A fault detection method for a redundant relay voting system

Abstract

This invention belongs to the field of electrical parameter testing and electrical fault diagnosis technology, and discloses a fault detection method for a redundant relay voting system, including: acquiring real-time contact voltage and equivalent capacitance between contacts, and analyzing to obtain real-time electric field strength and effective spacing; analyzing to obtain quantum tunneling probability and quantum tunneling current based on real-time electric field strength and effective spacing; connecting a current mirror circuit in parallel between contacts to acquire the main current, and combining it with the quantum tunneling current to obtain a current signal; statistically analyzing the current signal within a preset time period to obtain the Fano factor; judging the state signal of the pre-built redundant architecture and performing state voting, triggering frequency domain verification based on the state voting result, and combining it with the Fano factor analysis to obtain the fault location result; and making a judgment based on the fault location result using a fault decision tree to obtain the fault judgment result. This invention provides a full-process technical guarantee for the operation of redundant equipment systems, from current signal acquisition to fault decision-making.

Description

Technical Field

[0001] This invention relates to the field of electrical parameter testing and electrical fault diagnosis technology, and more specifically, to a fault detection method for a redundant relay voting system. Background Technology

[0002] Traditional methods for detecting electrical component faults have many drawbacks. For example, relying on manual inspection and experience-based preventative maintenance strategies is not only time-consuming and labor-intensive, but also makes it difficult to detect faults in real time, which can easily lead to sudden equipment shutdowns and affect production efficiency.

[0003] Chinese patent application CN117092452B discloses a method for isolating high-resistance grounding faults in distribution networks based on traveling wave signal detection: Step S1: Extract the three-phase traveling wave current and three-phase traveling wave voltage signals of phases A, B, and C through broadband voltage transformers (PT) and current transformers (CT), and transmit them to the intelligent feeder terminal (FTU) via multiplexed signal cables to achieve traveling wave quantity acquisition; Step S2: Transient signal processing; Determine the characteristic frequency band range (SFB). When the intelligent feeder terminal (FTU) acquires the transient signal, filter the sampled zero-sequence voltage and zero-sequence current within the SFB range. Calculate the direction coefficient D using the filtered zero-sequence transient voltage and zero-sequence transient current; Step S3: Traveling wave signal processing; Set the traveling wave voltage and current trigger thresholds. After traveling wave trigger acquisition, record... Simultaneously satisfy the triggering conditions for traveling wave voltage and current; calculate the phase information of the traveling wave, retain the traveling wave data that meets the conditions, calculate the average time of the traveling wave pile, determine the power frequency periodicity between adjacent traveling wave piles, for effective piles that meet the power frequency periodicity, the average power needs to be calculated once for each discharge, and the sum of all average power within the statistical time period dt is denoted as Ptotal; Step S4: Fault isolation; within the specified statistical time period dt, count the number of traveling wave piles with power frequency periodicity, denoted as Num; using the direction coefficient D, the number of traveling wave piles with power frequency periodicity Num and the sum of all average power within the statistical time period dt Ptotal, determine whether a line fault has occurred and whether the fault point is upstream or downstream of the detection point, and perform graded protection and isolation. This invention is based on a traveling wave positioning-type primary and secondary deep integration pole-mounted circuit breaker device. It proposes a high-resistance fault detection and isolation method that combines high-frequency traveling waves with traditional transient signal detection, which greatly improves the detection and handling capabilities of weak characteristic faults such as high-resistance grounding and hidden discharge. Since high-frequency traveling waves are not affected by line operation mode and fault transition resistance, the 10kV distribution network single-phase fault detection technology based on high-frequency traveling wave detection has broad application prospects.

[0004] While the above methods can meet the needs of most scenarios, research and practical application of these methods and existing technologies have revealed at least the following shortcomings:

[0005] The above methods are based on macroscopic electrical quantities for fault isolation only and lack the ability to identify quantum tunneling effects. For example, under micro-spacing and high-voltage conditions, transient currents caused by quantum tunneling may be misjudged as real faults, thereby triggering unnecessary circuit breaker actions and affecting system reliability.

[0006] In view of this, the present invention proposes a fault detection method for a redundant relay voting system to solve the above problems. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a fault detection method for a redundant relay voting system, comprising the following steps:

[0008] Real-time contact voltage and equivalent capacitance between contacts are collected, and real-time electric field strength and effective distance are analyzed.

[0009] Based on the real-time electric field strength and effective spacing, the quantum tunneling probability and quantum tunneling current are obtained through analysis.

[0010] A current mirror circuit is connected in parallel between the contacts to collect the main current, and the current signal is obtained by combining it with quantum tunneling current.

[0011] The current signal within a preset time period is statistically analyzed to obtain the Fano factor;

[0012] The status signals of the pre-built redundant architecture are judged, a status vote is performed based on the status signals, a frequency domain verification is triggered based on the status vote results, and the fault location results are obtained by combining Fano factor analysis.

[0013] Based on the fault location results, a fault judgment result is obtained by using a fault decision tree.

[0014] Furthermore, methods for obtaining the Fano factor include:

[0015] The average current value is calculated by statistically analyzing the current signal within a preset time period.

