A method for monitoring aging of a connector of an electric energy metering box based on loop impedance identification

Abstract

This invention discloses a method for monitoring the aging of connectors in energy metering boxes based on loop impedance identification. The steps are as follows: Step S01, injecting a broadband composite excitation signal into the loop containing the connector; Step S02, synchronously capturing the reflected and transmitted signals generated at the loop interface and collecting the loop load current in real time to construct an original impedance response dataset; Step S03, identifying the chemical composition and growth thickness of the oxide layer at the contact interface; Step S04, calculating the effective contact point area density using an interface quantum tunneling effect mapping model; Step S05, calculating the bolt tightening torque retention rate using a thermomechanical deformation evolution model; Step S06, matching the oxide layer growth thickness, effective contact point area density, and bolt tightening torque retention rate with a preset health fingerprint grading standard to output a diagnostic conclusion. The beneficial effect of this invention is that it can distinguish the causes of aging and detect early aging, significantly improving the accuracy and reliability of monitoring the aging of connectors in energy metering boxes.

Description

Technical Field

[0001] This invention belongs to the field of power equipment condition monitoring and fault diagnosis technology. Specifically, it relates to an aging monitoring method for power metering box connectors based on loop impedance identification. Background Technology

[0002] Electricity metering boxes are core equipment for accurate electricity metering and power management at the end of low-voltage distribution networks, and are widely deployed in various power supply and distribution locations such as industrial plants, commercial buildings, and residential communities. The connectors inside the metering box (including DIN rail circuit breaker terminals, CT secondary circuit terminal blocks, and meter pins) bear the responsibility of reliable conduction of high-density current, and the conductivity of their contact interfaces directly determines metering accuracy and equipment safety. With increasing service life, connectors inevitably undergo contact aging under the combined effects of humid and hot environments, load current thermal cycling, and mechanical vibration, leading to increased contact resistance, localized overheating, and even burn-out accidents. This results in abnormal power loss and inaccurate metering, and in severe cases, threatens personnel and property safety. Therefore, accurate and real-time aging monitoring of metering box connectors has significant engineering and economic value.

[0003] Existing methods for monitoring connector aging mainly fall into two categories: one is based on direct measurement of contact resistance, which involves applying a DC current using a four-wire micro-ohmmeter under power-off conditions, measuring the voltage drop across the contact point, and calculating the contact resistance value to determine the contact condition; the other is based on abnormal temperature detection, which uses an infrared thermal imager or an attached temperature sensor to monitor the connector surface temperature, using excessive temperature rise as an aging warning signal. In addition, some studies have attempted to place vibration sensors near the terminals to detect the characteristic frequency of vibrations caused by loosening, thereby assisting in determining the mechanical fastening condition.

[0004] However, the above methods have significant limitations in practical applications. The contact resistance method only reflects the macroscopic resistance value of the current contact state and cannot distinguish whether the increase in contact resistance stems from a thickening of the interface oxide layer, a reduction in the effective contact area, or a decrease in contact pressure caused by loose bolts; furthermore, this method requires a power outage, making it difficult to implement online monitoring while the circuit is live. The temperature anomaly detection method is limited by the resolution of thermal images and on-site obstruction conditions, lacking sensitivity to early, subtle aging. It can only detect anomalies when the contact state has severely deteriorated and the temperature rise is significant, lacking preventative diagnostic capabilities. The vibration detection method only reflects degradation in the mechanical fastening dimension and has no response to contact deterioration caused by chemical corrosion and oxidation. Summary of the Invention

