Online detection method for medium and low voltage direct current circuit breaker

Abstract

The invention discloses an online detection method for a medium and low voltage direct current circuit breaker, and relates to the technical field of power system protection control, and the method comprises the following steps: S001, building a radiation echo spectrum distribution model during the switching operation of the direct current circuit breaker, calculating an interference intensity estimated value through collecting a current drop rate and a fracture voltage climbing rate, and calculating the interference intensity estimated value according to the current drop rate and the fracture voltage climbing rate; and extracting a disturbance frequency band with concentrated interference energy and generating a characteristic signal set for representing an interference propagation path. According to the method, a closed-loop anti-interference mechanism is constructed through feature modeling, interference recognition, signal correction and parameter optimization, the stability and accuracy of state judgment are improved, misjudgment and response failure caused by high-frequency interference are prevented, the method is particularly suitable for a complex parallel system and a frequent on-off scene, and the safety and the intelligent level of a direct-current system are effectively enhanced.

Description

Technical Field

[0001] This invention relates to the field of power system protection and control technology, specifically to an online detection method for medium and low voltage DC circuit breakers. Background Technology

[0002] Online monitoring of medium- and low-voltage DC circuit breakers refers to an online analysis and evaluation technology used in medium- and low-voltage DC distribution networks or DC power terminal systems to continuously monitor the operating status and fault symptoms of circuit breakers based on real-time collected operating data and status change trends. This allows for the identification of potential risks before anomalies evolve into serious faults and provides decision support for subsequent protection actions. This monitoring method typically integrates multi-dimensional diagnostic parameters such as current change rate analysis, bus voltage drop detection, and local energy accumulation identification. Through high-speed signal acquisition devices and local intelligent analysis units, it continuously observes the circuit breaker contact status, conduction path integrity, and transient response characteristics. Simultaneously, it is combined with high-breaking-speed actuators (such as vacuum arc-extinguishing devices, semiconductor circuit breakers, or composite breaking structures) and energy absorption units to form a closed-loop correlation identification of abnormal behaviors, helping to determine whether there are potential faults caused by poor contact, mechanical hysteresis, sensor drift, or external interference. By using microsecond-level signal response and periodic-level trend tracking, the operational safety and accuracy of protection actions of medium and low voltage DC systems in complex application scenarios can be significantly improved, overcoming the technical bottlenecks of response delay, inaccurate identification, and action failure in traditional detection methods.

[0003] The existing technology has the following shortcomings: In existing technologies, after a DC circuit breaker performs a high-speed breaking operation, the sudden drop in DC current often generates high-frequency electromagnetic interference waves. These electromagnetic echoes undergo multiple reflections and superpositions in local circuits, easily interfering with the circuit breaker's internal state judgment components. This is particularly true during the critical transition between a sharp rise in the break voltage and the system's zero current threshold. State sensing devices (such as current sensors, voltage detectors, and auxiliary contact feedback circuits) are susceptible to the coupling effect of instantaneous high-frequency interference signals, causing their outputs to enter unstable or critical transition ranges. This leads to ambiguity, contradictions, or erroneous identification of the current state by the circuit breaker control unit. This problem has not been effectively suppressed in existing technologies. Especially in scenarios involving multiple breaking operations and parallel device coordinated protection, it can easily cause logic control link failures, missing the response window for rapid fault clearing, and resulting in serious consequences such as fault propagation, power backflow, or bus voltage runaway, severely threatening system stability and the real-time performance of protection actions.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an online detection method for medium and low voltage DC circuit breakers to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an online detection method for medium and low voltage DC circuit breakers, comprising the following steps: S001 During the interruption operation of the DC circuit breaker, a radiation echo spectrum distribution model is constructed. The interference intensity is estimated by collecting the current drop rate and the voltage rise rate at the break point. The disturbance frequency band where the interference energy is concentrated is extracted and a set of characteristic signals is generated to represent the interference propagation path. S002, based on the feature signal set, sets a dynamic time window observation mechanism in the input link of the state judgment component, and uses the sampling signal within a continuous period to match the preset reference signal sequence to identify sampling points with amplitude jitter, waveform ghosting or phase jump characteristics, and generate interference indicator index; S003, based on the interference indicator index, performs multi-segment fitting residual comparison operation, continuously analyzes the instantaneous change trajectory of the output signal of the state judgment component, marks the residual abnormal segment and generates a disturbance interference mapping map to determine the unstable interval of the state judgment response. S004. Based on the disturbance and interference mapping diagram, perform local self-correction operation, introduce synchronous redundant sensing paths for detection nodes with jump output in the state judgment link, determine the reliable output channel by comparing the consistency of the output signals of each path, and form an anti-interference dynamic evaluation feedback mechanism. S005, based on the anti-interference dynamic evaluation feedback mechanism, sends adjustment signals to the logic control system to dynamically set the judgment threshold for fault identification and the parameter weights of the response strategy, thereby limiting the fault judgment process to the time period with the least interference, so as to enhance the accuracy of state identification and the stability of response. S006, based on the interference intensity information and feedback residual distribution data during the opening and closing processes of multiple circuit breakers, updates the parameters of the interference identification model, and dynamically adjusts the signal gain of each sensing path through learning optimization, forming a closed-loop control process that coordinates interference identification, anomaly calibration and response strategy adjustment.

[0007] Preferably, step S001 includes: After the DC circuit breaker issues the interruption control command, the current and voltage changes at both ends of the break are collected in the main conduction path. The current sensor has a response time of less than 100 nanoseconds and is installed 5 to 15 centimeters away from the break. The voltage acquisition uses a differential probe with a bandwidth of not less than 20 MHz and is wired in a shielded manner. The collected current descent rate and break voltage ramp rate were analyzed in both the time and frequency domains to identify the frequency range where the interference energy density was concentrated, and the center frequency, upper and lower bounds of the bandwidth, maximum amplitude point and total energy were recorded. The frequency analysis results are mapped to the original time series, and interference waveforms that meet the conditions of amplitude, duration, and repetition are extracted to form a set of characteristic signals containing spectral distribution, duration, sampling channels, and original time points. Based on the internal electrical wiring structure of the circuit breaker, the occurrence order, propagation mode and influence path of characteristic signals in different sampling channels are analyzed, and an interference propagation path diagram is constructed to indicate the starting position, propagation direction and range of the interference signal.

[0008] Preferably, step S002 includes: Status judgment signals are collected in the current detection circuit, voltage detection circuit and auxiliary contact feedback circuit respectively. The sampling device used has high bandwidth and high precision capabilities and is configured with an independent clock synchronization structure to ensure the consistency of multi-channel sampling data on the time axis. A dynamic time window observation process is introduced for each sampling channel. Within a preset time window, the real-time sampling signal is compared point by point with the reference signal sequence of the corresponding channel to identify abnormal sampling features such as continuous amplitude jitter, waveform ghosting, and phase jump at feature points. For all identified abnormal sampling points, interference indicator information is generated, including channel identifier, abnormality type, abnormal time, duration, sampling point index and deviation value, and cross-channel interference events are identified by cross-comparison. By binding interference indicator information with the original sampling data, and inserting labels for anomaly type, location, difference, and suggested handling method into the original data, a structured interference indicator index data table is constructed for subsequent link judgment.