[0016] Perform a Fourier transform on the current signal to obtain the power spectral density;

[0017] The Fano factor is calculated based on the average current and the power spectral density at zero frequency, combined with the electron charge.

[0018] Furthermore, methods for state voting based on state signals include:

[0019] The current and voltage signals within a preset time period are statistically analyzed to obtain the time-domain signal. The time-domain signals of each redundant device are acquired, and then filtered, denoised, and baseline corrected to obtain the pre-processed time-domain signal.

[0020] Obtain the sampled value of the i-th redundant device at time t, the total number of redundant devices N, and the sum of the sampled values ​​of all other modules except the i-th module at time t. Calculate the time-domain voting value and compare it with the dynamic threshold. If the time-domain voting value is greater than the dynamic threshold for M consecutive periods, mark the corresponding state signal as 1; otherwise, mark it as 0.

[0021] The status signals of the statistical redundant equipment architecture, if exceed If the status signal of the redundant device is 1, then the status voting result is abnormal, triggering frequency domain verification; where, This indicates rounding up; otherwise, it is considered normal and no operation is performed.

[0022] Furthermore, methods for obtaining fault location results include:

[0023] For a branch with a state signal of 1, check if the Fano factor is greater than a preset first threshold. If it is, determine that the fault location result is a false alarm caused by quantum tunneling and ignore the state of the corresponding branch. Check if the Fano factor is less than a preset second threshold. If it is, determine that the fault location result is a real fault and mark the corresponding branch as abnormal.

[0024] Furthermore, methods for obtaining fault diagnosis results include:

[0025] Time-domain analysis, frequency-domain analysis, and coupling analysis are performed on time-domain signals to obtain their time-domain characteristics, frequency-domain characteristics, and coupling characteristics.

[0026] By using time-domain features, frequency-domain features, and coupling features as inputs to the decision tree model, the types of faults can be obtained.

[0027] When the decision tree outputs an unknown fault type, a feature archiving operation is triggered: the dual-domain features of the unknown fault and the manually labeled fault type are recorded.

[0028] Furthermore, methods for obtaining time-domain features, frequency-domain features, and coupling features include:

[0029] Calculate the deviation between the mean time-domain signal of the faulty branch and the mean time-domain signal of the normal branch to obtain the mean offset.

[0030] Calculate the ratio of peak value to RMS value of the time-domain signal to obtain the peak factor;

[0031] The frequency of abrupt changes is obtained by counting the number of times the slope of the time-domain signal exceeds a preset slope threshold.

[0032] The temporal characteristics are obtained by splicing together the mean offset, peak factor, and mutation frequency.

[0033] Perform a Fourier transform on the time-domain signal to obtain the frequency-domain signal, calculate the difference between the main frequency of the faulty branch's frequency-domain signal and the main frequency of the normal frequency-domain signal, and obtain the main frequency offset.

[0034] The harmonic distortion rate is obtained by calculating the ratio of the total harmonic power to the fundamental power of the frequency domain signal.

[0035] Calculate the ratio of the noise energy in the frequency domain signal to the total energy to obtain the noise energy ratio;

[0036] Frequency domain characteristics are obtained by splicing the main frequency offset, harmonic distortion rate, and noise energy ratio;

[0037] The energy concentration is obtained by calculating the ratio of the frequency domain amplitude at each mutation moment to the total frequency domain amplitude at all mutation moments.

[0038] Calculate the Pearson correlation coefficient between the main frequency offset and the mean offset. Determine whether they are correlated based on the coefficient threshold. When the Pearson correlation coefficient is greater than the first coefficient threshold, they are considered to have the same trend and the trend label is marked as 1. When the Pearson correlation coefficient is not greater than the first coefficient threshold and not less than the second coefficient threshold, they are considered to be correlated and the trend label is marked as 0. When the Pearson correlation coefficient is less than the second coefficient threshold, they are considered to be unrelated and the trend label is marked as -1.

[0039] The coupling characteristics are obtained by splicing together energy concentration and trend labels.

[0040] Furthermore, methods for obtaining real-time electric field intensity and effective spacing include:

[0041] The thickness of the dielectric layer of the relay contact is obtained, and the real-time electric field strength is calculated by combining the real-time contact voltage.

[0042] The initial spacing is calculated based on the equivalent capacitance of the contact and the dielectric constant of the dielectric.

[0043] The corrected spacing is calculated based on the initial spacing, real-time contact voltage, and spacing compensation factor.

[0044] The absolute spacing is calculated based on the corrected spacing, real-time electric field strength, and initial spacing.

[0045] The effective spacing is calculated based on the initial spacing and the absolute spacing.

[0046] Furthermore, methods for obtaining the quantum tunneling probability include:

[0047] The barrier height is calculated based on the real-time electric field strength, electron charge, effective spacing, and the fitted value of the barrier height.

[0048] The equilibrium energy is calculated based on temperature; the local voltage is calculated based on real-time electric field strength, effective spacing, and total insulation layer thickness; and the particle energy is calculated based on equilibrium energy, electron charge, and local voltage.

[0049] The quantum tunneling probability is calculated by combining the reduced Planck constant, effective spacing, particle mass, barrier height, and particle energy; the reduced Planck constant is obtained by calculating the Planck constant.

