a fault state, and the trip command is issued.

[0067] In this step, the sampling data of each sampling channel, such as the A-phase current, B-phase current, C-phase current, zero-sequence current, and voltage channel, are extracted one by one from the output set of suspected fault transient events. The fundamental effective value of each channel at the current moment is calculated quickly using the half-wave Fourier algorithm.

[0068] For example, a system contains three-phase current sampling channels (A, B, and C) and one zero-sequence current channel. The processor iterates through the suspected fault events near the 102nd cycle of the output, extracts the corresponding data segments for each channel, and calculates the effective values ​​for each channel using a half-wave Fourier algorithm: Phase A current IA = 7.48A, Phase B current IB = 7.52A, Phase C current IC = 5.01A, and zero-sequence current I0 = 1.89A.

[0069] Furthermore, based on the dynamic judgment criteria corresponding to each channel, and according to the amplitude-duration joint condition, the effective value of the AC quantity of each channel is judged in a graded manner to obtain the graded over-limit judgment result of each channel.

[0070] In this step, the construction of the dynamic judgment benchmark includes: A new valid value is determined based on the normal monitoring phase where the multi-level mutation detection results are empty. The dynamic judgment benchmark is updated by combining the current valid value with the exponentially weighted moving average algorithm. The weight of the new valid value is determined by a preset forgetting factor. When the deviation between the updated dynamic judgment benchmark and the previous dynamic judgment benchmark exceeds a preset deviation ratio, the update of the dynamic judgment benchmark is paused until the update resumes after a preset number of consecutive normal sampling cycles. The current value of the dynamic judgment benchmark is stored in a power-off storage memory and backed up periodically. When determining the level of limit exceedance, the limit exceedance threshold is set to the larger of the product of the preset limit exceedance multiple and the dynamic determination benchmark and the preset minimum protection setting.

[0071] Specifically, when the multi-level mutation detection result is empty, meaning no mutation event was detected, the system is determined to be in the normal monitoring phase. At this time, the new effective value calculated by the half-wave Fourier algorithm for the current cycle is extracted, and an exponentially weighted moving average is performed between this value and the historical dynamic judgment benchmark to update the benchmark value. The update formula can be expressed as: Updated benchmark value = Forgetting factor × New effective value + (1 - Forgetting factor) × Old benchmark value.

[0072] The forgetting factor typically ranges from 0.01 to 0.20. A larger forgetting factor value carries a higher weight for the new effective value, making the dynamic judgment benchmark more sensitive to load changes, but also weakening its anti-interference capability. A smaller forgetting factor value results in a smoother and more stable benchmark, but a slower response to actual load changes. Typical setting recommendations: In industrial power distribution scenarios with stable loads, a forgetting factor of 0.05 to 0.10 is recommended for smoothing; in renewable energy power plant scenarios with frequent load fluctuations, a forgetting factor of 0.10 to 0.20 is recommended for tracking. The forgetting factor value can be adjusted online adaptively: when the standard deviation of the effective value fluctuation over several consecutive periods is less than a preset threshold, the forgetting factor is automatically reduced to improve smoothness; when the standard deviation of the fluctuation increases, the forgetting factor is automatically increased to accelerate tracking. An example using a forgetting factor of 0.10 represents a compromise between smoothness and tracking performance.

[0073] For example, in a previous normal monitoring cycle, the current effective value of phase A current was 5.12A, the previous dynamic judgment benchmark stored value was 5.05A, and the preset forgetting factor was 0.10. According to the update formula: updated benchmark value = 0.10 × 5.12 + 0.90 × 5.05 = 5.057A, and the updated dynamic judgment benchmark is approximately 5.06A, smoothly tracking minute changes in the load current.

[0074] Secondly, the updated dynamic judgment benchmark is compared with the previous dynamic judgment benchmark before the update, and the relative deviation between the two is calculated, i.e., relative deviation = |updated benchmark value - old benchmark value| / old benchmark value × 100%. When the relative deviation exceeds the preset deviation ratio, it indicates that an abnormal change in the load level has occurred, which may be due to sensor malfunction, sudden load change, or missed transient events. At this time, the automatic update of the benchmark is suspended, and the previous stable benchmark value is kept unchanged to avoid abnormal data contaminating the benchmark. Subsequently, the exponentially weighted moving average update of the benchmark is only resumed after a preset number of normal sampling periods.

[0075] Setting the preset deviation percentage too low can lead to frequent and erroneous pauses in normal load adjustments, while setting it too high will reduce the ability to protect against abnormal data. A typical setting range is 10% to 20%, with 15% as an example; the specific setting can be determined based on the normal fluctuation range of the system load. The preset quantity is usually set to 3 to 10 normal sampling periods, with 5 as an example. Setting it too low may result in interference persisting during recovery updates, while setting it too high will prolong the adaptation time of the baseline to real load changes.

[0076] For example, the preset deviation ratio is 15%. Assuming the previous reference value was 5.06A, and the current effective value jumps to 7.80A due to instantaneous sensor interference, with the forgetting factor remaining at 0.10, the calculated updated reference value is 0.10 × 7.80 + 0.90 × 5.06 = 5.334A. The deviation is |5.334 - 5.06| / 5.06 ≈ 5.4%, which does not exceed 15%, and the reference is updated normally. If an instantaneous extreme interference causes the effective value to become 15.0A, the updated reference value is 0.10 × 15.0 + 0.90 × 5.334 = 6.434A, and the deviation is |6.434 - 5.334| / 5.334 ≈ 20.6%, which exceeds 15%. The system immediately pauses the reference update, locks the previous stable reference value of 5.334A, and resumes updating only after five consecutive normal sampling cycles.

[0077] To prevent the dynamic judgment benchmark from being permanently locked when a real step change occurs in the load, a maximum lockout duration is set. If the dynamic judgment benchmark is suspended from updating due to excessive deviation, and the lockout duration exceeds the maximum lockout duration, the lockout is forcibly released regardless of whether the current deviation has decreased, and the dynamic judgment benchmark is reinitialized with the current valid value. The maximum lockout duration is adjusted according to the load characteristics of the protected object, typically set to 100ms to 500ms, to ensure that the benchmark eventually tracks the actual load change.

[0078] Secondly, after each dynamic benchmark update, the current value is written to non-volatile memory, such as EEPROM or ferroelectric memory, to ensure that the device can immediately read the most recent valid benchmark value after a power outage or restart, without having to start accumulating from zero again. In addition, a periodic backup mechanism is set up, for example, every hour or when the benchmark value changes by a certain margin, the current benchmark value and its corresponding timestamp are written to the backup storage area for later analysis and debugging.

