A high-voltage circuit breaker arc monitoring system and method

Abstract

This invention discloses a high-voltage circuit breaker arc monitoring system and method, belonging to the field of power equipment operation status detection technology, including the following steps: S1, synchronously acquiring arc spectral signals and electromagnetic interference vector data under a unified time reference, and performing primary differential processing to generate a stable baseline signal; S2, inputting the stable baseline signal into an interference identification model, generating a multi-dimensional spectrum diagram through time-frequency mapping, extracting the interference time sequence boundary, and outputting an interference weight matrix. This invention constructs an arc monitoring method based on synchronous acquisition of spectral and electromagnetic signals, integrating interference identification, signal purification, energy analysis, and control command generation, and achieving parameter self-optimization through closed-loop feedback, effectively improving the accuracy of arc identification and the efficiency of arc extinguishing response, and enhancing the operational safety and intelligence level of the circuit breaker in a strong interference environment.

Description

Technical Field

[0001] This invention relates to the field of power equipment operation status detection technology, specifically to a high-voltage circuit breaker arc monitoring system and method. Background Technology

[0002] High-voltage circuit breaker arc monitoring refers to the real-time capture, identification, and analysis of high-temperature plasma arcs generated between the contacts of a high-voltage circuit breaker during opening and closing operations or abnormal operation. This is achieved through various sensing methods, including optical, acoustic, electromagnetic, thermal imaging, and gas composition analysis. Arcing is a transient discharge phenomenon accompanied by strong light radiation, impact sound waves, severe temperature rise, and electromagnetic pulses. If not detected and assessed in a timely manner, its high temperature and strong current can damage contacts, insulation media, and mechanical components, and may even lead to equipment failure or power grid accidents. Arc monitoring technology collects characteristic parameters of arcing signals, such as spectral distribution, sound pressure waveform, current distortion rate, infrared temperature rise curve, and partial discharge intensity. Combined with data processing and pattern recognition algorithms, it can determine the time, location, duration, and energy intensity of arcing, enabling health assessment of the high-voltage circuit breaker's operating status and early warning of potential faults, thereby improving the safety and reliability of the power system.

[0003] The existing technology has the following shortcomings:

[0004] During arcing monitoring of high-voltage circuit breakers, when external environmental events such as lightning strikes, electromagnetic pulse (EMP) radiation impacts, or high-current short-circuit accidents occur in the power system, transient interference signals with extremely high amplitudes are generated within a very short time. These transient signals enter the arcing monitoring link through electromagnetic coupling, conductor induction, or reverse conduction to grounding, and superimpose with the actual arcing signal generated inside the circuit breaker in both the time and frequency domains, thus distorting the waveform characteristics, amplitude distribution, and energy density of the monitored signal. Because the instantaneous amplitude of the interference signal is usually much higher than that of the arcing signal, and although its duration is short, it is sufficient to cover the early characteristic range of arcing, the monitoring system may be unable to correctly identify the occurrence of arcing during the feature extraction and event discrimination stages, leading to delayed monitoring response or even complete loss of arcing identification capability. Under these circumstances, the arc will continue to exist and release high-temperature plasma and strong electromagnetic radiation, causing the metal on the contact surface to melt and weld together. The insulation and metal structural components inside the arc extinguishing chamber will be continuously eroded, and the chamber pressure will rise sharply. This may eventually induce shell rupture or explosive failure, which will not only lead to the destruction of the circuit breaker, but also cause serious power grid accidents such as busbar undervoltage and system cascading tripping.

[0005] 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

[0006] The purpose of this invention is to provide a high-voltage circuit breaker arc monitoring system and method to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring arcing in a high-voltage circuit breaker, comprising the following steps:

[0008] S1, synchronously acquires arc spectral signals and electromagnetic interference vector data under a unified time reference, and performs primary differential processing to generate a stable base surface signal;

[0009] S2 inputs the stable base plane signal into the interference identification model, generates a multi-dimensional spectrum diagram through time-frequency mapping, extracts the interference time sequence boundary, and outputs the interference weight matrix;

[0010] S3, based on the interference weight matrix, adaptively adjust the filter kernel coefficients to enhance and suppress the signal, and obtain a dynamically purified arc characteristic signal sequence;

[0011] S4, continuously analyzes the energy of the arc characteristic signal sequence, and predicts the early triggering of protection actions based on the energy growth trend and preset threshold;

[0012] S5 generates control commands based on the trigger prediction results, adjusts the tripping pressure and arc extinguishing airflow to achieve rapid arc extinguishing;

[0013] S6 transmits the entire process operation data back to the identification model and controller, updates the identification parameters and control strategies, and builds a closed-loop optimized arc monitoring mechanism.

[0014] Preferably, step S1 includes:

[0015] A miniature fiber optic spectrometer is installed at the observation window of the arc-extinguishing chamber of the high-voltage circuit breaker to collect spectral signals during the arc-ignition process. The spectral response range is set to 350 nm to 1050 nm, and the sampling frequency is set to 100 kHz.

[0016] A three-axis electric field induction antenna is installed on the outside of the circuit breaker housing, and current transformers are installed at the incoming and outgoing line circuits to collect electromagnetic interference vector data respectively. The three-axis electric field antenna measures the frequency band from 10 Hz to 10 MHz.

[0017] A GPS timing chip is used to uniformly time-mark the collected data, and a unified time reference is used to achieve dual-channel synchronous acquisition and alignment of spectral signals and electromagnetic vector data;

[0018] The primary differential processing is performed, and non-cooperative electromagnetic signals are smoothed or eliminated by linear interpolation within a local time window. Time reconstruction and amplitude normalization are completed to generate a stable base plane signal.

[0019] Preferably, step S2 includes:

[0020] The stable base plane signal is divided into time sequences using a sliding time window of 500 microseconds per segment, and the frequency distribution map in the range of 1 kHz to 5 MHz is obtained by fast Fourier transform.

[0021] The amplification relationship between the energy at each frequency point and the reference energy is calculated based on the sliding window spectrogram, and the start and end times of high-amplitude transient interference signals are identified.

[0022] The timing and frequency range of interference events are mapped to the time axis and frequency band axis to construct an interference weight matrix. The interference level is set according to the interference amplitude and duration and weight coefficients are assigned to guide subsequent signal processing.

[0023] Preferably, step S3 includes:

[0024] Each time point of the interference weight matrix is ​​matched with the sampling point of the stable base plane signal, and the filter window length and weight distribution corresponding to different interference levels are assigned.

[0025] Based on the interference level, the window length and center weight are adjusted in the sliding weighted average filter to achieve smooth suppression of high interference signal segments and detail enhancement of low interference signal segments.

[0026] Amplitude-weighted amplification is performed on signals in the 250 kHz to 2 MHz frequency band, attenuation processing is performed on signals in non-target frequency bands, and amplitude normalization and dynamic overlap reconstruction are performed after filtering in each frequency band.

[0027] The processed signals are sorted and reconstructed while retaining their original timestamps to form a continuously and dynamically purified arc characteristic signal sequence.

[0028] Preferably, the specific steps for performing amplitude-weighted amplification on signals in the 250 kHz to 2 MHz frequency band, attenuation processing on signals in non-target frequency bands, and amplitude normalization and dynamic overlap reconstruction after filtering in each frequency band are as follows:

[0029] The 250 kHz to 2 MHz frequency band signal is extracted from the stable base signal, and amplitude weighting is performed by setting an amplification factor between 1.25 and 1.5 according to the frequency range.