[0050] Furthermore, methods for obtaining quantum tunneling current include:

[0051] The carrier concentration was collected.

[0052] The thermal velocity is calculated by combining temperature and electron mass.

[0053] The tunnel area was collected.

[0054] The quantum tunneling current is calculated based on carrier concentration, thermal velocity, tunneling area, and quantum tunneling probability.

[0055] Furthermore, methods for obtaining current signals include:

[0056] The main current is calculated based on the mirror current and mirror ratio.

[0057] The current signal is obtained by superimposing the main current and the quantum tunneling current.

[0058] The technical effects and advantages of the fault detection method for a redundant relay voting system of the present invention are as follows:

[0059] This invention achieves quantitative analysis of the virtual current caused by quantum tunneling through quantum tunneling probability calculation, providing a fundamental physical basis for distinguishing virtual current from real fault current in micro-pitch high-voltage environments. Simultaneously, this invention accurately acquires the main current through a current mirror circuit and superimposes it with the tunneling current to form a complete current signal. Combined with the analysis of signal statistical characteristics using the Fano factor and the state voting mechanism of the redundant architecture, it effectively distinguishes between tunneling virtual current and real fault current, avoiding voter misjudgment. This invention achieves accurate fault type identification through multi-dimensional extraction of time-domain, frequency-domain, and coupling features and decision tree classification. The unknown fault archiving mechanism also continuously optimizes the model's generalization ability. This invention solves the problem of insufficient identification of quantum tunneling effects in micro-pitch high-voltage scenarios using traditional methods, providing a full-process technical guarantee for the reliable operation of redundant equipment systems, from current signal acquisition to fault decision-making. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of a fault detection method for a redundant relay voting system according to the present invention;

[0061] Figure 2 This is a schematic diagram of the method for obtaining effective spacing according to the present invention;

[0062] Figure 3 This is a schematic diagram of the method for obtaining time-domain features, frequency-domain features, and coupling features according to the present invention;

[0063] Figure 4 This is a schematic diagram of the method for obtaining the corrected quantum tunneling probability according to the present invention. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Example 1:

[0066] Please see Figure 1 As shown in the figure, this embodiment provides a fault detection method for a redundant relay voting system, including the following steps:

[0067] Real-time contact voltage and equivalent capacitance between contacts are collected, and real-time electric field strength and effective distance are analyzed.

[0068] Methods for obtaining real-time electric field strength and effective spacing include:

[0069] The thickness of the relay contact dielectric layer is obtained, and the real-time electric field strength is calculated by combining this with the real-time contact voltage; for example, the real-time electric field strength... ,in, This refers to the real-time contact voltage. It is the vacuum permittivity; The relative permittivity of the medium; The contact area; For dielectric layer capacitance;

[0070] Reference Figure 2 The initial spacing is calculated based on the equivalent capacitance of the contacts and the dielectric constant of the dielectric material; for example, the initial spacing... ,in, The equivalent capacitance of the contact in its initial state;

[0071] The corrected spacing is calculated based on the initial spacing, real-time contact voltage, and spacing compensation factor; such as the corrected spacing. ,in, The spacing compensation factor is obtained by fitting the actual spacing and the corrected spacing using the least squares method. Rated operating voltage;

[0072] The absolute spacing is calculated based on the corrected spacing, real-time electric field strength, and initial spacing; such as the absolute spacing. ,in, The reference electric field strength is specifically the electric field strength under rated voltage.

[0073] The effective spacing is calculated based on the initial spacing and the absolute spacing. (The effective spacing is...) .

[0074] The aforementioned method accurately reflects the barrier electric field strength under high voltage conditions through real-time electric field strength calculation, providing electric field parameters for the quantitative analysis of quantum tunneling probability. By calculating the effective spacing, it realistically depicts the actual physical gap of the contacts, solving the problem of spacing measurement errors caused by mechanical deformation and temperature drift under micro-gap conditions. The accurate acquisition of these two key parameters directly improves the accuracy of subsequent quantum tunneling probability and tunneling current calculations, enabling effective differentiation between virtual connection current and real fault current. Based on this, combined with subsequent steps such as Fano factor analysis and state voting, false alarms caused by quantum tunneling can be accurately identified, avoiding the misjudgment of "virtual current" as contact adhesion faults. Ultimately, this provides a reliable physical basis for fault branch location and fault type determination, fundamentally solving the problem of misjudgment in micro-gap high-voltage scenarios.

[0075] Based on the real-time electric field strength and effective spacing, the quantum tunneling probability and quantum tunneling current are obtained through analysis.

[0076] Methods for obtaining the quantum tunneling probability include:

[0077] The barrier height is calculated based on the real-time electric field strength, electron charge, effective spacing, and fitted value of the barrier height; such as the barrier height. ,in, The potential barrier height is the fitted value, which is obtained by measuring the quantum tunneling current at different voltages and then back-calculating it through curve fitting. The electrons carry a charge; Real-time electric field strength;

[0078] The equilibrium energy is calculated based on temperature; such as equilibrium energy. ,in, Boltzmann's constant; The ambient temperature of the contacts; the local voltage is calculated based on the real-time electric field strength, effective spacing, and total insulation thickness; such as the local voltage. ,in, This is the total insulation layer thickness. , The total insulation capacitance; particle energy is calculated based on balance energy, electron charge, and local voltage; such as particle energy. ,in, To balance energy; Local voltage;

[0079] The quantum tunneling probability is calculated by combining the reduced Planck constant, effective spacing, particle mass, barrier height, and particle energy; such as the quantum tunneling probability. ,in, To reduce Planck's constant, which is obtained through calculations using Planck's constant, The particle mass is the rest mass of the electron, and its value is 9.109 × 10⁻⁶. −31 kg.