[0079] For example, after each benchmark update, the processor uses I... 2 The C bus writes the latest reference value of 5.06A to address 0x0100 of the ferroelectric memory. Simultaneously, a backup operation is triggered every hour on the clock, writing the current reference value of 5.06A along with the timestamp 14:00:00 to the backup record area. When the device loses power and is powered on again, the initialization program first reads 5.06A from the ferroelectric memory as the initial dynamic judgment reference and immediately begins operation without waiting for the reference to converge again.

[0080] Finally, when determining the level of over-limit, the dynamic reference is multiplied by the preset level-one over-limit multiple and compared with the preset minimum protection setting. The larger of the two values ​​is taken as the actual over-limit threshold for that level. This ensures that when the dynamic reference is too low and the product of the multiple is too low under extremely low load or no-load conditions, the protection threshold will not be lower than the basic safety boundary of the equipment, thus preventing abnormal loss of protection sensitivity.

[0081] The preset first-level over-limit multiple indicates the maximum multiple by which the effective value of the AC quantity is allowed to exceed the dynamic judgment benchmark. When this multiple is exceeded and the corresponding duration condition is met, it is judged as a first-level over-limit. The setting needs to comprehensively consider the balance between protection sensitivity and anti-interference capability: setting too low may easily trigger over-limit due to small fluctuations in normal load, while setting too high may lead to missed detection of small-amplitude fault currents. The typical setting range is usually 1.2 to 1.5 times, and 1.3 times can be used as an example.

[0082] The preset minimum protection setting refers to a fixed current or voltage lower limit value that is pre-set and stored in the device during the system design or initialization phase, based on the inherent physical characteristics of the protected equipment, such as transformers, lines, and busbars, including their rated parameters, insulation levels, and thermal stability limits. The setting typically refers to relevant power system relay protection setting guidelines. For example, for current protection, it can be set to 1.2 to 1.5 times the rated current of the protected equipment or the minimum operating current value determined based on short-circuit calculations. For instance, the preset minimum protection setting can be 6.50A.

[0083] For example, the preset first-level over-limit multiple is 1.3 times, and the preset minimum protection setting is 6.50A. The current dynamic judgment benchmark is 5.06A, and the calculated product is 5.06 × 1.3 = 6.578A. Comparing 6.578A with the minimum protection setting of 6.50A, the larger value of 6.578A is taken as the actual threshold for the first-level over-limit. If, under extremely low load and no-load conditions, the dynamic benchmark is only 1.20A, the product is 1.20 × 1.3 = 1.56A. In this case, the actual threshold is taken as the preset minimum protection setting of 6.50A, ensuring that the protection threshold is always not lower than the basic boundary for safe operation of the equipment.

[0084] In this step, the amplitude-duration joint condition includes: The ratio of the effective value of the AC quantity of each channel to the value of the dynamic judgment benchmark of the corresponding channel is defined as the limit-crossing multiple; When the over-limit multiple exceeds the highest level threshold, the channel is determined to have experienced the highest level over-limit, and the over-limit determination result is immediately output as true without any delay. When the over-limit multiple is between the intermediate level threshold and the highest level threshold, the first duration timer is started. When the duration between the intermediate level threshold and the highest level threshold exceeds the first preset duration, the over-limit determination result is output as true. When the over-limit multiple is between the starting level threshold and the intermediate level threshold, the second duration timing is started. When the duration between the starting level threshold and the intermediate level threshold exceeds the second preset duration, the over-limit judgment result is output as true, wherein the second preset duration is greater than the first preset duration. When the over-limit multiple does not exceed the initial level threshold, the channel is determined to be within the normal range, the over-limit determination result is false, and the timer for each duration is reset.

[0085] Specifically, for each sampling channel, the effective value of the AC quantity calculated by the half-wave Fourier algorithm is divided by the dynamic judgment benchmark value corresponding to that channel. The resulting ratio is defined as the limit-breaking multiple. The limit-breaking multiple is a dimensionless relative quantity used to eliminate dimensional differences between different channels and different load levels, so that the limit-breaking thresholds at all levels can be set in a uniform proportional form, facilitating cross-channel collaborative judgment.

[0086] For example, the current effective value of phase A current is 7.48A, the dynamic judgment benchmark is 5.06A, and the calculated over-limit multiple is 7.48 / 5.06≈1.48. The effective value of phase B current is 7.52A, the benchmark is 5.05A, and the over-limit multiple is 7.52 / 5.05≈1.49. The effective value of phase C current is 5.01A, the benchmark is 5.04A, and the over-limit multiple is 5.0 / 5.04≈0.99.

[0087] Secondly, a maximum threshold is set. When the over-limit multiple of a channel exceeds the maximum threshold, it indicates that the AC quantity has reached an extremely dangerous level. Without waiting for any delay confirmation, the over-limit judgment result of that channel is immediately set to true, and the highest level over-limit signal is output. This design ensures absolute speed of protection action in the event of a severe fault.

[0088] The highest-level threshold is typically set based on the maximum fault current level that the protected equipment can withstand or the short-circuit current multiple under the system's maximum operating conditions, ensuring absolute speed of protection action in the event of a severe fault. A typical setting range is 1.80 to 2.50 times, with 2.00 times being an example.

[0089] For example, the preset highest level threshold is 2.0 times. The current effective value of the zero-sequence current channel is 1.89A, and its dynamic judgment benchmark is 0.15A. The over-limit multiple is 1.89 / 0.15=12.60, which far exceeds the 2.0 times threshold. The over-limit judgment result of the zero-sequence channel is immediately set to true, and the judgment level is the highest level over-limit, without any delay.

[0090] Next, the interval formed by the intermediate-level threshold and the highest-level threshold is set as the second-level over-limit interval. When the over-limit multiple falls into the second-level over-limit interval, the first duration timer is started. Only when the over-limit multiple is continuously maintained within this interval for more than the first preset duration is the over-limit determination result of the channel set to true. If the over-limit multiple falls back below the intermediate-level threshold during the timing period, the timer is reset.

[0091] The intermediate-level threshold should be set higher than the initial-level threshold to form a tiered range, while being lower than the highest-level threshold to retain a channel for no-delay response to severe faults. A typical setting range is 1.40 to 1.80 times the initial threshold, with 1.50 times being an example. The selection of the first preset duration should comprehensively consider the protection's speed requirements and the protected equipment's short-time withstand capability to moderate overcurrents. A typical setting range is 1 ms to 5 ms, with 2 ms being an example.

