A circuit breaker break voltage non-contact measurement method and system

Abstract

The application discloses a circuit breaker fracture voltage non-contact measurement method and system, and relates to the technical field of power system state monitoring; the application obtains discrete voltage signals of a double end of a circuit breaker; in a closed steady state stage of the circuit breaker, a known power grid power frequency voltage is introduced as a reference, a recursive least square method is used to update a space capacitance cross coupling matrix in real time, and dynamic self-calibration of parameters immune to environmental interference is realized; when a breaking instruction and voltage mutation are detected, a high-frequency transient sequence is intercepted; an inverse matrix of the coupling matrix is used for algebraic decoupling, and double-end cross electric field interference is stripped; a regularization inversion filtering algorithm is introduced in a frequency domain to eliminate high-frequency attenuation caused by hardware parasitic parameters, and a real fracture transient recovery voltage waveform is reconstructed, and breaking performance is judged according to the real fracture transient recovery voltage waveform; the application solves the problems of large environmental drift influence, waveform aliasing distortion and the like of traditional non-contact measurement, and realizes safe and high-precision TRV online monitoring.

Description

Technical Field

[0001] This invention belongs to the field of power system condition monitoring technology, and specifically relates to a non-contact measurement method and system for circuit breaker break voltage. Background Technology

[0002] Circuit breakers are core control devices in power systems, performing control and protection tasks. At the instant a circuit breaker interrupts a short-circuit fault current, the moving and stationary contacts separate, and the arc is extinguished at the current zero-crossing point. At this moment, a high-frequency, high-amplitude transient recovery voltage (TRV) is generated across the break. The waveform characteristics of the TRV (such as peak voltage and rate of rise) directly determine whether the insulation strength between the breaks can be successfully restored. It is the most critical indicator for evaluating the arc-extinguishing performance of circuit breakers and preventing catastrophic power grid accidents caused by arc re-breakdown.

[0003] Currently, traditional contact-based measurement methods for obtaining break voltage mainly rely on parallel voltage dividers or voltage transformers. These methods require direct electrical connection between the measuring equipment and the high-voltage conductor at the 10,000-volt level, which not only results in high insulation modification costs and large size, but also poses serious safety hazards of ground discharge and insulation breakdown during long-term operation. To address this issue, non-contact measurement technologies (such as plate sensors based on the principle of capacitive coupling) are gradually being applied. However, existing non-contact measurement technologies often only deploy a single-sided plate for rough estimation, ignoring the strong cross-electric field coupling interference between the moving and stationary contacts inside the arc-extinguishing chamber; more critically, the spatial coupling capacitance is highly susceptible to random drift due to ambient temperature, air humidity, and the degree of contamination on the insulator surface, leading to measurement coefficient failure; in addition, the megahertz-level high-frequency TRV signal, when passing through the sensor leads, is subject to severe low-pass filtering effects due to parasitic inductance, resulting in high-frequency waveform distortion and peak attenuation.

[0004] It can be seen that existing non-contact measurement technologies have the following problems: 1. It is difficult to overcome the problem of random drift of spatial stray capacitance caused by environmental factors, resulting in systematic errors in long-term measurements; 2. There is a lack of effective means to isolate the interference of electric fields between moving and stationary contacts, resulting in severe waveform aliasing during double-end measurements; 3. Due to the parasitic parameters of the hardware link, it is difficult to truly reproduce the high-frequency peak characteristics of the megahertz-level transient recovery voltage. Summary of the Invention

[0005] (a) Technical problems to be solved To address the problems in related technologies, this invention provides a non-contact measurement method and system for circuit breaker break voltage, thereby overcoming the aforementioned technical problems existing in the prior art.

[0006] (II) Technical Solution To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: Firstly, a non-contact method for measuring the voltage at the circuit breaker contacts is provided, specifically as follows: S1. Extract the original induced charge signals output by the external non-contact electric field sensors on the stationary and moving contact sides of the circuit breaker's arc-extinguishing chamber, and convert them into discrete time-series voltage signals. S2. When the circuit breaker is in the closed-current steady state stage, the discrete time series voltage signal and the power frequency reference voltage signal of the power grid bus are used as optimization benchmarks to construct the spatial capacitance cross-coupling matrix equation. The self-coupling coefficient characteristics and mutual coupling coefficient characteristics in the spatial capacitance cross-coupling matrix equation are calculated and updated in real time using the recursive least squares method to generate the optimal steady-state spatial decoupling matrix. S3. When a tripping command is detected and the amplitude change rate of the discrete time sequence voltage signal exceeds the limit, the high-frequency transient induced voltage sequence during the process of the arc burning to extinguishing at the break point is extracted. S4. Obtain the inverse matrix of the optimal steady-state space decoupling matrix as the transient decoupling operator; perform matrix multiplication on the high-frequency transient induced voltage sequence and the transient decoupling operator to reconstruct the absolute ground transient voltage sequence of the stationary contact and the absolute ground transient voltage sequence of the moving contact. S5. In the frequency domain, the absolute ground transient voltage sequence of the stationary contact and the absolute ground transient voltage sequence of the moving contact are respectively inverted and filtered and compensated using the inverse function of the wideband transfer function of the sensor to obtain the compensated absolute ground transient voltage sequence of the stationary contact and the compensated absolute ground transient voltage sequence of the moving contact. S6. Subtract the compensated transient voltage sequence of the moving contact from the compensated transient voltage sequence of the stationary contact to ground to generate a transient recovery voltage time sequence of the circuit breaker. Calculate the peak voltage characteristics and transient recovery voltage rise rate characteristics based on the transient recovery voltage time sequence of the circuit breaker to determine the circuit breaker's breaking performance. Preferably, step S1 includes the following steps: S11. On the stationary contact side and the moving contact side of the circuit breaker insulation bushing flange, a double-shielded D-dot non-contact electric field sensor based on the differential geometry principle is symmetrically arranged; the original induced charge signal output by the sensor on the stationary contact side is extracted as the first current signal, and the original induced charge signal output by the sensor on the moving contact side is extracted as the second current signal. S12. Establish hardware integral conversion equations; substitute the first current signal and the second current signal extracted in S11 into the hardware integral conversion equations respectively, and calculate and output the first continuous analog voltage signal and the second continuous analog voltage signal. S13. The first continuous analog voltage signal and the second continuous analog voltage signal are continuously discretized and sampled using a high-precision analog-to-digital converter, and the DC bias characteristics are deducted using a moving average filtering algorithm to obtain a pure AC discrete time series voltage signal. Preferably, the specific calculation process of the moving average filtering algorithm in S13 includes the following steps: S131, Set the sliding time window length to N window A discrete sampling point; S132, in the n Extracting data from discrete sampling points from the original discrete sequence... n N window +1 to n Continuous N window Individual voltage amplitude data; for the continuous N window The voltage amplitude data are summed and then divided by the given value. N window Calculate the arithmetic mean of the current time step; S133. The arithmetic mean is used as the dynamic quantization value of the input offset voltage characteristic Voffset of the operational amplifier; the current... n The discrete time sequence voltage signal is output by subtracting the dynamic quantization value from the original discrete sequence voltage amplitude of each discrete sampling point. Preferably, step S2 includes the following steps: S21. Extract the discrete-time voltage signal generated in S13 and the steady-state voltage sequence to obtain the absolute voltage to ground of the stationary contact and the absolute voltage to ground of the moving contact; establish the physical constraints of the circuit breaker's closed state: during the closed current-carrying period, set the absolute voltage to ground of the stationary contact and the absolute voltage to ground of the moving contact to be electrically equipotential. S22. Establish the equation for the second-order spatial capacitance cross-coupling matrix; S23. Substitute the physical constraints in S21 into the second-order space capacitance cross-coupling matrix equation in S22 to obtain a system of linear equations. S24. Using the recursive least squares method based on the forgetting factor, the linear equation system is iteratively solved within the set power frequency cycle time window to calculate the first equivalent aggregation coefficient and the second equivalent aggregation coefficient; the a priori proportional constant set by the three-dimensional spatial symmetry geometry of the circuit breaker arc-extinguishing chamber is called, and the first equivalent aggregation coefficient and the second equivalent aggregation coefficient are substituted into the algebraic equation system for decoupling, and the optimal steady-state space decoupling matrix containing four independent elements is output. Preferably, step S24, which uses a recursive least squares method based on a forgetting factor to iteratively solve the problem within a set power frequency cycle time window, includes the following steps: S241. Initialize the parameter vector of the equivalent aggregation coefficients; initialize the error covariance matrix; set the forgetting factor; S242. At any discrete time step, extract the discrete sequence of the power frequency reference voltage of the power grid bus in S21 to construct the input regression vector; extract the discrete time sequence voltage signal generated in S13 to construct the actual observation vector. S243. Using the error covariance matrix from the previous time step and the current input regression vector, calculate the Kalman gain feature matrix: S244. Calculate the prior prediction error characteristics using the formula error calculation formula; S245. Multiply the Kalman gain feature matrix by the prior prediction error feature, add the parameter vector of the previous time step, and update the current parameter vector. S246. Update the current error covariance matrix using the formula for calculating the error covariance matrix; S247. Calculate the rate of change of the Euclidean norm of the parameter vectors of adjacent time steps; when the rate of change of the Euclidean norm is less than the convergence tolerance within a set number of consecutive steps, stop the iteration; extract the elements in the parameter vector at this time as the first equivalent aggregation coefficient and the second equivalent aggregation coefficient. Preferably, step S3 includes the following steps: S31. Monitor the status signal of the mechanical auxiliary contact of the circuit breaker operating mechanism. When a level transition characteristic from closed to open is detected, start the set mechanical debounce timer. When the mechanical debounce timer overflows and the level transition characteristic is maintained, generate a pre-triggering command. S32. After receiving the pre-triggered arming command, extract the discrete time sequence voltage signal generated in S13 and calculate the absolute value of the difference between two adjacent sampling points. S33. Extract the maximum value of the differential absolute value feature within the continuous power frequency cycle during the steady-state phase of the closed-circuit flow, and set it as the steady-state noise reference; multiply the steady-state noise reference by the set multiplier parameter to calculate the transient over-limit threshold. S34. When the real-time calculated absolute value of the difference is greater than the transient over-limit threshold, a hardware-level synchronous sampling interrupt is triggered; based on the current trigger point, extract several historical sampling points from the ring buffer before the current trigger point, increase the sampling rate of the analog-to-digital converter, and continuously collect several sampling points after the trigger point; using the first continuous analog voltage signal and the second continuous analog voltage signal output from S12, splice them together to generate a high-frequency transient induced voltage sequence matrix with a total length of the number of pre-trigger sampling points plus the number of subsequent sampling points after the trigger point. Preferably, step S4 includes the following steps: S41. Extract the optimal steady-state space decoupling matrix output from S24; calculate the determinant feature of the optimal steady-state space decoupling matrix using the determinant feature calculation formula; S42. Compare the absolute value of the determinant feature with the set singular value tolerance threshold; if the absolute value is greater than the singular value tolerance threshold, use the adjoint matrix method to calculate the inverse matrix of the optimal steady-state space decoupling matrix using the determinant feature as the denominator, and generate the transient decoupling operator. S43. Extract each discrete time step vector from the high-frequency transient induced voltage sequence matrix generated in S34; The discrete time step vector is multiplied on the left by the transient decoupling operator generated by S42, and the matrix multiplication operation equation is executed; the absolute ground transient voltage sequence of the stationary contact and the absolute ground transient voltage sequence of the moving contact are output. Preferably, step S5 includes the following steps: S51. Extract the parameters of the discrete difference equation of the entire measurement system obtained in the offline calibration stage, and construct the sensor broadband transfer function in the discrete complex frequency domain and Z domain. S52. Using fast Fourier transform, the static contact absolute to ground transient voltage sequence and the moving contact absolute to ground transient voltage sequence output by S43 are mapped to the frequency domain to obtain the frequency domain complex sequence. S53. Calculate the frequency response of the wideband transfer function of the sensor on the unit circle to obtain the complex frequency response matrix at discrete frequency points; introduce Wiener filter regularization parameters and calculate the inversion compensation operator using the inversion compensation operator calculation formula. S54. In the frequency domain, perform element-wise multiplication operations on the frequency domain complex sequence generated in S52 and the inversion compensation operator generated in S53 to generate a frequency domain compensation sequence. The frequency domain compensation sequence is mapped back to the time domain using the inverse fast Fourier transform, and the compensated static contact absolute transient voltage sequence to ground and the compensated moving contact absolute transient voltage sequence to ground are output. Preferably, step S6 includes the following steps: S61. Under the same discrete time step index, subtract the compensated transient voltage sequence of the stationary contact to ground from the compensated transient voltage sequence of the moving contact to ground output by S54 to generate the transient recovery voltage time sequence of the circuit breaker. S62. Perform a global extreme value search on the time series of transient recovery voltage at the circuit breaker contact, and extract the discrete data point with the largest absolute value as the peak point; record the voltage amplitude corresponding to the peak point as the peak voltage feature. S63. Extract the voltage zero-crossing point corresponding to the instant of arc extinction in the transient recovery voltage time series of the circuit breaker contact; calculate the time elapsed from the voltage zero-crossing point to the peak point, and record it as a time interval feature; S64. Divide the peak voltage feature extracted in S62 by the time interval feature extracted in S63 to calculate the transient recovery voltage rise rate feature. S65. Extract the rated peak voltage threshold and rated rise rate threshold from the pre-stored rated TRV envelope parameters on the circuit breaker nameplate; perform logical comparisons between the peak voltage characteristic and the transient recovery voltage rise rate characteristic and the rated peak voltage threshold and the rated rise rate threshold, respectively; if the peak voltage characteristic > the rated peak voltage threshold or the transient recovery voltage rise rate characteristic > the rated rise rate threshold, generate and output a circuit breaker interruption failure risk alarm message. Secondly, a non-contact measurement system for circuit breaker break voltage is also provided to implement the aforementioned non-contact measurement method for circuit breaker break voltage. The system includes a signal sensing and integration preprocessing module, a steady-state adaptive parameter decoupling and optimization module, a multi-dimensional transient triggering and cross-stripping module, and a frequency domain inversion compensation and state assessment module; wherein: The aforementioned signal sensing and integration preprocessing module is used to extract the original induced charge signals from the external electric field sensors of the stationary and moving contacts, convert the original induced charge signals into continuous analog voltage signals using an active integration circuit, and generate discrete time series voltage signals through discretization and debiasing processing. The aforementioned steady-state adaptive parameter decoupling optimization module is used to extract the steady-state voltage sequence from the discrete-time voltage signal and combine it with the power frequency reference voltage signal of the power grid bus when the circuit breaker is closed. It then uses the recursive least squares method to update the coefficients of the spatial capacitance cross-coupling matrix equation in real time to generate the optimal steady-state spatial decoupling matrix. The aforementioned multidimensional transient triggering and cross-stripping module is used to trigger high-frequency sampling and extract the high-frequency transient induced voltage sequence matrix based on mechanical commands and the absolute value characteristics of voltage differences; and to perform multiplication operations using the inverse matrix of the optimal steady-state space decoupling matrix to strip the cross electric field interference and reconstruct the absolute ground transient voltage sequence. The aforementioned frequency domain inversion compensation and state assessment module is used to call the inverse function of the sensor's broadband transfer function model to perform filtering compensation on the absolute ground transient voltage sequence in the frequency domain; generate the circuit breaker break transient recovery voltage time sequence by subtraction; and determine the circuit breaker breaking performance based on the peak voltage characteristics and transient recovery voltage rise rate characteristics.