[0080] The above method combines the calculation of barrier height with parameters such as real-time electric field strength and effective spacing, and uses the tunneling current under different voltages to back-calculate the fitted value, ensuring dynamic matching between the barrier height and actual working conditions, and avoiding the deviation between the fixed theoretical value and the real barrier. The accurate calculation of balance energy and particle energy quantifies the ability of electrons to overcome the barrier, and provides accurate particle energy parameters for the physical modeling of tunneling probability.

[0081] Based on this, the quantum tunneling probability calculated by combining key physical quantities such as the reduced Planck constant and effective spacing can accurately characterize the possibility of electrons passing through the micro-pitch contact gap, directly determining the magnitude of the virtual connection current. The accurate acquisition of this probability makes the subsequent calculation of the quantum tunneling current more realistic. Furthermore, through comparison with the main current and Fano factor analysis, the virtual current can be clearly distinguished from the real fault current, fundamentally preventing the voting device from misjudging the tunneling effect as a contact adhesion fault. Ultimately, this provides a reliable quantitative basis for accurately locating faulty branches and determining the type of fault.

[0082] Methods for obtaining quantum tunneling current include:

[0083] The carrier concentration is collected. If it is a metal electrode, such as gold or copper, the carrier concentration is directly obtained from the material handbook value. If it is a semiconductor / insulator, the carrier concentration is obtained through Hall effect testing.

[0084] The thermal velocity is calculated by combining temperature and electron mass; such as thermal velocity. ,in, Boltzmann's constant; The ambient temperature of the contact point; For electronic quality;

[0085] The tunneling area can be obtained by capacitance method, combined with dielectric constant, effective spacing, and measured capacitance; for example, the tunneling area... ,in, This is the equivalent capacitance corresponding to the tunnel path;

[0086] The quantum tunneling current is calculated based on carrier concentration, thermal velocity, tunneling area, and quantum tunneling probability. (The quantum tunneling current is then used as an example.) ,in, This represents the carrier concentration.

[0087] The aforementioned method ensures that the number of electrons participating in tunneling matches the actual material properties by specifically obtaining the carrier concentration; the thermal motion velocity, combined with temperature and electron mass, quantifies the initial mobility of electrons, reflecting the influence of temperature on tunneling; the tunneling area is accurately calculated using the capacitance method, avoiding geometric area measurement errors at micro-pitches and realistically characterizing the physical range of tunneling. These parameters, combined with the quantum tunneling probability, result in a quantum tunneling current that accurately reflects the actual magnitude of the "virtual connection current," making it distinguishable from the main fault current in terms of magnitude and characteristics. This precise quantization lays the foundation for subsequent steps such as Fano factor analysis and state voting to identify false alarms. By comparing the characteristics of the tunneling current with the main current, the virtual current interference caused by quantum tunneling can be clearly eliminated, preventing it from being misjudged as a contact adhesion fault, ultimately achieving accurate location of the faulty branch and precise judgment of the fault type.

[0088] A current mirror circuit is connected in parallel between the contacts to collect the main current, and the current signal is obtained by combining it with quantum tunneling current.

[0089] Methods for connecting current mirror circuits in parallel between contacts include:

[0090] Obtain constraint data, including main current range, image accuracy, contact impedance, and environmental conditions. Specifically, the main current range is obtained by recording the current waveform when the contacts are open and closed using an oscilloscope and a current probe, extracting the peak and valley values, and using this data. Image accuracy is obtained based on the maximum percentage error between the predefined output current and the theoretical value. Contact impedance is obtained by averaging multiple measurements when the contacts are open. Environmental conditions include actual temperature, humidity, and vibration frequency.

[0091] Select a basic topology circuit from the preset circuit topology library. The basic topology circuits include BJT mirror circuits and MOS mirror circuits. When the main current range is greater than the main current threshold, select the BJT mirror circuit; otherwise, select the MOS mirror circuit. The core structure of the BJT mirror circuit consists of two NPN transistors, with their base-collector shorted and their output transistors sharing a common ground with the emitters of the input transistors, resulting in an output current equal to the input current. The core structure of the MOS mirror circuit consists of two NMOS transistors, with their gates shared and their sources sharing a common ground. The mirror ratio is controlled by the aspect ratio, resulting in an output current to input current ratio equal to the output aspect ratio to the input aspect ratio. The main current threshold is obtained by calculating the midpoint of the sum of the minimum stable current of the BJT mirror circuit and the maximum rated current of the MOS transistors.

[0092] By connecting X transistors in parallel, the main current of the current mirror circuit is scaled to the range of the main current, where X is calculated by the maximum rated current of a single transistor and the maximum value of the main current.