[0030] Signals from the 1 kHz to 250 kHz frequency band and the 2 MHz to 5 MHz frequency band are extracted, and amplitude reduction is performed by setting attenuation coefficients of 0.5, 0.7 and 0.9 according to the interference level at the corresponding time point in the interference weight matrix.

[0031] The signals processed in different frequency bands are normalized and then superimposed and reconstructed to form a frequency-selectively enhanced signal for subsequent arc feature extraction and analysis.

[0032] Preferably, step S4 includes:

[0033] The arc characteristic signal sequence is divided into multiple overlapping sliding windows with a time length of 1 millisecond. The energy of the absolute value of the signal amplitude is calculated within each window to obtain the energy value of the electrical signal for each segment.

[0034] Construct an energy time series, extract the energy change trend of a continuous window, and correct it by combining the amplitude ratio before and after signal enhancement to obtain an accurate energy growth curve;

[0035] If the current energy value exceeds the energy threshold of 2.5 millijoules and the growth ratio exceeds 1.5 times for three consecutive windows, output the early trigger prediction instruction.

[0036] Preferably, step S5 includes:

[0037] Receive the arc energy value, growth rate and risk level information from the early trigger prediction results, and transmit them to the control response structure in a period of no more than 1 millisecond;

[0038] Based on the risk level, retrieve the corresponding tripping drive pressure and arc extinguishing airflow control parameters from the preset response strategy table;

[0039] The corresponding control commands are output to the tripping drive execution unit and the airflow control unit respectively. The drive contacts complete the separation within 10 milliseconds and control the high-pressure airflow to be ejected to form a high-speed airflow along the arc direction.

[0040] Record and transmit the tripping action time, airflow start time, control command issuance time, and feedback status for subsequent correction and optimization of the control strategy.

[0041] Preferably, step S6 includes:

[0042] The system collects operational data including tripping time, execution pressure, airflow velocity, arc energy and waveform, and packages them into feedback data frames after adding timestamps.

[0043] The relevant data in the feedback data frame is input into the recognition structure. If the recognition deviation exceeds the set threshold, the spectrum window width, interference energy threshold and slope factor are adjusted and updated to the default enabled parameters.

[0044] The control-related data in the feedback data frame is input to the response control structure. If the response does not achieve the expected effect, the drive pressure is increased, the airflow is increased, and the action delay is shortened. The corrected version is then set as the default control strategy for the next time.

[0045] A high-voltage circuit breaker arc monitoring system includes a synchronous sensing and baseline construction module, an interference identification and timing extraction module, a signal purification and feature extraction module, an energy analysis and early warning judgment module, a response control execution module, and a data feedback and strategy optimization module.

[0046] The synchronous sensing and datum construction module synchronously acquires arc spectral signals and electromagnetic interference vector data under a unified time reference, and performs primary differential processing to generate stable datum signals.

[0047] The interference identification and timing extraction module inputs the stable base plane signal into the interference identification model, generates a multi-dimensional spectrum diagram through time-frequency mapping, extracts the interference timing boundary, and outputs the interference weight matrix.

[0048] The signal purification and feature extraction module adaptively adjusts the filter kernel coefficients according to the interference weight matrix to enhance and suppress the signal, thereby obtaining a dynamically purified arc characteristic signal sequence.

[0049] The energy analysis and early warning judgment module performs continuous energy analysis on the arc characteristic signal sequence and completes the early triggering prediction of protection actions based on the energy growth trend and preset threshold.

[0050] The response control execution module generates control commands based on the trigger prediction results, adjusts the tripping pressure and arc-extinguishing airflow to achieve rapid arc extinguishing;

[0051] The data feedback and strategy optimization module feeds back the entire process operation data to the identification model and controller, updates the identification parameters and control strategies, and builds a closed-loop optimized arc monitoring mechanism.

[0052] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0053] This invention is based on the synchronous acquisition of spectral and electromagnetic vector signals. It extracts stable baseline signals through primary differential extraction, effectively stripping away underlying noise fluctuations. Combining an interference identification model and multi-dimensional spectrograms, it accurately extracts the temporal boundaries of transient interference and constructs an interference weight matrix, achieving precise separation of arcing and interference during signal processing. Furthermore, based on dynamically purified arcing characteristic sequences, it conducts time-by-time energy progression analysis and pre-converts the prediction results into drive control commands, ensuring that arc extinguishing actions are completed before the arcing energy threshold is reached. Finally, through feedback correction of the entire process's operational data, it achieves continuous iterative optimization of identification parameters and control strategies. This closed-loop mechanism establishes a complete technical chain from "sensing—identification—control—optimization," enabling early perception, accurate identification, proactive response, and dynamic self-learning adjustment of arcing risks even under complex conditions such as multi-source strong interference and high grid load. This effectively avoids circuit breaker damage and systemic power accidents caused by identification failures or response delays, significantly enhancing the operational safety and intelligence level of power equipment. Attached Figure Description

[0054] 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.

[0055] Figure 1 This is a flowchart of a high-voltage circuit breaker arc monitoring method according to the present invention.

[0056] Figure 2 This is a schematic diagram of a high-voltage circuit breaker arc monitoring system according to the present invention. Detailed Implementation

[0057] 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.

[0058] This invention provides, for example Figure 1 The method for monitoring arcing in a high-voltage circuit breaker, as shown, includes the following steps:

[0059] S1. Construct a synchronous dual-domain sensing link to synchronously acquire the spectral signal of the arc inside the high-voltage circuit breaker and the vector data of external electromagnetic interference under the same time reference. Perform primary differential separation processing on the acquired spectral signal and vector data to generate a stable base surface signal for subsequent analysis.

[0060] To achieve high-precision and high-stability identification and monitoring of arcing behavior of high-voltage circuit breakers during operation, a dual-channel synchronous sensing method is adopted to separately acquire the spectral signal at the time of arcing and the vector signal of external electromagnetic interference. Signal differential and separation processing is then performed under a unified time reference to construct a stable base signal required for subsequent data analysis. The specific process includes the following steps:

[0061] A miniature fiber optic spectrometer is installed at the observation window of the arc-extinguishing chamber of the high-voltage circuit breaker to collect the strong light radiation signal generated during the arc discharge process. The spectrometer's spectral response range is set to 350 nm to 1050 nm to cover the common emission bands of arcing, and a quartz lens with an incident light aperture of 1.0 mm is configured to enhance the incident light intensity. The sampling frequency of the spectral data is set to 100 kHz to ensure the capture of the abrupt waveform characteristics at the initial stage of arcing. Simultaneously, a triaxial electric field induction antenna is installed outside the circuit breaker housing to collect electric field changes caused by external events such as lightning strikes, electromagnetic pulses, and transient short circuits in cables; and current transformers are installed at the incoming and outgoing circuits of the circuit breaker to detect conductor coupling interference signals caused by sudden changes in large currents in the line. The triaxial electric field antenna's measurement frequency band covers 10 Hz to 10 MHz, and the current transformer adopts a high-frequency toroidal ferrite structure with excellent high-frequency response characteristics.