[0092] For example, the preset intermediate level threshold is 1.5 times, the highest level threshold is 2.0 times, and the first preset duration is 2ms. The over-limit multiple of phase A current is 1.48, and the over-limit multiple of phase B current is 1.49. Neither of them reaches 1.5 times, so they are not within the judgment range of this level. Suppose that in another scenario, the over-limit multiple of a certain channel is 1.72, falling into the range [1.5, 2.0), then a 2ms timer is started; if 1.72 is maintained for more than 2ms, the timer ends, and the judgment is true; if it only lasts for 1.5ms and then falls back to 1.45, the timer is reset, and no over-limit is output.

[0093] Furthermore, a first-level over-limit interval is set, formed by the initial level threshold and the intermediate level threshold. When the over-limit multiple falls into the first-level over-limit interval, a second duration timer is started. Since the over-limit degree in this interval is relatively mild, a longer duration is needed to confirm that it is not an instantaneous fluctuation; therefore, the second preset duration should be greater than the first preset duration. Only when the over-limit multiple is continuously maintained for more than the second preset duration within this interval is the over-limit determination result set to true.

[0094] The initial threshold should be set slightly higher than the normal load fluctuation range to ensure that minor fluctuations in normal operation do not frequently trigger over-limit timing, while also sensitively capturing the initial characteristics of minor faults, providing a tiered entry point for subsequent more serious over-limit judgments. A typical setting range is 1.20 to 1.40 times, with 1.30 times being a possible example.

[0095] The second preset duration setting must be longer than the first preset duration. This reflects the inverse time-limit protection characteristic that the less severe the fault, the longer the required confirmation time. This ensures reliable identification of minor faults while effectively avoiding misjudgments caused by instantaneous disturbances. The typical setting range is 3ms to 10ms, and 5ms can be used as an example.

[0096] For example, the preset initial threshold is 1.3 times, the intermediate threshold is 1.5 times, and the second preset duration is 5ms. If the phase A current over-limit multiple of 1.48 falls within the range [1.3, 1.5), a 5ms timer is started; if the phase B current over-limit multiple of 1.49 also falls within this range, a 5ms timer is also started. If the over-limit multiple for both channels continues to exceed 5ms, the timer ends, and the first-level over-limit judgment results for both phase A and phase B are set to true.

[0097] Finally, when the over-limit multiple is lower than the initial level threshold, the channel is determined to be within the normal operating range, and the over-limit determination result is false. At the same time, to ensure that the timer status is consistent with the current operating conditions, all duration timers that may have been started but have not yet completed their timing are reset to zero, including the first duration timer and the second duration timer.

[0098] For example, if the current over-limit multiple for phase C is 0.99, which is lower than the initial threshold of 1.3, the over-limit determination for phase C is false. Since no timers were previously started in phase C, no additional reset is required. If a timer for a certain channel was started due to a brief over-limit multiple exceeding 1.3, but subsequently dropped back below 1.3, the timer will be immediately reset to ensure that the timing restarts from zero the next time the over-limit zone is entered.

[0099] Furthermore, after all sampling channels have completed the graded over-limit judgment, the over-limit judgment result flags of each channel are traversed, and the total number of channels with true judgment results, i.e., those that have met the amplitude-duration joint conditions of the corresponding grade, is counted.

[0100] For example, after the graded over-limit judgment, the first-level over-limit of phase A current channel is true, the first-level over-limit of phase B current channel is true, the third-level over-limit of zero-sequence current channel is true, the over-limit of phase C current channel is false, and the over-limit of voltage channel is false. The number of channels with true over-limit judgment results is 3.

[0101] Furthermore, the number of channels exceeding the limit is compared with a preset channel number threshold. Only when the number of channels exceeding the limit is greater than or equal to the preset channel number threshold can the process proceed to the next step of multi-channel collaborative condition judgment. A single channel exceeding the limit may be due to sensor malfunction or local disturbance, failing to meet the physical characteristics of a fault, and will directly return to the monitoring state.

[0102] The preset threshold for the number of channels is based on the physical characteristics of power system faults: real faults, such as phase-to-phase short circuits and ground faults, usually cause abnormal changes in multiple electrical quantity channels simultaneously, while single-channel overruns are mostly caused by sensor malfunctions, electromagnetic interference, or local disturbances. Setting a threshold can filter out the vast majority of false tripping risks caused by single-point interference at the source, serving as the first safety barrier for tripping decisions. The channel number threshold is typically set to 2.

[0103] For example, the preset threshold for the number of channels is 2. The current count is 3, including phase A, phase B, and zero sequence, satisfying the condition of being greater than 2, and proceeding to the multi-channel coordination condition judgment stage. If only the phase A current channel exceeds the limit (true), and the remaining channels are normal, the count is 1, which does not meet the threshold requirement. Therefore, no trip command is issued, and monitoring continues.

[0104] Further, check whether the preset multi-channel coordinated over-limit conditions are met. Multi-channel coordinated over-limit conditions are a set of logical judgment rules based on the physical laws of power system faults, used to verify whether the correlation between multiple over-limit channels is consistent with the characteristics of a real fault. Typical coordinated conditions may include any one or more combinations of the following: Condition A: At least two phase current channels are simultaneously determined to be over-limit true; Condition B: At least one phase current channel and the zero-sequence current channel are simultaneously determined to be over-limit true; Condition C: At least one current channel and its corresponding voltage channel are simultaneously determined to be over-limit true; Condition D: When the protected object includes multiple lines, at least two different circuit current channels on the same busbar segment are simultaneously determined to be over-limit true. Meeting any of the above conditions is considered passing the multi-channel coordinated verification.

[0105] For example, the current over-limit channel combination is phase A current, phase B current, and zero-sequence current. Check the coordination conditions: Condition A is met: both phase A and phase B currents exceed the limit simultaneously, consistent with phase-to-phase short circuit characteristics; Condition B is also met: both phase A and phase B are phase current channels, exceeding the limit simultaneously with the zero-sequence current channel, consistent with ground fault characteristics. At least one condition must be met to pass the coordination verification.

[0106] Finally, when the number of channels reaches the preset channel number threshold and at least one multi-channel coordinated over-limit condition is met, it is confirmed that the protected object is in a real fault state, rather than single-channel interference or transient disturbance. At this time, a trip command is immediately generated and sent to the trip coil drive circuit of the corresponding circuit breaker through the digital output interface or communication bus to perform fault isolation. At the same time, all judgment information of the trip event, including the over-limit channel, over-limit multiple, duration, dynamic reference value, etc., is packaged and recorded in the event log.

