Modularized reclosing circuit breaker protection system based on intelligent electrical cloud cooperation

Abstract

The invention discloses a modular reclosing circuit breaker protection system based on intelligent electrical cloud coordination, particularly relates to the technical field of intelligent power distribution network relay protection, and is used for solving the problems that fault diagnosis is delayed and a protection strategy cannot be generated in a fault clearing time window due to resource competition in a cloud coordination protection system. A dynamic preemption priority is generated based on power grid topology vulnerability and associated circuit breaker collaborative requirements by monitoring the load state of cloud computing node resources in real time; credibility attenuation control is carried out in combination with a fault current waveform distortion rate; when the resource preemption necessity exceeds a threshold value, freezing a periodic task thread and releasing computing resources, and opening up a high-priority communication channel at the same time; sending a fault recording compression instruction to a target circuit breaker terminal to obtain an edge preprocessing feature vector; and protection strategy optimization is executed in the released resources to generate reclosing parameters, and the reclosing parameters are issued and executed through a high-priority channel, so that quick response of a fault diagnosis task is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of relay protection for intelligent distribution networks, and more specifically, to a modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration. Background Art

[0002] In the field of smart distribution network protection, the use of cloud-coordinated circuit breaker protection systems can improve fault handling efficiency. Usually, the electrical quantity data collected by distributed terminals is uploaded to the cloud platform, and protection strategy optimization (such as adaptive reclosing parameter adjustment) is achieved through centralized computing. The system is then sent to the local circuit breaker for execution, relying on the cloud to complete core fault diagnosis and decision-making, taking into account both protection accuracy and intensive operation and maintenance requirements.

[0003] However, the existing cloud-collaborative circuit breaker protection system faces a fundamental conflict between dynamic resource allocation and real-time performance assurance: when the cloud platform simultaneously processes multiple computing tasks (such as periodic equipment status monitoring and batch terminal data cleaning), sudden fault diagnosis requests may enter a queue waiting state due to instantaneous saturation of computing resources. In this case, the cloud cannot complete policy generation and distribution within the power system fault clearing time window, causing the local circuit breaker to be forced to downgrade and execute non-optimized protection logic, resulting in the failure of collaborative efficiency. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration, including: The load monitoring module is used by the cloud platform to monitor the resource load status of each computing node in real time; A priority generation module is used to respond to fault diagnosis requests and generate dynamic preemption priorities based on the grid topology vulnerability of the protection area where the target circuit breaker is located and the degree of coordinated action requirements of the associated circuit breakers; Credibility attenuation module, used to extract the waveform distortion rate of the fault current. If it does not fall within the threshold range of the historical fault library, the credibility of the dynamic preemption priority is attenuated. The resource preemption module is used to determine the necessity of resource preemption based on the resource load status and the processed dynamic preemption priority. If it exceeds the resource preemption threshold, the periodic task thread is frozen, computing resources are released, and a high-priority communication channel is opened; A feature extraction module is used to send a fault recording compression instruction to the target circuit breaker terminal through a high-priority communication channel and receive the fault feature vector after edge preprocessing; The strategy execution module is used to perform protection strategy optimization within the released resources based on the fault feature vector, generate reclosing parameters and send them to the target circuit breaker terminal through a high-priority communication channel for execution.

[0006] Furthermore, the cloud platform monitors the resource load status of each computing node in real time, including: Get the curve of the CPU usage of the computing node over time; Get a curve showing the memory usage of a computing node over time; Calculate the percentage of cumulative time that exceeds a preset CPU usage threshold based on the CPU usage curve; Calculate the maximum duration of memory usage exceeding the preset threshold based on the memory usage curve; The cumulative duration ratio and the maximum duration are combined to form a quantitative indicator of resource load status.

[0007] Furthermore, in response to a fault diagnosis request, a dynamic preemption priority is generated based on the grid topology vulnerability of the protection area where the target circuit breaker is located and the degree of coordinated action required of the associated circuit breakers, including: After responding to the fault diagnosis request, the grid topology vulnerability value is multiplied by the collaborative action demand value, and then multiplied by the preset priority scaling factor to generate the dynamic preemption priority.

[0008] Furthermore, the numerical method for obtaining the power grid topology vulnerability is: Obtain the real-time topological connection relationship of the protection area where the target circuit breaker is located; Perform N-1 security checks on real-time topology connections, and use the percentage of branches that fail the check as the first factor. Calculate the ratio of the critical load power to the total load power in the protection area as the second factor; Multiplying the first factor by the second factor yields the grid topology vulnerability value.

[0009] Furthermore, the method for obtaining the collaborative action requirement value is as follows: Measure the actual operation timing margin of the associated circuit breakers; Read the preset standard time difference; The ratio of the absolute difference between the actual action timing margin and the standard coordination time difference to the standard coordination time difference is calculated as the collaborative action requirement value.

[0010] Furthermore, the waveform distortion rate of the fault current is extracted. If it does not fall within the threshold range of the historical fault library, the dynamic preemption priority is subjected to credibility attenuation processing, including: Extracting the fault current and performing discrete sampling on the fault current waveform to obtain a current sequence; Calculate the root mean square error between the current sequence and the standard sine wave as the waveform distortion rate; Read the upper and lower threshold values ​​of the waveform distortion rate for each type of fault in the historical fault database; Determine whether the waveform distortion rate is greater than the waveform distortion rate threshold lower limit and less than the waveform distortion rate threshold upper limit at the same time; When the waveform distortion rate is less than the lower threshold value of the waveform distortion rate or greater than the upper threshold value of the waveform distortion rate, the dynamic preemption priority is multiplied by the preset attenuation coefficient; When the waveform distortion rate is greater than the waveform distortion rate lower threshold and less than the waveform distortion rate upper threshold at the same time, the dynamic preemption priority remains unchanged.

[0011] Furthermore, the necessity of resource preemption is determined based on the resource load status and the processed dynamic preemption priority. If the resource load exceeds the resource preemption threshold, the periodic task thread is frozen, computing resources are released, and a high-priority communication channel is opened, including: Mapping the processed dynamic preemption priority to the resource preemption demand value; Determine the current resource tension coefficient based on resource load status; Multiply the resource preemption demand value by the resource tension coefficient to obtain the quantitative value of resource preemption necessity; Comparing the resource preemption necessity quantified value with the preset resource preemption threshold; When the resource preemption necessity quantified value is greater than the resource preemption threshold, a thread freeze instruction is sent to the operating system to suspend the periodic task thread; Reclaim computing resources occupied by suspended threads; Creates a high-priority communication channel independent of regular channels in the communication protocol stack.

[0012] Furthermore, a fault recording compression instruction is sent to the target circuit breaker terminal through a high-priority communication channel, and the fault feature vector after edge preprocessing is received, including: Transmitting the fault recording compression instruction carrying the compression format identifier to the target circuit breaker terminal through the established high-priority communication channel; The target circuit breaker terminal performs format conversion processing on the original fault recording data according to the compression format identifier; The target circuit breaker terminal performs feature extraction calculation on the fault recording data after format conversion to generate a fault feature vector; The target circuit breaker terminal encapsulates the fault feature vector into a data message; receiving a data message carrying a fault feature vector returned by a target circuit breaker terminal through a high-priority communication channel; Parse the data packet to extract the fault feature vector after edge preprocessing.

[0013] Furthermore, based on the fault feature vector, protection strategy optimization is performed within the released resources, reclosing parameters are generated and sent to the target circuit breaker terminal through a high-priority communication channel for execution, including: Input the fault feature vector into the preset protection action rule library for pattern matching; When the match is successful, the corresponding reclosing parameters are directly output; When the matching fails, the transient stability simulation calculation module is started; Establish an equivalent circuit model of the protection area where the target circuit breaker is located in the transient stability simulation calculation module; Inject the current and voltage corresponding to the fault characteristic vector into the equivalent circuit model; Solve the dynamic response curve of the equivalent circuit model within a preset time window; Calculate the optimal reclosing delay parameters based on the zero-crossing time difference of the dynamic response curve; Encapsulate reclosing parameters or optimal reclosing delay parameters as control instructions; Transmit control instructions to target circuit breaker terminals via high-priority communication channels; The target circuit breaker terminal interprets the control command and performs the reclosing operation.