[0009] Preferably, step S003 includes: Based on the abnormal sampling point number and time information recorded in the interference indicator index, the output signal sequence corresponding to the time period is extracted from the output data of the state judgment component, and the sampling point range is extended forward and backward to ensure coverage of the entire interference effect process. The extracted signal sequence is divided into continuous data segments. In each segment, a fitting process is performed according to the physical change law of the actual working state of the circuit breaker to generate an ideal response curve. The residual value of each sampling point is calculated to represent the deviation between the actual output and the fitted output. Based on the degree of continuous deviation and abrupt change characteristics of the residual values, abnormal residual segments associated with the interference behavior are identified, and cross-validation is performed in combination with the time records in the interference indicator index. The effectiveness of the interference is confirmed when the time overlap between the two is significant. The confirmed abnormal segments are organized in chronological order and by channel category, and a disturbance mapping diagram is drawn. Different levels of disturbance intensity are identified by color or graphic style to indicate unstable areas in the state judgment and response process.

[0010] Preferably, step S004 includes: Based on the response interval marked with high residuals in the disturbance mapping diagram, extract the channel number information, the time segment of the anomaly occurrence and the residual intensity level, set the target channel that needs to perform self-correction operation, and extract the corresponding sampling frequency, sampling quantity and data format parameters to ensure that the original signal and the redundant signal are comparable. In each judgment channel that needs to be corrected, a synchronous redundant sampling channel with a different structural path is introduced to keep the sampling control logic time base of the main channel and the redundant channel consistent, and to complete the redundant sampling of the original judgment signal to obtain reference signal data. The judgment signals collected by the original channel and the redundant channel in the interference section are compared point by point. The evaluation is carried out according to the amplitude difference, waveform trend change and feature point time difference. The channel with smaller signal fluctuation, more stable change trend and more accurate feature point positioning is selected as the reliable signal output path for the current time period. The path selection results and comparison data are recorded as anti-interference feedback content, and the channel number, interference time segment, residual information and signal stability level are updated as the basis for subsequent judgment logic adjustment, redundant path activation priority allocation and response threshold dynamic setting.

[0011] Preferably, step S005 includes: Read the anti-interference dynamic evaluation feedback data, extract the stable signal segment, the segment with the smallest residual, the consistency score of the redundant path and the corresponding timestamp of each judgment channel, and identify the smallest interference segment that has not fluctuated, jumped or changed abruptly. Based on the actual signal fluctuation range of the minimum interference section, the current identification threshold and voltage trigger threshold are adjusted respectively, and the response judgment operation is limited to the time range of the minimum interference section. Calculate the confidence score of the output signal of each judgment channel, combine the residual amplitude and the stable time interval, set the channel output weight, and generate adjustment instructions including threshold correction parameters, response time window and path weight settings; Adjustment instructions are injected into the control logic decision-making process to complete online updates of control strategies. Fault identification results and response effects are tracked, recorded, and compared with historical strategies to improve judgment accuracy and response stability.

[0012] Preferably, step S006 includes: Extract the current drop rate, break voltage rise rate, interference main frequency range, interference duration and channel residual information within multiple operation cycles, and construct a time series sample set containing interference intensity and judgment bias. Based on the time series sample set, the detection sensitivity, judgment frequency range and disturbance duration threshold parameters in the existing interference identification model are corrected, and the updated parameters are used synchronously for the identification logic structure in the next cycle. Based on the residual level in the historical period of each sensing channel, the gain configuration of the signal output is adjusted. The influence weight of the reliable channel is increased by adjusting the gain, thereby reducing the degree of misleading interference path. The updated identification parameters and gain configuration are bound to the current action cycle number and control strategy version, and then called during the next judgment initialization process. Iterative optimization is performed based on feedback from subsequent cycle evaluations.

[0013] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention constructs a radiation echo spectrum distribution model during the circuit breaker's breaking process, extracts key disturbance frequency bands to form a set of characteristic signals, and combines dynamic time window observation, interference indicator index analysis, multi-segment residual fitting comparison, synchronous redundant signal correction, adaptive adjustment of control parameters, and cross-cycle model parameter optimization to form a complete closed-loop anti-interference identification and response control mechanism. This method not only significantly enhances the ability of the state judgment link to identify high-frequency transient interference and improves output stability, but also effectively reduces the risk of protection action delays, false triggering, or control link failures caused by sensor signal distortion or judgment errors. Compared with traditional static judgment or passive filtering strategies, this invention has stronger adaptability and real-time performance, and is particularly suitable for complex operating environments such as high voltage and high current, multiple devices in parallel, and frequent switching. It can significantly improve the operational reliability and fault handling accuracy of DC circuit breakers under extreme electromagnetic interference conditions, ensuring the overall safety and intelligence level of the system. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0015] Figure 1 This is a flowchart of the online detection method for low-voltage DC circuit breakers in this invention. Detailed Implementation

[0016] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0017] This invention provides, for example Figure 1 The online testing method for medium and low voltage DC circuit breakers shown includes the following steps: S001 During the interruption operation of the DC circuit breaker, a radiation echo spectrum distribution model is constructed. The interference intensity is estimated by collecting the current drop rate and the voltage rise rate at the break point. The disturbance frequency band where the interference energy is concentrated is extracted and a set of characteristic signals is generated to represent the interference propagation path. To address the high-frequency electromagnetic interference (EMI) caused by sudden current drops and voltage surges during high-speed breaking operations of DC circuit breakers, a method based on transient process physical quantity extraction and interference propagation path modeling is proposed. This method collects and analyzes key electrical parameters during circuit breaker breaking, constructs a set of characteristic signals for identifying interference behavior, and determines the propagation path of the interference in different electrical branches, providing a basis for ensuring the stability of subsequent state judgment signals. The specific steps are as follows: During the period when the DC circuit breaker issues the breaking control command and enters the actual breaking action, high-speed acquisition of key electrical parameters is performed. These parameters mainly include the current flowing through the main contact path and the voltage across the main break. The current shows a significant decreasing trend at the moment of contact separation, and its rate of decrease is quantified in amperes per microsecond. The voltage rises sharply from near zero volts to the DC bus voltage or experiences a spike overshoot, and its rate of increase is expressed in volts per microsecond. When measuring current, a current sensor with a response time of less than 100 nanoseconds is used. The sensor is installed within 5 to 15 centimeters of the break in the main circuit of the circuit breaker to ensure that the measured current data accurately reflects the abrupt change in the initial breaking behavior. When measuring voltage, a differential probe with a bandwidth of not less than 20 MHz is connected to both ends of the break, with the lead-out measurement line not exceeding 20 centimeters in length, and shielded wiring is used to reduce power frequency noise interference. The triggering points for current and voltage sampling should be synchronized uniformly through the on / off control trigger signal to ensure alignment of the acquisition time axis. The sampling frequency should be set at no less than 100 megabits per second to meet the requirements for accurate recording of rapidly changing processes.

[0018] The measured current descent rate and voltage rise rate data samples were projected into the time and frequency domains for joint analysis. By comparing the rate of change of continuous sampling points within different time windows, the locations of abrupt changes were preliminarily identified. Within this time period, the frequency response of the measured signal was evaluated, and the energy density distribution in each frequency range was analyzed to identify frequency bandwidths with significant energy accumulation, such as obvious energy accumulation peaks between 3 MHz and 7 MHz. For each frequency band, its center frequency, upper and lower bounds of bandwidth, maximum amplitude point, and total energy in that band were recorded and compared with the reference spectrum under non-interference conditions to determine whether the frequency band belongs to the disturbance output unique to the transient behavior of circuit breaker interruption. Unlike traditional preset interference models, this analysis is entirely based on measured data and has the ability to adaptively extract transient interference characteristics under actual interruption behavior, thereby avoiding the problem of mismatch with actual operating conditions.