[0107] For example, in the scenario described above, if the number of channels (3) exceeds the preset channel number threshold (2) and conditions A and B are met, a phase-to-phase short circuit accompanied by a ground fault is confirmed in the protected object, and a trip command is immediately issued. The command is output to the circuit breaker tripping circuit via optocoupler isolation, and fault isolation is completed within the protection output time requirement. Simultaneously, the processor writes the event details—timestamp, over-limit channel identifier, over-limit multiples for each channel (1.48, 1.49, 12.60), and dynamic reference values ​​(5.06A, 5.05A, 0.15A)—into the non-volatile event log area for post-event fault analysis.

[0108] Step S300 in the method provided by the present invention further includes: Obtain the limit violation level information determined for each channel in the tiered limit violation judgment; When the number of over-limit channels reaches the preset channel number threshold, and the over-limit level of each over-limit channel is the highest level of over-limit, the protected object is directly confirmed to be in a fault state, and the trip command is issued. When the number of over-limit channels reaches the preset channel number threshold, and the over-limit level of each over-limit channel is not all the highest level over-limit, verify the consistency of the over-limit direction of each over-limit channel. If the direction is consistent, confirm that the protected object is in a fault state and issue the trip command. When the over-limit directions of the various over-limit channels are inconsistent, it is determined to be a non-fault state and returns to the monitoring state.

[0109] Specifically, after all sampling channels have undergone graded limit violation determination, the limit violation level corresponding to each channel that is determined to be true is extracted. The limit violation levels include at least Level 1, Level 2, and the highest level, corresponding to cases where the limit violation multiple exceeds the starting level threshold, the intermediate level threshold, and the highest level threshold, respectively. For channels that are determined to be false, their limit violation level record is empty or zero.

[0110] For example, after the graded over-limit judgment, the over-limit multiple of the A phase current channel is 1.48, which meets the first-level over-limit condition, and the over-limit level is recorded as first-level; the over-limit multiple of the B phase current channel is 1.49, which is also recorded as first-level; the over-limit multiple of the zero-sequence current channel is 12.60, which far exceeds the highest level threshold, and the over-limit level is recorded as the highest level; the over-limit of the C phase current and voltage channels is false, and the over-limit level is empty.

[0111] Secondly, when the number of channels that have been determined to have exceeded the limit reaches a preset channel number threshold, and all of the aforementioned channels have exceeded the limit at the highest level, it indicates that multiple channels have simultaneously detected extremely severe limit exceedances. This scenario corresponds to an extremely severe fault condition, such as a multi-phase metallic short circuit accompanied by grounding. No additional verification such as directional consistency is required; the protected object is directly confirmed to be in a fault state, and a trip command is immediately issued.

[0112] For example, in a severe fault scenario, the A-phase current exceedance level is the highest level, the B-phase current exceedance level is the highest level, the zero-sequence current exceedance level is the highest level, the number of channels (3) exceeds the preset channel number threshold (2), and all channels are at the highest level. The fault is directly confirmed without further verification, and a trip command is immediately issued.

[0113] Secondly, when the number of over-limit channels reaches the threshold, but at least one of these channels does not have the highest over-limit level (i.e., there are level one or two over-limit channels), the over-limit direction consistency verification step begins. Over-limit direction refers to the direction in which the AC quantity deviates from the dynamic judgment benchmark, and is divided into positive over-limit (effective value greater than the benchmark, i.e., overcurrent direction) and negative over-limit (effective value less than the benchmark, i.e., undercurrent direction). For common overcurrent protection scenarios, under normal fault conditions, all phase currents and zero-sequence currents should be positive over-limits. The verification method for direction consistency is: check whether the over-limit directions of all over-limit channels are the same. If all directions are consistent, it conforms to the physical characteristics of the fault; if there are channels with inconsistent directions, it may be due to sensor malfunction or superposition of multi-source interference.

[0114] For example, in the current scenario, the out-of-limit channels are: Phase A (Level 1 out-of-limit, positive direction), Phase B (Level 1 out-of-limit, positive direction), and Zero-sequence (Highest level, positive direction). The number of channels (3) is greater than the preset channel number threshold (2), and not all are at the highest level, so directional consistency verification is initiated. The out-of-limit directions of the three channels are checked; all are positive overcurrent, indicating consistent direction, and the verification passes.

[0115] Finally, when the over-limit direction consistency verification passes, the protected object is confirmed to be in a fault state, and a trip command is issued. When the verification fails, i.e., at least one channel has an over-limit direction opposite to other channels, the current event is determined to not conform to the physical characteristics of a real fault, and is more likely to be interference or abnormal operating conditions. It is confirmed as a non-fault state, no trip command is issued, and the system returns to normal monitoring state. At the same time, all over-limit flags and timers are reset.

[0116] For example, in another scenario, phase A current exceeds the positive limit at the first stage, phase B current exceeds the positive limit at the first stage, but phase C voltage channel detects a negative limit exceedance. The direction of voltage drop and the direction of current increase may be physically the same, but if a unified direction is defined, they may be judged as inconsistent, or a channel may have an abnormal direction due to reverse sensor connection. If an inconsistency in direction is detected, it is judged as a non-fault state, no trip command is issued, and the system returns to monitoring.

[0117] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0118] In summary, this step fundamentally solves the problem of poor adaptability of fixed thresholds under load fluctuation and frequency deviation scenarios through dynamic judgment benchmarks; the three-level graded delay mechanism ensures rapid action without delay during severe faults, and delays confirmation during minor over-limit situations to prevent erroneous tripping due to instantaneous interference; the multi-channel quantity threshold and over-limit level combination and directional consistency multiple collaborative verification fully leverage the physical laws of synchronous changes of multiple electrical quantities during power system faults, completely eliminating false tripping caused by single-channel sensor anomalies or electromagnetic interference, and achieving the optimal balance of tripping decision in terms of speed, reliability and safety.

[0119] S400: In conjunction with the trip command, perform action status monitoring of the target power supply and the protected object, and perform feedback supplementary actions and record based on the action status monitoring results.