[0014] Compared with the prior art, the present invention has the following beneficial effects: 1. Through dynamic resource preemption and edge collaborative processing, the timeliness and reliability of cloud protection strategy generation are both improved. Different from the traditional cloud system's passive waiting for resources processing mode, the system introduces grid topology vulnerability assessment and circuit breaker collaborative demand analysis to generate dynamic preemption priority, so that high-value fault diagnosis tasks can immediately trigger resource reorganization mechanisms. Combined with the waveform distortion rate-driven credibility attenuation algorithm, it effectively filters out the ineffective occupation of resources by suspicious fault information. When the necessity of resource preemption reaches the threshold, the system actively freezes periodic tasks and releases computing resources, and simultaneously opens an independent communication channel to ensure full-link resource guarantee from fault feature extraction to policy issuance, eliminating the risk of protection degradation caused by diagnostic delays.

[0015] 2. While ensuring real-time performance, the adaptability of the protection strategy is significantly optimized. Fault recording compression and feature preprocessing are performed by the target circuit breaker terminal, which greatly reduces the cloud computing load, allowing the released resources to deeply execute complex calculations such as transient stability simulation. The dual decision-making mechanism based on the protection action rule library and dynamic response curve not only uses historical fault modes to achieve fast parameter matching, but also accurately calculates the optimal reclosing delay through equivalent circuit simulation. The high-priority communication channel ensures that the optimized parameters are quickly issued, so that the circuit breaker can perform adaptive reclosing within the first voltage cycle after the fault is cleared, effectively suppressing the risk of false operation of the traditional fixed parameter mode in complex fault scenarios, and improving the reliability of distribution network fault recovery. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a structural diagram of a modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to the present invention; Figure 2 This is a flow chart of the present invention for performing credibility attenuation processing on dynamic preemption priority. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0018] Example: Figure 1 The present invention provides a structural diagram of a modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration, which includes: The load monitoring module is used by the cloud platform to monitor the resource load status of each computing node in real time; A priority generation module is used to respond to fault diagnosis requests and generate dynamic preemption priorities based on the grid topology vulnerability of the protection area where the target circuit breaker is located and the degree of coordinated action requirements of the associated circuit breakers; Credibility attenuation module, used to extract the waveform distortion rate of the fault current. If it does not fall within the threshold range of the historical fault library, the credibility of the dynamic preemption priority is attenuated. The resource preemption module is used to determine the necessity of resource preemption based on the resource load status and the processed dynamic preemption priority. If it exceeds the resource preemption threshold, the periodic task thread is frozen, computing resources are released, and a high-priority communication channel is opened; A feature extraction module is used to send a fault recording compression instruction to the target circuit breaker terminal through a high-priority communication channel and receive the fault feature vector after edge preprocessing; The strategy execution module is used to perform protection strategy optimization within the released resources based on the fault feature vector, generate reclosing parameters and send them to the target circuit breaker terminal through a high-priority communication channel for execution.

[0019] The cloud platform collects real-time operational data from each compute node through the performance monitoring interface provided by the operating system kernel. To monitor CPU utilization, the operating system's process management API is called to obtain the CPU utilization percentage for all logical cores at a fixed sampling interval. The sampling interval is set to, for example, 100 milliseconds, and the timestamp and corresponding CPU utilization value are recorded for each sampling. This continuous collection operation generates a curve showing CPU utilization over time. This curve is composed of a chronological sequence of timestamps and corresponding utilization values.

[0020] To monitor memory utilization, the real-time physical memory usage of the compute node is read through the operating system's memory management API. For each sampling period, the ratio of used memory to total memory capacity is calculated and converted into a percentage. The sampling interval is also set to, for example, 100 milliseconds. During continuous data collection, the timestamps and corresponding memory utilization values ​​are synchronously recorded, forming a curve showing memory utilization over time. This curve consists of a sequence of timestamps and corresponding memory utilization values.

[0021] When calculating the cumulative duration ratio based on the curve of the CPU occupancy rate changing over time, the following operations are performed: the CPU occupancy rate threshold is set to, for example, 80%, and the threshold is determined based on the stress test results of the computing node in the power protection business scenario. The specific determination method is: when simulating the peak load of power fault diagnosis, when the CPU occupancy rate exceeds the threshold for a long time, the system response delay exceeds the acceptable range. Traverse all sampling points of the CPU occupancy rate curve and identify sampling points where the occupancy rate value exceeds the set threshold. Count the time periods that continuously exceed the threshold, and the overlimit duration of each time period is the number of sampling points in the time period multiplied by the sampling interval time. The overlimit duration of all time periods is accumulated to obtain the total overlimit duration. The cumulative duration ratio is the total overlimit duration divided by the preset time window length. The time window length is set to, for example, 5 seconds, which corresponds to the typical time requirement for clearing high-voltage power grid faults.

[0022] To calculate the maximum duration based on a time-varying memory usage curve, perform the following operations: Set a memory usage threshold, for example, to 75%, based on the compute node's memory resource reservation policy. Traverse the sequence of sampling points in the memory usage curve and identify consecutive periods in which the usage value consistently exceeds the threshold. Calculate the duration of each consecutive period, where the duration is the difference between the end and start timestamps of the period. Compare the durations of all consecutive periods and record the maximum value. This maximum value represents the maximum duration in which the memory usage consistently exceeds the preset threshold.

[0023] When the cumulative duration ratio and the maximum duration are combined to form a resource load state quantitative index, a piecewise function is used for fusion processing. When the cumulative duration ratio of the CPU occupancy curve is less than, for example, 30% and the maximum duration of the memory occupancy curve is less than, for example, 2 seconds, the resource load state quantitative index is equal to the cumulative duration ratio multiplied by, for example, 0.5 plus the maximum duration multiplied by, for example, 0.5. When the cumulative duration ratio is greater than or equal to, for example, 30% or the maximum duration is greater than or equal to, for example, 2 seconds, the resource load state quantitative index is equal to the cumulative duration ratio multiplied by, for example, 0.7 plus the maximum duration multiplied by, for example, 0.3. The weight distribution of the piecewise function is determined based on the resource scheduling simulation test results: in the high-load scenario of the CPU, its weight coefficient is increased to reflect its greater impact on real-time performance. The value of the resource load state quantitative index is standardized to an integer interval of zero to one hundred through a linear mapping function.

[0024] The CPU occupancy threshold is set based on the following specific test process: in a simulated line short-circuit fault scenario, the diagnostic task load is gradually increased until the CPU occupancy reaches different levels, the task response delay change curve is recorded, and the occupancy corresponding to the point where the response delay suddenly increases is selected as the threshold. The memory occupancy threshold is set based on the following: for example, one-quarter of the total memory is reserved for emergency cache to avoid performance degradation caused by memory swapping. The weight coefficient in the piecewise function is determined through resource scheduling simulation tests: different load combination scenarios are constructed, and parameter optimization is performed with the accuracy of resource preemption decisions as the optimization goal. The conversion method of the linear mapping function is: multiply the original calculation result by one hundred and round it up.

[0025] The calculation process for quantitative resource load indicators is implemented by an independent monitoring agent. This program, deployed on each compute node, collects and calculates data at a fixed interval and writes the results to a shared memory area. The monitoring agent automatically reduces the sampling frequency when the CPU is under high load. For example, when the CPU utilization exceeds 90%, the sampling interval is adjusted to 200 milliseconds. Memory utilization monitoring only counts the physical memory directly used by applications and ignores the operating system cache memory.