[0007] (III) Beneficial Effects The present invention has the following beneficial effects: This invention achieves dynamic self-calibration of spatial stray capacitance, solving the accuracy degradation problem caused by environmental drift. It utilizes the physical constraint of equipotentiality between moving and stationary contacts during circuit breaker closure and current flow, and uses the known power frequency voltage of the power grid bus as a natural optimization benchmark. Combined with the recursive least squares method based on the forgetting factor, it tracks and updates spatial capacitance parameters in real time without interrupting power. This mechanism eliminates the dependence on manual offline calibration, enabling non-contact sensors to have adaptive calibration capabilities that are immune to environmental interference such as temperature and humidity.

[0008] This invention constructs a spatial cross-decoupling matrix model to eliminate electric field aliasing interference in double-ended measurements. To address crosstalk generated by the compact space inside the arc-extinguishing chamber, this invention establishes a second-order capacitor cross-coupling matrix and obtains its inverse matrix as a decoupling operator at the instant of transient interruption. Through matrix multiplication in the digital domain, the complex electric field overlap artifacts between the moving and stationary contacts are successfully removed, restoring the true absolute ground potential at both ends and ensuring the accuracy of differential voltage at the break.

[0009] This invention introduces a frequency domain regularization inversion compensation algorithm to accurately reproduce the high-frequency spikes of the megahertz-level TRV. Addressing the high-frequency attenuation defects caused by parasitic inductance in the hardware link, this invention uses the inverse model of the sensor's wideband transfer function in the frequency domain for filtering compensation. Furthermore, by setting Wiener filter regularization parameters using objectively calculated signal-to-noise ratio, it effectively suppresses the unbounded amplification of high-frequency white noise while eliminating waveform phase delay and amplitude attenuation, providing high-fidelity data support for the accurate determination of circuit breaker re-breakdown risk.

[0010] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For those skilled in the art, the drawings can be obtained from these drawings without creative effort.

[0012] Figure 1 This is a flowchart illustrating a non-contact measurement method for circuit breaker break voltage according to the present invention. Figure 2 This is a schematic diagram of a non-contact measurement system for circuit breaker break voltage according to the present invention. Detailed Implementation

[0013] The technical solutions of the embodiments of the invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the invention, and not all embodiments. Based on the embodiments of the invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the invention.