[0005] Currently, methods for monitoring the aging of connectors in electricity metering boxes, whether contact resistance or temperature anomaly monitoring, have significant limitations. The former cannot distinguish the cause of aging and requires power outage operation, while the latter has poor sensitivity to early, subtle aging. To address these issues, the following invention is proposed: A method for monitoring the aging of connectors in an energy metering box based on loop impedance identification includes the following steps: Step S01: Inject a wideband composite excitation signal with timing coding characteristics into the circuit where the connector under test is located; Step S02: For the broadband composite excitation signal injected in step S01, synchronously capture the reflected and transmitted signals generated at the loop interface and collect the loop load current in real time, extract the amplitude and phase of each frequency component, and construct the original impedance response dataset. Step S03: The original impedance response dataset constructed in step S02 is processed using a dielectric polarization spectrum decoupling analysis algorithm to identify the chemical composition and growth thickness of the oxide layer at the contact interface. Step S04: Based on the chemical composition and growth thickness of the oxide layer identified in step S03, calculate the electron penetration probability and the effective contact point area density of the contact interface using the interface quantum tunneling effect mapping model. Step S05: Using the effective contact point area density obtained in step S04 and the loop load current collected in step S02 as inputs, the bolt tightening torque retention rate under load fluctuation thermal stress is dynamically calculated through the thermomechanical deformation evolution model coupled with the thermoelastic equation of the material. Step S06: Match the oxide layer growth thickness identified in step S03, the effective contact point area density obtained in step S04, and the bolt tightening torque retention rate calculated in step S05 with the preset health fingerprint grading standard, and output a diagnostic conclusion that includes aging degree rating, remaining life prediction, and maintenance suggestions.

[0006] Furthermore, in step S01, the wideband composite excitation signal is generated by a combination of multi-frequency sine wave superposition and pseudo-random binary sequence modulation, covering a frequency band of 1 Hz to 1 MHz. The amplitude of each frequency component is weighted according to a predetermined amplitude spectrum distribution to ensure a uniform power spectral density across the entire frequency band.

[0007] Furthermore, in step S02, reflected and transmitted signals are independently acquired under at least three different load conditions to eliminate background noise introduced by load fluctuations; the impedance response under each load condition is organized in matrix form to form a multi-load state impedance matrix, which serves as the input for subsequent decoupling analysis.

[0008] Further, in step S03, the dielectric polarization spectrum decoupling analysis algorithm uses an improved Cole-Cole model to perform layered fitting of the relaxation arcs of each frequency band, extracts the relaxation time constant and arc width parameters corresponding to each polarization mechanism, and performs similarity matching with the pre-stored standard spectrum feature library of oxide components such as copper oxide, cuprous oxide, and copper sulfide to determine the mole fraction of each component, and then calculates the comprehensive equivalent thickness of the oxide layer at the contact interface.

[0009] Further, in step S04, the interface quantum tunneling effect mapping model uses the Wenzel-Clemers-Brillouin approximation method to calculate the probability of electrons penetrating the oxide barrier, and uses the equivalent thickness of the oxide layer as the barrier width parameter and the weighted average value of the bandgap energy of each oxide component as the barrier height parameter to establish a monotonic mapping function between the penetration probability and the effective contact point area density.

[0010] Further, in step S05, the thermomechanical deformation evolution model calculates the temperature rise of the contact area based on Joule's law using the current density within the effective contact area, uses the thermal expansion coefficient and elastic modulus of the contact material as thermoelastic parameters, and uses the Prandtl-Royce elastoplastic constitutive equation to calculate the cumulative plastic strain caused by periodic temperature changes, and then uses the strain-preload conversion relationship to deduce the staged relaxation amount of the bolt tightening torque.

[0011] Furthermore, in step S06, the health fingerprint grading standard includes three levels: normal, warning, and dangerous. The grading judgment is based on the combined criteria of the oxide layer equivalent thickness threshold, the effective contact point area density threshold, and the bolt tightening torque retention rate threshold. If all three indicators are within the normal threshold range, it is judged as normal; if any one reaches the warning threshold, it is judged as warning; and if any one touches the dangerous threshold, it is directly judged as dangerous.

[0012] Furthermore, step S01 also includes adaptive frequency band adjustment, which dynamically adjusts the weight of each frequency band component based on the equivalent thickness of the oxide layer identified in the previous measurement in step S03; when the equivalent thickness of the oxide layer is lower than a preset thin film threshold, the weight of the high-frequency band component is increased to improve the spectral resolution of the thin oxide layer; when the equivalent thickness of the oxide layer exceeds a preset thick film threshold, the weight of the low-frequency band component is increased to ensure the penetration depth and signal-to-noise ratio of deep polarization information.