[0026] The CPU usage curve's sampling points are stored using a ring buffer structure, with a buffer capacity set to, for example, 1000 sampling points. The memory usage curve's buffer capacity is set to, for example, 500 sampling points. Over-limit period identification uses a state machine algorithm: Initially, the state is within the limit. When a sampling point exceeds the threshold, the period begins and the state switches until a sampling point below the threshold is detected, marking the end of the period. Maximum duration calculation uses a dynamic update mechanism: When a new continuous period is detected, it is immediately compared and updated with the current maximum value.

[0027] The calculation of resource load status quantitative indicators includes an exception handling mechanism: if the cumulative duration exceeds 100% or the maximum duration exceeds the time window length, the indicator value is directly set to the maximum value of 100. Before the indicator is output, smoothing filtering is performed: the arithmetic mean of the three most recent calculation results is used as the final output value. The filter window size is set based on the inertial characteristics of load changes. When the monitoring agent starts, it performs a self-calibration process: 100 consecutive no-load data collections are performed to determine the baseline offset. This offset is deducted from all subsequent data collections to eliminate system errors.

[0028] Data reads from the shared memory area use atomic operations to ensure consistency. The resource scheduling module accesses metric values ​​through memory mapping, verifying the data checksum with each read to prevent transmission errors. Three consecutive read failures trigger an alarm and switch to a backup compute node. The monitoring agent regularly outputs resource load logs containing raw sampled data and intermediate values ​​during the calculation process, which are used for offline analysis to verify metric accuracy. Log files are stored in a rolling manner, with a maximum file size of, for example, 10 megabytes.

[0029] When responding to a fault diagnosis request, the system retrieves the real-time topological connectivity of the target circuit breaker's protection zone from the grid energy management system. This real-time topological connectivity is represented by a node-branch model and includes the connection status and switch positions of electrical equipment such as busbars, transformers, and transmission lines. This acquisition is accomplished using a common power system communication protocol, for example, synchronizing topology change information once per second. The protection zone boundary is determined by the electrical distance upstream and downstream of the target circuit breaker, for example, including the busbar directly connected to the target circuit breaker and two adjacent layers of equipment.

[0030] When performing an N-1 safety check on the real-time topological connection relationship, the system traverses all branches within the protection area, disconnecting only one branch at a time to simulate a fault. AC power flow calculations are performed for each disconnection condition, using the Newton-Raphson method for iterative solution. The convergence conditions are that the node power imbalance is less than 0.01 MW and the maximum number of iterations is 20. The verification standard is: after disconnecting any branch, the power of the remaining branches does not exceed 90% of the rated transmission capacity, and the node voltage deviation does not exceed 5% of the nominal voltage. The number of branches that failed the verification is counted, and their proportion to the total number of branches is calculated as the first factor. For example, if the protection area contains 20 branches and an over-limit occurs when disconnecting 4 of them, the first factor is 20%. The line impedance matrix and transformer ratio data required for the power flow calculation are obtained in real time from the power grid parameter database.

[0031] When calculating the ratio of critical load power to total load power within a protected area, critical loads are defined as loads where a power outage would result in significant safety or economic losses. The system identifies critical loads from a library of load attribute tags preset by the grid dispatching system. These tags include three categories: "medical emergency load," "national security load," and "continuous production load." The total active power value of the protected area is obtained in real time, while the total active power of the critical loads is read through the smart meter using the DLMS / COSEM protocol. The second factor is the percentage of the total active power of the critical loads divided by the total active power of all loads. The data collection period is set to, for example, 200 milliseconds.

[0032] When multiplying the first and second factors to obtain the grid topology vulnerability value, the system performs numerical normalization: both the first and second factors are expressed as percentages, and the product is divided by 10,000 to convert it to a real number between 0 and 1. When the protected area includes a hub substation, the first factor is multiplied by a factor of 1.5; when it includes primary loads, the second factor is multiplied by a factor of 1.2. The upper limit of the corrected value is capped at 1.0.

[0033] When measuring the actual timing margin of associated circuit breakers, the system collects the timestamps of their actuation events using synchronized phasor measurement devices. The actual timing margin is the millisecond difference between the time the trip command is issued and the time fault detection is initiated. PTP is used for clock synchronization to ensure an error of less than 1 microsecond. If there is no historical actuation record for an associated circuit breaker, the average actuation time for similar faults over the past three months is used as an estimate.

[0034] When reading the preset standard coordination time difference, the system retrieves the parameters from the relay protection setting list database by device number index. For ring network structures, the minimum coordination time difference principle is used. For example, the default value for 110 kV equipment is 250 milliseconds, and for 220 kV equipment, it is 300 milliseconds. If no record is found in the database, the default rule for voltage level is generated.

[0035] To calculate the coordination requirement, the system performs three steps: calculating the absolute difference between the actual timing margin and the standard coordination time difference; dividing this difference by the standard coordination time difference; and converting it to a percentage. For example, if the standard coordination time difference is 300 milliseconds and the actual timing margin is 250 milliseconds, the absolute difference is 50 milliseconds, and the ratio is 16.67%.

[0036] The dynamic preemption priority is generated by multiplying the grid topology vulnerability value by the coordination action requirement value (converted to a real number between 0 and 1), and then multiplying it by the preset priority scaling factor. For example, a grid topology vulnerability value of 0.12 and a coordination action requirement value of 0.1667 multiply by the value of 0.02, which, when multiplied by the scaling factor of 100, results in a priority of 2. The preset priority scaling factor starts at 100 and is dynamically adjusted based on the resource preemption success rate over the last hour. When the success rate is below 90%, the scaling factor increases by 5 steps; when it is above 95% and the system load is below 70%, the scaling factor decreases by 5 steps. The adjustment range is limited to 50 to 150.

[0037] The dynamic preemption priority generation process implements boundary protection: calculated results less than 1 are forcibly set to 1, and those greater than 100 are forcibly set to 100. Median filtering is performed before output, taking the median of the current result and the two previous values. The calculation timeout threshold is set to 50 milliseconds; if a timeout occurs, the previous valid value is output.

[0038] The system records a complete calculation log, including timestamp, protection zone number, primary factor, secondary factor, grid topology vulnerability value, associated circuit breaker number, actual action timing margin, standard coordination time difference, coordinated action requirement value, preset priority scaling factor, and final priority value. Log entries are stored in chronological order and can be retrieved by fault type for decision effectiveness analysis.

[0039] Critical loads are defined based on the terms of the power supply contract, and hub substations are determined based on historical grid operational data. During N-1 safety verification, when the AC power flow calculation times out or oscillates, the algorithm automatically switches to a DC power flow approximation algorithm. Timing margin measurement data is stored using a ring buffer structure, retaining the last 10 records for data quality verification. Coordinated action requirement calculation supports multi-level coordination. When multiple associated circuit breakers are involved, the parameters of the nearest parent circuit breaker are used as the benchmark.

[0040] Figure 2 A flowchart of the present invention for performing credibility attenuation processing on the dynamic preemption priority is provided, and the original waveform data of the fault current of the target circuit breaker is obtained from the fault recording device. The fault current waveform is collected by a current transformer and discretely sampled at a fixed sampling rate by an analog-to-digital converter. The sampling rate is set to, for example, 128 points per cycle based on the transient process characteristics of the power system, corresponding to a sampling interval of 156.25 microseconds for a 50 Hz system. The discrete sampling process continuously records the complete data of three cycles after the fault occurs, generating a current sequence including a timestamp sequence and a current amplitude sequence. The current amplitude unit is uniformly converted to kiloamperes, and the conversion coefficient is determined based on the current transformer ratio parameter.