[0014] To address the technical problems raised in the background section, please refer to [link / reference]. Figure 1 This invention provides a non-contact method for measuring circuit breaker break voltage, comprising: S1. Extract the original induced charge signals output by the external non-contact electric field sensors on the stationary and moving contact sides of the circuit breaker arc-extinguishing chamber, and use an active integrator circuit to convert the original induced charge signals into discrete time-series voltage signals. S2. When the circuit breaker is in the closed-current steady state stage, the discrete time series voltage signal and the power frequency reference voltage signal of the power grid bus are used as optimization benchmarks to construct the spatial capacitance cross-coupling matrix equation. The self-coupling coefficient characteristics and mutual coupling coefficient characteristics in the spatial capacitance cross-coupling matrix equation are calculated and updated in real time using the recursive least squares method to generate the optimal steady-state spatial decoupling matrix. S3. Establish a transient triggering time window model for circuit breaker interruption; when a tripping command is detected and the amplitude change rate of the discrete time sequence voltage signal exceeds the limit, trigger a high-frequency synchronous sampling interruption to capture the high-frequency transient induced voltage sequence during the process of arc combustion to extinction at the break point. S4. Obtain the inverse matrix of the optimal steady-state space decoupling matrix as the transient decoupling operator; perform matrix multiplication on the high-frequency transient induced voltage sequence and the transient decoupling operator to remove the cross electric field interference between the moving and stationary contacts, and reconstruct the absolute ground transient voltage sequence of the stationary contact and the absolute ground transient voltage sequence of the moving contact. S5. Call the sensor broadband transfer function that was measured in advance during the offline calibration stage; in the frequency domain space, use the inverse function of the sensor broadband transfer function to perform inversion filtering compensation on the absolute ground transient voltage sequence of the stationary contact and the absolute ground transient voltage sequence of the moving contact respectively, to eliminate the high-frequency phase delay and amplitude attenuation characteristics caused by parasitic inductance, and obtain the compensated absolute ground transient voltage sequence of the stationary contact and the compensated absolute ground transient voltage sequence of the moving contact. S6. Subtract the compensated transient voltage sequence of the moving contact from the compensated transient voltage sequence of the stationary contact to ground to generate a transient recovery voltage time sequence of the circuit breaker. Calculate the peak voltage characteristics and transient recovery voltage rise rate characteristics based on the transient recovery voltage time sequence of the circuit breaker to determine the circuit breaker's breaking performance. The above embodiments, by introducing a known grid voltage for self-calibration before the circuit breaker breaks (during the closed steady-state phase), overcome the predicament of random drift of spatial capacitance due to temperature, humidity, and adjacent phase conductors in traditional non-contact measurements, and realize dynamic real-time updating of coupling parameters. By establishing and inverting the cross-coupling matrix, the complex spatial electric field overlap interference between the stationary and moving contacts is mathematically eliminated, restoring the true absolute potential of both ends. At the same time, for the high-frequency TRV signal at the megahertz level, a transfer function inversion compensation algorithm is introduced to eliminate the low-pass filtering effect of the hardware link, ensuring the true restoration of waveform spikes and providing high-confidence data support for the evaluation of the circuit breaker's arc-extinguishing capability. The above embodiment S1 includes the following steps: S11. On the stationary contact side and the moving contact side of the circuit breaker insulating bushing flange, symmetrically arrange double-shielded D-dot non-contact electric field sensors based on differential geometry principles; extract the original induced charge signal output by the sensor on the stationary contact side as the first current signal. i 1( t The original induced charge signal output by the sensor on the moving contact side is extracted as the second current signal. i 2( t ); In specific implementation, the above embodiment S11 is as follows: Since the high-voltage conductor inside the circuit breaker is wrapped by SF6 gas or vacuum insulating medium and an external epoxy resin sleeve, direct electrical contact with it is impossible; this system uses flexible PCB printing technology to fabricate the D-dot electric field sensor, whose inner induction plate has an effective area of... A eff Set to 50cm 2 It is tightly attached to the outer surface of the insulating sleeve corresponding to the stationary and moving contacts; the outer layer is equipped with a grounded shield to block the transverse electromagnetic interference of the conductors of adjacent phases (for example, blocking phases A and C when measuring phase B); according to Gauss's law of electromagnetic fields, the transient electric field in space will induce a displacement current on the surface of the induction plate, and the first current signal output by the sensor is... i 1( t Mathematically, it is equal to the product of the plate area, the dielectric constant, and the time derivative of the electric field strength in space, i.e. ;in, e 0 represents the vacuum permittivity (physical constant value is 8.854 × 10⁻⁶). 12 F / m), e r The relative permittivity of the insulating bushing material of the circuit breaker. A eff is the effective physical contact area of ​​the inner sensing electrode plate of the sensor, and is the transient electric field intensity generated in space on the surface of the stationary contact. dE 1( t) / dt This indicates that the electric field strength is a variable that changes over a continuous time period. t The first derivative; similarly, the second current signal is obtained. i 2( t ); Furthermore, regarding the problem of inverting the electrostatic field inverse from the single-point electric field intensity to the absolute ground potential of the conductor, this invention is based on the Equivalent Charge Method (ECM) and a spatial capacitance network model: since the conductor and insulating bushing of the circuit breaker arc-extinguishing chamber are axisymmetric rigid geometric bodies, their spatial electric field distribution is macroscopically uniquely determined; the local electric field intensity measured by the sensor is essentially the integral of the displacement current generated by the charge distributed on the surface of the high-voltage conductor at that point in space; the second-order spatial capacitance cross-coupling matrix established through S22 is physically equivalent to integrating the complex spatial electric field path (U= ∫E*dl; where U represents the potential difference between the high-voltage conductor and ground, E represents the spatial electric field intensity vector, and dl represents the differential displacement vector on the integration path) is dimensionally reduced to a lumped parameter capacitance network; that is, the spatial integration effects such as the geometric distance between the sensor placement position and the conductor surface, and the dielectric constant have all been tightly encapsulated and absorbed in the coupling coefficient; therefore, as long as the coupling matrix parameters are accurate, the absolute potential of the conductor can be rigorously reconstructed from the single-point induction signal through matrix operations; S12. Establish the hardware integral transformation equation: ;in, k ∈{1,2}, V k ( t ) represents the first output of the integrator circuit. k Continuous analog voltage signals from each sensor, C int To have the characteristic value of a precision integrating capacitor, V offset This refers to the input offset voltage characteristics of the operational amplifier; t For continuous time variables, ∫ dt This represents continuous integration in the time domain; the first current signal extracted in S11... i 1( t ) and the second current signal i 2( t Substituting each value into the hardware integral conversion equation, calculate the first continuous analog voltage signal output. V 1( t ) and the second continuous analog voltage signal V 2( t ); In specific implementation, the above embodiment S12 is as follows: due to the output of S11 ik ( t The electric field strength is a differential form and cannot directly reflect the voltage amplitude; the system design incorporates an active hardware integrator circuit based on the OPA656 broadband operational amplifier; a high-frequency ceramic capacitor made of NPO material (with an extremely low temperature drift coefficient) is selected as the integrating capacitor. C int Its characteristic value is precisely calibrated to 10nF; the first current signal i 1( t When the circuit is connected to the inverting input of the integrator circuit, according to Kirchhoff's current law and the volt-ampere characteristic of a capacitor, the hardware circuit performs the integral conversion equation at the physical level. Due to the asymmetry of the transistors inside the operational amplifier, there will inevitably be an input offset voltage. This offset voltage will generate a slowly drifting DC component under the action of integration. V offset ； The hardware circuit ultimately outputs a first continuous analog voltage signal proportional to the space potential of the stationary contact. V 1( t ), and a second continuous analog voltage signal proportional to the space potential of the moving contact. V 2( t ); S13. Use a high-precision analog-to-digital converter to process the first continuous analog voltage signal. V 1( t ) and the second continuous analog voltage signal V 2( t Continuous discretization sampling is performed, and the DC bias characteristics caused by the offset voltage are subtracted using a moving average filtering algorithm to obtain a pure AC discrete-time series voltage signal. V 1( n )and V 2( n ),in n For discrete sampling point index; The specific calculation process of the moving average filtering algorithm in the above embodiment S13 includes the following steps: S131, Set the sliding time window length to N window A discrete sampling point, the N window This corresponds to a complete cycle of the power grid frequency of 50Hz. S132, in the n Extracting data from discrete sampling points from the original discrete sequence... n N window +1 to n Continuous N window Individual voltage amplitude data; for the continuousN window The voltage amplitude data are summed and then divided by the given value. N window Calculate the arithmetic mean of the current time step; S133. The arithmetic mean is used as the dynamic quantization value of the input offset voltage characteristic Voffset of the operational amplifier; the current... n The original discrete sequence voltage amplitude of each discrete sampling point is subtracted from the dynamic quantization value to output a zero-mean discrete time sequence voltage signal. In specific implementation, the above embodiment S13 specifically involves: the main control system being configured with a 16-bit high-precision analog-to-digital converter (ADC); during the steady-state monitoring phase, the sampling rate of the ADC is set to 10 kS / s (i.e., one point is collected every 0.1 milliseconds); the ADC samples the first continuous analog voltage signal... V 1( t Continuous sampling is performed to generate the original discrete sequence; Furthermore, in order to eliminate the DC bias generated in S12 V offset The system sets the sliding time window length. N window =200, corresponding to one complete cycle (20ms) of the power grid frequency of 50Hz; in the... n Extracting data from discrete sampling points from the original discrete sequence... n 199 to n The arithmetic mean of the current time step is calculated by summing 200 consecutive voltage amplitude data points and dividing the sum by 200. Since the integral of a standard AC power frequency signal over one cycle is theoretically strictly zero, this arithmetic mean physically accurately represents the input offset voltage characteristic of the operational amplifier at the current moment. V offset The quantization value; will be the current number. n Subtracting the dynamic quantization value from the original discrete-time voltage amplitude at each point removes DC drift interference, resulting in a pure AC, zero-mean one-dimensional discrete-time voltage signal. V 1( n )and V 2( n This signal will serve as the sole data source for subsequent steady-state optimization and transient triggering. Furthermore, for the system startup phase or the initial stage of switching from transient mode back to steady-state mode, when the sampling time window is not yet full... N window n < 1 point (i.e., n < 1 point) N windowTo address the negative index edge effect caused by [missing information], this system employs a hardware pre-filling and static baseline locking strategy: within the initial 20ms of ADC power-on, the differential trigger logic is blocked, and the [missing information] is [missing information]. N window The arithmetic mean of each sampling point is used for a one-time static calculation, which serves as the initial DC bias baseline. V oi ; in n< N window Within the invalid index range, directly subtract the static baseline for bias removal; when n≥ N window Then, the algorithm is switched to dynamic moving average filtering to ensure the integrity and distortion-free nature of waveform data under any operating condition; The above embodiment S1 uses a dual-shielded sensor based on differential geometry principles combined with an active integrator circuit to convert the non-directly contactable megahertz-level high-voltage transient electric field into a safe analog voltage signal on the low-voltage side, thus avoiding the risk of insulation breakdown caused by traditional contact measurement from the physical source. Furthermore, by setting a sliding time window algorithm that is strictly aligned with the power grid frequency, the DC offset voltage caused by temperature drift of the operational amplifier is dynamically calculated and subtracted in real time in the digital domain, eliminating the inherent baseline drift defect of the hardware integrator and providing a zero-mean, high signal-to-noise ratio discrete data source for subsequent high-precision mode decoupling. The above embodiment S2 includes the following steps: S21. Extract the discrete-time series voltage signal generated in S13. V 1( n )and V 2( n From the steady-state voltage sequence in the data, the absolute voltage to ground of the stationary contact is obtained. U 1( n The absolute voltage to ground of the moving contact U 2( n Establish physical constraints for the circuit breaker's closed state: During the current-carrying period, set the absolute voltage to ground of the stationary contact. U 1( n The absolute voltage to ground of the moving contact U 2( n Electrically equipotential, that is U 1( n )= U 2( n )= U grid ( n ),in U grid ( n () represents the discrete sequence of the power frequency reference voltage of the power grid bus, acquired synchronously. In specific implementation, the above embodiment S21 is as follows: During the steady-state phase when the circuit breaker is operating normally and has not tripped, since the internal moving and stationary contacts are in a tightly pressed state, the contact resistance is usually at the micro-ohm level. Therefore, the absolute voltage to ground of the stationary contact is relatively low. U 1( n The absolute voltage to ground of the moving contact U 2( n They are completely equipotential in the electrical topology; The system synchronously receives digital sampled value (SMV) messages from the bus voltage transformer (PT) merging unit through the IEC 61850 process layer network within the substation, and extracts the accurate discrete sequence of the power frequency reference voltage of the grid bus. U grid ( n The above constraints provide crucial known input excitations for the subsequent solution of unknown space capacitance parameters. S22. Establish the second-order spatial capacitance cross-coupling matrix equation: ;in, K 11 The self-coupling coefficient characteristic of the stationary contact to the first sensor is given. K 22 The self-coupling coefficient characteristic of the moving contact to the second sensor; K 12 and K 21 The characteristic of the mutual coupling coefficient caused by the stray electric field in space; In specific implementation, the above embodiment S22 is as follows: Due to the extremely compact internal space of the circuit breaker arc-extinguishing chamber, the electric field lines emitted from the high-voltage stationary contact not only vertically penetrate the insulating shell but also couple to the first sensor on the stationary contact side (forming the main capacitor C). 