[0062] To achieve synchronous acquisition of the two signals, GPS timing chips are configured in both the spectrometer and the electromagnetic induction device. A dedicated time reference control unit writes a unified nanosecond-level timestamp into each frame of signal data. Data acquired from each signal channel is sent to a high-throughput edge computing terminal in real time and stored in dual-channel aligned pairs: "spectral signal-timestamp" and "electromagnetic vector-timestamp." UTC satellite time is used as the time reference to achieve microsecond-level synchronization accuracy, ensuring complete alignment of subsequent signals in the time dimension. With this structure, even if the start times of the spectral signal and the electromagnetic interference signal are slightly different, complete synchronization can be achieved through a unified time reference.

[0063] Differential separation processing was performed on the synchronously acquired data. The arcing spectral signal exhibits a rapid increase in intensity, a shift in the main peak, and multi-order spectral line expansion at the moment of discharge, while electromagnetic interference signals manifest as pulse packets, short-period high-amplitude jumps, or broadband harmonic enhancement. By comparing the time intervals in the spectral signal showing continuous intensity increases with the corresponding electromagnetic signals within those time intervals, electromagnetic signal variations with non-corresponding slope changes, extremely large amplitudes, but extremely short durations were identified. These variations typically characterize external interference pulses rather than actual arcing behavior. In specific processing, the electromagnetic vector signal was subjected to split-axis projection, and the correlation between the rate of change of energy density in each direction and the spectral change trend was calculated to determine whether it constituted actual arcing coordination characteristics. For the identified non-coordinated signal components, a linear interpolation method based on a local time window was used for smoothing correction or removal, thereby preserving the original observation data strongly correlated with arcing behavior.

[0064] The differentially processed spectral and electromagnetic signals are subjected to time reconstruction and amplitude normalization to form a stable baseline signal. This stable signal has three key characteristics: first, it is completely aligned with the original signal in the time dimension, avoiding data jitter or loss of synchronization; second, it eliminates high-frequency abrupt interference components in amplitude, exhibiting a smoother characteristic transition; and third, it retains the true changing trend in the early stage of arcing, including the initial increase in spectral intensity, the peak shift process, and the continuous characteristics of electromagnetic disturbances. This stable baseline signal will serve as the basic input signal for subsequent interference identification, feature extraction, energy calculation, and control response, providing high-quality, low-interference, and highly continuous data support for arcing monitoring.

[0065] The core function of this step is to provide a high-quality, low-interference, and time-consistent signal input foundation for arc detection in high-voltage circuit breakers, thereby significantly improving the accuracy and reliability of subsequent identification, analysis, and response processes. During opening and closing operations or abnormal operation, high-voltage circuit breakers may generate high-temperature plasma arcing between their contacts. This process is accompanied by strong light radiation, transient electromagnetic disturbances, and complex nonlinear signal characteristics. However, in actual operating environments, external disturbances such as lightning strikes, electromagnetic pulses, and power system short circuits can simultaneously generate transient strong interference signals with huge amplitudes and wide spectrums. Once these interference signals are superimposed on the arc detection link, they can easily mask the key characteristics of the true arc signal, making it difficult for the monitoring system to accurately identify whether an arc has occurred, thus causing response delays or even misjudgments.

[0066] This step constructs a synchronous dual-domain sensing link, acquiring raw signals from the spectral domain (characterizing arcing) and the electromagnetic vector domain (characterizing interference) respectively. Relying on a unified time reference, microsecond-level synchronous sampling is achieved, effectively preserving the temporal characteristics of each signal. Based on this, primary differential separation processing, using correlation analysis between spectral intensity variation trends and electromagnetic signal amplitude jump characteristics, effectively eliminates external interference components not belonging to the arcing process, thereby extracting the true and pure arcing characteristic signal. The resulting stable baseline signal possesses good continuity, anti-interference capability, and alignment capability, serving as the core data source for subsequent energy analysis, protection trigger prediction, and response control decisions. It is the key starting point and technical foundation of the entire arcing monitoring process.

[0067] S2, input the stable base plane signal into the transient interference identification model, generate the corresponding multi-dimensional spectrum diagram through time-frequency mapping, determine the time boundary of the transient interference signal based on the change characteristics in the spectrum diagram, and output the interference weight matrix consistent with the time axis;

[0068] To accurately identify and quantitatively analyze transient interference signals generated during the operation of high-voltage circuit breakers, and to improve the effectiveness and anti-interference capability of arc monitoring signals, the stable baseline signal after primary differential processing undergoes analytical processing in both frequency and time dimensions. This extracts the distribution range of the interference signal along the time axis and constructs interference weight information corresponding one-to-one with the signal data for subsequent dynamic enhancement and filtering of characteristic signals. This processing includes the following steps:

[0069] The stable baseline signal is input in segments according to continuous time sequence and then subjected to frequency expansion processing. In specific implementation, a sliding time window of 500 microseconds is used to divide the stable baseline signal into multiple non-overlapping time segments. Each signal segment is converted into frequency domain form using Fast Fourier Transform (FFT) to obtain a frequency distribution map in the range of 1 kHz to 5 MHz. During the conversion, a Hamming window function is used for window smoothing to suppress spectral leakage and maintain good frequency resolution. To improve signal continuity, a 250-microsecond overlap zone is set between adjacent processing windows to achieve sliding analysis in the time domain. The final output spectrum results are arranged in chronological order to form a two-dimensional spectrum, where the horizontal axis represents the passage of time and the vertical axis represents the frequency distribution. The grayscale or color intensity in the spectrum represents the signal energy value at each frequency point, in decibels per milliwatt (dBm).

[0070] Time-series analysis of energy distribution changes in the two-dimensional spectrogram is performed to identify the occurrence period and impact range of transient interference. In practical processing, a reference energy baseline value is first determined, which can be taken from the average spectral intensity of the signal 5 milliseconds before the disturbance. The energy amplification factor is then calculated based on this baseline value. If the energy value of a certain frequency at a certain time point exceeds twice the baseline value and lasts for more than 300 microseconds, that time period is identified as a suspected interference period. If the spectral coverage bandwidth of this abrupt change region exceeds 20 kHz, and similar amplification occurs simultaneously in more than three frequency bands, the event is further identified as high-amplitude transient interference. All suspected interference events are recorded one by one by marking the start time, end time, and frequency coverage range of each interference event, and are numbered and categorized. To verify the actual impact range of the interference, superposition analysis can be performed in conjunction with the energy change trend of the arcing spectral signal to confirm the existence of corresponding abrupt changes. The above method can effectively distinguish spectral disturbances caused by typical interference behaviors such as lightning strikes, current surges, and electromagnetic radiation shocks.

[0071] Based on the identified interference periods and frequency ranges, an interference weight matrix is ​​constructed to guide the dynamic processing of subsequent signals. The horizontal axis of this matrix represents the time series with the same number of sampling points as the stable baseline signal, and the vertical axis represents the defined frequency band division interval, for example, dividing 1 kHz to 5 MHz into 50 equal-width bands. Each cell in the matrix represents the interference intensity level of a specific frequency band within a given time segment. Based on the measured amplitude and duration of energy fluctuations, the interference intensity is divided into three levels: low interference (amplitude increase less than 1.5 times the baseline value), medium interference (amplitude increase between 1.5 and 3 times), and high interference (amplitude increase exceeding 3 times and duration exceeding 1 millisecond). Each level is assigned a different weight coefficient, such as 0.2, 0.5, and 0.8, to adjust the degree of subsequent signal filtering and enhancement. The resulting interference weight matrix will dynamically change over time, fully covering the energy evolution process of each transient interference event. It provides both spectral distribution characteristics and temporal correspondence capabilities, providing accurate basic data support for subsequent arcing characteristic signal extraction, energy assessment, and protection triggering decisions.