[0041] When calculating the RMS error (RMS) between the current sequence and a standard sine wave, the standard sine wave is generated according to the following rules: the frequency is consistent with the rated system frequency, for example, 50 Hz; the amplitude is the peak RMS current value in the half-cycle before the fault occurs; and the phase angle is synchronized with the voltage zero crossing at the time of the fault. The RMS error is calculated using a discrete point-to-point comparison method: for each sampling point in the current sequence, the square of the difference between the current sequence and the standard sine wave value at the corresponding time is calculated; the squared differences across all sampling points are accumulated; the accumulated value is divided by the total number of sampling points and the square root is taken. The result is the waveform distortion rate (WDR) expressed in kiloamperes.

[0042] When reading waveform distortion rate threshold intervals from the historical fault database, the database uses a relational database storage structure. Each fault record contains a fault type code, a waveform distortion rate lower threshold value, and a waveform distortion rate upper threshold value. Fault type codes are categorized according to common fault classification standards. For example, a single-phase ground fault is coded as TP1, and a phase-to-phase short circuit fault is coded as TP2. The lower and upper threshold values ​​are determined by statistically analyzing historical data: for each fault type, typical event cases are selected and their waveform distortion rate quantiles are calculated as thresholds. Database queries use the current fault type code as the index field.

[0043] To determine whether the waveform distortion rate is simultaneously greater than the lower threshold and less than the upper threshold, a numerical comparison logic operation is performed. This comparison uses double-precision floating-point calculations, with accuracy retained to four decimal places. For example, if the waveform distortion rate value is 2.354 kA, and the lower threshold for the single-phase grounding fault type is 1.200 kA and the upper threshold is 3.500 kA, the range condition is met.

[0044] When the waveform distortion rate is less than the lower threshold or greater than the upper threshold, the dynamic preemption priority is multiplied by a preset attenuation factor. The preset attenuation factor is initially set to, for example, 0.7, which is determined based on historical data analysis. Multiplication is performed using floating-point arithmetic, and the result is rounded to two decimal places. For example, the original priority of 85.00 is multiplied by 0.7 to obtain 59.50.

[0045] If the waveform distortion rate is both greater than the lower threshold and less than the upper threshold, the dynamic preemption priority remains unchanged. No value modification is performed on this path, and the original priority is directly passed to subsequent processing.

[0046] The fault current acquisition process includes data preprocessing: the raw sampled values ​​are digitally filtered to eliminate DC components using a high-pass filter with a cutoff frequency of 0.5 Hz. Outliers are removed at each sampling point. If the current sudden change rate exceeds, for example, 10 kiloamperes per millisecond, it is replaced by linear interpolation of the previous two sampling points. After the waveform distortion rate is calculated, the current fault waveform and calculation results are automatically stored in a historical fault database for 24 hours.

[0047] The threshold update mechanism for the historical fault database involves regularly analyzing new fault case data and recalculating the waveform distortion rate quantiles for each type of fault. When the number of new cases for a particular fault type exceeds, for example, 20, a threshold recalculation process is triggered. Before the new threshold takes effect, it must pass a review process. If the review fails, the original threshold remains.

[0048] The preset attenuation coefficient is dynamically adjusted by regularly evaluating the effectiveness of the credibility attenuation decision and statistically analyzing the change in fault diagnosis accuracy after priority attenuation. If the accuracy drops by more than 5%, the attenuation coefficient is increased by 0.1; if the accuracy increases by more than 5%, the attenuation coefficient is decreased by 0.1. The adjustment range is limited to 0.5 to 0.9.

[0049] Temperature compensation is incorporated into waveform distortion rate calculations: The sampled value is proportionally corrected based on the reading of the current transformer's ambient temperature sensor. Temperature sensor data is acquired in real time via a communication protocol, and the compensation coefficient is stored in the device calibration parameter table.

[0050] The phase synchronization method for the current sequence and the standard sine wave is to extract the zero-crossing point of the voltage waveform 1 millisecond before the fault occurs as the phase reference. If the voltage channel fails, the full-second pulse of the system timing clock is used as the synchronization signal. If the synchronization error exceeds 200 microseconds, the backup synchronization method is activated.

[0051] The RMS error calculation uses a segmented processing mechanism: the current sequence is divided into multiple subsegments by cycle. The distortion factor is calculated independently for each subsegment and then averaged. This process suppresses localized distortion interference caused by transient processes. The subsegments are automatically marked based on voltage zero crossings, and each cycle contains, for example, 128 sampling points.

[0052] Dynamically preempted priority attenuation operations are logged, including the fault waveform characteristic value, waveform distortion rate, lower threshold value, upper threshold value, original priority, attenuation coefficient, and new priority value. Log entries are linked to the fault recording file number, supporting waveform data retrospective review.

[0053] Special handling when the waveform distortion rate exceeds the threshold: If multiple consecutive faults exceed the threshold, a transformer calibration alarm is triggered. A device calibration work order is automatically generated, including the percentage deviation of the current waveform distortion rate from the historical average. The calibration cycle is dynamically adjusted based on the equipment's operating life. For example, new equipment will be calibrated once a year after its first year of operation, and every six months after five years of operation.

[0054] Version management for preset attenuation coefficients: A new version number is generated after each coefficient adjustment, and the old coefficient value is retained in the historical parameter library. Rollback is supported, allowing you to revert to the previous version if the new coefficient causes a diagnostic error. Version change records include the modification time, adjustment basis, and expected impact.

[0055] Fault current waveforms are stored using a lossy compression algorithm, with a compression ratio of, for example, 4 to 1. A checksum is added to the compressed data, and if the checksum fails, the original data is automatically retransmitted. Waveform files are stored by device partition, with a maximum file size limit of 10 megabytes.

[0056] Standard sine wave amplitude calibration method for RMS error calculation: If the first half cycle before a fault occurs is invalid, use 1.5 times the rated current peak value as the default amplitude. If phase synchronization fails, use the fault onset as the zero-phase reference. If sub-segment division occurs across cycles, adjust the calculation weight proportionally based on the actual number of sampling points.

[0057] Verify the validity of the waveform distortion rate results: When the calculated result is greater than 20 kiloamperes, the review process is initiated. The review uses a dual parallel algorithm: the primary algorithm is the root mean square error method, and the secondary algorithm is the total harmonic distortion method. If the difference between the two exceeds 10%, the average value is taken as the final result.

[0058] The historical fault database's case selection criteria only includes cases within the transformer verification period and with a fault resistance less than 10 ohms. Waveform quality testing is performed before new cases are added to the database, and cases with a signal-to-noise ratio below 40 decibels are automatically removed. Threshold recalculation uses a rolling window mechanism, retaining the most recent 100 valid records at a time.

[0059] The attenuation coefficient evaluation cycle is set at 30 days, and the evaluation data is derived from the most recent 100 fault diagnosis records. Accuracy statistics exclude invalid cases caused by communication interruptions. The coefficient adjustment operation is automatically performed during system idle time, and a data snapshot is created before execution for rollback purposes.

[0060] Temperature compensation coefficient calibration method: Set 11 calibration points within the range of -10°C to +60°C. Collect 100 sets of data at each calibration point to fit a linear compensation curve. The compensation formula is: Compensated current value = original sample value × [1 + 0.001 × (25 - current temperature)].

[0061] Calibration alarms are triggered by three consecutive threshold violations or a single deviation exceeding 50% of the historical average. Work orders are issued within two hours on weekdays and on the next business day on non-business days. Calibration result feedback requires that new parameters be uploaded within 24 hours after calibration, otherwise an upgrade notification will be triggered.