[0013] Compared with existing technologies, this invention has the following advantages: It achieves online measurement of live circuits by actively injecting a wideband composite excitation signal, eliminating the need for power outages and ensuring uninterrupted power supply; it independently identifies the chemical composition and equivalent thickness of the oxide layer at the microscopic level using a dielectric polarization spectrum decoupling algorithm, overcoming the fundamental deficiency of traditional contact resistance methods in distinguishing oxidation mechanisms; it introduces a quantum tunneling effect mapping model, transforming the unobservable microscopic contact point density into a calculable macroscopic characterization, achieving cross-scale correlation from electronic transport mechanisms to macroscopic electrical performance; the thermomechanical deformation evolution model comprehensively considers the local current density increase effect caused by the decrease in contact point density and the nonlinear relaxation characteristics of bolt torque, providing a more physically accurate lifetime decay prediction trajectory; the three-dimensional health fingerprint joint criterion significantly reduces the false alarm and false alarm rates compared to a single threshold method; and the multi-connector time-division multiplexing scanning architecture enables the method to achieve full-coverage monitoring at the box level, meeting the needs of large-scale deployment. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the overall process of the present invention. Detailed Implementation

[0015] The following is in conjunction with the appendix Figure 1 The present invention will be further described in detail below to enable those skilled in the art to more clearly understand the invention. The steps are as follows: A method for monitoring the aging of connectors in power metering boxes based on loop impedance identification is proposed.

[0016] This method relies on a dedicated monitoring device integrated into a standardized DIN rail module, installed in an empty rail position within the power metering box. The device includes the following functional modules: 1. A signal generation module, containing an arbitrary waveform generator and drive amplifier circuit, responsible for generating a wideband composite excitation signal and controlling the injection power; 2. A coupling injection and acquisition module, containing a dual-channel CT coupler and a high-speed synchronous ADC, used for non-invasive injection of the excitation signal and synchronous acquisition of reflected and transmitted signals, respectively; 3. A signal processing and calculation module, containing an embedded processor, running a dielectric polarization spectrum decoupling algorithm, a quantum tunneling mapping model, a thermomechanical deformation evolution model, and a health fingerprint matching algorithm; 4. A communication and storage module, containing a 4G / NB-IoT wireless communication unit and non-volatile memory, responsible for uploading diagnostic results and storing local historical data; 5. An auxiliary power supply module, drawing power from the power distribution bus within the metering box to provide the isolated regulated DC power required by each module.

[0017] The connection between the monitoring device and the connector under test uses a standardized probe adapter. The probe adapter is available in different models according to different DIN rail types and is attached to the busbars or input / output terminals on both sides of the connector under test via a snap-fit ​​installation. The adapter has a built-in current sensing coil that senses the high-frequency broadband excitation signal on the conductor segments on both sides of the connector, effectively isolating it from the power frequency circuit and ensuring operational safety. During probe adapter installation, there is no need to remove the connector or disconnect the load current; the entire process is completed while the circuit is energized, meeting the requirements for uninterrupted power supply installation.

[0018] Step S01: Inject a wideband composite excitation signal with timing coding characteristics into the circuit where the connector under test is located. The aging characteristics of connector interfaces are distributed across a very wide frequency range: the thin oxide layer exhibits significant capacitive polarization response in the high-frequency range (10 kHz to 1 MHz), while the deep oxide layer and interface double-layer effect are mainly manifested in the low-frequency range (1 Hz to 1 kHz). Mechanical contact inhomogeneity caused by bolt loosening produces specific inductive-resistive mixed characteristics in the mid-frequency range (100 Hz to 10 kHz). Therefore, effectively extracting information from all aging dimensions requires the excitation signal to cover a wide frequency range of 1 Hz to 1 MHz, and each frequency band must have a sufficient signal-to-noise ratio.

[0019] This step generates a wideband composite excitation signal by combining multi-frequency sine wave superposition with pseudo-random binary sequence (PRBS) modulation. Specifically, in the low-frequency band of 1 Hz to 1 kHz, 20 discrete frequency points are selected at logarithmic intervals and superimposed to generate a low-frequency sine wave mixture; in the mid-high frequency band of 1 kHz to 1 MHz, 30 discrete frequency points are selected at logarithmic intervals and superimposed to generate a mid-high frequency sine wave mixture. The synthesized signal is time-modulated using a PRBS code sequence to give the signal a time-domain identifiable coding feature, ensuring orthogonality between the excitation signals of each loop during multi-loop scanning and avoiding crosstalk. The amplitude of each frequency component is weighted according to a predetermined amplitude spectrum distribution, with the amplitude of the low-frequency components appropriately increased (to compensate for the output power limitation of the low-frequency excitation source) and the high-frequency components moderately attenuated (to prevent interference with the normal operation of the circuit under test), ultimately ensuring a uniform power spectral density distribution across the entire frequency band.