[0019] After identifying the high-energy-density interference frequency band, it is necessary to further extract the corresponding time-domain waveform to construct a signal set representing the interference characteristics under this interruption action. The selection of characteristic signals should consider the signal duration, waveform morphology, and repetition characteristics across multiple sampling points. For example, if a short pulse with a rise time of less than 500 nanoseconds, an amplitude exceeding three times the baseline, accompanied by localized periodic high-frequency oscillations, and repeating across multiple time periods during the interruption, this type of waveform can be defined as a typical interference characteristic signal in the current channel. Similarly, if a waveform exhibiting more than three reverse deflections within a short period is observed in the voltage detection channel, accompanied by transient patterns where superimposed noise exceeds 20% of the peak signal, it can be identified as a typical interference waveform in the voltage channel. All waveforms meeting the above conditions are extracted and numbered to form a characteristic signal set under the current interruption action. Each characteristic signal should include the original time point, duration, spectral distribution, and sampling channel identifier for subsequent interference propagation path localization analysis.

[0020] Based on the extracted set of characteristic signals and the internal wiring information of the circuit breaker, the propagation path and effective area of ​​interference signals in different electrical branches are analyzed. Specifically, the characteristic signals are arranged according to their order of appearance on the time axis, and the first appearance time and duration of characteristic signals of the same frequency in each channel are compared to infer their propagation sequence. For example, if an interference waveform appears earlier in the current detection branch than in the voltage detection branch, and their frequencies are similar, it can be inferred that its propagation path starts in the main circuit and couples to the voltage detection branch through a common-mode interference path. Further reference to the control cable length, grounding point location, cable laying method, and the relative layout of internal structural metal components allows determination of whether the interference signal propagates through electromagnetic radiation or electromagnetic induction. Based on the above analysis, a correspondence is established between all characteristic signals and their propagation paths, forming a complete interference propagation path diagram. This path diagram not only clearly shows the starting point, propagation direction, and effective range of the interference signal, but also clearly indicates the state judgment links that it may affect.

[0021] To effectively analyze interference propagation paths, this step, based on the acquisition of the main current and voltage of the main circuit breaker, further combines the internal electrical wiring and sensor configuration of the circuit breaker to perform collaborative analysis of interference characteristic waveforms collected from multiple signal channels (including but not limited to current detection channels, voltage detection channels, and auxiliary contact feedback channels). Specifically, high-bandwidth synchronous sampling devices are deployed in different sampling channels, enabling each channel to independently record high-frequency disturbances under transient interruption behavior. In the extracted set of characteristic signals, each interference waveform is accompanied by its corresponding sampling channel identifier and original sampling timestamp. By comparing the occurrence time, duration analysis, and waveform similarity calculation of characteristic signals with consistent frequencies in different channels, the propagation sequence and path can be inferred, thereby reconstructing the interference propagation path.

[0022] This step aims to accurately identify and analyze the propagation path of transient electromagnetic interference (EMI) generated during the high-speed breaking process of a DC circuit breaker, providing fundamental support for subsequent state-sensing stability assurance and anti-interference judgment. At the moment of breaking, the sudden drop in current and the sharp rise in voltage at the break point trigger high-frequency radiated interference waves. These interference waves easily couple into current sensors, voltage detection circuits, and auxiliary control signal loops, causing distortion of state identification information. This step first obtains the characteristic changes in representative physical quantities to construct a spectrum model under actual operating conditions; secondly, it identifies the energy-concentrated frequency bands and extracts typical waveforms with interference characteristics; and thirdly, combined with the internal wiring structure of the circuit breaker, it analyzes the propagation path and timing relationship of interference signals between channels, clarifying the source, transmission direction, and affected area of ​​the interference. This step effectively compensates for the problem that traditional methods cannot dynamically sense the actual distribution of electromagnetic disturbances, achieving a systematic identification from the interference behavior itself to its path of action. This is a prerequisite for ensuring the accuracy of subsequent control judgments of the circuit breaker and improving its anti-interference capability.

[0023] S002, based on the feature signal set, sets a dynamic time window observation mechanism in the input link of the state judgment component, and uses the sampling signal within a continuous period to match the preset reference signal sequence to identify sampling points with amplitude jitter, waveform ghosting or phase jump characteristics, and generate interference indicator index; Based on the feature signal set constructed in the previous stage, a dynamic time window observation process is set up on the DC circuit breaker state judgment signal acquisition channel. By comparing it with the reference signal sequence cycle by cycle, abnormal signal sampling points affected by transient electromagnetic interference are identified, and finally, an interference indicator index is generated for subsequent reliability assessment and signal correction of the judgment data processing link. This method is carried out through the following steps: High-speed sampling devices are deployed in the current detection circuit, voltage detection circuit, and auxiliary contact feedback circuit of the DC circuit breaker to acquire raw signals related to state judgment. Each sampling channel uses an analog-to-digital converter with a bandwidth of no less than 10 MHz and a sampling accuracy of more than 12 bits, and is equipped with an independent clock synchronization module to ensure strict consistency of timestamps for multi-channel data. The sampling frequency should be set between 100 MHz and 200 MHz per second so that each data cycle can cover microsecond-level interference waveform changes. Throughout the entire process of the DC circuit breaker entering the breaking state, continuous data acquisition is maintained from the start of contact action until 10 milliseconds after the arc is extinguished. To improve the accuracy of subsequent interference judgment, a reference signal sequence is introduced into the sampling channel. This reference signal sequence is a stable output curve obtained after the circuit breaker has run for multiple cycles under the same action command conditions in an environment without electromagnetic interference. After filtering out peak values ​​and normalizing the amplitude, it is fixed and stored as a comparison object for the actual sampled data.

[0024] A dynamic time-window observation mechanism is implemented in all sampling channels. The time span of each observation window is set from 10 microseconds to 50 microseconds, and the number of sampling points can be automatically adjusted according to different sampling frequencies. Within each dynamic window, the original signal is compared point by point with the reference signal of the corresponding channel, mainly examining three aspects of deviation characteristics: First, the degree of amplitude jitter, that is, whether the difference between the current sampling point and the previous and subsequent sampling points is continuously greater than twice the difference of the reference signal at the same position; second, waveform ghosting, that is, checking whether the signal curve has two or more extremely close maximum or minimum values ​​within a certain short period of time. This double-peak or double-valley waveform does not exist in the reference waveform and is an abnormal signal characteristic; third, phase jump performance, that is, comparing the positional difference between the sampled waveform and the reference waveform at the main characteristic points (such as peaks, troughs, or zero intersections). If it exceeds one sampling period, it is marked as a phase jump. In the above three judgment processes, all data comparisons are based on the complete data segment within the actual observation window, and a preset threshold value is used to determine anomalies. Each data point determined to be abnormal is recorded in time.