[0062] The mapping of the processed dynamic preemption priority to the resource preemption demand value is achieved using piecewise linear interpolation. The processed dynamic preemption priority input range is an integer from 1 to 100, and the resource preemption demand value output range is an integer from 0 to 100. When the processed dynamic preemption priority is less than or equal to 30, the resource preemption demand value is equal to the processed dynamic preemption priority multiplied by 0.5. When the processed dynamic preemption priority is between 31 and 70, the resource preemption demand value is equal to 30 × 0.5 + (processed dynamic preemption priority - 30) × 1.0. When the processed dynamic preemption priority is greater than 70, the resource preemption demand value is equal to 30 × 0.5 + 40 × 1.0 + (processed dynamic preemption priority - 70) × 1.5. This mapping relationship is determined based on historical system load stress test results: in simulated high-priority task scenarios, when the processed dynamic preemption priority exceeds 70, the resource demand shows a nonlinear growth characteristic.

[0063] When determining the current resource strain coefficient based on resource load, the resource load is represented by a resource strain index. The current resource strain coefficient is calculated as: Current resource strain coefficient = Resource strain index divided by 100. The resource strain index obtains its latest value in real time from the resource monitoring process via shared memory. When the resource strain index exceeds 95, the current resource strain coefficient is forcibly set to 1.0. This process prevents calculation anomalies under extreme load conditions.

[0064] When multiplying the resource preemption demand value by the current resource strain coefficient to obtain the quantified value of resource preemption necessity, a floating-point multiplication operation is performed. The result is rounded to two decimal places. For example, a resource preemption demand value of 80.00 multiplied by the current resource strain coefficient of 0.85 yields 68.00. Set bounds on the calculation result: if the product is greater than 100.00, 100.00 is output; if it is less than 0.00, 0.00 is output. Boundary protection prevents subsequent comparison operations from failing.

[0065] When comparing the resource preemption necessity quantification value with the preset resource preemption threshold, the preset resource preemption threshold is initially set to 50.00. This threshold setting is based on system performance test data: when the resource preemption necessity quantification value is greater than 50, the benefit of accelerated fault diagnosis due to resource preemption outweighs the loss of periodic task suspension. The comparison operation uses IEEE 754 standard floating-point comparison instructions, with an error tolerance of 0.001.

[0066] When the quantified value of resource preemption necessity exceeds the preset resource preemption threshold, a thread freeze instruction is sent to the operating system kernel. This thread freeze instruction is implemented through the operating system application program interface (API) and targets periodic task threads. Periodic task threads are background tasks with fixed execution cycles, such as a data backup thread with a 5-minute cycle or a log archiving thread with a 1-hour cycle. The freezing target selection rule is to select the top three periodic task threads started within the last 5 minutes and ranked by CPU utilization, excluding pre-marked critical system threads.

[0067] When reclaiming computing resources occupied by a suspended thread, memory and processor core release operations are performed. Memory reclaim is implemented through the operating system's memory management interface, marking the suspended thread's heap and stack memory as reclaimable. Processor core release is accomplished through the scheduler interface, unbinding the suspended thread from the processor core. Resource reclaimed amounts are recorded in real time: memory reclaimed amounts are accurate to the megabyte, and core reclaimed numbers are accurate to the unit digit.

[0068] When creating a high-priority communication channel in the communication protocol stack, independent of regular channels, network resource configuration is performed. A dedicated port number range, such as 60000 to 60010, is reserved in the transport layer protocol stack. A User Datagram Protocol socket is created and the Type of Service field is set to the highest priority. The channel establishment process includes a bandwidth guarantee mechanism: a fixed percentage of network bandwidth, such as 30% of the total bandwidth, is reserved for the high-priority communication channel.

[0069] Resource preemption demand value mapping table maintenance mechanism: Monthly statistical analysis of resource preemption success rates is performed. When the success rate is below 85%, the slope of the second segment of the mapping curve is increased by 0.1. When the success rate is above 95%, the slope is decreased by 0.1. The slope adjustment range is limited to 0.3 to 2.0. Updates are performed during system idle time. Version change records include the modification time, original slope, new slope, and adjustment rationale.

[0070] Periodic task thread freeze exception handling: If the operating system returns a freeze failure, the next candidate thread is tried. After three consecutive failures, failover is triggered, switching to a backup compute node and performing the same operation. If the freeze is successful, a resource release callback function is registered, which automatically wakes up the frozen thread after the failure handling is completed.

[0071] Memory reclamation triggers defragmentation: When the amount of memory reclaimed in a single operation exceeds 50 megabytes, memory compression is automatically invoked. Processor core release uses a load balancing strategy, prioritizing cores bound to non-real-time tasks. Resource reclamation reports record items including timestamp, amount of reclaimed memory, list of released core numbers, and list of frozen thread identifiers.

[0072] The high-priority communication channel establishment process includes a fault-tolerant mechanism: if the preferred port is occupied, subsequent ports are automatically tried in the port sequence. Channel parameters are configured as follows: the lifetime value is set to 1, and the window size is fixed at 8192 bytes. Channel health monitoring: Heartbeat packets are sent every 100 milliseconds, and three consecutive losses trigger channel reestablishment.

[0073] Resource stress coefficient calculation adds a trend factor: When the resource load status quantitative indicator shows an upward trend for five consecutive samplings, multiply the current resource stress coefficient by 1.2. Trend determination uses linear regression to calculate the slope of the five most recent sampling points. An upward trend is determined when the slope is greater than 0.5.

[0074] The resource preemption necessity quantification value is decayed over time: at every hour, the current value is multiplied by 0.95. If the decayed value is less than 10.00, it returns to zero. The decay coefficient is dynamically adjusted based on the system load factor: when the resource load status quantification index is less than 30, the decay coefficient is increased to 0.90.

[0075] Flow control on high-priority communication channels uses a token bucket algorithm: the token bucket capacity is set to 1,048,576 bytes, and the token generation rate is set to 3,145,728 bits / second. Data transmission is subject to burst limits: the amount of data sent in a single transaction does not exceed 80% of the bucket capacity. Channel usage metrics are monitored in real time: the number of bytes sent per second is counted, and rate limiting is initiated when the threshold is exceeded.

[0076] The reallocation strategy for reclaimed resources: Freed memory is preferentially allocated to the data buffers of fault diagnosis tasks, and released processor cores are bound to high-priority fault calculation threads. Reallocation is performed within 50 milliseconds after resource reclaim is completed, and the allocation results are written to the resource allocation registration table.

[0077] Communication channel security: Bidirectional X.509 certificate authentication is performed upon channel establishment, with a validity period of 30 days. Data transmission is encrypted using 256-bit Advanced Encryption Standard. Key management: New keys are generated hourly, and seamless updates are achieved through Elliptic Curve Diffie-Hellman key exchange.

[0078] Resource analysis of frozen threads: This tool uses the operating system's performance interface to obtain detailed thread resource usage, including thread identifier, memory usage (in megabytes), processor cycle consumption, and the last execution timestamp. This analysis report serves as a basis for resource recovery.

[0079] Quality of service assurance for high-priority communication channels: Configurable latency caps of 10 milliseconds, jitter caps of 2 milliseconds, and packet loss caps of 0.1%. Routing priority is achieved through the Differentiated Services Code Point (DSCP) field. Channel performance monitoring: End-to-end latency is measured every 5 seconds, and channel optimization is triggered if it exceeds 10 milliseconds three times in a row.

[0080] Resource preemption decision-making process audit: This log records the timestamp, processed dynamic preemption priority, resource preemption demand value, resource load status quantitative indicators, current resource intensity coefficient, resource preemption necessity quantitative value, preset resource preemption threshold, decision result, frozen thread list, reclaimed resource amount, and channel port number. Audit logs are stored with AES encryption and retained for 90 days.

[0081] To calibrate the preset resource preemption threshold, perform stress testing quarterly, gradually increasing the simulated task load. Record the relationship between the quantified value of resource preemption necessity and system response, and use the response latency inflection point as the new threshold. The test report includes the test time, number of simulated tasks, response latency at each load point, and the calculation process for the inflection point.