[0020] The excitation signal is applied to the circuit under test in a non-invasive manner through a coupling injection device. The injection point is selected on the busbars or conductors on both sides of the connector, and the injection power is controlled within a safe range that does not affect the measurement accuracy (typically 1 mW to 10 mW). The introduction of timing coding characteristics enables subsequent multiple acquisition data to have mutually verifiable timing identifiers, effectively distinguishing response drift caused by aging evolution and random fluctuations caused by measurement noise within different measurement cycles.

[0021] Furthermore, based on the equivalent oxide layer thickness identified in the subsequent step S03 of the previous measurement, adaptive frequency band adjustment can be performed on the current excitation signal: if the equivalent oxide layer thickness is below the thin film threshold of 0.5 nm, the weight of each component in the high-frequency band (100 kHz to 1 MHz) is increased by 30% to improve the spectral resolution of the dielectric relaxation arc of the thin oxide layer; if the equivalent oxide layer thickness exceeds the thick film threshold of 5 nm, the weight of each component in the low-frequency band (1 Hz to 100 Hz) is increased by 40% to ensure that the penetration depth and signal-to-noise ratio of the deep polarization mechanism meet the effective identification requirements. For the first measurement, adaptive adjustment is not performed; a uniformly weighted full-band signal is used directly.

[0022] Step S02: For the broadband composite excitation signal injected in step S01, simultaneously capture the reflected and transmitted signals and collect the loop load current in real time to construct the original impedance response dataset. After the excitation signal is injected, due to impedance mismatch at the connector contact interface, a reflected signal returning towards the injection point and a transmitted signal continuing to propagate across the interface will be generated simultaneously along the signal propagation path. In this step, while the excitation signal is injected, high-speed synchronous sampling modules distributed on both sides of the connector are used to synchronously capture the reflected and transmitted signals. The sampling rate is no less than 10 times the highest frequency component of the excitation signal to ensure aliasing-free acquisition under the Nyquist criterion.

[0023] Perform a Fast Fourier Transform (FFT) on the acquired time-domain signal to extract the complex amplitude (amplitude and initial phase) of the component corresponding to each excitation frequency point, and calculate the reflection coefficient accordingly. With transmission coefficient :

[0024] in, , , The injected reference signal, reflected signal, and transmitted signal are respectively located at the frequency... The complex amplitude at the point is calculated. Based on the reflection and transmission coefficients, the complex impedance spectrum of the contact interface is calculated according to the two-port network theory. :

[0025] in, This is the characteristic impedance of the loop (usually taken as 50 Ω).

[0026] To eliminate the impact of background current noise introduced by load fluctuations on the accuracy of the impedance spectrum, the above acquisition and calculation process was independently performed under three different load conditions: low load (20% of rated current), medium load (60% of rated current), and high load (100% of rated current), resulting in three sets of complex impedance spectra. Simultaneously, the corresponding real-time loop load current value was recorded for each load condition. These serve as the load input parameters for the thermomechanical deformation model in step S05. The three sets of data are organized in matrix form to construct a multi-load state impedance matrix. ,in The total number of excitation frequency points. The median is calculated for each column of the matrix (i.e., the impedance value at the same frequency point under three operating conditions). Outliers caused by instantaneous load disturbances are removed, and the median sequence is used as the final raw impedance response dataset. It is stored in two forms: a Nyquist plot (real-imaginary part plot) and a Bode plot (amplitude-frequency, phase-frequency plot), providing input for the spectral decoupling analysis in step S03. For example... Figure 1 As shown, the above two steps constitute the front-end signal acquisition stage of the entire aging monitoring process.