[0025] Once a sampling point within the observation window is identified that meets any of the characteristics of amplitude jitter, waveform ghosting, or phase jump, it is included in the interference indicator information set. Each interference indicator information includes the following fields: channel identifier (e.g., current detection channel 1), interference identification type (amplitude anomaly, morphological anomaly, phase anomaly), anomaly occurrence time (relative microsecond value with the trigger interrupt signal as zero), anomaly duration (total duration of sampling points continuously identified as anomalies), sampling point index number range (used to mark the position in the original sampling array), and the numerical difference between the offset and the reference signal. Simultaneous anomalies in multiple channels will be recorded as cross-channel interference events through cross-comparison, along with a consistency judgment result to indicate the presence of a common-source interference signal. All identified anomaly information will be uniformly compiled into an interference indicator index data table, and categorized into three levels—mild disturbance, moderate disturbance, and severe disturbance—based on the anomaly type to distinguish interference behaviors of different intensities.

[0026] After generating the interference indicator index, it is structurally bound to the original signal data of the corresponding channel. This involves attaching a label to each data point identified as interference in the original data sequence. The label includes the identification type of the point, its position within the abnormal interval, the difference from the reference signal, and a suggested processing method (such as temporary ignoring, substitution, or redundant data comparison and replacement). This binding operation is performed on a per-sampling-channel basis, appending an interference identification tag list to the original data queue to ensure that each suspicious data point has a clear source and a corresponding processing tag. In subsequent processing, the interference location can be quickly determined using the index information without re-analyzing the original waveform. Based on this, the judgment logic can be adjusted, redundant signal paths switched, or identification threshold settings optimized, significantly improving the accuracy of circuit breaker status judgment and its resistance to false positives in interference scenarios.

[0027] This step aims to accurately identify and quantitatively label the interference effects in the DC circuit breaker status judgment signals, ensuring that subsequent fault determination processes are based on stable and reliable signal input. During the high-speed interruption of the circuit breaker, high-frequency electromagnetic interference generated by sudden current drops and voltage surges may couple into signal channels such as current detection, voltage detection, and auxiliary contact feedback, causing signal amplitude jitter, abnormal waveform morphology, or phase jumps, thereby triggering misjudgments or action delays in the judgment logic. Through this step, dynamic observation windows are established in each signal channel, and sampling points with abnormal characteristics are identified in a timely manner by comparing them cycle by cycle with the reference signal sequence. An interference indicator index is generated, marking the anomaly type, location, and duration. This feature-based behavior recognition processing method not only improves the accuracy of anti-interference identification but also provides a basis for subsequent data correction, path switching, or threshold adjustment. Compared with existing techniques that rely on static hardware filtering or preset threshold judgment, this step has stronger adaptability and identification accuracy, and is a key prerequisite for building a highly robust status judgment link.

[0028] S003, based on the interference indicator index, performs multi-segment fitting residual comparison operation, continuously analyzes the instantaneous change trajectory of the output signal of the state judgment component, marks the residual abnormal segment and generates a disturbance interference mapping map to determine the unstable interval of the state judgment response. Based on the interference indicator index data constructed in the previous stage, continuous behavioral analysis is performed on the output signal of the DC circuit breaker state judgment component. The change trajectory of the output signal is compared using a multi-segment fitting method to identify abnormal response segments caused by electromagnetic interference, and a disturbance interference mapping diagram is drawn to visually represent the unstable response intervals existing in the state judgment link. This process is completed sequentially using the following steps: Based on the interference indicator index generated in the previous stage, the output signal sequence corresponding to the interference time period is extracted from the historical output data of the status judgment component. The output of the status judgment component usually reflects the operating status of the circuit breaker in digital or analog form, such as whether the main circuit current has reached the action threshold, whether the voltage has recovered to a safe value, and whether the auxiliary contacts have changed according to the predetermined logic. The status judgment component is the core processing module in this system used to identify, classify, and determine the operating status of the circuit breaker. It mainly includes a high-speed data receiving unit, a signal processing logic unit, a judgment model execution unit, and a status output interface. The input of this component is the current, voltage, and other signals collected in step S001, which are sent to the judgment component via a synchronous sampling channel. The signal processing logic unit first performs preprocessing on the raw sampled data, including denoising, normalization, and feature point extraction, and then enters the judgment model execution module. Based on preset criteria such as current jump threshold, voltage descent rate, and auxiliary contact logic, this module determines whether the current state is in a fault precursor or interference state through threshold comparison, trend recognition, etc., and outputs the corresponding status signal according to the judgment result. The output signal can be an analog value (such as residual amplitude) or a digital value (such as 0 representing normal and 1 representing a fault trend). This output is then used as the input for residual comparison in S003. The status judgment component can be implemented using an FPGA, DSP, or embedded processing chip, or deployed in software within a high-performance control chip. Before extracting data, the abnormal sampling point number, interference start time, duration, and channel type recorded in the interference indicator index are first read. Then, with each abnormal sampling point as the center, the time range is extended forward and backward by no less than 200 consecutive sampling points, and the output signal data sequence within the corresponding time period is extracted to ensure coverage of the possible changes in the trend before and after the interference. It is recommended that the sampling frequency be consistent with the input signal, i.e., no less than 100 mega-sampling times per second, to ensure that the time resolution of the residual comparison is sufficiently fine.

[0029] The extracted output data sequence is divided into multiple non-overlapping continuous segments, each with a recommended length of 200 to 500 sampling points. During the segmentation process, critical transition zones should be avoided, and waveform behavior within each segment should be kept as consistent as possible. Within each segment, a corresponding reference trend should be set based on the physical laws governing the actual operating state of the circuit breaker. For example, during the voltage rise phase, the trend should be continuous; during the rapid current decline phase, the trend should be rapid decay; and during the stable maintenance phase, the trend should be slow or constant. Based on these expected behaviors, a fitting process is performed on each segment of actual data, constructing an ideal response curve using manually defined standard linear or nonlinear slope intervals. Then, the actual output value at each sampling point is compared with the expected value on its corresponding fitted curve, calculating the residual for each sampling point—the difference between the two. The residual results for all sampling points are used to assess the deviation of the data segment from the ideal state.

[0030] Based on the distribution of residuals, areas exhibiting consecutive large deviations are identified as residual abnormal segments. The specific criteria are as follows: when the residual value of more than 20 consecutive sampling points exceeds the set threshold of the expected value of the fitted curve (e.g., more than 10%), or when the residual shows a significant sharp increase within a short period (e.g., the change exceeds 20% of the original amplitude within 5 sampling points), this data segment is marked as a residual abnormal segment. To improve accuracy, the time location of each abnormal segment should be cross-validated with the original interference sampling points recorded in the interference indicator index. If the time overlap exceeds 80%, the residual abnormal segment is considered to be associated with interference behavior. Otherwise, it is necessary to further determine whether it is caused by the natural transient behavior of the circuit breaker or the noise characteristics of the equipment itself, and it can be marked as pending verification. Each confirmed valid abnormal segment needs to record the following data fields: abnormal start time, end time, channel type, maximum residual value, average residual value, waveform variation trend description, and the degree of matching with the interference indicator index.