[0072] The transient interference identification model is an interference judgment structure built based on time series spectrum analysis. Its input is the frequency distribution data of a stable baseline signal within a unit time window, and its output is the interference intensity weight value corresponding one-to-one with the time axis. This identification process relies on the following physical logic: the signal is segmented into fixed time windows, and the frequency energy spectrum of each segment is obtained through Fast Fourier Transform (FFT). Subsequently, the rate of energy change in adjacent time windows within a specific frequency band is compared in the spectrum, and it is determined whether there is a sudden increase in amplitude. If the energy jumps significantly within a short period of time and occurs synchronously in multiple frequency bands, it is judged as transient interference. This identification structure does not rely on deep learning or machine learning model training processes; it is entirely built based on frequency analysis and the continuity of time, possessing good repeatability and engineering feasibility.

[0073] The main function of this step is to identify and quantify the impact of transient electromagnetic interference during the arcing monitoring of high-voltage circuit breakers, and to map this impact into weighted parameters that can be used for subsequent signal processing in a time-consistent manner. This provides clear boundary and control criteria for enhancing the arcing characteristic signal and suppressing interference components. In the actual operating environment of high-voltage circuit breakers, disturbances such as lightning strikes, electromagnetic pulses, and high-current short circuits in power systems can introduce high-amplitude, high-frequency transient interference signals in a very short time. These interference signals often superimpose on the original arcing signal in the form of pulse waves, harmonic clusters, or broadband abrupt changes, causing spectral energy distortion and time characteristic drift, which seriously interferes with the accuracy of arcing monitoring.

[0074] By inputting the differentially processed stable baseline signal into the transient interference identification model and performing spectral analysis under a unified time reference, the performance characteristics of the aforementioned interference in both the time and frequency domains can be effectively captured. Specifically, indicators such as energy abrupt change regions, frequency band jump trends, and peak drift rates in the spectrum can accurately reflect the start time, duration, and frequency coverage of the interference signal. Further, by locating and marking these abrupt change features, the boundary positions of the interference signal on the time axis can be extracted. Then, this boundary information is mapped to specific frequency bands to construct an "interference weight matrix" synchronized with the original signal. This matrix not only clearly marks when and at what frequency the interference occurred but also quantifies the interference intensity level, forming a "risk guidance map" for subsequent signal processing.

[0075] Therefore, the role of this step is not limited to identifying the interference itself, but more importantly, expressing the interference event in a structured, quantified, and aligned manner, thereby enabling precise support for subsequent dynamic filtering, adaptive enhancement, and protection action triggering of the arcing signal. This proactive interference identification mechanism significantly enhances the immunity and practicality of the entire arcing monitoring system in complex electromagnetic environments, and is a key link from "perception" to "reliable judgment."

[0076] S3, based on the interference weight matrix, adaptively and dynamically adjust the filter kernel coefficients in the signal processing process, perform enhancement processing of the arcing signal and suppression processing of the interference signal at the microsecond time resolution, and obtain a continuously dynamically purified arcing characteristic signal sequence.

[0077] To achieve accurate extraction of arcing signals from high-voltage circuit breakers under complex interference backgrounds, the filtering kernel parameters used in signal processing are dynamically adjusted based on the interference weight matrix constructed in the previous steps, combined with the signal's temporal and frequency domain characteristics. This achieves coordinated processing of arcing signal enhancement and interference signal suppression, ultimately yielding a continuous, clear, and low-noise arcing characteristic signal sequence. The technical process includes the following steps:

[0078] The two-dimensional structure of the interference weight matrix is ​​strictly matched to the stable baseline signal. In the specific implementation, the time sampling frequency of the stable baseline signal is set to 1 MHz, forming 1 million sampling points per second; the interference weight matrix is ​​generated with the same time step, so that each sampling point corresponds to an interference intensity coefficient. The frequency axis is divided into 100 consecutive bandwidth intervals from 1 kHz to 5 MHz, each interval being 49 kHz. The interference weight values ​​are divided into three levels: 0.1, 0.5, and 0.9, representing low interference, medium interference, and high interference states, respectively. These weight values ​​will serve as a processing reference and be used as input for kernel function parameter tuning in the subsequent filtering stage.

[0079] Multiple fixed-structure sliding weighted average filters are configured in the signal processing channel for real-time processing of arcing signals. The basic structure of the filter is a symmetrical window structure, initially set as an 11-point smooth weighted structure. Based on the interference weight value at the corresponding time point, the kernel length and weight distribution within each time period are dynamically adjusted: when the interference level of a certain frequency band in the corresponding time interval is 0.9, the window length is expanded to 21 points, and the center weight value is relatively reduced to enhance filtering smoothness; when the interference level is 0.1, the window length remains at 11 points, and the center weight value is increased to improve detail retention. Before each filtering process, the kernel parameters are updated once according to the interference level corresponding to the current time point to ensure that the signal processing has high real-time performance and adaptability.

[0080] While enhancing the arcing signal, selective frequency band suppression is performed. Based on field experiments, the arcing signal energy is mainly concentrated between 250 kHz and 2 MHz. Therefore, for this frequency band, amplitude-weighted amplification is applied after filtering, with an amplification factor set between 1.25 and 1.5. For the 1 kHz to 250 kHz and 2 MHz to 5 MHz frequency bands, attenuation is performed in stages according to interference weight values, with attenuation factors set to 0.5, 0.7, and 0.9, respectively. In practice, all frequency bands are extracted using bandpass filters, and then amplitude scaling and normalization are performed before re-superimposition to ensure a smooth dynamic transition between the enhancement and suppression sections, without introducing abrupt distortion at cross-band intersections.

[0081] The processed signal data is reconstructed into a time-series arcing characteristic signal sequence. An amplitude normalization process is performed before output to standardize the signal between 0 and 1, facilitating subsequent quantification analysis and feature extraction. To ensure the signal's temporal consistency with the original sampling, all data retains the original timestamp information during filtering. After processing, the signal is reconstructed in time-axis order to ensure no phase drift or sampling misalignment occurs. The final output signal possesses the following technical characteristics: 1) It retains the abrupt change edges and peak positions of the arcing signal; 2) It removes high-amplitude interference in non-target frequency bands; 3) It has high dynamic response capability, suitable for real-time identification of microsecond-level events. This arcing characteristic signal sequence will serve as direct input data for energy progression analysis and protection action prediction.

[0082] The purpose of this step is to dynamically adjust the filter kernel function parameters of the signal processing stage on a microsecond-scale using the interference weight matrix extracted in the previous step. This enhances the real arcing signal and selectively suppresses superimposed interference signals, ultimately obtaining a continuous, stable, and low-noise arcing characteristic signal sequence. During the operation of high-voltage circuit breakers, arcing, as a typical transient discharge phenomenon, exhibits characteristics such as rapid rise times, abrupt changes in high-frequency components, and concentrated spectral lines. However, external interference sources such as lightning strikes, electromagnetic pulses, and cable-induced interference generate a large amount of high-amplitude, high-energy noise signals on the same time scale. These interference signals easily overlap with the real arcing signal in both the time and frequency domains, significantly interfering with arcing identification.