11 Its edge-divergent electric field will also obliquely penetrate the space and couple to the second sensor on the moving contact side (forming stray mutual capacitance C). 12 Similarly, the moving contact also has a main capacitance C. 22 With stray mutual capacitance C 21 ; According to the principle of electrostatic field superposition, the total induced charge received by the first sensor is the result of the combined effect of the electric fields of the stationary and moving contacts; therefore, the voltage signal generated by S13 V 1( n Not only with U 1( n Related to, and also affected by U 2( n Interference from matrix equations; coefficients in matrix equations K ij In physical dimensions, it is equal to the space coupling capacitance. C ij The integrating capacitor set in S12C int The ratio (i.e.) K ij = C ij / C int Due to changes in external environmental temperature, humidity, and the degree of contamination on the insulator surface, C ij It is a time-varying parameter that drifts slowly with the environment; Furthermore, in this embodiment, it is set K ij = C ij / C int Furthermore, the integrating capacitor exhibits ideal capacitive impedance, remaining unchanged with frequency, resulting in constant hardware gain. However, in transient high-frequency measurements at the MHz level, the operational amplifier bandwidth is limited and... C int The drastic impedance changes caused by parasitic inductance (i.e., the physical effect of gain varying with frequency) are completely removed from the spatial coupling matrix in the mathematical framework. K The process involves stripping the data from the middle and then uniformly transferring it to the broadband transfer function in step S5. H ( z Frequency domain lumped modeling and inversion compensation are performed; this orthogonal decoupling strategy of fixing spatial parameters at low frequencies and compensating for hardware distortion at high frequencies ensures the physical rigor of the algorithm across the entire frequency band. S23. Substituting the physical constraints in S21 into the second-order space capacitance cross-coupling matrix equation in S22, we obtain a system of linear equations: V 1( n )=( K 11 + K 12 ) U grid ( n )and V 2( n )=( K 21 + K 22 ) U grid ( n ); S24. Using the recursive least squares method based on the forgetting factor, iteratively solve the linear equation system within a set power frequency cycle time window to calculate the first equivalent aggregation coefficient. K A = K 11 + K 12 With the second equivalent polymerization coefficientK B = K 21 + K 22 ; The dynamic scaling factor set by invoking the three-dimensional spatial symmetry geometry of the circuit breaker's arc-extinguishing chamber. c ( t )= K 12 / K 11 = K 21 / K 22 The first equivalent aggregation coefficient K A Second equivalent polymerization coefficient K B Substitute into the system of algebraic equations K 11 + γK 11 = K A and K 22 + γK 22 = K B Decoupling in the middle, calculating independent K 11 , K 12 , K 21 , K 22 Eigenvalues, outputting the optimal steady-state space decoupling matrix containing four independent elements. ; In specific implementation, the above embodiment S24 is as follows: Since the reduced-dimensional linear equations can only yield the aggregate sum and cannot directly determine the independent self-coupling coefficient and cross-coupling coefficient, this is a typical underdetermined equation problem. Therefore, the system calls the insulation structure data of a specific model circuit breaker extracted using finite element electric field simulation software during the factory manufacturing stage. Because the geometry of the circuit breaker's arc-extinguishing chamber and the sensor installation position are physically rigidly fixed, the ratio of the cross-coupling capacitance to the self-coupling capacitance of its stationary contact to the opposite sensor is a constant determined solely by the structure in three-dimensional space geometry (denoted as ). c 0), but in actual operation, changes in ambient temperature and humidity will cause changes in the dielectric constant of the insulating bushing surface, thus causing the ratio to drift. Therefore, dynamic correction needs to be introduced. The optimal steady-state spatial decoupling matrix is ​​continuously refreshed during the circuit breaker closing stage, breaking the dilemma of random drift of spatial capacitance due to environmental influence in traditional non-contact measurement, and realizing dynamic adaptive calibration of coupling parameters. Furthermore, considering that changes in ambient temperature and humidity can cause non-uniform changes in the dielectric constant of the insulating sleeve surface, thereby increasing the self-capacitance... K 11 Mutual capacitance K 12 The drift ratios are inconsistent; this system introduces an environmental dynamic correction mechanism: high-precision temperature and humidity sensors are installed at the perimeter of the circuit breaker; the correction factor κ is calculated in real time using an environmental mapping surface model established by factory finite element simulation. T , H The fixed prior proportionality constant is upgraded to a dynamic proportionality coefficient. c ( t )= c 0*[1+κ( T , H )],in c 0 represents the base ratio under standard operating conditions; Furthermore, the aforementioned basic ratio c The finite element electric field simulation modeling process for the circuit breaker is as follows: Import the 1:1 real CAD model of the circuit breaker into the ANSYS Maxwell 3D electrostatic field solver; set the high-voltage conductor inside the arc-extinguishing chamber as a first-type Dirichlet boundary condition (known potential), and the air domain at infinity as a zero-potential absorbing boundary; apply adaptive mesh refinement (mesh size lower limit set to 0.1mm) to the sensor electrode edge and the interface of the insulating bushing; and extract the electrostatic field energy matrix by applying unit excitation voltage to the stationary and moving contacts respectively, thereby calculating the electrostatic field energy matrix under standard operating conditions. K 11 and K 12 And obtain the precise c 0 value; In the above embodiment S24, the recursive least squares method based on the forgetting factor is used to iteratively solve the problem within a set power frequency cycle time window, including the following steps: S241. Initialize the parameter vector of the equivalent aggregation coefficients. i (0)=[ K A (0), K B (0)] T =[0,0] T ; K A (0) K B (0) represents the initial guesses of the first and second equivalent aggregation coefficients set at the initial time of the iteration (step 0) of the recursive least squares method; initialize the error covariance matrix P(0) = I ,in d To define the characteristics of a large constant,I Let be a second-order identity matrix; set the forgetting factor. l f This is used to assign higher weights to recently sampled data; S242, in any... n At each discrete time step, extract the discrete sequence of the power frequency reference voltage of the power grid bus in S21 to construct the input regression vector. ( n )=[ U grid ( n ), U grid ( n )] T Extract the discrete-time series voltage signal generated by S13 to construct the actual observation vector. Y ( n )=[ V 1( n ), V2 ( n )] T ; T Indicates transpose; S243, Call the error covariance matrix of the previous time step. P ( n 1) With the current input regression vector ( n ), calculate the Kalman gain eigenvalue matrix: ;in, K gain ( n ) is the first n The Kalman gain characteristic matrix for each discrete time step is used to dynamically adjust the step size of parameter updates. P ( n 1) is the previous time step (the first time step) n The error covariance matrix of step 1); ( n () represents the input regression vector for the current step; T ( n () represents the transpose of the input regression vector; S244. Calculate the formula using formula error. Calculate the prior prediction error characteristics; where, e ( n ) represents the prior prediction error characteristics; Y ( n () represents the sensor voltage vector actually observed in the current step; i T ( n 1) is the transpose of the parameter vector updated in the previous time step; i T ( n 1) ( n This represents the theoretical prediction of the current system output using historical parameters; this error reflects the deviation between the sensor voltage predicted by the current model and the actual sampled voltage. S245, the Kalman gain feature matrix K gain ( n Multiply by the aforementioned prior prediction error characteristics e ( n ), plus the parameter vector from the previous time step i ( n 1) Update the current parameter vector: i ( n )= i ( n 1)+ K gain ( n ) e ( n );in, i ( n ) is the current number n The parameter vector output after each step update contains equivalent aggregation coefficients that approximate the real physical state. S246. Calculate using the formula for the error covariance matrix. Update the current error covariance matrix; where, P ( n ) is the current number n The updated error covariance matrix represents the uncertainty of the current parameter estimate. As the number of iterations increases, the elements of this matrix approach zero. S247. Calculate the rate of change of the Euclidean norm of the parameter vectors at adjacent time steps. i ( n ) i ( n 1)|| / | i ( n- 1) ||; When the rate of change of the Euclidean norm is less than the convergence tolerance within a set number of consecutive steps, stop the iteration; extract the parameter vector at this point. i ( n The two elements in ) are used as the first equivalent aggregation coefficient. K AWith the second equivalent polymerization coefficient K B ; Furthermore, the convergence tolerance setting is calculated based on the hardware quantization limit of the analog-to-digital converter (ADC): this system uses a 16-bit ADC, whose minimum quantization error resolution is 1 / 2. 16 ≈1.52×10 5 To ensure that the accuracy of iterative calculations is not lower than the hardware physical noise floor and to avoid invalid infinite loops, the convergence tolerance is objectively set to a value slightly smaller than the quantization error, i.e., 1.0 × 10⁻⁶. 5 ; The above embodiments construct a mathematical closed loop for parameter adaptive optimization by introducing recursive least squares (RLS) based on the forgetting factor. By assigning higher weights to recently sampled data through the forgetting factor, the algorithm can capture and track the slow drift of spatial stray capacitance caused by the diurnal temperature difference and air humidity changes in the substation. This mechanism uses the known grid bus voltage when the circuit breaker is closed as a natural excitation source, eliminating the need for manual power outage recalibration. It achieves dynamic self-calibration of the coupling matrix coefficients during steady-state operation, solving the problem of long-term accuracy degradation in non-contact measurements. The above embodiment S3 includes the following steps: S31. Monitor the status signal of the mechanical auxiliary contact of the circuit breaker operating mechanism. When a level transition characteristic from closed to open is detected, start the set mechanical debounce timer. When the mechanical debounce timer overflows and the level transition characteristic is maintained, generate a pre-triggering command. In specific implementation, the above embodiment S31 is as follows: The system uses a high-speed optocoupler isolation circuit to continuously monitor the status of the normally open auxiliary contact 52a of the circuit breaker operating mechanism; in the closed steady state, the optocoupler outputs a high level of 3.3V; when the relay protection device issues a trip command, the circuit breaker mechanism linkage begins to move, the auxiliary contact opens, and the optocoupler output terminal experiences a falling edge interruption from a high level of 3.3V to a low level of 0V; after the microprocessor captures this falling edge, it starts the internal hardware timer for mechanical debouncing; the debouncing time window is set to 10ms, and if the level remains at 0V without rebound within 10ms, the mechanical trip command is officially confirmed to be valid, the system state machine flag bit flips from IDLE (idle steady state) to ARMED (pre-triggered armed state), and a pre-triggered armed command is generated; S32. After receiving the pre-triggered arming command, extract the discrete time sequence voltage signal generated in S13. V 1( n ), calculate the absolute value of the difference Δ between two adjacent sampling points. V =| V 1( n ) V 1( n- 1)|; S33. Extract the maximum value of the differential absolute value characteristic within 10 consecutive power frequency cycles during the steady-state phase of the closed-circuit flow, and set it as the steady-state noise reference Δ. V base The steady-state noise reference Δ V base Multiply by the set multiplier parameter l The transient over-limit threshold is calculated. In specific implementation, the above embodiment S33 is as follows: Due to the presence of various high-frequency electromagnetic interferences at the substation site, directly setting a fixed trigger threshold is very likely to cause false triggering of the system; during the closed-circuit steady-state phase, the system records in real time the data within 10 consecutive power frequency cycles (200ms). V 1( n The maximum absolute value of the difference between adjacent points of the signal is used as the steady-state noise reference Δ. V base Set the multiplier parameters and calculate the transient threshold. l ×Δ V base This dynamic threshold strategy ensures that the trigger level can adapt to changes in ambient background noise. Furthermore, the multiplier parameters l The setting is based on the normal distribution in statistics. s The criteria and transient characteristics of the circuit breaker are calculated: under the assumption that the pure steady-state background noise follows a Gaussian distribution, we take... l =3.0 can filter out 99.73% of random noise; further combined with the theoretically calculated minimum initial voltage change rate when the circuit breaker interrupts capacitive / inductive loads ( you / dt The theoretical minimum transient difference is obtained by multiplying the product of the system sampling period and the theoretical minimum transient difference; this theoretical minimum transient difference is then divided by the historical steady-state noise reference Δ. V base The lower limit integer value is taken as the final multiplier parameter. In this embodiment, λ is set to 5.0 after calculation. S34. When the difference absolute value characteristic Δ is calculated in real time V When the transient threshold is exceeded, a hardware-level synchronous sampling interrupt is triggered; based on the current trigger point... n Using 0 as the baseline, extract from the circular buffer. n 0 pre-triggered sampling points N pre The number of historical sampling points is increased, and the sampling rate of the analog-to-digital converter is improved. After continuous sampling, the number of sampling points triggered is increased. N postEach sampling point utilizes the first continuous analog voltage signal output by S12. V 1( t ) and the second continuous analog voltage signal V 2( t The total length generated by splicing is N pre + N post High-frequency transient induced voltage sequence matrix V trans =[ V 1( n trans ), V 2( n trans )] T ,in n trans This is an index for high-frequency transient discrete sampling points. T Indicates transpose; N pre Used to fully preserve the historical background waveform containing the power frequency phase before the circuit breaker breaks; N post Used to cover the entire process of transient recovery voltage after the arc at the break point is extinguished, from high-frequency oscillation to decay and subsidence; In specific implementation, the above embodiment S34 is as follows: when the moving and stationary contacts of the circuit breaker separate and the arc is extinguished at the moment the current crosses zero, the potential at both ends of the break will undergo violent interphase high-frequency transient recovery oscillations (the frequency can reach hundreds of kHz to several MHz), at which time the voltage change rate dV / dt Maximum; when the calculated absolute value of the difference characteristic Δ V When the transient threshold is exceeded, the main control chip immediately triggers a hardware-level synchronous sampling interrupt, and instantly jumps the ADC sampling rate from the steady-state 10kS / s to the high-speed mode 100MS / s (i.e., a sampling