[0027] In terms of the engineering implementation of synchronous acquisition, independent analog-to-digital converters (ADCs) are configured at the reflecting and transmitting ends, respectively. The two ADCs share the same clock source to ensure that the phase synchronization error does not exceed 5 ns, meeting the requirement that the phase error does not exceed 1.8° at the highest frequency of 1 MHz. Before being written to the memory, the acquired time-domain signal is processed by an anti-aliasing low-pass filter to eliminate the aliasing effect of high-frequency noise. For environmental power frequency noise (50 Hz and its harmonics), a high-resolution spectrum analysis with a frequency resolution of not less than 0.1 Hz is used in the FFT calculation to effectively separate the power frequency component from the excitation frequency point. If the excitation frequency point happens to fall near the power frequency harmonics (frequency difference less than 0.5 Hz), the complex impedance data at that frequency point is marked as low confidence and is weighted down in the subsequent decoupling analysis.

[0028] After the multi-load state impedance matrix is ​​constructed, a consistency check must be performed on the matrix data: calculate the standard deviation of the real part of the impedance at the same frequency point under the three load conditions. If the standard deviation exceeds 15% of the corresponding median value, it is determined that the frequency point is contaminated by transient disturbances during load condition switching, and resampling is performed on the point. If the standard deviation still exceeds the standard after resampling, the frequency point is excluded from the dataset used for subsequent fitting, and the effective frequency coverage range is noted in the diagnostic report. Under normal circumstances, the fitting accuracy of step S03 can be guaranteed if the number of effective frequency points is not less than 85% of the total number of frequency points.

[0029] Step S03: Using the original impedance response dataset constructed in step S02, identify the chemical composition and growth thickness of the oxide layer at the contact interface using the dielectric polarization spectrum decoupling analysis algorithm. The oxide layer formed on the surface of copper or aluminum conductors in connectors during long-term service is not a single chemical component, but rather a coexistence of multiple oxidation and corrosion products, such as copper oxide (CuO), cuprous oxide (Cu2O), and copper sulfide (CuS), in a layered or mixed manner. Each component exhibits a unique relaxation arc in the electrochemical impedance spectroscopy due to its different crystal structure, band gap width, and dielectric relaxation mechanism. Traditional Cole-Cole models can only fit single relaxation processes and cannot effectively decouple complex oxide layers with multiple coexisting components. This step employs an improved multi-relaxation Cole-Cole model, decomposing the complex impedance spectrum in the original impedance response dataset into a superposition of several independent polarization mechanisms.

[0030] The improved expression for the total impedance of the Cole-Cole model is as follows:

[0031] in, It is the bulk series resistance. The number of relaxation arcs involved in the fitting (determined by the number of identifiable half-arcs in the Nyquist plot of the measured impedance spectrum, typically...) ), For the first The characteristic resistance of each polarization process Its relaxation time constant, For the arc width parameter ( It degenerates into standard Debye relaxation.

[0032] A nonlinear least squares algorithm (Levenberg-Marquardt method) was used to perform multi-arc layered fitting of the measured complex impedance spectrum, with the optimization objective being the weighted mean square error of the fitting residuals.

[0033] Among them, weight To normalize the weights, we ensure that the fitting residuals of the high-frequency band (smaller impedance magnitude) and the low-frequency band (larger impedance magnitude) are balanced in magnitude.

[0034] After fitting, extract the characteristic parameter sets of each relaxation arc. The sample is matched against a pre-stored standard spectral feature library for similarity. This library contains typical relaxation time constant ranges and arc width parameter distributions for various oxide components (CuO, Cu2O, CuS, Al2O3, etc.) calibrated under accelerated aging tests at different temperature and humidity conditions for copper and aluminum contact materials. The matching algorithm uses a normalized Euclidean distance metric to calculate the distance between the parameter set of the sample to be tested and the parameter prototypes of each component in the library. The three components with the smallest distances are selected as possible chemical compositions, and the matching is based on the relaxation arc area of ​​each component (proportional to...). Calculate the mole fraction of each component. .

[0035] Based on the relationship between the mole fraction of each component and its conductivity, the comprehensive equivalent thickness of the oxide layer is calculated using the effective medium theory. :

[0036] in, The effective dielectric constant is weighted by the mole fraction of each component. The vacuum permittivity, This represents the macroscopic area of ​​the contact interface. The interface equivalent capacitance extracted from the high-frequency band of the impedance spectrum (by fitting the high-frequency arc corresponding to...) , (Parameter conversion). Step S03 finally outputs the mole fraction of each component. Equivalent thickness of oxide layer This serves as the input for step S04.