[0083] This step dynamically adjusts the filter parameters for each time point using the time-frequency dual-dimensional interference intensity information provided by the interference weight matrix. This includes adjusting the sliding window length, weighting coefficient distribution, and frequency band selection range. For periods with strong interference, the system increases filter smoothness to suppress sudden non-target fluctuations; while for periods with weak interference, it maintains a smaller window size to preserve signal details. This adaptive filtering mechanism, unlike traditional static filtering methods, not only identifies and responds to sudden and nonlinear changes in interference behavior but also maximizes the preservation of arcing intrinsic characteristics at key signal structures such as rising edges, main peak regions, and energy increments, avoiding excessive suppression of the target signal.

[0084] The final output of a continuous dynamic purification arc characteristic signal sequence can accurately reflect the start time, duration, and energy intensity changes of an arc event. It boasts comprehensive advantages of high time accuracy, high spectral fidelity, and low interference error, providing a highly reliable data foundation for subsequent arc energy analysis, fault level assessment, and active protection response. Therefore, this step is the core link in transforming the sensed data from raw, mixed signals into clear, interpretable arc information, and is a key technological bridge for realizing high-voltage arc monitoring from sensing to identification.

[0085] S4 performs continuous energy analysis on a time-by-time basis on the continuously dynamically purified arc characteristic signal sequence, calculates the energy growth trend of the arc in real time in combination with the signal enhancement results, and completes the early triggering prediction of protection actions based on the relationship between the trend and the preset energy threshold.

[0086] To achieve dynamic energy assessment and risk trend identification of the arcing process inside high-voltage circuit breakers, after obtaining a continuously dynamically purified arcing characteristic signal sequence, time-by-time continuous energy analysis is performed. This analysis is then combined with the energy growth trend and a pre-set energy threshold for judgment, thereby enabling early warning of protection actions. This technical process includes the following steps:

[0087] The arcing characteristic signal sequence is segmented into time windows of 1 millisecond length, each containing 1000 sampling points. Windows are updated using a 50% overlap sliding method, meaning an evaluation is performed every 500 microseconds. Within each window, the absolute values ​​of the signal amplitudes are summed and multiplied by the time interval (1 microsecond) to calculate the integrated energy of the electrical signal for that time period. The resulting energy unit is volt-second (V·s), which can be converted to an equivalent relative energy value. Each energy value is used as a reference point for the arcing energy within that time period to continuously track energy changes.

[0088] An energy time series was constructed, recording the energy values ​​over 10 consecutive sliding windows to create a trend chart of energy changes based on actual measurements. At the start of each new window, the latest energy data was updated, and the oldest data point was discarded, maintaining a constant sliding window length. The energy increase per unit time was extracted by calculating the energy difference between two consecutive time points. Simultaneously, the direction and rate of this increase were recorded. If the energy increase is positive and gradually increases over three consecutive windows, it indicates that the arcing state is in a rapid growth phase.

[0089] The energy value of each window is compared with the average energy value of the previous 10 windows to calculate the relative growth ratio. If this ratio exceeds 1.5 times and the number of signal peaks within that window is more than three, it indicates that the signal segment has exhibited typical behavior of intensified arcing. Based on this, to avoid numerical errors caused by signal enhancement processing, a restorative correction is performed on the energy value of the signal enhancement segment. The correction method is as follows: the ratio of the average amplitude of the original signal before enhancement processing to the average amplitude after enhancement processing is used as a correction factor. This factor is multiplied by the energy value obtained after enhancement to obtain the corrected energy reference value for that segment, ensuring the accuracy of the actual assessment.

[0090] Early triggering judgment for protective actions. The critical criterion for energy change is set as follows:

[0091] 1) In three consecutive sliding windows, the energy growth ratio exceeded 1.5 times;

[0092] 2) The total energy value of the current window exceeds the set absolute energy threshold of 2.5 millijoules. This threshold is determined based on the melting energy threshold of the circuit breaker contact metal material, ensuring that the value has an engineering basis. When both of the above conditions are met simultaneously, it is determined that the current arcing has a risk of continuous aggravation and may cause structural damage, and an early trigger prediction command is issued to provide a basis for the subsequent control structure to issue tripping or enhanced arc extinguishing actions.

[0093] The current judgment result is output as a signal, and the original signal, calculated energy value, growth ratio, growth rate, and correction factor of the current time window are recorded simultaneously and stored in the operation log table. This data table serves as the traceability basis for subsequent operation and maintenance systems and is also used for future optimization and adjustment of energy threshold parameters and judgment criteria. Through continuous updates and verification, this method can construct a dynamic energy identification system with self-adjustment and engineering adaptability, further improving the intelligence level and reliability of the entire arc monitoring and protection mechanism.

[0094] The main function of this step is to achieve real-time dynamic monitoring and trend judgment of the arcing state inside the high-voltage circuit breaker, and to make advance predictions of protection actions before the arcing energy reaches a disaster-causing level, providing a decision-making basis for response measures such as rapid contact tripping and enhanced arc-extinguishing airflow. Since arcing is essentially an instantaneous high-temperature, high-energy discharge process, its energy release behavior has obvious stages and progression. Often, in the initial stage, the arc intensity is still within a controllable range, but if it is not identified and handled in time, its continuous growth may lead to serious faults such as contact melting, insulation breakdown, or even cavity explosion. Therefore, simply judging whether arcing has "occurred" is insufficient; more crucial is the ability to capture the trend of rapidly increasing arcing energy.

[0095] This step involves dividing the continuously dynamically purified arc characteristic signal sequence into time windows and performing energy integral calculations at the millisecond level to construct an arc energy change curve that evolves over time. Simultaneously, by calculating the energy growth rate between adjacent time periods and combining it with the amplitude amplification ratio during signal enhancement processing, energy correction is applied to the signal to ensure the final judgment has high physical accuracy and engineering reference value. Next, this growth trend is compared with preset energy thresholds (such as the energy corresponding to the material's melting point or the equipment's critical load-bearing capacity) to determine whether the current arc state is rapidly evolving towards a dangerous direction.

[0096] If the assessment indicates that the arcing energy continues to rise within a short period and exceeds the set growth threshold and absolute energy limit, the system will provide an estimated signal to trigger protection actions in advance. This signal is not a simple passive response, but is generated based on "trend" and "prediction" logic, possessing feedforward control capabilities. This helps the power system to perform rapid current interruption and enhanced arc extinguishing operations before the arcing causes substantial damage. In summary, this step not only enhances the ability of arcing monitoring to move from identification to prediction, but also builds a bridge from data to behavioral control, making it a key link in achieving intelligent and highly reliable power protection.

[0097] S5 transmits the early trigger prediction results to the hierarchical response controller. The hierarchical response controller generates corresponding drive control commands based on the real-time arc energy threshold and controls the contact opening execution pressure and arc extinguishing gas output flow to achieve rapid arc extinguishing.