[0031] All confirmed residual abnormality segments are integrated according to time sequence and channel type to create a disturbance / interference mapping diagram. This diagram uses time as the horizontal axis and signal channel type (e.g., current judgment output, voltage judgment output, auxiliary contact output) as the vertical axis, plotting the start and end positions of each residual abnormality segment and using color or shading to indicate different levels of abnormality intensity. For example, residual values ​​exceeding the baseline value by more than 30% are marked in red, those between 10% and 30% in orange, and those below 10% in yellow. When multiple channels experience abnormalities during the same time period, a cross-over pattern is used in the diagram, indicating the possible propagation of common-source interference. After completion, this diagram serves as the basis for describing the interference behavior of the current action cycle and is stored in a dataset used for fault diagnosis or response strategy adjustment. Subsequent operations such as rapid fault identification strategies, adaptive threshold adjustments, and data path switching can be based on this diagram for area limitation and time window control, improving anti-interference response efficiency and judgment accuracy.

[0032] This step, based on the identification of input interference signals, further identifies the response deviations caused by interference in the judgment link by performing residual analysis on the output signal of the DC circuit breaker's status judgment component, thereby accurately delineating the unstable region of the judgment result. Because the input signal is affected by transient electromagnetic interference, it may cause abrupt changes, drift, or distortion in the status judgment output, which, if not identified and processed, can easily lead to misjudgments, missed judgments, or protection response delays. Through this step, the status judgment output signal is piecewise fitted, and the actual output is compared with the expected change trend segment by segment residual, effectively identifying the abnormal response region caused by interference. After cross-validation with the interference indicator index, a complete disturbance interference mapping map is formed, intuitively showing the interference impact and severity on each signal channel at different times. This mapping map not only confirms the reliability of the current status judgment result but also provides a key reference for subsequent data correction, judgment link optimization, and response strategy adjustment, making it one of the core components of achieving anti-interference closed-loop control.

[0033] S004. Based on the disturbance and interference mapping diagram, perform local self-correction operation, introduce synchronous redundant sensing paths for detection nodes with jump output in the state judgment link, determine the reliable output channel by comparing the consistency of the output signals of each path, and form an anti-interference dynamic evaluation feedback mechanism. Based on the disturbance mapping map constructed in the previous stage, a local self-correction process based on synchronous redundant sampling paths is implemented for the state judgment signal nodes identified as exhibiting unstable output behavior. Through real-time comparison and filtering of the output signals from redundant paths, judgment signals with stability and reliability under the current disturbance conditions are identified, and a feedback mechanism is constructed based on this to correct subsequent judgment logic and adjust response strategies. This process unfolds sequentially through the following steps: Based on the generated disturbance mapping map, target detection nodes requiring self-calibration are selected. In the previous stage, through multi-segment fitting and residual comparison, the degree of interference impact of different judgment channels within a specific time window has been clearly marked. In the mapping map, nodes in the high residual range, such as those experiencing short-term abrupt changes in the main circuit current judgment output during the initial circuit breaker opening phase, or those showing a jump in the voltage judgment signal after the arc is extinguished, are considered high-interference nodes with reduced stability. The judgment channel number, sampling time interval, anomaly degree, and residual statistics corresponding to these nodes are extracted as input conditions to determine the channels that need to enter the calibration process, and an independent calibration task is established for each channel. During task initialization, the total number of sampling points, sampling frequency, and data storage format should also be extracted to ensure that the sampling data of redundant paths are completely comparable to the original channels.

[0034] In each judgment channel requiring calibration, an independently configured synchronous redundant sensing path is introduced to obtain reference signals that originate from the same source but have different structural paths. Taking the current judgment channel as an example, if the original channel uses a Hall effect-based current sensor to collect the instantaneous current signal of the main circuit, the redundant path can be configured with a fiber optic current transformer, placed within 10 mm of the physical position offset from the main sensor, and uses an independent sampling circuit to synchronously record the current waveform. For the voltage judgment channel, if the original channel uses a resistor voltage divider method to sample the voltage across the main break, the redundant path is configured with a capacitively coupled signal extraction method, and a differential amplifier is used to eliminate common-mode interference signals. For the auxiliary contact judgment channel, redundant signals can be obtained through an opto-isolated detection circuit to ensure higher signal stability under strong electromagnetic interference. All redundant paths use independent sampling control logic, with their clock reference consistent with the main channel, and the sampling frequency set to 100 to 200 mega-cycles per second to ensure complete recording of microsecond-level signal fluctuations.

[0035] After sampling, the output signals of the original channel and the redundant channel within the interference section are compared point by point to determine the signal stability and reliability of the current judgment channel. The comparison method includes three aspects: First, amplitude consistency comparison, that is, the difference between the sampled values ​​of the two paths at the same timestamp does not exceed the set tolerance range, for example, the current error is not greater than 5 amperes and the voltage error is not greater than 3 volts; Second, waveform trend consistency comparison, that is, the two signals should have the same rising or falling trend within the set time window, and should not have opposite trends; Third, key feature point position comparison, that is, the time difference between the occurrence of peaks, troughs, zero intersections, etc. in the two paths should not exceed 500 nanoseconds. If the redundant path shows more stable, more continuous, or closer to the reference shape in the above three dimensions, the path is judged as a reliable output path; if the original path and the redundant path are consistent, the output of the original path is retained as a valid signal; if both paths show instability, a multi-round sampling verification mechanism is entered or it is marked as temporarily uncorrectable.

[0036] The signal corresponding to the identified reliable output path is used as the basis for judgment within the current time window, and a dynamic feedback recording structure is established to continuously evaluate the impact of interference and the effect of correction. Specifically, after completing path switching or redundancy confirmation, the current channel number, time segment, selection result of the redundant path, consistency comparison score, and residual change value before and after the switch are recorded, and updated in conjunction with the corresponding segment in the interference mapping diagram. If the residual value of the corrected judgment result decreases by more than 30% compared to the original path, it is marked as a successful correction, and the redundant path is prioritized for use in subsequent similar interference scenarios. If the improvement in correction result is limited or switching failures occur repeatedly, the path is marked as a low-reliability path and its usage frequency is reduced. All feedback information serves as input parameters for subsequent judgment logic, path activation strategies, and interference threshold adjustments, forming a closed-loop anti-interference capability evolution link to enhance the system's adaptive response capability under different interference intensities.

[0037] Simultaneously, an interference propagation path diagram is introduced as an auxiliary basis during the path selection process. This propagation path diagram records the propagation path, starting position, and propagation direction of typical interference signals between channels, used to pre-determine the probability that certain signal paths may be affected by interference. When multiple redundant paths are available, the system prioritizes channels not on the main interference propagation path in the propagation path diagram as reliable paths to improve the robustness of the judgment results against interference. Therefore, the interference propagation path diagram is used as a reference basis for path selection priority in this step, forming part of the anti-interference dynamic feedback mechanism.

[0038] This step aims to compare and dynamically correct the output results in real time using redundant signal paths after the status judgment link is affected by interference, ensuring that the circuit breaker can still make accurate judgments based on reliable signals at critical moments. During the high-speed breaking process of a DC circuit breaker, the main signal path may experience output jumps, drifts, or short-term failures due to electromagnetic interference, directly affecting the accuracy of status identification and the timeliness of protection actions. This step first locates abnormal output nodes using a disturbance mapping map, then introduces a redundant sensing path parallel to the main path for synchronous sampling. Within the interference range, the amplitude, trend, and characteristic point consistency of the two signals are compared, ultimately determining the reliable output channel and replacing the original signal for status judgment. Simultaneously, the path selection results are fed back and recorded, providing a basis for subsequent judgment optimization and strategy adjustment. This method not only achieves dynamic self-healing of the fault judgment link but also improves overall anti-interference robustness. Unlike traditional schemes that rely on static redundancy switching or signal filtering, it possesses higher adaptability and precision control capabilities, making it a key technical link in ensuring the reliable operation of circuit breakers in complex electromagnetic environments.