[0037] Regarding algorithm robustness, two special cases need to be considered: first, when the contact interface is in the early, mild oxidation stage ( Below 0.5 nm, the relaxation arc features in the Nyquist plot are extremely weak, and the signal-to-noise ratio is low. Therefore, the Cole-Cole model is improved to a single-arc form for fitting. The oxide composition is tentatively set as the initial native oxide layer of copper (mainly Cu2O), and the uncertainty of the current measurement is marked as high. It is recommended to appropriately increase the high-frequency signal amplitude in the next measurement to improve the recognizability of the arc features. Secondly, when the contact interface has developed into a heavily oxidized state (… (Beyond 10 nm), the high resistance of the oxide layer limits the penetration depth, preventing the effective excitation of low-frequency relaxation arcs. Consequently, the fitting range automatically shrinks to the mid-to-high frequency range. The estimation accuracy decreased, but it was sufficient to determine the level of danger, and no precise numerical value was required. The standard spectral feature library is maintained using an open update mechanism. When a sufficient number of field calibration samples are accumulated, data-driven methods can be used to refine the prototype relaxation parameters of each component in the library online, further improving the algorithm's adaptability to different climate and operating conditions.

[0038] The initial establishment process of the standard spectral feature library is as follows: First, standard samples of contact interfaces containing known chemical components are prepared. Through heat treatment and chemical treatment, oxide layers mainly composed of single components (CuO, Cu2O, CuS, Al2O3, etc.) are generated on the surface of copper or aluminum contact sheets. The molar fraction of components and film thickness of each sample are determined by X-ray photoelectron spectroscopy (XPS) and transmission electron microscopy (TEM). Then, a broadband excitation signal with the same parameters as in step S01 is applied to each standard sample, and complex impedance spectra are collected in the corresponding frequency bands to extract the relaxation time constant. With arc width parameter Statistical analysis was performed on the collected data under various temperature and humidity conditions (temperature range 20℃ to 60℃, relative humidity range 40% to 95%) to establish the mean and variance of each parameter, serving as the parameter distribution prototypes of each component in the feature library under different environmental conditions. In actual matching calculations, based on the environmental parameters acquired in real time by the device's built-in temperature and humidity sensors, parameter prototypes under corresponding conditions were selected from the feature library for similarity comparison, further improving the robustness of component identification under varying temperature and humidity environments.

[0039] Step S04: Based on the oxide layer chemical composition and growth thickness identified in step S03, calculate the penetration probability and map the effective contact point area density using the interface quantum tunneling effect mapping model. The contact interface between the copper (or aluminum) conductors in the connector is not a perfectly tight contact at the microscopic scale. Instead, the entire current conduction task is carried by a finite number of randomly distributed real micro-protrusion contact points (asperiti). As the oxide layer grows, the surface of most of the micro-protrusion contact points is covered by oxide, causing the carrier conduction mechanism at the contact points to shift from classical ohmic conduction to quantum tunneling conduction. When the equivalent thickness of the oxide layer is less than about 10 nm, the quantum tunneling effect is the dominant mechanism for electrons to cross the oxide barrier, and the magnitude of its penetration probability directly determines the effective contribution of the contact point to macroscopic current conduction.

[0040] This step uses the Wenzel-Clemers-Brillouin (WKB) approximation method to establish a quantum tunneling probability calculation model. For a rectangular potential barrier, the tunneling probability given by the WKB approximation is:

[0041] in, To reduce Planck's constant, The effective mass of electrons in the oxide layer (the effective mass parameters of materials such as CuO and Cu2O are weighted according to the mole fraction of each component). The potential barrier height is the weighted average of the mole fractions of the band gap energy of each oxidizing component. For copper conductors, the Fermi level is given.

[0042] For the actual contact interface, let the total number of micro-protrusion contact points per unit area be... (Estimated from the macroscopic morphology of the contact surface and Hertzian contact theory), then the effective contact point area density Defined as:

[0043] in, The average contact area of ​​a single microprotrusion contact point (determined by material hardness and contact pressure according to the Bowden-Tabor model). With equivalent oxide layer thickness The value increases and then monotonically decreases. Beyond approximately 8nm When the effective contact area density approaches zero, the contact interface is completely dominated by the high-resistivity state controlled by the tunneling mechanism.