[0098] To achieve rapid response and proactive control of the arcing state, the advance triggering prediction results are transmitted in real time to the hierarchical response control structure. Specific control commands are dynamically generated based on the arcing energy level assessment results, thereby adjusting the driving pressure for contact tripping and the output flow rate of the arc-extinguishing gas to quickly extinguish the arc and prevent further energy accumulation. The entire control response process includes the following steps:

[0099] The early trigger prediction results output by the arcing energy analysis module are transmitted to the control response structure. These prediction results include the current arcing duration (in milliseconds), the current cumulative energy value (in millijoules), the energy growth rate (in millijoules / millisecond), and the risk level determination result (labeled as Level I, Level II, or Level III). Information transmission is completed via an industrial Ethernet communication interface, with a transmission cycle of no more than 1 millisecond to ensure real-time data transmission. The risk level determination criteria are as follows: Level I: energy not exceeding 60% of the total threshold; Level II: between 60% and 90%; Level III: exceeding 90%. This level classification is used to determine the strength and priority of the response strategy.

[0100] Based on the received risk level and current energy value, the control structure internally selects the corresponding control scheme from the preset response strategy table. The control strategy table presets three fixed response levels: At risk level I, the system maintains normal response parameters, the contact tripping drive pressure is set to the standard rated value (e.g., 3.5 MPa), and the arc-extinguishing airflow maintains a normal flow rate (e.g., 25 L / s); at risk level II, the contact tripping pressure is increased to 1.4 times the rated value (e.g., 4.9 MPa), and the arc-extinguishing airflow increases to 35 L / s; at risk level III, the contact tripping pressure is set to 2.0 times the rated value (e.g., 7.0 MPa), and the airflow increases to 50 L / s. All response strategies are triggered via a lookup table, and the response parameters are immediately loaded into the control output port. The command signal is transmitted to the tripping execution unit and the airflow control unit via the digital output interface.

[0101] After receiving the control signal, the tripping drive actuator starts the high-voltage electromagnetic thruster or servo motor located along the contact axis. This actuator is connected to the transmission assembly and begins to generate a separation force within 1 millisecond of receiving the high-level signal, completing the tripping action with a contact opening distance of 8 mm within no more than 10 milliseconds. During this process, the tripping speed is no less than 1 m / s, ensuring that the arc channel is rapidly extended and the arc voltage is increased, ultimately leading to the natural extinction of the arc. Simultaneously, after receiving the flow command, the arc-extinguishing gas flow unit controls the high-pressure gas in the gas tank to be rapidly released to the nozzle through the solenoid valve. The nozzle diameter is set to 2.5 mm, and the guide channel length is 100 mm. The released high-pressure gas forms a high-speed gas flow along the arc direction in the arc-extinguishing chamber, with a flow velocity greater than 80 m / s. This immediately reduces the arc temperature after contact with the arc and disrupts the plasma sustaining environment.

[0102] Throughout the entire control process, the control structure records the contact opening time, the arc-extinguishing airflow response start time, the command issuance time, the feedback confirmation signal, and the response result. This data is archived in a timing log file and uploaded to the central monitoring equipment every 10 milliseconds for long-term archiving. If the feedback signal indicates that the contacts did not separate within the set time or the airflow did not reach the set flow rate, the response is marked as "failure," and a fault diagnosis process is initiated. All response data will be used to subsequently revise the parameter reference values ​​in the control strategy table, improving the targeting and success rate of the next round of arcing event responses.

[0103] This step aims to establish a real-time linkage control mechanism between the predicted arcing energy and the execution of arc extinguishing actions. This enables high-voltage circuit breakers to proactively extinguish arcs before the risk of arcing escalates into a fault, thereby preventing equipment damage, system tripping, and even grid cascading accidents. During high-voltage circuit breaker operation, once an arc forms, it rapidly affects the contacts, insulation, and metal components with extremely high temperature and energy density. Although previous steps have determined through signal identification and energy assessment that the arc is continuously growing, effective protection cannot be achieved if this information cannot be promptly translated into a clear and executable control action. Traditional overcurrent or overvoltage protection typically has a certain delay, and its action logic is based on the premise that a fault has already formed. This step, however, emphasizes triggering the execution response in advance based on "trend prediction," possessing stronger feedforward control capabilities.

[0104] This step involves inputting the "early trigger prediction result" into the response control structure via data communication. Based on the determined arcing energy level (e.g., Level I, Level II, Level III), the corresponding drive control strategy is selected in real time, and clear control parameter instructions are output, including the required gas pressure for tripping, the electromagnetic drive intensity, and the specific flow rate of the arc-extinguishing gas. For example, when a high-risk Level III arc is detected, the control structure can rapidly generate a drive signal increased to twice the rated pressure, ensuring high-speed separation of the contacts in a very short time and simultaneously releasing a high-speed, high-pressure gas flow to form an arc-extinguishing gas field that disrupts the plasma arcing channel. This linkage not only ensures that tripping and arc-extinguishing actions are triggered simultaneously but also automatically adjusts the action intensity according to the risk level, achieving "on-demand response."

[0105] Furthermore, this step also includes the function of monitoring and providing feedback on the response process. The control structure can determine whether the response action was accurately completed based on parameters such as the actual tripping time, airflow response time, and execution feedback signal, and mark and record the control strategy to provide a data foundation for subsequent optimization. In summary, this step is a key link in the entire arc monitoring and control chain that efficiently transforms "early warning information" into "execution behavior," achieving a closed loop between monitoring, judgment, and response. It is one of the core technologies for ensuring the safe operation of high-voltage equipment and the efficiency of protection actions.

[0106] S6 synchronously transmits the operating data of the entire process of contact tripping and arc extinguishing back to the transient interference identification model and hierarchical response controller, and iteratively updates the interference identification parameters and response control strategies respectively, forming a continuously closed-loop optimized high-voltage circuit breaker arc monitoring mechanism.

[0107] To improve the accuracy of arc monitoring and the effectiveness of response in high-voltage circuit breakers, a closed-loop optimization mechanism was designed to synchronously transmit and use the operational data generated throughout the entire process of contact tripping and arc extinguishing for parameter updates and response control strategy correction. This mechanism drives the continuous evolution of the monitoring link through operational feedback, enabling the monitoring structure to adapt to changes in field conditions and improving the system's stability and adaptability under multiple disturbances and scenarios. The optimization process includes the following steps:

[0108] The system comprehensively collects all key operational data from the early trigger judgment to the end of the arc extinction process, and packages the data into feedback data frames in a unified timestamp format. The specific data collected includes the following nine categories:

[0109] 1) Contact tripping start time (in microseconds);

[0110] 2) Contact tripping completion time (in microseconds);

[0111] 3) Actual output pressure of the drive actuator (unit: megapascals);

[0112] 4) The initial release time of the arc-extinguishing gas flow;

[0113] 5) Actual flow velocity of the arc-extinguishing airflow (unit: liters / second);

[0114] 6) The time for determining the extinction of the arc;

[0115] 7) Total arc extinction duration (in milliseconds);

[0116] 8) Current total energy value of the arcing signal (in millijoules);

[0117] 9) Final arc signal waveform curve segment.

[0118] The above data is recorded synchronously through local acquisition devices and uploaded to the data processing area of ​​the main control structure within 30 milliseconds after the action is completed. To ensure the consistency and matching of the data sequence, each data entry is accompanied by a 13-digit timestamp, identified in the format of year-month-day-hour-minute-second-millisecond, to ensure the correct timing of subsequent processing.