[0093] Based on the reference voltage, main current range, and bias voltage, the resistance range is calculated, and the optimal resistance value is obtained through a multi-objective optimization algorithm; for example, the minimum resistance value within the main current range. Main current range, maximum resistance Minimum value of bias voltage resistance maximum value of bias voltage resistance Minimum resistance Maximum resistance ,in, The reference voltage; Minimum value of main current; The maximum value of the main current; The safety factor is dimensionless and is typically set to 1.05-1.2. It is used to compensate for fluctuations in circuit parameters and is determined based on the design redundancy. This is the bias voltage; This is the minimum bias current. This represents the maximum value of the bias current.

[0094] Methods for obtaining the optimal resistance value include:

[0095] The current tracking error target is calculated based on the number of test conditions, the measured current of the original relay circuit under the i-th condition, the output current of the current mirror circuit, and the operating current of the original relay circuit; such as the current tracking error target. ,in, This represents the total number of test conditions. For the original redundant equipment circuit in the first Measured current under various operating conditions; For the current mirror circuit in the first Output current under various operating conditions; For the original relay circuit in the first Rated operating current under various operating conditions;

[0096] The stability error target is calculated based on the number of disturbance conditions, the output current under rated temperature and rated load, and the reference output current; such as the stability error target. ,in, This represents the total number of disturbance conditions. For the first The output current at rated temperature and rated load under various disturbance conditions; This is the reference output current of the circuit under standard operating conditions (no disturbance).

[0097] The current tracking error target and the stability error target are normalized respectively to obtain the standardized tracking target and the standardized stability target. The standardized tracking target and the standardized stability target are then weighted to obtain the comprehensive evaluation function.

[0098] The fitness function is minimized by the comprehensive evaluation function. The particle swarm optimization algorithm is used to find the optimal solution, and the particle position is updated iteratively. Finally, the solution converges to obtain the optimal resistance value.

[0099] Methods for obtaining current signals include:

[0100] The main current is calculated based on the mirror current and mirror ratio; such as the main current. ,in, It is the mirror current; It is a mirror image ratio;

[0101] The current signal is obtained by superimposing the main current and the quantum tunneling current.

[0102] The aforementioned method ensures high-precision acquisition of the main current through the design of a current mirror circuit. It maintains stability under high-voltage, high-current conditions using a BJT topology and enhances sensitivity under low-current conditions using a MOS topology. Simultaneously, environmental considerations are addressed by optimizing resistor values ​​to compensate for external interference, further improving mirror accuracy. Based on this, the main current acquired by the mirror circuit is superimposed with the quantum tunneling current to form a complete current signal, allowing both current components to be analyzed in the same dimension. This superposition is not a simple addition but is based on precise physical modeling and circuit design. Subsequent steps, such as Fano factor analysis and time / frequency domain feature extraction, effectively distinguish between the quantum tunneling current and the actual fault current, fundamentally solving the problem of the voter misjudging the tunneling effect as a fault. Ultimately, this achieves accurate fault location and reliable fault type identification.

[0103] The current signal within a preset time period is statistically analyzed to obtain the Fano factor;

[0104] Methods for obtaining the Fano factor include:

[0105] The average current value is calculated by statistically analyzing the current signal within a preset time period.

[0106] Perform a Fourier transform on the current signal to obtain the power spectral density;

[0107] The Fano factor is calculated based on the average current and the power spectral density at zero frequency, combined with the electron charge. (Fano factor) ,in, This is the average value of the current signal within a preset time period; Let be the power spectral density of the current signal at zero frequency; The measurement bandwidth of the current signal is determined by the sampling rate of the measuring device; It represents the electron charge.

[0108] The above method reflects the correlation between noise fluctuations and the mean of the current signal through the Fano factor. The virtual current generated by quantum tunneling originates from the random tunneling behavior of microscopic electrons, resulting in significant noise fluctuations and a large Fano factor. In contrast, the main current of a real fault is a macroscopic current-carrying current with more stable noise characteristics, leading to a smaller Fano factor. By analyzing the current signal over a preset time period to obtain the Fano factor, the statistical difference between the virtual current and the real fault current can be accurately identified through threshold comparison. This eliminates interference from quantum tunneling at the signal characteristic level, avoids misjudgment by the voting device, and provides a reliable basis for subsequent fault branch location and fault type determination.

[0109] The status signals of the pre-built redundant equipment architecture are judged, status voting is performed based on the status signals, frequency domain verification is triggered according to the status voting results, and fault location results are obtained by combining Fano factor analysis.

[0110] Methods for state voting based on state signals include:

[0111] The current and voltage signals within a preset time period are statistically analyzed to obtain the time-domain signal. The time-domain signals of each redundant device are acquired, and then filtered, denoised, and baseline corrected to obtain the pre-processed time-domain signal.

[0112] Obtain the sampled value of the i-th redundant device at time t, the total number of redundant devices N, and the sum of the sampled values ​​of all modules except the i-th module at time t. Calculate the time-domain voting value, such as the time-domain voting value. ,in, For the first A redundant device at time The sampled values; This represents the total number of redundant devices. For a moment At that time, except for the first Besides the redundant equipment, the rest The sum of the sampled values ​​of each redundant device is compared with a dynamic threshold. If the time-domain voting value for M consecutive cycles is greater than the dynamic threshold, the corresponding state signal is marked as 1; otherwise, it is marked as 0. The dynamic threshold can be adjusted according to the signal amplitude, such as within ±5% of the signal amplitude. M is determined according to the relay operating frequency.