interval of 10 nanoseconds); with the trigger point n Using 0 as a baseline, extract the previous data from the memory circular buffer. N pre =5000 historical sampling points (capturing the background waveform 50μs before the interruption), and continuously acquiring data backwards. N post =45,000 sampling points (capturing the transient evolution process of 450μs after the interruption), spliced ​​to generate a high-frequency transient induced voltage sequence matrix with a total length of 50,000 points; the mechanical trigger of S31 and the electrical rate of change trigger of S34 constitute a multi-condition fusion logic, which eliminates false triggering caused by slight ripples generated by normal switching operations of the power grid. Furthermore, to avoid the loss of detail on the first transition edge of the transient waveform due to PLL (phase-locked loop) locking delay caused by ADC clock source switching, this system adopts an architecture of continuous high-speed sampling, cyclic coverage, and software sampling: the ADC hardware always operates continuously at a maximum sampling rate of 100MS / s, and pushes the data into the high-speed ring buffer inside the FPGA in real time; during the closed steady-state phase, the system uses a software algorithm to sample one point every 10,000 points (i.e., software downsampling to 10kS / s) for steady-state optimization in S2; when the hardware comparator of S34 detects a differential mutation triggering an interrupt, the system does not change the physical clock of the ADC, but directly freezes and locks the ring buffer, using the high-frequency historical data already stored before the trigger point as... N pre Extract and continue writing. N post High-frequency data; this mechanism replaces clock switching with trigger locking, achieving zero-delay and zero-loss capture of transient transition frontiers; The above embodiment S3 constructs a multi-dimensional transient triggering time window model that integrates mechanical state and electrical characteristics; through the debouncing judgment of mechanical auxiliary contacts, the system is woken up from dormancy to armed state, eliminating false triggering caused by normal steady-state fluctuations of the power grid; using a dynamic differential threshold calculated based on objective statistics, the microsecond-level voltage change at the moment of arc extinction at the break point is accurately captured; this composite logic combined with the mechanism of dynamically adjusting the sampling rate not only saves the storage overhead during steady-state operation, but also ensures the capture of megahertz-level high-frequency transient recovery waveforms; The above embodiment S4 includes the following steps: S41. Extract the optimal steady-state space decoupling matrix output from S24. K opt ; Calculate using the determinant characteristics formula | K opt |= K 11 K 22 K 12 K 11 Calculate the determinant characteristics of the optimal steady-state space decoupling matrix; S42, the determinant feature | K opt The absolute value of | is compared with a set singular value tolerance threshold; if the absolute value is greater than the singular value tolerance threshold, the adjoint matrix method is used to divide the determinant feature | K opt | As the denominator, calculate the inverse matrix of the optimal steady-state space decoupling matrix to generate the transient decoupling operator. ; In specific implementation, the above embodiment S42 specifically involves: setting a singular value tolerance threshold; if the determinant feature | K opt If the absolute value of | is less than the singular value tolerance threshold, it indicates that the coupling coefficient obtained in S24 has a serious linear correlation (physically manifested as severe sensor failure or detachment), and the matrix tends to be ill-conditioned and cannot be inverted; at this time, the system will throw an abnormal command and block the decoupling calculation; if normal, the transient decoupling operator is rigorously calculated using the adjoint matrix method. K opt 1 ; Furthermore, the singular value tolerance threshold is set based on the machine precision of the microprocessor's floating-point unit (FPU): the system uses IEEE 754 standard single-precision floating-point numbers (Float32) for matrix inversion, with an effective digital precision of approximately 1.19 × 10⁻⁶. 7 To prevent division-to-zero anomalies or severe truncation error amplification when the matrix determinant approaches machine zero, the singular value tolerance threshold is objectively set as a safety boundary that is a multiple of the single-precision machine precision, i.e., 1.0 × 10⁻⁶. 6 ; S43. Extract the high-frequency transient induced voltage sequence matrix generated in S34. V trans Each discrete time step vector in [ V 1( n trans ), V 2( n trans )] T ; The transient decoupling operator generated by left-multiplying the discrete time step vector by S42 K opt 1 Perform matrix multiplication equations: The output is a static contact absolute-to-ground transient voltage sequence stripped of spatial cross-field interference. U 1raw absolute ground transient voltage sequence of moving contact U 2raw ; In specific implementation, the above embodiment S44 specifically involves: performing matrix multiplication point-by-point for the 50,000 discrete time steps in the high-frequency transient matrix; the expanded algebraic equation is: The equation clearly states in a physical sense that the true voltage to ground of the stationary contact cannot be determined solely by the output of its own sensor. V1. When converting to a single-ended ratio, the high-frequency oscillation voltage of the moving contact must be subtracted. V 2. Through spatial stray capacitance K 12 The spurious interference component from the oblique coupling; through the above linear algebra inverse operation, the system achieves physical space isolation in the digital domain, eliminates the waveform aliasing distortion that inevitably occurs during double-ended non-contact measurement, and restores the true absolute potential of the moving and stationary contacts; The above embodiment S4 uses linear algebra inverse matrix operations to transform the invisible cross electric field interference in physical space into a decoupling operator in the digital domain. Since the distance inside the arc-extinguishing chamber is extremely small, the high-frequency oscillation of the moving contact will inevitably produce strong pull-down or lift artifacts on the stationary contact sensor through spatial stray capacitance. This step uses the inverse matrix of the optimal steady-state spatial decoupling matrix to multiply with the high-frequency transient sequence, which mathematically removes this double-ended crosstalk and accurately reconstructs the independent absolute transient potentials of the moving and stationary contacts relative to ground, eliminating peak misjudgment caused by waveform aliasing. The above embodiment S5 includes the following steps: S51. Extract the parameters of the discrete difference equation of the entire measurement system obtained during the offline calibration phase, and construct the sensor broadband transfer function in the discrete complex frequency domain and Z-domain. ,in b i Indicates the characteristics of the feedforward zero coefficient. a j For the characteristics of feedback pole coefficients; where, z For complex variables in the discrete complex frequency domain (Z-domain); M The order of the feedforward zeros of the transfer function; N The order of the feedback pole; z i and z j These represent the signal in the discrete time domain. i Step and the first j Step delay operator; In specific implementation, S51 of the above embodiment is as follows: During the offline calibration stage of the system at the factory, a standard step voltage test signal with a rising edge of nanosecond is injected into the sensor plate. Due to the parasitic inductance (about 50nH) of the sensor leads, this inductance will resonate with the integrating capacitor set in S12 at high frequencies, and produce a low-pass filtering attenuation effect, causing the MHz-level TRV spike to be flattened in the hardware link and the waveform to produce a phase delay; using the system identification algorithm (such as ARMAX model fitting), the parameters of the second-order discrete difference equation of the hardware link are extracted, and the sensor broadband transfer function model H(z) in the Z domain is constructed and stored in non-volatile memory; Furthermore, to address the engineering challenge of offline calibration failing to accurately reflect on-site parasitic parameters, the sensor's broadband transfer function model... H ( z The in-situ online calibration method is used to obtain the following: After the sensor is installed in the circuit breaker bushing, a broadband swept-frequency signal with a known spectrum is injected into the gap between the sensor plate and the bushing using a non-contact vector network analyzer (VNA) with an impedance matching network or a high-frequency pulse injection loop; the output response of the integrator circuit is acquired simultaneously, and the parameters of the entire-link discrete difference equation, including the actual parasitic inductance and lead capacitance in the field, are extracted using the least squares frequency domain identification algorithm, thereby constructing an accurate calibration result. H ( z ); S52. Use Fast Fourier Transform to convert the static contact absolute-to-ground transient voltage sequence output from S43. U 1raw absolute ground transient voltage sequence of moving contact U 2raw Mapping these values ​​to the frequency domain yields a sequence of complex numbers in the frequency domain. U 1raw ( e hω )and U 2raw ( e hω );in, e hω Let represent the complex frequency variable in the discrete-time Fourier transform (DTFT), where e The base is the natural number. h The imaginary unit ( h 2 = 1), oh Normalized digital angular frequency (value range is ) (π to π) S53, Transfer function of the sensor broadband H (z) Perform frequency response calculations on the unit circle to obtain the complex frequency response matrix at discrete frequency points. H ( e hω Introducing Wiener filter regularization parameters α reg ( oh The formula is calculated using the inversion compensation operator. Calculate the inversion compensation operator, where H 1 ( e hω ) is the complex frequency response matrix H ( e hω The conjugate complex matrix of ). In specific implementation, the above embodiment S53 specifically involves: calculating the inversion compensation operator. G When, if we directly take the reciprocal of the transfer function, 1 / H 1 ( e hω Because the amplitude-frequency response |H| in the high-frequency range (above the sensor cutoff frequency) is extremely small and approaches zero, the denominator is extremely small, which causes the high-frequency white noise to be infinitely amplified during inversion. Therefore, this system introduces frequency-dependent adaptive dynamic parameters. α reg ( oh When the signal frequency is low, |H| 2 Much larger α reg ( oh ), operator G Approximately equal to the complete inverse model; when the signal frequency is extremely high and exceeds the cutoff frequency, |H| 2 Approaching 0, in the denominator α reg ( oh It plays a leading protective role and effectively suppresses the unbounded amplification of high-frequency noise; the formula achieves the optimal mathematical balance between restoring the true TRV broadband waveform and suppressing high-frequency white noise. Furthermore, the frequency-related adaptive dynamic parameters α reg ( oh The method for obtaining ) is as follows: Extract the trigger point segment from S34. N pre The noise power spectral density distribution was estimated using the Welch periodogram method from historical background sampling points (pure noise range). P n ( oh After extracting the trigger point N post Similarly, the power spectral density of the noisy signal can be estimated from each transient sampling point (containing both signal and noise). P x ( oh Based on the Wiener filter optimal estimation theory, the adaptive regularization parameter matrix at each discrete frequency point is objectively calculated. When the denominator is extremely small or negative, the lower threshold is set to 10. 4 This method automatically increases the regularization parameter in the high-frequency roll-off region with low signal-to-noise ratio to suppress noise; and automatically decreases the regularization parameter in the dominant frequency band where transient energy is concentrated to achieve inversion; the lower threshold is 10. 4Based on the principle of numerical calculation stability, the settings can be dynamically adjusted to ensure that the power spectral density is much smaller than that of a typical signal (10). 2 ~10 0 The order of magnitude is sufficient, which can prevent numerical overflow caused by the denominator approaching zero, and does not affect the inversion accuracy of the effective frequency band; S54. In the frequency domain, the frequency domain complex sequence generated in S52 is... U 1raw ( e hω )and U 2raw ( e hω ) respectively with the inversion compensation operator generated in S53 G ( e hω Perform element-wise multiplication to generate a frequency domain compensated sequence; The frequency domain compensation sequence is mapped back to the time domain using the inverse fast Fourier transform, and the compensated static contact absolute-to-ground transient voltage sequence is output. U 1comp ( n trans The absolute ground transient voltage sequence of the moving contact after compensation U 2comp ( n trans ); In the above embodiment S5, a frequency domain inversion compensation algorithm based on Wiener filtering theory is introduced to address the signal low-pass attenuation problem caused by the resonance of the parasitic inductance of the sensor leads and the integrating capacitor at high frequencies. By using objectively calculated regularization parameters, the algorithm compensates for the phase delay and amplitude attenuation of the transient recovery voltage in the megahertz band when performing the inverse operation of the transfer function, and effectively suppresses the explosive amplification of high-frequency white noise, ensuring the true restoration of nanosecond-level steep waveform spikes. The above embodiment S6 includes the following steps: S61. Indexing the same discrete time step n trans Below, the compensated static contact absolute-to-ground transient voltage sequence output by S54 U 1comp ( n trans Subtract the compensated absolute ground transient voltage sequence of the moving contact U 2comp ( n trans Generate the time series of transient recovery voltage at the circuit breaker contact. U TRV ( n trans ); S62. The transient recovery voltage time series of the circuit breaker contact. U TRV ( n trans Perform a global extremum search to extract the discrete data point with the largest absolute value as the peak point; record the voltage amplitude corresponding to the peak point as the peak voltage feature. U c ; S63. Extract the voltage zero-crossing point corresponding to the instant of arc extinction from the transient recovery voltage time series of the circuit breaker contact; calculate the time elapsed from the voltage zero-crossing point to the peak point, and record it as a time interval feature. t 3; S64. Extract the peak voltage characteristics from S62. U c Divide by the time interval feature extracted by S63 t 3. Calculate the transient recovery voltage rise rate characteristics. RRRV ; S65. Extract the rated peak voltage threshold from the pre-stored rated TRV envelope parameters on the circuit breaker nameplate. U crated With the rated rise rate threshold RRRV rated Peak voltage characteristics U c Characteristics of transient recovery voltage rise rate RRRV Each with the rated peak voltage threshold U crated With the rated rise rate threshold RRRV rated Perform a logical comparison; if the peak voltage characteristics... U c >Rated peak voltage threshold U crated or transient recovery voltage rise rate characteristics RRRV > Rated rise rate threshold RRRV rated If so, a circuit breaker failure risk alarm message will be generated and output; The above embodiment S6 constructs a complete decision-making closed loop from bottom-level waveform reconstruction to top-level state assessment; by performing time-domain subtraction on the compensated double-terminal absolute potential, a real break TRV sequence is generated, and the peak voltage and rise rate are accurately extracted using global extreme value search; these are compared with the rated nameplate parameters specified in the IEC international standard using hard logic, realizing an objective and quantitative determination of whether there is dielectric recovery hysteresis or re-breakdown risk during the circuit breaker breaking process, directly generating a high-confidence risk warning message for power grid dispatching and equipment maintenance.