[0039] S005, based on the anti-interference dynamic evaluation feedback mechanism, sends adjustment signals to the logic control system to dynamically set the judgment threshold for fault identification and the parameter weights of the response strategy, thereby limiting the fault judgment process to the time period with the least interference, so as to enhance the accuracy of state identification and the stability of response. Based on the anti-interference dynamic evaluation feedback process completed in the previous stage, the identified signal stability data is applied to the adaptive adjustment of the control logic strategy. By actively setting the criterion threshold, response trigger time period, and judgment signal channel weight, the fault identification process is accurately triggered within the time period of minimum interference, improving the judgment accuracy and control response stability of the circuit breaker in high-interference environments. The steps are as follows: The anti-interference assessment feedback data generated in the previous self-calibration process is read. This includes key parameters such as the effective signal segment, the stable output segment of the redundant path, the minimum residual segment and its timestamp range, the channel consistency score, and the residual change before and after signal drift correction for each status judgment channel within the current circuit breaker operation cycle. The feedback data for each judgment channel is sorted according to the sampling timestamp and cross-validated with the interference mapping map from the previous stage. If no jump, fluctuation, or abrupt residual segments appear in multiple channels within a certain time period, and the redundant path output maintains high consistency with the original path (e.g., the average residual of 500 consecutive sampling points does not exceed 2% of the total amplitude), then this time period is identified as the minimum interference segment within the current operation cycle. The start and end points of this time period are recorded with microsecond-level timestamps to provide accurate boundaries for defining the subsequent fault identification window.

[0040] After identifying the minimum interference segment, the trigger threshold used for fault judgment in the control logic is recalculated based on the signal stability parameters of that time period, and the response judgment is restricted to be completed only within that segment. Taking current identification as an example, the fault current jump threshold set in the original control logic is 15 amps. Within the minimum interference segment, the actual amplitude fluctuation range of the current change is between 10 amps and 18 amps, and the waveform is continuous without glitches, jump signals, or sudden increases in residuals. Therefore, the fault judgment threshold can be lowered to 13 amps based on the consistency between the redundant path and the original path, improving the fault identification sensitivity. In voltage identification, if the original voltage drop threshold is 200 volts, and the actual voltage change range within the minimum interference segment is between 245 volts and 210 volts, the voltage trigger threshold can be temporarily lowered to 220 volts. Meanwhile, to ensure that the identification results are based solely on reliable signal triggers, the trigger response window in the control logic is limited to the period between the start and end times of the minimum interference segment. For example, the judgment is made within the range of the 4000th to 6000th sampling points after the action is triggered, to avoid false triggering in the area affected by interference.

[0041] Based on the aforementioned calculation results, a set of adjustment instructions is generated, including fault identification threshold correction parameters, response time period limitation parameters, and channel output weight adjustment values. This set of parameters is then injected into the control decision-making process through the control logic interface. The adjustment parameters include: the reliability level of the output signal of each judgment channel within the current action cycle (e.g., a 0-100 score), the residual amplitude and stability range of each signal path, and the judgment priority ranking within the time window. If the reliability score of the current channel exceeds 90 points, the voltage channel score is 85 points, and the auxiliary signal score is 95 points, then the current and auxiliary signals can be set as the primary judgment channels, with the voltage signal used as an auxiliary judgment basis, and their weights set to 0.45, 0.45, and 0.10, respectively. All adjustment values ​​are updated in the control logic configuration table in the form of a parameter file and bound to the control strategy instance of the current action cycle to achieve online updates of the dynamic control strategy.

[0042] After applying the above adjustment strategy in the control logic, the actual fault identification results and action response process are tracked, recorded, and their effects evaluated. This includes recording indicators such as whether the trigger time point is within the minimum interference zone, whether the weight distribution of the channel used for final triggering is consistent with the setting, whether the control command is executed on schedule, whether the protection action is effective, whether the action time delay is shortened, and whether false actions are suppressed. The current periodic evaluation results are compared with the response effects when using the default fixed strategy in the past. If the judgment accuracy is improved, the response time is shortened, or the false action rate is reduced, the current adjustment strategy is confirmed to be effective, retained, and promoted for use in subsequent similar circuit breaker opening actions. If the strategy effect is not obvious, it is fed back to the evaluation mechanism for automatic optimization of threshold, weight, or response window parameters. This closed-loop feedback process ensures that each adjustment is based on real-time interference environment data and historical evaluation effects, continuously improving the intelligence, adaptability, and engineering stability of the control strategy through iteration.

[0043] This step applies the results of the previous stage's identification of interference behavior and signal reliability assessment to the fault identification and control decision-making process, achieving dynamic optimization of the fault judgment logic. By analyzing the stability of the state judgment signal over different time periods, the time window with the least interference is selected as the trigger interval for fault judgment. Simultaneously, based on the output consistency and residual changes of each channel, the fault identification threshold and parameter weights of the judgment channels are dynamically set. The control strategy no longer relies on fixed thresholds and static judgment times, but instead makes judgments during the period of least interference and most reliable signals, thereby improving the accuracy of fault identification and the stability of the control response. This method differs from traditional strategy adjustment schemes that use static criteria; it can adaptively adjust according to the interference characteristics in each actual action cycle, ensuring that the circuit breaker maintains a stable, fast, and accurate protection response even under electromagnetic interference. This step is a crucial component in constructing an anti-interference closed-loop control chain and has key technical value for improving the robustness of control decisions and the accuracy of actions.

[0044] S006, based on the interference intensity information and feedback residual distribution data during the opening process of multiple circuit breakers, updates the parameters of the interference identification model, and dynamically adjusts the signal gain of each sensing path through learning optimization, forming a closed-loop control process of coordinated linkage between interference identification, anomaly calibration and response strategy adjustment. To continuously enhance the anti-interference capability of the circuit breaker during multiple consecutive breaking processes, interference intensity information collected in multiple operating cycles and residual distribution data from corresponding state judgment results are summarized and analyzed. Based on this, key judgment parameters used to identify interference behavior are updated. Simultaneously, the signal gain level of the sensing path is dynamically adjusted according to the residual information, achieving closed-loop linkage control between interference identification, signal anomaly calibration, and dynamic adjustment of the response strategy. This constructs an anti-interference judgment mechanism with adaptive optimization capabilities. The steps are as follows: Interference behavior characteristics and state judgment residual information recorded in the actual breaking operation cycles of multiple circuit breakers were extracted to construct a time-series-based behavior dataset. The interference behavior characteristics specifically include the current drop rate (amperes per microsecond), the break voltage rise rate (volts per microsecond), the dominant frequency range of the high-frequency signal (hertz), the interference duration (microseconds), the type of affected signal channel (e.g., current detection, break voltage detection, auxiliary contact feedback), and their start and end positions on the time axis for each cycle. The corresponding state judgment residual information includes the deviation of the judgment output from the baseline curve within that cycle (percentage), the maximum residual value (absolute value of the physical quantity), the average residual value, the duration of continuous deviation, and the affected channel number. All data were organized chronologically and by device number to construct a structured sample set. Each sample should contain complete interference characteristics and corresponding judgment link residual information.