[0113] The status signals of the statistical redundant equipment architecture, if exceed If the status signal of the redundant device is 1, then the status voting result is abnormal, triggering frequency domain verification; where, This indicates rounding up; otherwise, it is considered normal and no operation is performed.

[0114] Methods for obtaining fault location results include:

[0115] For branches with a state signal of 1, the Fano factor is checked to see if it is greater than a preset first threshold. If so, the fault location result is determined to be a false alarm caused by quantum tunneling, and the state of the corresponding branch is ignored. The Fano factor is checked to see if it is less than a preset second threshold. If so, the fault location result is determined to be a real fault, and the corresponding branch is marked as abnormal. The preset first threshold and the preset second threshold are obtained by simulating quantum tunneling scenarios and real fault scenarios to obtain tunneling samples and fault samples. By statistically analyzing the sample distribution, the preset first threshold is set as the lower limit of the confidence interval for tunneling samples, such as the lower limit of the 95% confidence interval; the preset second threshold is set as the upper limit of the confidence interval for fault samples, such as the upper limit of the 95% confidence interval.

[0116] The aforementioned method utilizes state voting to compare time-domain signals from redundant devices, initially screening potentially abnormal branches based on the majority principle. This approach leverages redundancy to reduce the risk of false triggering caused by single-module noise, while dynamic thresholds and the M-value ensure sensitive detection of genuine anomalies. Furthermore, for branches marked as abnormal, a threshold judgment using the Fano factor directly and accurately distinguishes between the statistical characteristics of quantum tunneling virtual current and real fault current. False alarms caused by quantum tunneling are eliminated from the abnormal branches, retaining only truly faulty branches. This hierarchical logic of initial screening and precise verification leverages the anti-interference capability of the redundant architecture and directly addresses the essential difference between the tunneling effect and real faults through the Fano factor, ultimately achieving accurate location of faulty branches and completely resolving the core problem of virtual currents being misjudged as real faults in micro-pitch high-voltage scenarios.

[0117] Based on the fault location results, a fault judgment result is obtained by using a fault decision tree.

[0118] Methods for obtaining fault diagnosis results include:

[0119] Reference Figure 3 The deviation between the mean time-domain signal of the faulty branch and the mean time-domain signal of the normal branch is calculated to obtain the mean offset.

[0120] Calculate the ratio of peak value to RMS value of the time-domain signal to obtain the peak factor;

[0121] The frequency of abrupt changes is obtained by counting the number of times the slope of the time-domain signal exceeds a preset slope threshold. The preset slope threshold is obtained by simulating normal and fault scenarios, collecting slope samples, and setting the upper limit of the slope in the normal scenario.

[0122] The temporal characteristics are obtained by splicing together the mean offset, peak factor, and mutation frequency.

[0123] Perform a Fourier transform on the time-domain signal to obtain the frequency-domain signal, calculate the difference between the main frequency of the faulty branch's frequency-domain signal and the main frequency of the normal frequency-domain signal, and obtain the main frequency offset.

[0124] The harmonic distortion rate is obtained by calculating the ratio of the total harmonic power to the fundamental power of the frequency domain signal.

[0125] Calculate the ratio of the noise energy in the frequency domain signal to the total energy to obtain the noise energy ratio;

[0126] Frequency domain characteristics are obtained by splicing the main frequency offset, harmonic distortion rate, and noise energy ratio;

[0127] The energy concentration is obtained by calculating the ratio of the frequency domain amplitude at each mutation moment to the total frequency domain amplitude at all mutation moments.

[0128] The Pearson correlation coefficient between the dominant frequency offset and the mean offset is calculated. Correlation is determined based on a threshold: when the Pearson correlation coefficient is greater than the first threshold, the two are considered to have the same trend and are labeled with a trend label of 1; when the Pearson correlation coefficient is neither greater than the first threshold nor less than the second threshold, they are considered to be correlated and are labeled with a trend label of 0; when the Pearson correlation coefficient is less than the second threshold, they are considered to be unrelated and are labeled with a trend label of -1. The first and second thresholds are obtained by collecting correlation coefficient samples from different scenarios and using cluster analysis.

[0129] Coupled features are obtained by splicing together energy concentration and trend labels;

[0130] By using time-domain features, frequency-domain features, and coupling features as inputs to the decision tree model, the types of faults can be obtained.

[0131] When the decision tree outputs an unknown fault type, a feature archiving operation is triggered: the dual-domain features of the unknown fault and the manually labeled fault type are recorded.