[0120] After issuing the first trip command, this step waits for a first time interval and uses the status feedback of the circuit breaker's auxiliary contacts to determine whether the switch has tripped. If it has not tripped, a second trip command is issued, and a backup protection request message containing the device identifier, fault channel, and timestamp is sent to the upper-level protection device via the communication interface. After waiting for a second time interval, the determination is made again. If it still has not tripped, the trip output is locked to a continuous output state, the failure to operate event is recorded, and an alarm is uploaded. Once it is confirmed that the switch has tripped at any time, complete fault information is recorded and uploaded to the monitoring terminal.

[0121] Step S400 in the method provided by the present invention includes: After the first trip command is issued, wait for the first time interval and monitor the switching status of the protected object through the status feedback of the circuit breaker auxiliary contacts; If the switch does not trip, a second trip command is issued. At the same time, a backup protection request message containing the device identifier, fault channel information and timestamp is sent to the upper-level protection device through the communication interface. After waiting for the second time interval, the switch status is monitored again. If the switch still does not trip, the trip output will be locked to continuous output state, and the circuit breaker failure to operate event information will be written to the power-off storage memory. The failure to operate alarm information will be uploaded through the communication interface. When the switch status is detected as open, the fault information corresponding to the trip command is written into the power-off storage memory, and the fault information is uploaded to the monitoring terminal through the communication interface.

[0122] In this step, upon issuing the trip command, a tripping pulse is immediately sent to the trip coil drive circuit of the target circuit breaker via the digital output interface. Subsequently, the first time interval timer is started. After the timer expires, the status signals of the circuit breaker's auxiliary contacts are collected, such as normally open auxiliary contacts indicating the closed position and normally closed auxiliary contacts indicating the open position, to determine whether the switch has successfully tripped.

[0123] The first time interval must cover the circuit breaker's inherent opening time, arc extinguishing time, and a certain safety margin. This ensures sufficient time for the mechanical opening and arc extinguishing process to be completed during normal circuit breaker operation, preventing misjudgment of normal opening as failure to operate due to insufficient waiting time. The setting is based on the following: the typical circuit breaker's inherent opening time is approximately 30ms to 60ms, and the arc extinguishing time is approximately 10ms to 20ms, with an added margin of 20ms to 70ms to accommodate differences in operating characteristics between circuit breakers from different manufacturers and at different voltage levels. A typical setting range is 80ms to 150ms; an example of 100ms can be used.

[0124] For example, after determining that a trip command has been issued, the processor sends a trip pulse to the circuit breaker trip coil through the optocoupler isolation output circuit. The first time interval is preset to 100ms. After 100ms, the processor reads the status of the circuit breaker auxiliary contacts: if the normally open auxiliary contacts are still closed and the normally closed auxiliary contacts are still open, it indicates that the switch has not tripped and the circuit breaker may fail to operate.

[0125] Furthermore, if the switch fails to trip after the first time interval, the first supplementary action is immediately initiated. First, a second trip command is issued, re-activating the trip coil to prevent the first command from failing due to momentary interference or poor circuit contact. Simultaneously, a backup protection request message is sent to the higher-level protection device, such as the outgoing line protection or busbar protection of the next-level substation, via a communication interface such as RS-485, Ethernet, or a station control layer communication protocol. This message must include at least the device's unique identifier, current fault channel information (e.g., A-phase current, B-phase current, zero-sequence current exceeding limits, exceeding limit multiples, and fault occurrence timestamp), requesting the higher-level protection device to activate the backup trip logic. Subsequently, a second time interval timer is started, and after the timer expires, the status of the circuit breaker auxiliary contacts is read again.

[0126] The second time interval needs to cover the execution time of the second trip command, the communication delay of the upper-level protection device receiving and parsing the backup protection request message, the inherent operating time of the upper-level protection device initiating the backup trip logic and driving its corresponding circuit breaker to trip, and an appropriate coordination margin. The setting is based on the following: the communication delay is typically about 10ms to 50ms, depending on the communication protocol and network topology; the operating time of the upper-level protection device is similar to that of the circuit breaker at this level; the combined time needs to be significantly longer than the first time interval to ensure that the backup protection has a sufficient response window to complete fault isolation. A typical setting range is 200ms to 300ms, and 250ms can be used as an example.

[0127] For example, if the processor detects that the switch has not tripped, it immediately performs a supplementary action: sending a tripping pulse to the trip coil again, and simultaneously sending a backup protection request message to the upstream substation outgoing line protection device via the Ethernet interface. The message contains: Device ID: 0x05, Fault Channel: Phase A, Phase B, Zero Sequence, Over-limit Multiple: 1.48 / 1.49 / 12.60, Timestamp: 14:05:23.120. The second time interval is preset to 250ms. After 250ms, the auxiliary contacts are read again: the normally open auxiliary contact remains closed, the normally closed auxiliary contact remains open, and the switch still has not tripped.

[0128] Furthermore, if the switch still fails to trip after the second time interval, it indicates that the circuit breaker has experienced a serious failure to operate, and both trip commands have failed to take effect. At this point, the trip output signal is locked in a continuous output state, maintaining the tripping level to continuously drive the trip coil, hoping to finally trip when the operating mechanism's jamming is relieved or the secondary circuit is restored. Simultaneously, the information about this circuit breaker failure to operate is written into non-volatile memory, including the failure time, fault channel details, the issuance time of the two trip commands, and the time sequence of switch status feedback. In addition, the failure-to-operate alarm information is uploaded to the monitoring terminal or dispatch automation system via the communication interface to notify maintenance personnel for emergency intervention.

[0129] For example, if the circuit breaker fails to operate after the second time interval, the processor permanently sets the trip output pin level to an active state, continuously driving the trip coil. Simultaneously, the failure-to-operate event (timestamp 14:05:23.495), fault channel information, the timestamps of the two trip commands (14:05:23.000 and 14:05:23.120), and the fact that the switch never disconnected are written to the ferroelectric memory. Subsequently, an emergency alarm message is sent to the monitoring backend via the station control layer network: "Circuit breaker failure to operate alarm, device 0x05, please handle on-site immediately."

[0130] Finally, at any stage of the aforementioned steps, once a change in the status of the circuit breaker auxiliary contacts is detected indicating that the switch has successfully tripped (e.g., the normally open auxiliary contact changes from closed to open, the normally closed auxiliary contact changes from open to closed, and the effective value of the current path has dropped to near zero), it is immediately confirmed that the fault has been successfully isolated. At this time, the continuous output lockout is lifted, and the complete fault information corresponding to this trip command is written to the non-volatile memory. The information includes the fault occurrence time, the trip command issuance time, the switch trip confirmation time, the over-limit channel identifier, the over-limit multiple and over-limit level of each channel, the dynamic judgment benchmark value, and the number of re-operations. Simultaneously, the above fault information is packaged and uploaded to the monitoring terminal through the communication interface for dispatch personnel to view and for post-fault analysis.