[0044] When the oxide layer thickness increases from 0 nm to 10 nm, the tunneling probability decreases by approximately four orders of magnitude, demonstrating a high sensitivity to oxide layer thickness. This is the physical basis for the early warning provided by this method even with minimal oxidation. Step S04 outputs the effective contact point area density. This will be passed to step S05 as the core parameter for calculating the heat source distribution.

[0045] Sensitivity analysis of the WKB quantum tunneling model shows that the penetration probability For the barrier width (i.e. The sensitivity to the potential barrier height is much higher than the sensitivity to the potential barrier height, which means that in step S03... The measurement accuracy is crucial to the final The estimation of the oxidizing component has the most significant impact, while the mole fraction error of each oxidizing component has the most significant impact. The impact is relatively minor. Based on a measurement uncertainty of ±0.3 nm for the oxide layer thickness, the corresponding relative uncertainty for the effective contact area density is approximately ±18% (in...). The uncertainty level (around nm) is engineering-acceptable for classifying health status.

[0046] Effective contact point area density The absolute value depends on the prior parameters of the contact material and initial contact pressure (Hertz contact radius and Bowden-Tabor hardness parameters). These parameters are obtained from the standard parameter library by querying the connector model during the installation phase. If it is a non-standard connector, they are entered supplementarily from the factory technical documents. For multiple sets of connectors of the same model, these parameters remain consistent and only need to be entered once for calculation of all connectors of the same model in the entire box, improving the engineering convenience of batch monitoring.

[0047] Step S05: Using the effective contact point area density obtained in step S04 and the loop load current collected in step S02 as inputs, the bolt tightening torque retention rate is dynamically calculated through the thermomechanical deformation evolution model coupled with the thermoelastic equation of the material.

[0048] There is a strong bidirectional coupling relationship between the mechanical fastening state of the connector (characterized by bolt tightening torque) and its contact resistance: the lower the effective contact area density, the higher the current density carried per unit contact area, and the more concentrated the Joule heat; the periodic fluctuations of Joule heat (fluctuating with the load current) cause the contact parts and bolt materials to undergo repeated thermal expansion and contraction, resulting in accumulated plastic strain; the accumulated plastic strain causes the bolt preload (equivalent to bolt tightening torque) to gradually loosen, further reducing the contact pressure and forming a positive feedback loop that accelerates aging. The thermomechanical deformation evolution model constructed in this step is a quantitative description of this coupled physical process.

[0049] 1. Calculation of heat source distribution With effective contact point area density With circuit load current As the input, the actual conductive area carrying the current is The average current density in the contact area is:

[0050] The volumetric power density (joule heat per unit volume) of the contact region is:

[0051] in, The resistivity of the contact material (considering temperature correction): , (Resistivity temperature coefficient). Temperature rise in the contact area. Solving the heat conduction equation under the quasi-steady-state approximation:

[0052] in, For the volume of the contact area, The equivalent convective heat transfer coefficient, This represents the heat dissipation area.

[0053] 2. Elastic-plastic constitutive equation Treating the contact components and bolt materials as elastoplastic bodies, their mechanical response under thermal stress is described using the Prandtl-Reuss incremental constitutive relation. The total strain increment considering thermal strain is decomposed into elastic strain increments. Plastic strain increment With thermal strain increment sum:

[0054] in, This is the coefficient of linear thermal expansion. When the equivalent stress exceeds the material's current yield strength... At this time, plastic flow is triggered, and the increment of plastic strain is given by the Prandtl-Reuss flow law:

[0055] in, Let be the yield function. The plastic multiplier is determined by the consistency condition. Within each load fluctuation cycle, the strain increment under time-varying thermal stress is numerically integrated (time step is 1 / 100 of the load cycle), and the total plastic strain at the end of each cycle is obtained by summing the results. ( (Number of load cycles).

[0056] 3. Bolt torque relaxation trajectory Bolt preload The relationship between the cumulative plastic strain and the elastic unloading equation is described by:

[0057] in, For the initial preload stress, The elastic modulus of the bolt material. Bolt tightening torque retention rate. Defined as:

[0058] The output of step S05 is the bolt tightening torque retention rate. The complete decay sequence over time is passed to step S06 for health status determination.