[0132] The aforementioned method captures the amplitude deviation, pulse characteristics, and abrupt change patterns of current signals through time-domain features, characterizes the frequency distribution and noise characteristics of the signal through frequency-domain features, and connects the intrinsic relationship between the time and frequency domains through coupling features. These features can significantly distinguish the essential differences between quantum tunneling virtual currents and real faults. Based on this, the decision tree model learns the mapping relationship between these features and fault types, directly outputting specific fault types and avoiding misclassification of tunneling effects as real faults at the classification level. Furthermore, the archiving mechanism for unknown faults continuously accumulates new features, optimizes the model's generalization ability, and ensures accurate fault identification even in complex scenarios. This multi-feature characterization – intelligent classification – continuous optimization logic ultimately achieves accurate fault type identification, providing a closed-loop intelligent solution to address the misjudgment problem caused by quantum tunneling.

[0133] Example 2:

[0134] Reference Figure 4 This embodiment provides a method for obtaining the corrected quantum tunneling probability applied to Embodiment 1, comprising the following steps:

[0135] By monitoring micro-deformation of the contact points using distributed fiber optic sensing, and combining this with temperature sensor data, the initial spacing is corrected using the coefficient of thermal expansion and vibration offset to obtain a secondary correction distance; such as the secondary correction distance. ,in, For micro-deformation of contacts monitored by distributed fiber optic sensing; The coefficient of thermal expansion of the contact material; Real-time temperature; For reference temperature; This represents the vibration offset. This is the initial spacing;

[0136] The absolute distance is calculated based on the second-corrected distance, real-time electric field strength, and initial spacing; such as the absolute spacing. ;

[0137] The effective spacing is calculated based on the initial spacing and the absolute spacing; such as the effective spacing. ;

[0138] The barrier correction coefficient at different temperatures was calculated based on density functional theory, such as the barrier correction coefficient. ,in, and All are material barrier temperature coefficients; correct the barrier height fitting value to obtain the corrected fitting value; such as the corrected fitting value ;in, The original barrier height is fitted without correction.

[0139] The corrected equilibrium energy is calculated based on the Boltzmann constant, temperature, carrier concentration, and reference concentration; for example, the corrected equilibrium energy... ,in, Reference carrier concentration; The Boltzmann constant is used; the local voltage is calculated based on the real-time electric field strength, effective spacing, and total insulation thickness; for example, the local voltage... The corrected particle energy is calculated based on the corrected equilibrium energy, electron charge, and local voltage; for example, the corrected particle energy... ;

[0140] The corrected barrier height is calculated based on the real-time electric field strength, electron charge, effective spacing, and corrected fitted value; for example, the corrected barrier height... .

[0141] By combining the reduced Planck constant, effective spacing, corrected particle mass, corrected barrier height, and particle energy, the corrected quantum tunneling probability is calculated. (The corrected quantum tunneling probability is then used to calculate this probability.) . To correct particle mass, , These are material parameters, which can be obtained by consulting relevant materials based on the contact metal material.

[0142] The aforementioned method uses distributed fiber optic sensing and temperature data to correct the effective spacing, accurately capturing the dynamic changes of micro-spacing under environmental disturbances and avoiding miscalculations of tunneling probability caused by spacing measurement deviations. Based on density functional theory, the barrier correction coefficient and the equilibrium energy correction of the Boltzmann constant ensure that the barrier height and particle energy can match the influence of the temperature field on electron behavior in real time, solving the problem of insufficient adaptability of fixed parameters in extreme environments. Furthermore, correcting the particle mass further refines the microscopic model of quantum tunneling. These corrections make the calculation of quantum tunneling probability more closely resemble real physical scenarios, ensuring a reduction in the quantization error of dummy currents and making their characteristic differences from real fault currents more significant. Based on this, subsequent steps such as Fano factor analysis and state voting can more accurately identify false tunneling alarms, ultimately improving the reliability of fault location and type determination, and fundamentally solving the problem of misjudgment in micro-spacing high-voltage scenarios.

[0143] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0144] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A fault detection method for a redundant relay voting system, characterized in that, Includes the following steps: Real-time contact voltage and equivalent capacitance between contacts are collected, and real-time electric field strength and effective distance are analyzed. Based on the real-time electric field strength and effective spacing, the quantum tunneling probability and quantum tunneling current are obtained through analysis. A current mirror circuit is connected in parallel between the contacts to collect the main current, and the current signal is obtained by combining it with quantum tunneling current. The current signal within a preset time period is statistically analyzed to obtain the Fano factor; The status signals of the pre-built redundant architecture are judged, a status vote is performed based on the status signals, a frequency domain verification is triggered based on the status vote results, and the fault location results are obtained by combining Fano factor analysis. Based on the fault location results, a fault judgment result is obtained by using a fault decision tree.

2. The fault detection method for a redundant relay voting system according to claim 1, characterized in that, Methods for obtaining the Fano factor include: The average current value is calculated by statistically analyzing the current signal within a preset time period. Perform a Fourier transform on the current signal to obtain the power spectral density; The Fano factor is calculated based on the average current and the power spectral density at zero frequency, combined with the electron charge.