[0015] Please see Figure 2 A non-contact measurement system for circuit breaker break voltage, implementing the aforementioned non-contact measurement method for circuit breaker break voltage, the system includes a signal sensing and integration preprocessing module, a steady-state adaptive parameter decoupling optimization module, a multi-dimensional transient triggering and cross-stripping module, and a frequency domain inversion compensation and state assessment module; wherein: The aforementioned signal sensing and integration preprocessing module is used to extract the original induced charge signals from the external electric field sensors of the stationary and moving contacts, convert the original induced charge signals into continuous analog voltage signals using an active integration circuit, and generate discrete time series voltage signals through discretization and debiasing processing. The aforementioned steady-state adaptive parameter decoupling optimization module is used to extract the steady-state voltage sequence from the discrete-time voltage signal and combine it with the power frequency reference voltage signal of the power grid bus when the circuit breaker is closed. It then uses the recursive least squares method to update the coefficients of the spatial capacitance cross-coupling matrix equation in real time to generate the optimal steady-state spatial decoupling matrix. The aforementioned multidimensional transient triggering and cross-stripping module is used to trigger high-frequency sampling and extract the high-frequency transient induced voltage sequence matrix based on mechanical commands and the absolute value characteristics of voltage differences; and to perform multiplication operations using the inverse matrix of the optimal steady-state space decoupling matrix to strip the cross electric field interference and reconstruct the absolute ground transient voltage sequence. The aforementioned frequency domain inversion compensation and state assessment module is used to call the inverse function of the sensor's broadband transfer function model to filter and compensate the absolute ground transient voltage sequence in the frequency domain; generate the circuit breaker break transient recovery voltage time sequence by subtraction; and determine the circuit breaker breaking performance based on the peak voltage characteristics and transient recovery voltage rise rate characteristics. The technical concept of this invention lies in steady-state self-learning spatial determination, transient decoupling and interference removal, and frequency domain adaptive distortion compensation; it replaces static hardware calibration with data-driven dynamic calibration, and reshapes physical isolation with multi-dimensional mathematical decoupling; it breaks through the traditional thinking limitation of trying to resist environmental interference by optimizing sensor hardware materials, and utilizes the operating law of circuit breakers being equipotential when closed and transient when open: during the closed steady-state period, the power grid itself is used as a natural standard signal source, and the recursive least squares method is combined to allow the system to enter a self-learning state, locking stray capacitances that drift with the environment in real time; during the open transient period, the learned matrix is ​​immediately inverted and transformed into a digital isolation barrier to remove crosstalk between contacts. Meanwhile, Wiener filtering is introduced for frequency domain inverse compensation to address high-frequency distortion; the non-contact sensor, which was originally highly susceptible to environmental interference, is upgraded into a precision intelligent measurement system with self-calibration, cross-decoupling, and wideband compensation capabilities; this not only ensures the absolute safety of personnel and equipment in 10,000-volt high-voltage measurements, but also achieves the accurate capture of nanosecond-level characteristics of megahertz-level transient recovery voltage in the complex electromagnetic environment of substations, enabling a quantitative assessment of circuit breaker breaking risks; The following detailed explanation is provided with reference to specific embodiments: Example 1: This embodiment uses the actual process of a 110kV SF6 gas-insulated circuit breaker (nameplate rated voltage Ur=126kV) breaking a three-phase short-circuit fault in a substation as an example to explain in detail the measurement method described in this invention: After the system is powered on, it enters the steady-state monitoring phase; the power grid is operating normally, and the power frequency reference voltage discrete sequence is synchronously sent by the substation bus PT. U grid ( n The effective value is 110kV / 3≈63.5kV, and its peak voltage is 89.8kV; at this time, the circuit breaker is in the closed state, the moving and stationary contacts are physically tightly pressed together, and their potentials are absolutely equal. The ADC samples the first and second consecutive analog voltage signals at a rate of 10 kS / s; the system extracts 200 sampling points, performs a moving average filter, and subtracts the operational amplifier offset voltage bias. V offset Then, a one-dimensional discrete-time voltage signal is generated. V 1( n )and V 2( n At the peak of the positive half-wave of a certain power frequency cycle, the system extracts the discrete voltage sequence output of the stationary contact sensor as follows: V 1 = 2.52V, the discrete voltage sequence output of the moving contact sensor is V 2 = 2.48V; The system calls the RLS recursive least squares algorithm in the steady-state adaptive parameter decoupling optimization module, and inputs the regression vector. =[89800,89800] T With observation vector Y =[2.52,2.48] T Substituting into the equation; after 100 recursive iterations, the error covariance matrix... P ( n Convergence, output equivalent aggregation coefficients K A =2.806×10 5 , K B =2.761×10 5 ; The system extracts the factory-preset a priori scaling constants of the three-dimensional insulation structure. c =0.15; Perform algebraic decoupling operation: K 11 =2.806×10 5 / 1.15=2.44×10 5 ; K 12 =0.15× K 11 =0.366×10 5 Similarly, calculate K 22 =2.40×10 5 , K 21 =0.360×10 5 The system fills these four precisely quantified elements into the optimal steady-state space decoupling matrix. K opt In this step, the actual grid voltage is used as the reference, eliminating stray capacitance drift error caused by changes in the dielectric constant of the insulator surface due to thunderstorms on the day of the incident. When the system runs to t At 50.00ms, a short-circuit fault occurs on the transmission line, and the relay protection device activates; the microprocessor's optocoupler captures the falling edge signal generated by the opening of contact 52a of the circuit breaker operating mechanism, and starts a 10ms hardware debounce timer; t At 60.00ms, the trip command is deemed valid, and the state machine switches to the pre-triggered armed state; in this state, the system continues to calculate. V 1( n The absolute value feature of the difference between adjacent points in the sequence Δ V The system extracts the maximum difference value recorded in the first 10 power frequency cycles as the steady-state noise reference Δ. V base =0.01V; Set multiplier l =5.0, the transient over-limit threshold is determined to be 0.05V; exist t At 65.23ms, the moving and stationary contacts of the circuit breaker had separated, the arc was extinguished at the instant the short-circuit current crossed zero, and violent interphase high-frequency oscillations occurred at the potentials across the break points; the ADC detected the absolute value characteristic of the voltage difference between the two adjacent points. D V=0.18V, far exceeding the transient over-limit threshold of 0.05V; the main control chip immediately triggers a hardware interrupt, increasing the ADC sampling rate to 100MS / s (sampling interval 10ns); at the current trigger point n Using 0 as the baseline, extract the preceding data from the circular buffer. N pre =5000 data points, continuously collected from there. N post=45,000 data points are spliced ​​to generate a high-frequency transient induced voltage sequence matrix with a total length of 50,000 points. V trans ; Entering the transient decoupling phase, the system calculates the matrix. K opt Determinant | K opt |=(2.44×2.40) (0.366×0.360)×10 10 =5.72×10 10 Its absolute value is greater than the singular value tolerance threshold of 10. 6 The matrix is ​​determined to be non-singular; the inverse matrix is ​​obtained using the adjoint matrix method and used as a transient decoupling operator. K opt 1 ; against V trans The 10000th discrete sampling point in the matrix (i.e., near the transient peak) has the following original voltage reading from the sensor: V 1 = 3.50V V 2= 1.20V; the system multiplies the vector on the left by the transient decoupling operator. K opt 1 Perform matrix multiplication: U 1raw =[2.40×10 5 ×3.50 ( 0.366×10 5 )×( 1.20)] / (5.72×10 10 =139kV; Through decoupling multiplication, the negative potential of the moving contact was successfully isolated through the stray capacitance in space. K 12 The pull-down interference caused by the stationary contact sensor restored the true 139kV absolute ground potential. Entering the frequency domain compensation phase, the system performs compensation on 50,000 points. U 1raw The sequence is mapped to the frequency domain using an FFT transform; the pre-calibrated wideband transfer function model H(z) of the sensor is then called to calculate the complex frequency response matrix in the frequency domain. H ( ehω And combined with regularization parameters α reg ( oh )=0.005 to calculate the inversion compensation operator G ( e hω The frequency domain complex sequence is multiplied by the inversion compensation operator using a complex dot product, and then inversely transformed back to the time domain using IFFT. Waveform spikes that were originally smoothed due to parasitic inductance are physically compensated, and the transient voltage amplitude of the stationary contact is corrected from the reconstructed 139kV to a fully compensated 145kV. Similarly, the moving contact sequence is decoupled and compensated to obtain the true potential at the fully compensated discrete point. 45kV; The system performs a final differential evaluation: subtracting the moving contact voltage from the fully compensated stationary contact voltage at the same discrete time step generates a time series of transient recovery voltages at the break point. U TRV =145kV ( 45kV) = 190kV; for a system containing 50,000 points U TRV ( n trans A global extremum search is performed on the sequence to extract the discrete point with the largest absolute value, and the peak voltage characteristics are recorded. U c =190kV; Extract the time interval characteristic between the zero-crossing point and the peak point of the waveform. t 3 = 63 μs; U c Divide by t 3. Calculate the transient recovery voltage rise rate characteristics. RRRV =190 / 63=3.01kV / μs; The system extracts the limiting parameters set according to the IEC 62271-100 standard from the nameplate of the 126kV circuit breaker in the non-volatile memory: rated peak voltage threshold. U crated =216kV, rated rise rate threshold RRRV rated =2.0kV / μs; Logical comparison revealed that although the calculated peak value of 190kV < 216kV, the calculated rise rate of 3.01kV / μs was greater than the rated rise rate threshold of 2.0kV / μs; The excessively steep rise rate meant that the strength recovery speed of the insulation medium between the breaks was slower than the voltage rise rate; The system immediately generated and output an alarm message of "break failure risk / re-breakdown hazard" to the station control layer terminal through the IEC 61850 MMS protocol, providing quantitative data closed-loop support for subsequent power outage maintenance of equipment; In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0016] The preferred embodiments of the invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method of non-contact measurement of circuit breaker opening voltage, characterized by, Includes the following steps: S1. Extract the original induced charge signals output by the external non-contact electric field sensors on the stationary and moving contact sides of the circuit breaker's arc-extinguishing chamber, and convert them into discrete time-series voltage signals. S2. When the circuit breaker is in the closed-current steady state stage, the discrete time series voltage signal and the power frequency reference voltage signal of the power grid bus are used as optimization benchmarks to construct the spatial capacitance cross-coupling matrix equation. The self-coupling coefficient characteristics and mutual coupling coefficient characteristics in the spatial capacitance cross-coupling matrix equation are calculated and updated in real time using the recursive least squares method to generate the optimal steady-state spatial decoupling matrix. S3. When a tripping command is detected and the amplitude change rate of the discrete time sequence voltage signal exceeds the limit, the high-frequency transient induced voltage sequence during the process of the arc burning to extinguishing at the break point is extracted. S4. Obtain the inverse matrix of the optimal steady-state space decoupling matrix as the transient decoupling operator; perform matrix multiplication on the high-frequency transient induced voltage sequence and the transient decoupling operator to reconstruct the absolute ground transient voltage sequence of the stationary contact and the absolute ground transient voltage sequence of the moving contact. S5. In the frequency domain, the absolute ground transient voltage sequence of the stationary contact and the absolute ground transient voltage sequence of the moving contact are respectively inverted and filtered and compensated using the inverse function of the wideband transfer function of the sensor to obtain the compensated absolute ground transient voltage sequence of the stationary contact and the compensated absolute ground transient voltage sequence of the moving contact. S6. Subtract the compensated transient voltage sequence of the moving contact from the absolute transient voltage sequence of the stationary contact to ground to generate a transient recovery voltage time sequence of the circuit breaker. Calculate the peak voltage characteristics and transient recovery voltage rise rate characteristics based on the transient recovery voltage time sequence of the circuit breaker to determine the circuit breaker's breaking performance.