[0131] For example, if in the preceding steps, after 100ms, it is detected that the normally open auxiliary contact opens and the normally closed auxiliary contact closes, and the current in phase A drops from 7.48A to below 0.02A, the switch is confirmed to have successfully tripped. Complete fault record: timestamp 14:05:23.000 tripping, 14:05:23.095 switch disconnection, over-limit channels A, B, and zero sequence, over-limit multiples of 1.48, 1.49, and 12.60 for each channel are written to the ferroelectric memory fault event area and uploaded to the monitoring terminal via the network. A notification pops up on the monitoring interface: "Device 0x05 protection action, fault isolated."

[0132] It should be noted that the above values ​​are for illustrative purposes only and do not constitute a limitation on the present invention.

[0133] In summary, this step ensures that every protection action, whether successful or not, can be fully traced, providing a detailed data foundation for post-event fault analysis and protection device operation and maintenance. This comprehensively improves the system's reliability in three aspects: single-point fault tolerance, regional collaborative protection, and event traceability.

[0134] In summary, this invention achieves a shift from passive post-event response to proactive predictive control. It significantly improves the speed and adaptability of control through dynamic benchmarks and graded delays, while comprehensively ensuring the reliability and fault tolerance of tripping through multi-channel coordination and multi-layer compensation mechanisms. Ultimately, it achieves an optimal balance in four dimensions: speed, selectivity, reliability, and safety. This solves the problem that traditional AC quantity over-limit control methods can only respond passively after the event in frequency fluctuation scenarios, resulting in insufficient control speed and adaptability.

[0135] Example 2, as Figure 3 As shown, the present invention provides an AC quantity over-limit control system based on frequency prediction, the system comprising: The adaptive sampling and data acquisition module 11 is used to continuously collect the frequency of AC quantities through a frequency measurement circuit, and to perform adaptive sampling of the target power supply based on the frequency to obtain a sampling data sequence.

[0136] Specifically, a frequency measurement circuit continuously acquires the cycle frequency of the AC quantity, and adaptively samples the target power supply based on the cycle frequency to obtain a sampling data sequence, including: The frequency measurement circuit continuously collects the frequency of each cycle of the AC quantity of the target power supply to obtain the current cycle frequency value. Based on the current cycle frequency value, the current frequency change rate is calculated and obtained; Based on the current cycle frequency value and the current frequency change rate, calculate the predicted duration of the next cycle, and evenly divide the predicted duration into a preset number of sampling intervals; According to the sampling interval, the instantaneous values ​​of each sampling point in the next cycle are continuously collected, and the sampled data sequence is formed after analog-to-digital conversion.

[0137] The non-fault transient filtering and event screening module 12 is used to perform multi-level mutation detection based on the sampled data sequence, and to filter non-fault transient events in combination with the multi-level mutation detection results, and output the filtered mutation events as suspected fault transient events.

[0138] Among them, multi-level mutation detection is performed based on the sampled data sequence, including: Select the first sampling point group, the second sampling point group, and the third sampling point group that are sequentially arranged in time from the sampling data sequence; Based on the half-wave Fourier algorithm, the first effective value corresponding to the first sampling point group, the second effective value corresponding to the second sampling point group, and the third effective value corresponding to the third sampling point group are calculated respectively. When the ratio of the second effective value to the first effective value exceeds a first preset multiple, it is determined that there is an initial abrupt change in the AC quantity; When the initial mutation exists and the ratio of the third effective value to the second effective value exceeds a second preset multiple, it is confirmed that there is a continuous mutation in the AC quantity; When the ratio of the second effective value to the first effective value does not exceed the first preset multiple, it is determined that the AC quantity has not undergone a continuous sudden change, and the monitoring status is returned; When the initial mutation occurs and the ratio of the third effective value to the second effective value does not exceed the second preset multiple, it is determined that the AC quantity has not undergone a continuous mutation, and the monitoring status is returned.

[0139] Among them, non-fault transient events are filtered by combining multi-level mutation detection results, and the filtered mutation events are output as suspected fault transient events, including: Extract the sampled data sequence that confirms the existence of a continuous mutation in the AC quantity from the multi-level mutation detection results, and calculate the waveform characteristic parameters, wherein the waveform characteristic parameters include at least the mutation rise rate, the second harmonic content ratio, and the instantaneous attenuation ratio; Filtering non-fault transient events based on the waveform feature parameters includes: When the mutation rate exceeds a preset high threshold and the instantaneous decay ratio is lower than a preset low threshold, the corresponding mutation event is determined to be a fault-type transient event. When the second harmonic content ratio exceeds the preset harmonic content threshold, the corresponding mutation event is determined to be a non-fault type transient event. When the instantaneous attenuation ratio exceeds the preset high threshold of attenuation ratio, the corresponding mutation event is determined to be a non-fault type transient event. When a mutation event does not meet all classification conditions at the same time, it is merged with the fault type transient event and output as the suspected fault transient event, while non-fault type transient events are filtered out.

[0140] The graded over-limit judgment and trip decision module 13 is used to traverse the suspected fault transient events, perform graded over-limit judgment combining dynamic judgment benchmark and amplitude-duration joint conditions, and count the number of over-limit judgments that are true. When the preset multi-channel coordinated over-limit conditions are met, a trip command is issued.

[0141] Specifically, the suspected fault transient events are traversed, and a graded over-limit judgment is performed combining dynamic judgment criteria and amplitude-duration joint conditions. The number of over-limit judgments that are true is counted. When the preset multi-channel coordinated over-limit conditions are met, a trip command is issued, including: The suspected fault transient events confirmed in each sampling channel are traversed, and the effective value of the AC quantity of each channel is calculated based on the half-wave Fourier algorithm. Based on the dynamic judgment criteria corresponding to each channel, and according to the amplitude-duration joint condition, the effective value of the AC quantity of each channel is judged in a graded manner to obtain the graded over-limit judgment result of each channel. The number of channels with true over-limit determination results is counted. When the number of channels reaches a preset channel number threshold and meets the preset multi-channel collaborative over-limit conditions, the protected object is confirmed to be in a fault state, and the trip command is issued.