[0045] Based on the aforementioned dataset, the applicability of existing interference identification model parameters was evaluated, and key parameters such as the judgment threshold, judgment range, and recognition sensitivity were updated. Updates included, but were not limited to: adjusting the detection sensitivity for the onset of interference abrupt changes, for example, lowering the lower limit of the current change rate required to identify the interference start point from 50 amperes per microsecond to 40 amperes per microsecond; expanding the acceptable interference identification frequency band coverage, for example, extending the upper limit of the judgment frequency, originally limited to the 3 MHz to 10 MHz range, to 20 MHz to cover typical spike interference signals at higher frequencies; and optimizing the interference duration threshold criterion, for example, relaxing the minimum identifiable disturbance duration from 5 microseconds to 2 microseconds to improve the model's responsiveness to short-term, high-intensity disturbances. These parameter updates should focus on identified cases of incorrect judgments or response delays, identifying areas where the original model parameters were insufficiently adaptable to actual operating conditions, correcting each parameter item by item, and synchronously updating them in the recognition logic structure used in subsequent operation cycles.

[0046] After updating the identification model parameters, the signal gain of each sensing signal path is dynamically adjusted to ensure that the influence weight of high-reliability signals is increased under multi-path redundancy judgment conditions, and the misleading degree of interfered channels is reduced. Gain adjustment should be based on channel performance in historical cycles. For example, in the current channel, if the Hall current sensor output has high residuals across multiple cycles, while the fiber optic current sensor output residuals are significantly lower than the average, the Hall path gain should be reduced from 1.0 to 0.7, and the fiber optic path gain should be increased from 1.0 to 1.3. For parallel channels in the voltage detection path, if the RC voltage divider path has high residuals while the differential amplification path has low residuals, their outputs should be amplified and suppressed according to the same rules. Gain adjustment can be achieved by adjusting the feedback resistor value in the analog front-end circuit, the amplification factor of the signal conditioning unit, or the scaling factor in the digital signal processing. All gain adjustment values ​​must be bound to the current judgment residual record data and continuously updated over time to support the weight redistribution of subsequent response logic.

[0047] After completing the aforementioned parameter updates for the identification model and channel gain configuration adjustments, all changed data, along with the corresponding action cycle number, control strategy version, and identification result accuracy, are stored and loaded as input into the judgment initialization configuration for the next circuit breaker action cycle. This forms a closed-loop optimization control mechanism with historical experience-driven capabilities. In the next cycle, the identification model directly calls the optimized parameter set, and the signal sampling path loads the adjusted gain configuration according to the evaluation results of the most recent cycle. After the action begins, the default parameters are not re-initialized; instead, the optimized parameters participate in the judgment process. Simultaneously, after the cycle ends, the judgment correctness, response time, channel stability, and other indicators are re-evaluated and compared with the previous cycle. If there is further improvement, the new parameters are retained; if the judgment accuracy decreases, a rollback is performed and the strategy is readjusted. The entire closed-loop control process achieves an adaptive evolutionary process from data-driven to response-based adjustment through a logical chain of continuous acquisition—identification evaluation—parameter update—result verification.

[0048] This step aims to construct a closed-loop optimization mechanism with continuous learning capabilities to improve the accuracy of electromagnetic interference identification and the stability of state judgment in DC circuit breakers during frequent opening and closing operations. In actual circuit breaker operation, the frequency, amplitude, and duration of interference vary under different operating conditions. Using fixed identification parameters and static signal gain configurations can easily lead to identification lag, misjudgment, or signal failure, affecting the accuracy of fault clearing and the reliability of protection response. Through this step, after each circuit breaker opening operation, the system actively collects data such as interference intensity, residual changes, and misjudged sections within that cycle, and uses this data to optimize and correct key criteria parameters in the interference identification model. Simultaneously, based on the performance of each channel in historical cycles, the system dynamically adjusts the gain weight of its signal output to enhance the influence of stable signals and suppress the misleading effect of interference paths. Ultimately, a closed-loop process of interference identification, signal anomaly calibration, and judgment response linkage is constructed. The system can continuously optimize its strategy during multiple circuit breaker operations, forming adaptive evolution capabilities and significantly improving the intelligent protection level and engineering practical value of the DC circuit breaker.

[0049] This invention constructs a radiation echo spectrum distribution model during the circuit breaker's breaking process, extracts key disturbance frequency bands to form a set of characteristic signals, and combines dynamic time window observation, interference indicator index analysis, multi-segment residual fitting comparison, synchronous redundant signal correction, adaptive adjustment of control parameters, and cross-cycle model parameter optimization to form a complete closed-loop anti-interference identification and response control mechanism. This method not only significantly enhances the ability of the state judgment link to identify high-frequency transient interference and improves output stability, but also effectively reduces the risk of protection action delays, false triggering, or control link failures caused by sensor signal distortion or judgment errors. Compared with traditional static judgment or passive filtering strategies, this invention has stronger adaptability and real-time performance, and is particularly suitable for complex operating environments such as high voltage and high current, multiple devices in parallel, and frequent switching. It can significantly improve the operational reliability and fault handling accuracy of DC circuit breakers under extreme electromagnetic interference conditions, ensuring the overall safety and intelligence level of the system.

[0050] Table 1: Key Transient Parameter Data Acquired During Circuit Breaker Opening Process Action cycle number Current descent rate (A / μs) Voltage ramp-up rate (V / μs) High-frequency interference main frequency range (MHz) Interference duration (μs) Period 1 65 410 3–7 12 Period 2 72 435 4–8 15 Period 3 69 420 5–9 14 This table records the key physical quantities collected by the DC circuit breaker during three interruption cycles. It can be seen that the current descent rate exceeded 60 A / μs, and the break voltage rapidly increased at a rate of several hundred volts / μs, both triggering concentrated bursts of high-frequency interference (3–9 MHz). This provides a basis for subsequent interference propagation path modeling.

[0051] Table 2: Abnormal sampling points identified by the dynamic time window observation mechanism Channel Category Exception types Time of anomaly occurrence (μs) Duration of anomaly (μs) Maximum offset Abnormal level Current detection channel Amplitude jitter 4.5 1.2 15 A moderate Voltage detection channel Phase jump 6.8 0.8 20 V serious Auxiliary contact feedback Waveform ghosting 8.1 2.0 12% amplitude difference Mild By using dynamic time-window comparison, anomalies were identified in three channels: current, voltage, and auxiliary contacts. The table details the specific time, duration, and intensity of the anomalies, providing precise input for generating an interference indicator index and subsequent residual comparison.

[0052] Table 3: Comparison of Feedback Effects in Dynamic Evaluation of Anti-interference Project Indicators Default fixed strategy This method dynamically adjusts the strategy. Increase Fault identification accuracy 88% 96% +9% Average response delay 2.3 ms 1.5 ms -35% Error rate 7% 2% -71% The comparison shows that this method outperforms the fixed strategy in terms of identification accuracy, response delay, and false trip rate. In particular, the false trip rate is reduced to 2%, which significantly improves the operational reliability of the circuit breaker and the credibility of protection actions under strong interference scenarios.