[0082] A fault recording compression command is sent to the target circuit breaker terminal via an established high-priority communication channel. This command includes a compression format identifier field. The compression format identifier field is an 8-bit binary code, with the first three bits indicating the required sampling rate, the middle three bits indicating the compression algorithm type, and the last two bits indicating the checksum. For example, the code "00101011" indicates a sampling rate of 128 points per cycle, a discrete cosine transform compression algorithm, and a cyclic redundancy check. Command transmission utilizes the User Datagram Protocol (UDP). The quality of service (QoS) flag in the packet header is set to the highest priority, the timeout retransmission interval is set to 50 milliseconds, and the maximum number of retries is three. The encoding rules for the compression format identifier are generated based on the device capability configuration table, which is stored in the non-volatile memory of the target circuit breaker terminal.

[0083] After receiving the command, the target circuit breaker terminal parses the compression format identification field and performs format conversion on the original fault recording data based on the identification field. The original fault recording data is a sequence of continuous sample values ​​collected by the current transformer, with a sampling interval of 156.25 microseconds. The format conversion process includes: resampling according to the sampling rate requirements of the compression format identification, and applying an anti-aliasing filter with a cutoff frequency of 40 Hz when the sampling rate is required to be reduced; initializing the compression parameters according to the compression algorithm type, such as setting the retention coefficient of discrete cosine transform compression to 0.8, which is determined based on the typical fault waveform reconstruction error test; and adding a check data block, such as a cyclic redundancy check block length of 32 bits. The processed data is converted to a standard compression frame structure, and the frame header contains a timestamp and device identifier. When the command identification does not match the device configuration, the default configuration is automatically enabled: sampling rate 64 points / cycle, stroke length compression algorithm, and parity check.

[0084] The target circuit breaker terminal performs feature extraction on the converted fault recording data to generate a fault feature vector. This feature extraction process involves three steps: 1. The RMS value of the fundamental component of the fault current is calculated using a full-cycle Fourier transform algorithm with a fixed number of 128 points; 2. The percentage of the second harmonic content is extracted using a 1024-point Hanning windowed Fast Fourier Transform (FFT); and 3. The waveform asymmetry is calculated by comparing the difference between the positive and negative half-cycle integrated areas using the formula: |Positive half-cycle area - Negative half-cycle area| ÷ (positive half-cycle area + negative half-cycle area). The fault feature vector is a three-dimensional floating-point array stored in the order [fundamental RMS value (kiloamperes), harmonic content (percentage), asymmetry coefficient]. For example, a fault feature vector is [3.25, 15.7, 0.12]. Exception handling: When data length is less than a full cycle, zeros are automatically padded to a full cycle. When the FFT spectrum leakage exceeds 10%, the waveform is recalculated using the Multiple Signal Classification algorithm.

[0085] The target circuit breaker terminal encapsulates the fault signature vector into a data message. The encapsulation process involves the following steps: adding a message header containing a 4-byte sequence number and an 8-byte Precision Time Protocol timestamp; converting each floating-point number in the signature vector to a 4-byte single-precision format according to the IEEE 754 standard; appending a 4-byte cyclic redundancy check (CRC) code using the generator polynomial 0x04C11DB7; and adding a 2-byte length field to indicate the payload size. The message structure is defined as: 0xAA55 start delimiter (2 bytes) + sequence number (4 bytes) + timestamp (8 bytes) + signature vector data (12 bytes) + CRC code (4 bytes) + 0x55AA end delimiter (2 bytes). After encapsulation, the message is stored in a transmit buffer with a capacity of 10 messages.

[0086] Data packets returned from the target circuit breaker terminal are received via a high-priority communication channel. Flow control is implemented during the reception process: when network latency exceeds 10 milliseconds, the receive window size is adjusted to 4096 bytes. Message reassembly utilizes a sliding window protocol with a window size of 8 messages. Receive integrity checks include: verifying that the start character 0xAA55 matches the end character 0x55AA; calculating the cyclic redundancy check code and comparing it with the checksum within the message; and checking sequence number continuity. A retransmission request is triggered if the check fails, using an exponential backoff strategy: 50 milliseconds for the first retransmission, 100 milliseconds for the second, and 200 milliseconds for the third. After three consecutive failures, the communication channel switches to the backup channel.

[0087] Parse the data message to extract the fault feature vector after edge preprocessing. The parsing process is carried out in three steps: separating the message header and payload data; converting the 12-byte feature vector data into single-precision floating-point numbers in 4-byte groups; and performing numerical validity verification. The validity verification rules are as follows: the fundamental wave effective value range is limited to 0.1 kiloamperes to 50.0 kiloamperes (based on 1% to 200% of the equipment's rated current); the harmonic content percentage range is 0 to 100; and the asymmetry coefficient range is 0 to 1. When the value exceeds the limit, an error flag is added, and the error code is encoded according to the type of limit exceeded. After parsing, it is stored in the diagnostic database using incremental compression: when the rate of change of consecutive vectors is less than 5%, only the difference is stored.

[0088] Temperature compensation for feature extraction calculations: Based on the circuit breaker terminal temperature sensor readings, compensation is applied to the fundamental RMS value. The compensation formula is: Compensated value = Original value × [1 + 0.0005 × (25 - current temperature)]. The temperature data update frequency is 1 Hz, and the temperature sensor calibration cycle is 90 days. The compensation coefficient is determined based on the transformer temperature drift characteristic test data.

[0089] Fault-tolerance mechanism for data packet parsing: A 200-millisecond reassembly timer is started when packet fragmentation is detected; if the cyclic redundancy check fails but the length is valid, the Reed-Solomon error correction algorithm is attempted for recovery; if there are more than five sequence number breakpoints, a full synchronization request is triggered, with a full synchronization interval of at least 60 seconds.

[0090] Storage and management of fault feature vectors: The real-time area retains the latest 100 records and is stored in double-data-rate synchronous dynamic random access memory (SDRAM). The historical area is partitioned by hour and stored in flash memory, with a maximum capacity of 10 megabytes per partition. Retrieval uses a binary search method, with the timestamp as the index field.

[0091] Compression command timeout monitoring: A 500-millisecond response timer is initiated when a command is sent. If no response occurs within the timeout, the terminal is marked offline and a device status alarm is triggered. The terminal online rate is calculated hourly. If it falls below 90%, a device maintenance work order is generated with a response time limit of two hours.

[0092] Resource management for feature extraction: When processor utilization exceeds 80%, computational accuracy is reduced, for example, the number of Fast Fourier Transform points is reduced from 1024 to 512. Memory usage is capped at 10 megabytes; if exceeded, historical data is discarded on a first-in, first-out basis. Computational tasks are prioritized to real-time tasks.

[0093] Bandwidth guarantee for message reception: A fixed bandwidth quota of 50 kilobits per second is allocated to feature vector messages. A packet drop policy is activated in the event of network congestion, prioritizing heartbeat messages over historical data packets. Channel performance metrics are updated every second, including reception success rate, average latency, and packet loss rate.

[0094] Fault signature vector transmission encryption: The signature vector data is encrypted using the 128-bit Advanced Encryption Standard (AES) algorithm, with the encryption key updated hourly. The encryption operation adds processing latency of less than 1 millisecond, achieved through a hardware acceleration module. Key update instructions are issued through an independent control channel, which uses the elliptic curve digital signature algorithm for control channel authentication.

[0095] Eigenvector anomaly flagging rules: When the error flag indicates that the fundamental RMS value exceeds the limit, it is automatically replaced with the previous RMS value. When the harmonic content exceeds the limit, the original value is retained but marked as suspicious data. When the asymmetry coefficient exceeds the limit, the waveform review process is triggered. The review is performed by resampling the original waveform at a resampling rate of 256 points / cycle.