[0142] The construction of dynamic judgment criteria includes: A new valid value is determined based on the normal monitoring phase where the multi-level mutation detection results are empty. The dynamic judgment benchmark is updated by combining the current valid value with the exponentially weighted moving average algorithm. The weight of the new valid value is determined by a preset forgetting factor. When the deviation between the updated dynamic judgment benchmark and the previous dynamic judgment benchmark exceeds a preset deviation ratio, the update of the dynamic judgment benchmark is paused until the update resumes after a preset number of consecutive normal sampling cycles. The current value of the dynamic judgment benchmark is stored in a power-off storage memory and backed up periodically. When determining the level of limit exceedance, the limit exceedance threshold is set to the larger of the product of the preset limit exceedance multiple and the dynamic determination benchmark and the preset minimum protection setting.

[0143] The amplitude-duration joint condition includes: The ratio of the effective value of the AC quantity of each channel to the value of the dynamic judgment benchmark of the corresponding channel is defined as the limit-crossing multiple; When the over-limit multiple exceeds the highest level threshold, the channel is determined to have experienced the highest level over-limit, and the over-limit determination result is immediately output as true without any delay. When the over-limit multiple is between the intermediate level threshold and the highest level threshold, the first duration timer is started. When the duration between the intermediate level threshold and the highest level threshold exceeds the first preset duration, the over-limit determination result is output as true. When the over-limit multiple is between the starting level threshold and the intermediate level threshold, the second duration timing is started. When the duration between the starting level threshold and the intermediate level threshold exceeds the second preset duration, the over-limit judgment result is output as true, wherein the second preset duration is greater than the first preset duration. When the over-limit multiple does not exceed the initial level threshold, the channel is determined to be within the normal range, the over-limit determination result is false, and the timer for each duration is reset.

[0144] Among them, determining to issue a trip command when the preset multi-channel coordinated over-limit conditions are met also includes: Obtain the limit violation level information determined for each channel in the tiered limit violation judgment; When the number of over-limit channels reaches the preset channel number threshold, and the over-limit level of each over-limit channel is the highest level of over-limit, the protected object is directly confirmed to be in a fault state, and the trip command is issued. When the number of over-limit channels reaches the preset channel number threshold, and the over-limit level of each over-limit channel is not all the highest level, the consistency of the over-limit direction of each over-limit channel is verified. If the direction is consistent, it is confirmed that the protected object is in a fault state, and the trip command is issued. When the over-limit directions of the various over-limit channels are inconsistent, it is determined to be a non-fault state and returns to the monitoring state.

[0145] The action monitoring and feedback supplementary action module 14 is used to combine the trip command to perform action status monitoring of the target power supply and the protected object, and to perform feedback supplementary action and record based on the action status monitoring results.

[0146] Specifically, in conjunction with the trip command, the operation status of the target power supply and the protected object is monitored, and feedback and supplementary actions are performed and recorded based on the operation status monitoring results, including: After the first trip command is issued, wait for the first time interval and monitor the switching status of the protected object through the status feedback of the circuit breaker auxiliary contacts; If the switch does not trip, a second trip command is issued. At the same time, a backup protection request message containing the device identifier, fault channel information and timestamp is sent to the upper-level protection device through the communication interface. After waiting for the second time interval, the switch status is monitored again. If the switch still does not trip, the trip output will be locked to continuous output state, and the circuit breaker failure to operate event information will be written to the power-off storage memory. The failure to operate alarm information will be uploaded through the communication interface. When the switch status is detected as tripped, the fault information corresponding to the trip command is written into the power-off storage memory, and the fault information is uploaded to the monitoring terminal through the communication interface.

[0147] 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.

[0148] This specification and accompanying drawings are merely illustrative examples of the invention and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its scope. Therefore, if such modifications and modifications fall within the scope of the invention and its equivalents, the invention is intended to include these modifications and modifications.

Claims

1. A frequency-prediction-based AC quantity over-limit control method, characterized in that, include: The frequency of the AC quantity is continuously collected through the frequency measurement circuit, and adaptive sampling of the target power supply is performed based on the frequency to obtain the sampling data sequence. Based on the sampled data sequence, multi-level mutation detection is performed, and non-fault transient events are filtered in combination with the multi-level mutation detection results. The filtered mutation events are output as suspected fault transient events. The suspected fault transient events are traversed, and a graded over-limit judgment is performed by combining dynamic judgment benchmarks and amplitude-duration joint conditions. The number of over-limit judgments that are true is counted. When the preset multi-channel coordinated over-limit conditions are met, a trip command is issued. In conjunction with the trip command, the action status monitoring of the target power supply and the protected object is performed, and feedback supplementary actions and recording are performed based on the action status monitoring results; Based on the sampled data sequence, multi-level mutation detection is performed, including: Select the first sampling point group, the second sampling point group, and the third sampling point group that are sequentially arranged in time from the sampling data sequence; Based on the half-wave Fourier algorithm, the first effective value corresponding to the first sampling point group, the second effective value corresponding to the second sampling point group, and the third effective value corresponding to the third sampling point group are calculated respectively. When the ratio of the second effective value to the first effective value exceeds a first preset multiple, it is determined that there is an initial abrupt change in the AC quantity; When the initial mutation exists and the ratio of the third effective value to the second effective value exceeds a second preset multiple, it is confirmed that there is a continuous mutation in the AC quantity; When the ratio of the second effective value to the first effective value does not exceed the first preset multiple, it is determined that the AC quantity has not undergone a continuous sudden change, and the monitoring status is returned; When the initial mutation occurs and the ratio of the third effective value to the second effective value does not exceed the second preset multiple, it is determined that the AC quantity has not undergone a continuous mutation, and the monitoring status is returned. By combining the results of multi-level mutation detection, non-fault transient events are filtered, and the filtered mutation events are output as suspected fault transient events, including: Extract the sampled data sequence that confirms the existence of a continuous mutation in the AC quantity from the multi-level mutation detection results, and calculate the waveform characteristic parameters, wherein the waveform characteristic parameters include at least the mutation rise rate, the second harmonic content ratio, and the instantaneous attenuation ratio; Filtering non-fault transient events based on the waveform feature parameters includes: When the mutation rate exceeds a preset high threshold and the instantaneous decay ratio is lower than a preset low threshold, the corresponding mutation event is determined to be a fault-type transient event. When the second harmonic content ratio exceeds the preset harmonic content threshold, the corresponding mutation event is determined to be a non-fault type transient event. When the instantaneous attenuation ratio exceeds the preset high threshold of attenuation ratio, the corresponding mutation event is determined to be a non-fault type transient event. When a mutation event does not meet all classification conditions at the same time, it is merged with the fault type transient event and output as the suspected fault transient event, while non-fault type transient events are filtered out.