[0053] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An online testing method for medium and low voltage DC circuit breakers, characterized in that, Includes the following steps: S001 During the interruption operation of the DC circuit breaker, a radiation echo spectrum distribution model is constructed. The interference intensity is estimated by collecting the current drop rate and the voltage rise rate at the break point. The disturbance frequency band where the interference energy is concentrated is extracted and a set of characteristic signals is generated to represent the interference propagation path. S002, based on the feature signal set, sets a dynamic time window observation mechanism in the input link of the state judgment component, and uses the sampling signal within a continuous period to match the preset reference signal sequence to identify sampling points with amplitude jitter, waveform ghosting or phase jump characteristics, and generate interference indicator index; S003, based on the interference indicator index, performs multi-segment fitting residual comparison operation, continuously analyzes the instantaneous change trajectory of the output signal of the state judgment component, marks the residual abnormal segment and generates a disturbance interference mapping map to determine the unstable interval of the state judgment response. S004. Based on the disturbance and interference mapping diagram, perform local self-correction operation, introduce synchronous redundant sensing paths for detection nodes with jump output in the state judgment link, determine the reliable output channel by comparing the consistency of the output signals of each path, and form an anti-interference dynamic evaluation feedback mechanism. S005, based on the anti-interference dynamic evaluation feedback mechanism, sends adjustment signals to the logic control system to dynamically set the judgment threshold for fault identification and the parameter weights of the response strategy, thereby limiting the fault judgment process to the time period with the least interference, so as to enhance the accuracy of state identification and the stability of response. S006, based on the interference intensity information and feedback residual distribution data during the opening and closing processes of multiple circuit breakers, updates the parameters of the interference identification model, and dynamically adjusts the signal gain of each sensing path through learning optimization, forming a closed-loop control process that coordinates interference identification, anomaly calibration and response strategy adjustment.

2. The online detection method for medium and low voltage DC circuit breakers according to claim 1, characterized in that, Step S001 includes: After issuing the interruption control command through the DC circuit breaker, the current in the main conduction path and the voltage changes at both ends of the break are collected. The collected current descent rate and break voltage ramp rate were analyzed in both the time and frequency domains to identify the frequency range where the interference energy density was concentrated, and the center frequency, upper and lower bounds of the bandwidth, maximum amplitude point and total energy were recorded. The frequency analysis results are mapped to the original time series, and interference waveforms that meet the conditions of amplitude, duration, and repetition are extracted to form a set of characteristic signals containing spectral distribution, duration, sampling channels, and original time points. Based on the internal electrical wiring structure of the circuit breaker, the occurrence order, propagation mode and influence path of characteristic signals in different sampling channels are analyzed, and an interference propagation path diagram is constructed to indicate the starting position, propagation direction and range of the interference signal.

3. The online detection method for medium and low voltage DC circuit breakers according to claim 1, characterized in that, Step S002 includes: Status judgment signals are collected in the current detection circuit, voltage detection circuit and auxiliary contact feedback circuit respectively. The sampling device used has high bandwidth and high precision capabilities and is configured with an independent clock synchronization structure to ensure the consistency of multi-channel sampling data on the time axis. A dynamic time window observation process is introduced for each sampling channel. Within a preset time window, the real-time sampling signal is compared point by point with the reference signal sequence of the corresponding channel to identify abnormal sampling features such as continuous amplitude jitter, waveform ghosting, and phase jump at feature points. Interference indicator information is generated for all identified abnormal sampling points, and cross-channel interference events are identified by cross-comparison. By binding interference indicator information with the original sampling data, and inserting labels for anomaly type, location, difference, and suggested handling method into the original data, a structured interference indicator index data table is constructed for subsequent link judgment.

4. The online detection method for medium and low voltage DC circuit breakers according to claim 1, characterized in that, Step S003 includes: Based on the abnormal sampling point number and time information recorded in the interference indicator index, the output signal sequence corresponding to the time period is extracted from the output data of the state judgment component, and the sampling point range is extended forward and backward to ensure coverage of the entire interference process. The extracted signal sequence is divided into continuous data segments. In each segment, a fitting process is performed according to the physical change law of the actual working state of the circuit breaker to generate an ideal response curve. The residual value of each sampling point is calculated to represent the deviation between the actual output and the fitted output. Based on the degree of continuous deviation and abrupt change characteristics of the residual values, abnormal residual segments associated with the interference behavior are identified, and cross-validation is performed in combination with the time records in the interference indicator index. The effectiveness of the interference is confirmed when the time overlap between the two is significant. The confirmed abnormal segments are organized in chronological order and by channel category, and a disturbance mapping diagram is drawn. Different levels of disturbance intensity are identified by color or graphic style to indicate unstable areas in the state judgment and response process.

5. The online detection method for medium and low voltage DC circuit breakers according to claim 1, characterized in that, Step S004 includes: Based on the response interval marked with high residuals in the disturbance mapping diagram, extract the channel number information, the time segment of the anomaly occurrence and the residual intensity level, set the target channel that needs to perform self-correction operation, and extract the corresponding sampling frequency, sampling quantity and data format parameters to ensure that the original signal and the redundant signal are comparable. In each judgment channel that needs to be corrected, a synchronous redundant sampling channel with a different structural path is introduced to keep the sampling control logic time base of the main channel and the redundant channel consistent, and to complete the redundant sampling of the original judgment signal to obtain reference signal data. The judgment signals collected by the original channel and the redundant channel in the interference section are compared point by point. The evaluation is carried out according to the amplitude difference, waveform trend change and feature point time difference. The channel with smaller signal fluctuation, more stable change trend and more accurate feature point positioning is selected as the reliable signal output path for the current time period. The path selection results and comparison data are recorded as anti-interference feedback content, and the channel number, interference time segment, residual information and signal stability level are updated as the basis for subsequent judgment logic adjustment, redundant path activation priority allocation and response threshold dynamic setting.

6. The online detection method for medium and low voltage DC circuit breakers according to claim 1, characterized in that, Step S005 includes: Read the anti-interference dynamic evaluation feedback data, extract the stable signal segment, the segment with the smallest residual, the consistency score of the redundant path and the corresponding timestamp of each judgment channel, and identify the smallest interference segment that has not fluctuated, jumped or changed abruptly. Based on the actual signal fluctuation range of the minimum interference section, the current identification threshold and voltage trigger threshold are adjusted respectively, and the response judgment operation is limited to the time range of the minimum interference section. Calculate the confidence score of the output signal of each judgment channel, combine the residual amplitude and the stable time interval, set the channel output weight, and generate adjustment instructions including threshold correction parameters, response time window and path weight settings; The adjustment instructions are injected into the control logic decision-making process to complete the online update of the control strategy. The fault identification results and response effects are tracked, recorded, and compared with historical strategies to improve the accuracy of judgment and the stability of response.

7. The online detection method for medium and low voltage DC circuit breakers according to claim 1, characterized in that, Step S006 includes: Extract current drop rate, break voltage rise rate, interference main frequency range, interference duration and channel residual information within multiple operation cycles, and construct a time series sample set containing interference intensity and judgment bias. Based on the time series sample set, the detection sensitivity, judgment frequency range and disturbance duration threshold parameters in the existing interference identification model are corrected, and the updated parameters are used synchronously for the identification logic structure in the next cycle. Based on the residual level in the historical period of each sensing channel, the gain configuration of the signal output is adjusted. The influence weight of the reliable channel is increased by adjusting the gain, thereby reducing the degree of misleading interference path. The updated identification parameters and gain configuration are bound to the current action cycle number and control strategy version, and then called during the next judgment initialization process. Iterative optimization is performed based on feedback from subsequent cycle evaluations.

Images