[0096] When the fault feature vector is input into the preset protection action rule library for pattern matching, the preset protection action rule library uses a relational database to store matching rules. Each rule contains a feature vector range definition and a corresponding set of reclosing parameters. The feature vector range definition includes six dimensions: fundamental RMS lower limit, fundamental RMS upper limit, harmonic content lower limit, harmonic content upper limit, asymmetry lower limit, and asymmetry upper limit. The pattern matching process performs the following operations: iterates through all rule entries and determines a match is successful when the fundamental RMS value of the fault feature vector is between the fundamental RMS lower limit and fundamental RMS upper limit of a rule, the harmonic content is between the harmonic content lower limit and harmonic content upper limit, and the asymmetry is between the asymmetry lower limit and asymmetry upper limit. Matching accuracy is set within a floating point number tolerance range of ±0.5%. For example, if the fundamental RMS value is 3.25 kA, the matching range is 3.235 kA to 3.265 kA.

[0097] When a match is successful, the corresponding reclosing parameters are directly output. These parameters include the reclosing enable flag, reclosing delay, and reclosing count. The output process performs a parameter validity check: the reclosing delay is limited to 0.1 seconds to 5.0 seconds, and the reclosing count is limited to 1 to 3 times. If the check passes, a parameter output event is generated, containing a timestamp, rule number, and output parameter value. For example, if rule R15 is matched, the output parameters are {enable flag = 1, delay = 0.8 seconds, reclosing count = 2}.

[0098] If a match fails, the transient stability simulation module is activated. This is triggered by a continuous traversal of all rules without a hit. The simulation module initializes within 10 milliseconds, including loading the latest grid topology snapshot, allocating a 100-megabyte simulation memory pool, and setting a 50-microsecond simulation step size. If a startup fails, the module automatically retries, with a maximum of three retries and a 100-millisecond interval.

[0099] In the transient stability simulation module, an equivalent circuit model of the protection zone where the target circuit breaker is located is established. This equivalent circuit model uses a π-type equivalent circuit model. The modeling process includes: generating a node admittance matrix based on the real-time topological connectivity; equating the generator to a series voltage source with internal resistance, whose value is taken from the device parameter library; equating the load to a constant impedance, whose value is calculated based on the pre-fault power; and equating the line to a π-type network consisting of resistance, inductance, and capacitance, whose parameters are taken from the line parameter database. Model accuracy control: Shunt capacitance is retained for voltage levels of 220 kV and above, while capacitance effects are ignored for voltage levels of 110 kV and below.

[0100] The current and voltage corresponding to the fault characteristic vector are injected into the equivalent circuit model. The injected current is calculated using the fault sequence component method. For single-phase ground faults, the injected zero-sequence current is equal to the fundamental RMS value multiplied by 0.33; for phase-to-phase short-circuit faults, the injected negative-sequence current is equal to the fundamental RMS value multiplied by 0.5. The initial voltage phase is set to the pre-fault voltage zero crossing point, and the amplitude is set to the rated voltage multiplied by 0.9. The injection point is selected as the busbar node where the target circuit breaker is located, and the injection duration is set to 100 milliseconds.

[0101] The dynamic response curve of the equivalent circuit model is calculated within a preset 500 millisecond time window. The solution is performed using the variable-step Runge-Kutta method with a relative error tolerance of 0.001. The dynamic response curve contains three-phase voltage waveforms, with a sample output every 50 microseconds, for a total of 10,000 points. Numerical stability is monitored in real time during the solution, and an implicit integration algorithm is used when the matrix condition number exceeds 10,000.

[0102] The optimal reclosing delay parameter is calculated based on the zero-crossing time difference of the dynamic response curve. The calculation process involves the following steps: extracting the voltage waveform from 100 milliseconds to 500 milliseconds after the fault is cleared; detecting the zero-crossing time of each phase voltage (the point where the voltage value changes from negative to positive); calculating the standard deviation of the three-phase zero-crossing times; and taking the zero-crossing time of the phase with the smallest standard deviation as the reference time point. The optimal reclosing delay parameter is equal to the reference time point minus the fault-clearing time point. The calculated result is limited to a range of 0.1 seconds to 5.0 seconds; if it exceeds the range, the boundary value is used.

[0103] Encapsulate reclosing parameters or optimal reclosing delay parameters into control instructions. The encapsulation format is defined as: 2-byte start code 0x55AA + 1-byte instruction type + 4-byte delay parameter (in milliseconds) + 1-byte reclosing count + 2-byte checksum + 2-byte end code 0xAA55. The checksum is calculated using a 16-bit sum-and-complement algorithm. Parameters are verified before encapsulation: if the delay parameter is < 100 milliseconds, it is forced to 100 milliseconds; if the reclosing count is > 3, it is forced to 3.

[0104] Control commands are transmitted to the target circuit breaker terminal via a high-priority communication channel using the Reliable User Datagram Protocol with a retransmission mechanism. The retransmission strategy is as follows: a 50-millisecond confirmation timer is started after the initial transmission; if no confirmation is received, the message is retransmitted every 50 milliseconds; the maximum number of retransmissions is five. The service type field of the data packet is set to the highest priority, and the transmission timeout threshold is set to 500 milliseconds.

[0105] The target circuit breaker terminal interprets the control command and executes the reclosing operation. The parsing process includes: verifying the matching of the start and end codes; calculating the checksum and comparing it with the checksum value in the command; and extracting the delay parameter and the number of reclosing attempts. The execution process is as follows: after the fault is cleared, the delay timer starts; when the timer reaches the delay parameter value, a closing pulse is issued; if closing fails and the remaining number of attempts is greater than 0, the operation is repeated after a one-second interval. The operation results are recorded in non-volatile memory.

[0106] The update mechanism for the preset protection action rule library: New fault cases are analyzed weekly. When a certain fault pattern occurs more than 10 times and does not match an existing rule, a new rule draft is automatically generated. New rules require manual review and confirmation, and after passing the review, they are hot-updated to the online library. The rule validity period is set to one year.

[0107] Equivalent circuit model parameter update process: Changes to line parameters trigger model rebuilds, and parameter change notifications from the dispatch center are received. Notification processing latency is <100 milliseconds, and model rebuild time is <50 milliseconds. Parameter version management uses incremental tagging.

[0108] Acceleration method for solving dynamic response curves: Use a 50-microsecond step size for the first 100 milliseconds, followed by a 100-microsecond step size. If the curve oscillation coefficient is less than 0.1, terminate the solution early after 300 milliseconds. The solution results are cached in shared memory.

[0109] Security protection for control commands: A digital signature is added before transmission, using the elliptic curve digital signature algorithm. The terminal verifies the signature validity before parsing it, and the signature validity period is set to 5 minutes. If signature verification fails, the command is discarded and a security alert is reported.

[0110] Reclosing lockout conditions: Reclosing is prohibited when a permanent fault is detected; operation is prohibited when the circuit breaker mechanism temperature is greater than 90°C; and execution is delayed if the number of operations has exceeded 10 in the last 24 hours. The lockout status is fed back to the control center in real time.

[0111] Resource monitoring for the simulation module: Real-time processor utilization statistics are collected. When utilization exceeds 80%, load balancing is initiated, migrating 50% of computing tasks to the backup node. Memory usage is capped at 90 megabytes; upon exceeding this limit, the oldest simulation results are released. A watchdog timer is set for each computing task, with a timeout threshold of 5 seconds.

[0112] Prioritize command transmission: A separate transmission queue is allocated for control commands in the communication protocol stack, with a queue depth of 8 messages. Status monitoring messages are discarded first during network congestion, ensuring a control command transmission success rate >99.9%. The transmission path supports dual-route switching.