2. The method of claim 1, wherein, S1 includes the following steps: S11. On the stationary contact side and the moving contact side of the circuit breaker insulation bushing flange, a double-shielded D-dot non-contact electric field sensor based on the differential geometry principle is symmetrically arranged; the original induced charge signal output by the sensor on the stationary contact side is extracted as the first current signal, and the original induced charge signal output by the sensor on the moving contact side is extracted as the second current signal. S12. Establish hardware integral conversion equations; substitute the first current signal and the second current signal extracted in S11 into the hardware integral conversion equations respectively, and calculate and output the first continuous analog voltage signal and the second continuous analog voltage signal. S13. The first continuous analog voltage signal and the second continuous analog voltage signal are continuously discretized and sampled using a high-precision analog-to-digital converter, and the DC bias characteristics are deducted using a moving average filtering algorithm to obtain a pure AC discrete time series voltage signal.

3. The method of claim 2, wherein, The specific calculation process of the moving average filtering algorithm in S13 includes the following steps: S131, set the sliding time window length as N window discrete sampling points; S132, at the first n discrete sampling point, extracting the continuous n N window voltage amplitude data from the original discrete sequence n N window voltage amplitude data; and accumulating and summing the continuous N window voltage amplitude data, and dividing the accumulated sum by the N window arithmetic mean value of the current time step.​ S133, taking the arithmetic mean value as a dynamic quantization value of the input offset voltage characteristic Voffset of the operational amplifier; subtracting the dynamic quantization value from the original discrete sequence voltage amplitude of the current first n discrete sampling point, and outputting a discrete time sequence voltage signal.