[0119] The arc detection-related portions of the feedback data frame—primarily including signal waveform, signal strength, arc extinction duration, and trigger point time difference—are synchronously input into the arc detection structure for comparison of the accuracy of the detection with the interference separation parameters. If the actual arc extinction time differs from the predicted time by more than 3 milliseconds, or if the detection start point deviates from the actual tripping point by more than 5 milliseconds, the system will readjust the spectral window width, interference energy threshold, and time abrupt change slope factor based on the actual arc extinction time. Specific adjustments include: expanding the sliding window for spectral analysis from 1000 microseconds to 1200 microseconds; increasing the interference energy trigger threshold by 20% from the original average value; and extending the transient change judgment segment from two consecutive points to three points. These adjustments will be stored as "corrected detection parameters" in the detection parameter configuration table and updated to the default enabled parameters before the next arc event.

[0120] The portion of the runtime data frame related to the control execution effect is fed back to the response control structure to verify whether the actual action achieves the predetermined execution effect, and the control parameter table is corrected accordingly. The specific judgment criteria are as follows: if the contact tripping time exceeds 15 milliseconds, or the actual flow rate of the arc-extinguishing gas does not reach more than 90% of the set value, or the arc energy does not decrease by more than 50% rapidly within 10 milliseconds after the response action, the system will determine the current response strategy as "inefficient response" and upgrade the corresponding parameters. For example, the tripping drive pressure will be increased from the original setting of 5.0 MPa to 5.5 MPa; the arc-extinguishing gas flow rate will be increased from 35 L / s to 42 L / s; and the solenoid valve opening delay will be shortened from 2 milliseconds to 1 millisecond. The corrected response parameters are named with the version number "R_timestamp" and the correction basis, original value, and adjustment range are recorded. This version will be automatically loaded as the default strategy for the next event response. If the response effect is good for three consecutive times, it will be fixed as the basic control strategy.

[0121] This step aims to implement a data-driven adaptive optimization mechanism for the high-voltage circuit breaker arc monitoring system. By structurally collecting and feeding back the operational data generated throughout the entire process of circuit breaking and arc extinction, the accuracy of arc signal identification and the control strategy for response actions can be continuously iterated and updated based on real operational results, thereby constructing a monitoring and control system with closed-loop learning capabilities. In practical applications, the signal morphology and energy release characteristics generated by arcing are affected by various factors such as equipment structure, power grid disturbances, and climate, which have certain uncertainties and dynamic changes. If the monitoring and identification parameters and control response strategies rely on static settings for a long time, they may fail to adapt to changes in the field, leading to identification deviations or response lags, thereby reducing the overall system stability and protection efficiency.

[0122] This step involves collecting key operational data during each circuit breaker operation, including contact opening start time, operation time, arc-extinguishing airflow response delay, peak flow rate, arc-extinguishing completion time, arc-extinguishing curve slope, and total energy dissipation. This data is then structured and uploaded to the transient interference identification and response control structures. This enables automatic correction of key identification parameters (such as spectral window width, interference judgment threshold, and spectral abrupt change boundary) and dynamic adjustment of response strategies (such as driving pressure, airflow intensity, and operation rhythm). This feedback mechanism ensures that the system can optimize future monitoring, identification, and control execution logic based on "actual performance" rather than "theoretical presets."

[0123] For example, if the arcing duration during an arc extinguishing process is found to be significantly longer than estimated, the system will retrospectively identify whether the starting point was delayed and correct the starting position of the window used for spectrum identification or extend the identification duration. Similarly, if the arc-extinguishing gas flow fails to reach the target flow rate within a specified time during a response, the control strategy will automatically increase the gas flow pressure setpoint or shorten the solenoid valve response time. This "two-way update" mechanism based on measured results not only improves the success rate of handling single events but also forms a self-correcting monitoring and control closed-loop system that evolves with operation and continuously enhances robustness. This significantly enhances the response speed and extinguishing efficiency of high-voltage circuit breakers to arcing conditions under complex operating conditions, thereby improving the overall safety and reliability of the power grid operation.

[0124] The above-mentioned high-voltage circuit breaker arc monitoring method enables high-precision identification and rapid response control of arc signals under strong electromagnetic interference, significantly improving the stability, accuracy, and real-time performance of arc monitoring. This invention is based on the synchronous acquisition of spectral and electromagnetic vector signals. It effectively removes underlying noise fluctuations by extracting stable base-plane signals through primary differential extraction. Combining an interference identification model and multi-dimensional spectrum diagrams, it accurately extracts the temporal boundaries of transient interference and constructs an interference weight matrix, achieving precise separation of arc and interference during signal processing. Furthermore, it conducts time-by-time energy progression analysis based on dynamically purified arc characteristic sequences and pre-converts the prediction results into drive control commands, ensuring that arc extinguishing actions are completed before the arc energy threshold. Finally, through feedback correction of the entire process operation data, it achieves continuous iterative optimization of identification parameters and control strategies. This closed-loop mechanism connects the technical links of "perception-identification-control-optimization". Even under complex operating conditions such as multi-source strong interference and high grid load, it can still achieve early perception, accurate identification, proactive response and dynamic self-learning adjustment of arc risk, effectively avoid circuit breaker damage and systemic power accidents caused by identification failure or response delay, and significantly enhance the operational safety and intelligence level of power equipment.

[0125] This invention provides, for example Figure 2 The high-voltage circuit breaker arc monitoring system shown includes a synchronous sensing and baseline construction module, an interference identification and timing extraction module, a signal purification and feature extraction module, an energy analysis and early warning judgment module, a response control execution module, and a data feedback and strategy optimization module.

[0126] The synchronous sensing and datum construction module synchronously acquires arc spectral signals and electromagnetic interference vector data under a unified time reference, and performs primary differential processing to generate stable datum signals.

[0127] The interference identification and timing extraction module inputs the stable base plane signal into the interference identification model, generates a multi-dimensional spectrum diagram through time-frequency mapping, extracts the interference timing boundary, and outputs the interference weight matrix.

[0128] The signal purification and feature extraction module adaptively adjusts the filter kernel coefficients according to the interference weight matrix to enhance and suppress the signal, thereby obtaining a dynamically purified arc characteristic signal sequence.

[0129] The energy analysis and early warning judgment module performs continuous energy analysis on the arc characteristic signal sequence and completes the early triggering prediction of protection actions based on the energy growth trend and preset threshold.

[0130] The response control execution module generates control commands based on the trigger prediction results, adjusts the tripping pressure and arc-extinguishing airflow to achieve rapid arc extinguishing;

[0131] The data feedback and strategy optimization module feeds back the entire process operation data to the identification model and controller, updates the identification parameters and control strategies, and builds a closed-loop optimized arc monitoring mechanism.

[0132] The present invention provides a method for monitoring arcing in a high-voltage circuit breaker, which is implemented by the aforementioned high-voltage circuit breaker arcing monitoring system. For details of the specific method and process of the high-voltage circuit breaker arcing monitoring system, please refer to the embodiment of the above-mentioned method for monitoring arcing in a high-voltage circuit breaker, which will not be repeated here.

[0133] 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. A method for monitoring arcing in a high-voltage circuit breaker, characterized in that, Includes the following steps: S1, synchronously acquires arc spectral signals and electromagnetic interference vector data under a unified time reference, and performs primary differential processing to generate a stable base surface signal; S2 inputs the stable base plane signal into the interference identification model, generates a multi-dimensional spectrum diagram through time-frequency mapping, extracts the interference time sequence boundary, and outputs the interference weight matrix; S3, based on the interference weight matrix, adaptively adjust the filter kernel coefficients to enhance and suppress the signal, and obtain a dynamically purified arc characteristic signal sequence; S4, continuously analyzes the energy of the arc characteristic signal sequence, and predicts the early triggering of protection actions based on the energy growth trend and preset threshold; S5 generates control commands based on the trigger prediction results, adjusts the tripping pressure and arc extinguishing airflow to achieve rapid arc extinguishing; S6 transmits the entire process operation data back to the identification model and controller, updates the identification parameters and control strategies, and builds a closed-loop optimized arc monitoring mechanism.