3. The fault detection method for a redundant relay voting system according to claim 1, characterized in that, Methods for state voting based on state signals include: The current and voltage signals within a preset time period are statistically analyzed to obtain the time-domain signal. The time-domain signals of each redundant device are acquired, and then filtered, denoised, and baseline corrected to obtain the pre-processed time-domain signal. Obtain the sampled value of the i-th redundant device at time t, the total number of redundant devices N, and the sum of the sampled values ​​of all other modules except the i-th module at time t. Calculate the time-domain voting value and compare it with the dynamic threshold. If the time-domain voting value is greater than the dynamic threshold for M consecutive periods, mark the corresponding state signal as 1; otherwise, mark it as 0. The status signals of the statistical redundant equipment architecture, if exceed If the status signal of the redundant device is 1, then the status voting result is abnormal, triggering frequency domain verification; where, This indicates rounding up; otherwise, it is considered normal and no operation is performed.

4. The fault detection method for a redundant relay voting system according to claim 1, characterized in that, Methods for obtaining fault location results include: For a branch with a state signal of 1, check if the Fano factor is greater than a preset first threshold. If it is, determine that the fault location result is a false alarm caused by quantum tunneling and ignore the state of the corresponding branch. Check if the Fano factor is less than a preset second threshold. If it is, determine that the fault location result is a real fault and mark the corresponding branch as abnormal.

5. The fault detection method for a redundant relay voting system according to claim 1, characterized in that, Methods for obtaining fault diagnosis results include: Time-domain analysis, frequency-domain analysis, and coupling analysis are performed on time-domain signals to obtain their time-domain characteristics, frequency-domain characteristics, and coupling characteristics. By using time-domain features, frequency-domain features, and coupling features as inputs to the decision tree model, the types of faults can be obtained. When the decision tree outputs an unknown fault type, a feature archiving operation is triggered: the dual-domain features of the unknown fault and the manually labeled fault type are recorded.

6. The fault detection method for a redundant relay voting system according to claim 5, characterized in that, Methods for obtaining time-domain features, frequency-domain features, and coupling features include: Calculate the deviation between the mean time-domain signal of the faulty branch and the mean time-domain signal of the normal branch to obtain the mean offset. Calculate the ratio of peak value to RMS value of the time-domain signal to obtain the peak factor; The frequency of abrupt changes is obtained by counting the number of times the slope of the time-domain signal exceeds a preset slope threshold. The temporal characteristics are obtained by splicing together the mean offset, peak factor, and mutation frequency. Perform a Fourier transform on the time-domain signal to obtain the frequency-domain signal, calculate the difference between the main frequency of the faulty branch's frequency-domain signal and the main frequency of the normal frequency-domain signal, and obtain the main frequency offset. The harmonic distortion rate is obtained by calculating the ratio of the total harmonic power to the fundamental power of the frequency domain signal. Calculate the ratio of the noise energy in the frequency domain signal to the total energy to obtain the noise energy ratio; Frequency domain characteristics are obtained by splicing the main frequency offset, harmonic distortion rate, and noise energy ratio; The energy concentration is obtained by calculating the ratio of the frequency domain amplitude at each mutation moment to the total frequency domain amplitude at all mutation moments. Calculate the Pearson correlation coefficient between the main frequency offset and the mean offset. Determine whether they are correlated based on the coefficient threshold. When the Pearson correlation coefficient is greater than the first coefficient threshold, they are considered to have the same trend and the trend label is marked as 1. When the Pearson correlation coefficient is not greater than the first coefficient threshold and not less than the second coefficient threshold, they are considered to be correlated and the trend label is marked as 0. When the Pearson correlation coefficient is less than the second coefficient threshold, they are considered to be unrelated and the trend label is marked as -1. The coupling characteristics are obtained by splicing together energy concentration and trend labels.

7. The fault detection method for a redundant relay voting system according to claim 1, characterized in that, Methods for obtaining real-time electric field strength and effective spacing include: The thickness of the dielectric layer of the relay contact is obtained, and the real-time electric field strength is calculated by combining the real-time contact voltage. The initial spacing is calculated based on the equivalent capacitance of the contact and the dielectric constant of the dielectric. The corrected spacing is calculated based on the initial spacing, real-time contact voltage, and spacing compensation factor. The absolute spacing is calculated based on the corrected spacing, real-time electric field strength, and initial spacing. The effective spacing is calculated based on the initial spacing and the absolute spacing.

8. The fault detection method for a redundant relay voting system according to claim 7, characterized in that, Methods for obtaining the quantum tunneling probability include: The barrier height is calculated based on the real-time electric field strength, electron charge, effective spacing, and the fitted value of the barrier height. The equilibrium energy is calculated based on temperature; the local voltage is calculated based on real-time electric field strength, effective spacing, and total insulation layer thickness; and the particle energy is calculated based on equilibrium energy, electron charge, and local voltage. The quantum tunneling probability is calculated by combining the reduced Planck constant, effective spacing, particle mass, barrier height, and particle energy; the reduced Planck constant is obtained by calculating the Planck constant.

9. A fault detection method for a redundant relay voting system according to claim 8, characterized in that, Methods for obtaining quantum tunneling current include: The carrier concentration was collected. The thermal velocity is calculated by combining temperature and electron mass. The tunnel area was collected. The quantum tunneling current is calculated based on carrier concentration, thermal velocity, tunneling area, and quantum tunneling probability.

10. A fault detection method for a redundant relay voting system according to claim 1, characterized in that, Methods for obtaining current signals include: The main current is calculated based on the mirror current and mirror ratio. The current signal is obtained by superimposing the main current and the quantum tunneling current.

Images