[0113] Feedback mechanism for reclosing operations: An operation report is sent within 200 milliseconds after execution, including the actual operation time, success flag, and the number of remaining operations. The report uses the same encapsulation format as the command. If the report is not received, a requery is triggered, with a requery interval of 1 second, up to three times.

[0114] Dynamic response curve storage strategy: Only key feature point data, including zero-crossing moments, extreme points, and inflection points, is saved. Run-length encoding is used for compression, with a typical compression ratio of 8:1. Curve data is stored in association with fault numbers and retained for 30 days.

[0115] Optimized compensation for optimal reclosing delay parameters: Based on circuit breaker mechanical response test data, a fixed compensation time of 20 milliseconds is added. An additional compensation of 1 millisecond / °C is applied when the ambient temperature is <0°C, and 0.5 millisecond / °C when the ambient temperature is >40°C. The compensation value is capped at 100 milliseconds.

[0116] Pattern matching downgrade: When the rule base fails to load, a local simplified rule set is used. The simplified rule set only contains the single dimension of the fundamental wave effective value. The matching timeout threshold is set to 20 milliseconds. The simplified rule set is updated monthly.

[0117] The calculations involved in the embodiments are all dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to actual conditions.

[0118] It should be noted that the present invention can be deployed on the device itself to realize embedded applications, and can also be run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0119] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission. Wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission methods include infrared, microwave, etc. The computer-readable storage medium can be any available medium accessible by a computer, or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0120] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0121] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0122] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0123] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0124] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0125] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0126] Finally: 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 in the scope of protection of the present invention.

Claims

1. A modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration, characterized by: include: The load monitoring module is used by the cloud platform to monitor the resource load status of each computing node in real time; A priority generation module is used to respond to fault diagnosis requests and generate dynamic preemption priorities based on the grid topology vulnerability of the protection area where the target circuit breaker is located and the degree of coordinated action requirements of the associated circuit breakers; Credibility attenuation module, used to extract the waveform distortion rate of the fault current. If it does not fall within the threshold range of the historical fault library, the credibility of the dynamic preemption priority is attenuated. The resource preemption module is used to determine the necessity of resource preemption based on the resource load status and the processed dynamic preemption priority. If it exceeds the resource preemption threshold, the periodic task thread is frozen, computing resources are released, and a high-priority communication channel is opened; A feature extraction module is used to send a fault recording compression instruction to the target circuit breaker terminal through a high-priority communication channel and receive the fault feature vector after edge preprocessing; The strategy execution module is used to perform protection strategy optimization within the released resources based on the fault feature vector, generate reclosing parameters and send them to the target circuit breaker terminal through a high-priority communication channel for execution.

2. The modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to claim 1 is characterized in that: The cloud platform monitors the resource load status of each computing node in real time, including: Get the curve of the CPU usage of the computing node over time; Get a curve showing the memory usage of a computing node over time; Calculate the percentage of cumulative time that exceeds a preset CPU usage threshold based on the CPU usage curve; Calculate the maximum duration of memory usage exceeding the preset threshold based on the memory usage curve; The cumulative duration ratio and the maximum duration are combined to form a quantitative indicator of resource load status.

3. The modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to claim 2 is characterized in that: In response to fault diagnosis requests, a dynamic preemption priority is generated based on the grid topology vulnerability of the target circuit breaker's protection area and the need for coordinated action of associated circuit breakers, including: After responding to the fault diagnosis request, the grid topology vulnerability value is multiplied by the collaborative action demand value, and then multiplied by the preset priority scaling factor to generate the dynamic preemption priority.

4. The modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to claim 3 is characterized in that: The numerical method for obtaining the power grid topology vulnerability is: Obtain the real-time topological connection relationship of the protection area where the target circuit breaker is located; Perform N-1 security checks on real-time topology connections, and use the percentage of branches that fail the check as the first factor. Calculate the ratio of the critical load power to the total load power in the protection area as the second factor; Multiplying the first factor by the second factor yields the grid topology vulnerability value.

5. The modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to claim 3 is characterized in that: The method for obtaining the collaborative action requirement value is as follows: Measure the actual operation timing margin of the associated circuit breakers; Read the preset standard time difference; The ratio of the absolute difference between the actual action timing margin and the standard coordination time difference to the standard coordination time difference is calculated as the collaborative action requirement value.

6. The modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to claim 3 is characterized in that: Extract the waveform distortion rate of the fault current. If it does not fall within the threshold range of the historical fault library, perform credibility attenuation processing on the dynamic preemption priority, including: Extracting the fault current and performing discrete sampling on the fault current waveform to obtain a current sequence; Calculate the root mean square error between the current sequence and the standard sine wave as the waveform distortion rate; Read the upper and lower threshold values ​​of the waveform distortion rate for each type of fault in the historical fault database; Determine whether the waveform distortion rate is greater than the waveform distortion rate threshold lower limit and less than the waveform distortion rate threshold upper limit at the same time; When the waveform distortion rate is less than the lower threshold value of the waveform distortion rate or greater than the upper threshold value of the waveform distortion rate, the dynamic preemption priority is multiplied by the preset attenuation coefficient; When the waveform distortion rate is greater than the waveform distortion rate lower threshold and less than the waveform distortion rate upper threshold at the same time, the dynamic preemption priority remains unchanged.

7. The modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to claim 6 is characterized in that: The necessity of resource preemption is determined based on the resource load status and the processed dynamic preemption priority. If the resource load exceeds the resource preemption threshold, the periodic task thread is frozen, computing resources are released, and a high-priority communication channel is opened, including: Mapping the processed dynamic preemption priority to the resource preemption demand value; Determine the current resource tension coefficient based on resource load status; Multiply the resource preemption demand value by the resource tension coefficient to obtain the quantitative value of resource preemption necessity; Comparing the resource preemption necessity quantified value with the preset resource preemption threshold; When the resource preemption necessity quantified value is greater than the resource preemption threshold, a thread freeze instruction is sent to the operating system to suspend the periodic task thread; Reclaim computing resources occupied by suspended threads; Creates a high-priority communication channel independent of regular channels in the communication protocol stack.

8. The modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to claim 7 is characterized in that: Sending a fault recording compression command to the target circuit breaker terminal via a high-priority communication channel and receiving the fault feature vector after edge preprocessing, including: Transmitting the fault recording compression instruction carrying the compression format identifier to the target circuit breaker terminal through the established high-priority communication channel; The target circuit breaker terminal performs format conversion processing on the original fault recording data according to the compression format identifier; The target circuit breaker terminal performs feature extraction calculation on the fault recording data after format conversion to generate a fault feature vector; The target circuit breaker terminal encapsulates the fault feature vector into a data message; receiving a data message carrying a fault feature vector returned by a target circuit breaker terminal through a high-priority communication channel; Parse the data packet to extract the fault feature vector after edge preprocessing.

9. The modular reclosing circuit breaker protection system based on intelligent electrical cloud collaboration according to claim 8, characterized in that: Based on the fault feature vector, protection strategy optimization is performed within the released resources. Reclosing parameters are generated and sent to the target circuit breaker terminal through a high-priority communication channel for execution, including: Input the fault feature vector into the preset protection action rule library for pattern matching; When the match is successful, the corresponding reclosing parameters are directly output; When the matching fails, the transient stability simulation calculation module is started; Establish an equivalent circuit model of the protection area where the target circuit breaker is located in the transient stability simulation calculation module; Inject the current and voltage corresponding to the fault characteristic vector into the equivalent circuit model; Solve the dynamic response curve of the equivalent circuit model within a preset time window; Calculate the optimal reclosing delay parameters based on the zero-crossing time difference of the dynamic response curve; Encapsulate reclosing parameters or optimal reclosing delay parameters as control instructions; Transmit control instructions to target circuit breaker terminals via high-priority communication channels; The target circuit breaker terminal interprets the control command and performs the reclosing operation.

Images