2. The method for monitoring arcing in a high-voltage circuit breaker according to claim 1, characterized in that, Step S1 includes: A miniature fiber optic spectrometer is installed at the observation window of the arc-extinguishing chamber of the high-voltage circuit breaker to collect spectral signals during the arc-ignition process. The spectral response range is set to 350 nm to 1050 nm, and the sampling frequency is set to 100 kHz. A three-axis electric field induction antenna is installed on the outside of the circuit breaker housing, and current transformers are installed at the incoming and outgoing line circuits to collect electromagnetic interference vector data respectively. The three-axis electric field antenna measures the frequency band from 10 Hz to 10 MHz. A GPS timing chip is used to uniformly time-mark the collected data, and a unified time reference is used to achieve dual-channel synchronous acquisition and alignment of spectral signals and electromagnetic vector data; The primary differential processing is performed, and non-cooperative electromagnetic signals are smoothed or eliminated by linear interpolation within a local time window. Time reconstruction and amplitude normalization are completed to generate a stable base plane signal.

3. The method for monitoring arcing in a high-voltage circuit breaker according to claim 1, characterized in that, Step S2 includes: The stable base plane signal is divided into time sequences using a sliding time window of 500 microseconds per segment, and the frequency distribution map in the range of 1 kHz to 5 MHz is obtained by fast Fourier transform. The amplification relationship between the energy at each frequency point and the reference energy is calculated based on the sliding window spectrum, and the start and end times of high-amplitude transient interference signals are identified. The timing and frequency range of interference events are mapped to the time axis and frequency band axis to construct an interference weight matrix. The interference level is set according to the interference amplitude and duration and weight coefficients are assigned to guide subsequent signal processing.

4. The method for monitoring arcing in a high-voltage circuit breaker according to claim 1, characterized in that, Step S3 includes: Each time point of the interference weight matrix is ​​matched with the sampling point of the stable base plane signal, and the filter window length and weight distribution corresponding to different interference levels are assigned. Based on the interference level, the window length and center weight are adjusted in the sliding weighted average filter to achieve smooth suppression of high interference signal segments and detail enhancement of low interference signal segments. Amplitude-weighted amplification is performed on signals in the 250 kHz to 2 MHz frequency band, attenuation processing is performed on signals in non-target frequency bands, and amplitude normalization and dynamic overlap reconstruction are performed after filtering in each frequency band. The processed signals are sorted and reconstructed while retaining their original timestamps to form a continuously and dynamically purified arc characteristic signal sequence.

5. The method for monitoring arcing in a high-voltage circuit breaker according to claim 4, characterized in that, The specific steps for performing amplitude-weighted amplification on signals in the 250 kHz to 2 MHz frequency band, attenuation processing on signals in non-target frequency bands, and amplitude normalization and dynamic overlap reconstruction after filtering in each frequency band are as follows: The 250 kHz to 2 MHz frequency band signal is extracted from the stable base signal, and amplitude weighting is performed by setting an amplification factor between 1.25 and 1.5 according to the frequency range. Signals from the 1 kHz to 250 kHz frequency band and the 2 MHz to 5 MHz frequency band are extracted, and amplitude reduction is performed by setting attenuation coefficients of 0.5, 0.7 and 0.9 according to the interference level at the corresponding time point in the interference weight matrix. The signals processed in different frequency bands are normalized and then superimposed and reconstructed to form a frequency-selectively enhanced signal for subsequent arc feature extraction and analysis.

6. The method for monitoring arcing in a high-voltage circuit breaker according to claim 1, characterized in that, Step S4 includes: The arc characteristic signal sequence is divided into multiple overlapping sliding windows with a time length of 1 millisecond. The energy of the absolute value of the signal amplitude is calculated within each window to obtain the energy value of the electrical signal for each segment. Construct an energy time series, extract the energy change trend of a continuous window, and correct it by combining the amplitude ratio before and after signal enhancement to obtain an accurate energy growth curve; If the current energy value exceeds the energy threshold of 2.5 millijoules and the growth ratio exceeds 1.5 times for three consecutive windows, output the early trigger prediction instruction.

7. The method for monitoring arcing in a high-voltage circuit breaker according to claim 1, characterized in that, Step S5 includes: Receive the arc energy value, growth rate and risk level information from the early trigger prediction results, and transmit them to the control response structure in a period of no more than 1 millisecond; Based on the risk level, retrieve the corresponding tripping drive pressure and arc extinguishing airflow control parameters from the preset response strategy table; The corresponding control commands are output to the tripping drive execution unit and the airflow control unit respectively. The drive contacts complete the separation within 10 milliseconds and control the high-pressure airflow to be ejected to form a high-speed airflow along the arc direction. Record and transmit the tripping action time, airflow start time, control command issuance time, and feedback status for subsequent correction and optimization of the control strategy.

8. The method for monitoring arcing in a high-voltage circuit breaker according to claim 1, characterized in that, Step S6 includes: The system collects operational data including tripping time, execution pressure, airflow velocity, arc energy and waveform, and packages them into feedback data frames after adding timestamps. The relevant data in the feedback data frame is input into the recognition structure. If the recognition deviation exceeds the set threshold, the spectrum window width, interference energy threshold and slope factor are adjusted and updated to the default enabled parameters. The control-related data in the feedback data frame is input to the response control structure. If the response does not achieve the expected effect, the drive pressure is increased, the airflow is increased, and the action delay is shortened. The corrected version is then set as the default control strategy for the next time.

9. A high-voltage circuit breaker arc monitoring system, used to implement the high-voltage circuit breaker arc monitoring method according to any one of claims 1-8, characterized in that, It includes modules for synchronous sensing and baseline construction, interference identification and timing extraction, signal purification and feature extraction, energy analysis and early warning judgment, response control execution, and data feedback and strategy optimization. The synchronous sensing and datum construction module synchronously acquires arc spectral signals and electromagnetic interference vector data under a unified time reference, and performs primary differential processing to generate stable datum signals. The interference identification and timing extraction module inputs the stable base plane signal into the interference identification model, generates a multi-dimensional spectrum diagram through time-frequency mapping, extracts the interference timing boundary, and outputs the interference weight matrix. The signal purification and feature extraction module adaptively adjusts the filter kernel coefficients according to the interference weight matrix to enhance and suppress the signal, thereby obtaining a dynamically purified arc characteristic signal sequence. The energy analysis and early warning judgment module performs continuous energy analysis on the arc characteristic signal sequence and completes the early triggering prediction of protection actions based on the energy growth trend and preset threshold. The response control execution module generates control commands based on the trigger prediction results, adjusts the tripping pressure and arc-extinguishing airflow to achieve rapid arc extinguishing; The data feedback and strategy optimization module feeds back the entire process operation data to the identification model and controller, updates the identification parameters and control strategies, and builds a closed-loop optimized arc monitoring mechanism.

Images