2. The AC quantity over-limit control method based on frequency prediction as described in claim 1, characterized in that, The frequency of the AC signal is continuously acquired through a frequency measurement circuit, and adaptive sampling of the target power supply is performed based on the frequency to obtain a sampling data sequence, including: The frequency measurement circuit continuously collects the frequency of each cycle of the AC quantity of the target power supply to obtain the current cycle frequency value. Based on the current cycle frequency value, the current frequency change rate is calculated and obtained; Based on the current cycle frequency value and the current frequency change rate, calculate the predicted duration of the next cycle, and evenly divide the predicted duration into a preset number of sampling intervals; According to the sampling interval, the instantaneous values ​​of each sampling point in the next cycle are continuously collected, and the sampled data sequence is formed after analog-to-digital conversion.

3. The AC quantity over-limit control method based on frequency prediction as described in claim 1, characterized in that, The suspected fault transient events are traversed, and a graded over-limit judgment is performed combining dynamic judgment criteria and amplitude-duration joint conditions. The number of over-limit judgments that are true is counted. When the preset multi-channel coordinated over-limit conditions are met, a trip command is issued, including: The suspected fault transient events confirmed in each sampling channel are traversed, and the effective value of the AC quantity of each channel is calculated based on the half-wave Fourier algorithm. Based on the dynamic judgment criteria corresponding to each channel, and according to the amplitude-duration joint condition, the effective value of the AC quantity of each channel is judged in a graded manner to obtain the graded over-limit judgment result of each channel. The number of channels with true over-limit determination results is counted. When the number of channels reaches a preset channel number threshold and meets the preset multi-channel collaborative over-limit conditions, the protected object is confirmed to be in a fault state, and the trip command is issued.

4. The AC quantity over-limit control method based on frequency prediction as described in claim 3, characterized in that, The construction of dynamic judgment criteria includes: A new valid value is determined based on the normal monitoring phase where the multi-level mutation detection results are empty. The dynamic judgment benchmark is updated by combining the current valid value with the exponentially weighted moving average algorithm. The weight of the new valid value is determined by a preset forgetting factor. When the deviation between the updated dynamic judgment benchmark and the previous dynamic judgment benchmark exceeds a preset deviation ratio, the update of the dynamic judgment benchmark is paused until the update resumes after a preset number of consecutive normal sampling cycles. The current value of the dynamic judgment benchmark is stored in a power-off storage memory and backed up periodically. When determining the level of limit exceedance, the limit exceedance threshold is set to the larger of the product of the preset limit exceedance multiple and the dynamic determination benchmark and the preset minimum protection setting.

5. The AC quantity over-limit control method based on frequency prediction as described in claim 3, characterized in that, The amplitude-duration joint condition includes: The ratio of the effective value of the AC quantity of each channel to the value of the dynamic judgment benchmark of the corresponding channel is defined as the limit-crossing multiple; When the over-limit multiple exceeds the highest level threshold, the channel is determined to have experienced the highest level over-limit, and the over-limit determination result is immediately output as true without any delay. When the over-limit multiple is between the intermediate level threshold and the highest level threshold, the first duration timer is started. When the duration between the intermediate level threshold and the highest level threshold exceeds the first preset duration, the over-limit determination result is output as true. When the over-limit multiple is between the starting level threshold and the intermediate level threshold, the second duration timing is started. When the duration between the starting level threshold and the intermediate level threshold exceeds the second preset duration, the over-limit judgment result is output as true, wherein the second preset duration is greater than the first preset duration. When the over-limit multiple does not exceed the initial level threshold, the channel is determined to be within the normal range, the over-limit determination result is false, and the timer for each duration is reset.

6. The AC quantity over-limit control method based on frequency prediction as described in claim 1, characterized in that, In conjunction with the trip command, the operation status of the target power supply and the protected object is monitored, and feedback and supplementary actions are performed and recorded based on the monitoring results, including: After the first trip command is issued, wait for the first time interval and monitor the switching status of the protected object through the status feedback of the circuit breaker auxiliary contacts; If the switch does not trip, a second trip command is issued. At the same time, a backup protection request message containing the device identifier, fault channel information and timestamp is sent to the upper-level protection device through the communication interface. After waiting for the second time interval, the switch status is monitored again. If the switch still does not trip, the trip output will be locked to continuous output state, and the circuit breaker failure to operate event information will be written to the power-off storage memory. The failure to operate alarm information will be uploaded through the communication interface. When the switch status is detected as open, the fault information corresponding to the trip command is written into the power-off storage memory, and the fault information is uploaded to the monitoring terminal through the communication interface.

7. The AC quantity over-limit control method based on frequency prediction as described in claim 3, characterized in that, When the preset multi-channel coordinated over-limit conditions are met, the trip command is issued, which also includes: Obtain the limit violation level information determined for each channel in the tiered limit violation judgment; When the number of over-limit channels reaches the preset channel number threshold, and the over-limit level of each over-limit channel is the highest level of over-limit, the protected object is directly confirmed to be in a fault state, and the trip command is issued. When the number of over-limit channels reaches the preset channel number threshold, and the over-limit level of each over-limit channel is not all the highest level over-limit, verify the consistency of the over-limit direction of each over-limit channel. If the direction is consistent, confirm that the protected object is in a fault state and issue the trip command. When the over-limit directions of the various over-limit channels are inconsistent, it is determined to be a non-fault state and returns to the monitoring state.

8. An AC quantity over-limit control system based on frequency prediction, characterized in that, The method for implementing the frequency-prediction-based AC quantity over-limit control method according to any one of claims 1 to 7 includes: The adaptive sampling and data acquisition module is used to continuously collect the frequency of AC quantities through a frequency measurement circuit, and to perform adaptive sampling of the target power supply based on the frequency to acquire a sampling data sequence. The non-fault transient filtering and event screening module is used to perform multi-level mutation detection based on the sampled data sequence, and to filter non-fault transient events in combination with the multi-level mutation detection results, and output the filtered mutation events as suspected fault transient events. The graded over-limit judgment and trip decision module is used to traverse the suspected fault transient events, perform graded over-limit judgment combining dynamic judgment benchmark and amplitude-duration joint conditions, count the number of over-limit judgments that are true, and determine to issue a trip command when the preset multi-channel coordinated over-limit conditions are met. The action monitoring and feedback supplementary action module is used to combine the trip command to monitor the action status of the target power supply and the protected object, and to perform feedback supplementary action and record based on the action status monitoring results.

Images