4. The method of claim 2, wherein, S2 includes the following steps: S21. Extract the discrete-time voltage signal generated in S13 and the steady-state voltage sequence to obtain the absolute voltage to ground of the stationary contact and the absolute voltage to ground of the moving contact; establish the physical constraints of the circuit breaker's closed state: during the closed current-carrying period, set the absolute voltage to ground of the stationary contact and the absolute voltage to ground of the moving contact to be electrically equipotential. S22. Establish the equation for the second-order spatial capacitance cross-coupling matrix; S23. Substitute the physical constraints in S21 into the second-order space capacitance cross-coupling matrix equation in S22 to obtain a system of linear equations. S24. Using the recursive least squares method based on the forgetting factor, iteratively solve the linear equation system within the set power frequency cycle time window to calculate the first equivalent aggregation coefficient and the second equivalent aggregation coefficient; call the prior proportional constant set by the three-dimensional spatial symmetry geometry of the circuit breaker arc-extinguishing chamber, substitute the first equivalent aggregation coefficient and the second equivalent aggregation coefficient into the algebraic equation system for decoupling, and output the optimal steady-state space decoupling matrix containing four independent elements.

5. The non-contact measurement method for circuit breaker break voltage according to claim 4, characterized in that, The steps in S24, which utilize a recursive least squares method based on a forgetting factor to iteratively solve the problem within a set power frequency cycle time window, include the following: S241. Initialize the parameter vector of the equivalent aggregation coefficients; initialize the error covariance matrix; set the forgetting factor; S242. At any discrete time step, extract the discrete sequence of the power frequency reference voltage of the power grid bus in S21 to construct the input regression vector; extract the discrete time sequence voltage signal generated in S13 to construct the actual observation vector. S243. Using the error covariance matrix from the previous time step and the current input regression vector, calculate the Kalman gain feature matrix: S244. Calculate the prior prediction error characteristics using the formula error calculation formula; S245. Multiply the Kalman gain feature matrix by the prior prediction error feature, add the parameter vector of the previous time step, and update the current parameter vector. S246. Update the current error covariance matrix using the formula for calculating the error covariance matrix; S247. Calculate the rate of change of the Euclidean norm of the parameter vectors of adjacent time steps; when the rate of change of the Euclidean norm is less than the convergence tolerance within a continuously set number of steps, stop the iteration; extract the elements in the parameter vector at this time as the first equivalent aggregation coefficient and the second equivalent aggregation coefficient.

6. The non-contact measurement method for circuit breaker break voltage according to claim 2, characterized in that, S3 includes the following steps: S31. Monitor the status signal of the mechanical auxiliary contact of the circuit breaker operating mechanism. When a level transition characteristic from closed to open is detected, start the set mechanical debounce timer. When the mechanical debounce timer overflows and the level transition characteristic is maintained, generate a pre-triggering command. S32. After receiving the pre-triggered arming command, extract the discrete time sequence voltage signal generated in S13 and calculate the absolute value of the difference between two adjacent sampling points. S33. Extract the maximum value of the absolute value of the difference within the continuous power frequency cycle during the steady-state phase of the closed-circuit flow, and set it as the steady-state noise reference; multiply the steady-state noise reference by the set multiplier parameter to calculate the transient over-limit threshold. S34. When the real-time calculated absolute value of the difference is greater than the transient over-limit threshold, a hardware-level synchronous sampling interrupt is triggered. Based on the current trigger point, several historical sampling points of the pre-trigger sampling points before the current trigger point are extracted from the ring buffer, and the sampling rate of the analog-to-digital converter is increased. Several sampling points of the subsequent trigger sampling points are continuously collected. The first continuous analog voltage signal and the second continuous analog voltage signal output by S12 are spliced ​​together to generate a high-frequency transient induced voltage sequence matrix with a total length of the number of pre-trigger sampling points plus the number of subsequent trigger sampling points.

7. The non-contact measurement method for circuit breaker break voltage according to claim 3, characterized in that, S4 includes the following steps: S41. Extract the optimal steady-state space decoupling matrix output from S24; calculate the determinant feature of the optimal steady-state space decoupling matrix using the determinant feature calculation formula; S42. Compare the absolute value of the determinant feature with the set singular value tolerance threshold; if the absolute value is greater than the singular value tolerance threshold, use the adjoint matrix method to calculate the inverse matrix of the optimal steady-state space decoupling matrix using the determinant feature as the denominator, and generate the transient decoupling operator. S43. Extract each discrete time step vector from the high-frequency transient induced voltage sequence matrix generated in S34; The discrete time step vector is multiplied on the left by the transient decoupling operator generated by S42, and the matrix multiplication operation equation is executed; the absolute transient voltage sequence of the stationary contact to ground and the absolute transient voltage sequence of the moving contact to ground are output.

8. A non-contact measurement method for circuit breaker break voltage according to claim 7, characterized in that, S5 includes the following steps: S51. Extract the parameters of the discrete difference equation of the entire measurement system obtained in the offline calibration stage, and construct the sensor broadband transfer function in the discrete complex frequency domain and Z domain. S52. Using fast Fourier transform, the static contact absolute to ground transient voltage sequence and the moving contact absolute to ground transient voltage sequence output by S43 are mapped to the frequency domain to obtain the frequency domain complex sequence. S53. Calculate the frequency response of the wideband transfer function of the sensor on the unit circle to obtain the complex frequency response matrix at discrete frequency points; introduce Wiener filter regularization parameters and calculate the inversion compensation operator using the inversion compensation operator calculation formula. S54. In the frequency domain, perform element-wise multiplication operations on the frequency domain complex sequence generated in S52 and the inversion compensation operator generated in S53 to generate a frequency domain compensation sequence. The frequency domain compensation sequence is mapped back to the time domain using the inverse fast Fourier transform, and the compensated static contact absolute transient voltage sequence to ground and the compensated moving contact absolute transient voltage sequence to ground are output.

9. A non-contact measurement method for circuit breaker break voltage according to claim 8, characterized in that, S6 includes the following steps: S61. Under the same discrete time step index, subtract the compensated transient voltage sequence of the stationary contact to ground from the compensated transient voltage sequence of the moving contact to ground output by S54 to generate the transient recovery voltage time sequence of the circuit breaker. S62. Perform a global extreme value search on the time series of transient recovery voltage at the circuit breaker contact, and extract the discrete data point with the largest absolute value as the peak point; record the voltage amplitude corresponding to the peak point as the peak voltage feature. S63. Extract the voltage zero-crossing point corresponding to the instant of arc extinction in the transient recovery voltage time series of the circuit breaker contact; calculate the time elapsed from the voltage zero-crossing point to the peak point, and record it as a time interval feature; S64. Divide the peak voltage feature extracted in S62 by the time interval feature extracted in S63 to calculate the transient recovery voltage rise rate feature. S65. Extract the rated peak voltage threshold and rated rise rate threshold from the pre-stored rated TRV envelope parameters on the circuit breaker nameplate; perform logical comparisons between the peak voltage characteristic and the transient recovery voltage rise rate characteristic and the rated peak voltage threshold and the rated rise rate threshold, respectively; if the peak voltage characteristic > the rated peak voltage threshold or the transient recovery voltage rise rate characteristic > the rated rise rate threshold, generate and output a circuit breaker interruption failure risk alarm message.

10. A non-contact measurement system for circuit breaker break voltage, characterized in that, A non-contact measurement method for circuit breaker break voltage as described in any one of claims 1-9, the system comprising: The signal sensing and integration preprocessing module is used to extract the original induced charge signals from the external electric field sensors of the stationary and moving contacts, convert the original induced charge signals into continuous analog voltage signals using an active integration circuit, and generate discrete time series voltage signals through discretization and debiasing processing. The steady-state adaptive parameter decoupling optimization module is used to extract the steady-state voltage sequence from the discrete-time voltage signal when the circuit breaker is closed, and combine it with the power frequency reference voltage signal of the power grid bus. The module then uses the recursive least squares method to update the coefficients of the spatial capacitance cross-coupling matrix equation in real time to generate the optimal steady-state spatial decoupling matrix. The multidimensional transient triggering and cross-stripping module is used to trigger high-frequency sampling and extract the high-frequency transient induced voltage sequence matrix based on mechanical commands and the absolute value of voltage difference; and to perform multiplication operation using the inverse matrix of the optimal steady-state space decoupling matrix to strip the cross electric field interference and reconstruct the absolute ground transient voltage sequence. The frequency domain inversion compensation and state assessment module is used to call the inverse function of the sensor's wideband transfer function model to filter and compensate the absolute ground transient voltage sequence in the frequency domain; generate the circuit breaker break transient recovery voltage time series by subtraction; and determine the circuit breaker breaking performance based on the peak voltage characteristics and transient recovery voltage rise rate characteristics.

Images