[0059] Key material parameters (coefficient of thermal expansion) in thermomechanical deformation evolution models Elastic modulus Yield strength Temperature-dependent parameters (including hardening coefficient, etc.) are stored in the material database, covering commonly used conductive materials for electrical connectors such as copper, brass, and phosphor bronze, as well as bolt materials such as stainless steel and carbon steel. The parameters are stored as piecewise linear functions as a function of temperature, allowing the model to automatically interpolate the material parameters corresponding to the current contact area temperature within the normal operating temperature range of 25℃ to 120℃.

[0060] Load current time series There are two ways to obtain the data: for smart metering boxes with real-time monitoring capabilities, it is obtained directly from the real-time current sampling interface of the metering module; for traditional metering boxes, the typical daily load curve of the circuit (generated by statistical analysis of historical electricity consumption data, categorized into typical scenarios such as weekdays, holidays, and seasons) is used as input. The typical daily load curve is stored with a time resolution of one sampling point every 15 minutes, and is automatically expanded into a high-resolution input sequence in seconds (through linear interpolation) in the thermomechanical deformation evolution model.

[0061] To ensure the accuracy of model predictions, the model should be calibrated every 12 months or after each on-site maintenance using the measured bolt tightening torque values ​​(measured directly with a torque wrench). The deviation between the measured torque values ​​and the model predictions should be fed back to the plastic strain variable value for each cycle. The correction factor is adjusted multiplicatively to achieve online alignment between the model and the measured data, thereby continuously improving the long-term prediction accuracy.

[0062] Step S06: Match the multidimensional aging characteristic parameters obtained in steps S03 to S05 with the preset health fingerprint grading standard, and output a comprehensive diagnostic conclusion. Up to this step, the comprehensive equivalent thickness of the oxide layer has been obtained from step S03. Obtain the effective contact point area density from step S04. Obtain the bolt tightening torque retention rate from step S05. The three parameters together constitute a multidimensional feature vector reflecting the degradation state of the connector. This step matches the feature vector with a preset health fingerprint grading standard to complete the comprehensive output of aging degree rating, remaining life prediction and maintenance recommendations.

[0063] 1. Health fingerprint grading standard The health fingerprint grading standard includes three levels: normal, warning, and dangerous. The threshold for each level is determined based on a combination of large-scale accelerated aging test data and field operation statistics, as detailed below: Normal level: (Typical value 2 nm) and (Typical value 60% of initial density) and (Typical value 90%) Warning level: (Typical range 2–5 nm) or (Typical range 30%–60% of initial density) or (Typical range 70%–90%) Hazard level: (Typical value 5 nm) or (Typical value 30% of initial density) or (Typical value 70%).

[0064] If all three indicators are within the normal range, the system is classified as normal; if any one indicator reaches the warning threshold, the system is classified as warning; if any one indicator reaches the danger threshold, the system is classified as dangerous, regardless of the status of the other indicators.

[0065] 2. Remaining life prediction Under normal and warning levels, based on the accumulated historical feature vector time series... Fit a degradation trend model for each feature. Oxide layer thickness. Exponential growth fitting based on the Arrhenius accelerated aging model was employed:

[0066] in, This is the oxidation rate constant (determined by least-squares fitting of historical sequences). Predicted based on extrapolated curves. Reaching the danger threshold Required remaining time Similarly, for and Each degradation curve is fitted separately to obtain the corresponding remaining lifetime prediction. and The final remaining lifespan is the minimum of the three factors:

[0067] 3. Maintenance suggestion generation Based on the aging level rating and the type of characteristic quantity that first reaches the danger threshold, targeted maintenance recommendations are generated: If For primary degradation, it is recommended to focus on checking the degree of oxidation on the contact surface and clean or replace them; if If the main degradation is detected, it is recommended to check whether the contact pressure has decreased due to the failure of the elastic component; if so... The primary degradation is identified, and it is recommended to prioritize tightening the bolts or replacing the washers. The final output is a comprehensive diagnostic report including the rating label, remaining life value, and the above-mentioned targeted maintenance recommendations.

[0068] When monitoring multiple sets of connectors within the power metering box, the system employs a time-division multiplexing strategy, switching the circuits under test sequentially according to a predetermined sequence. Each set of connectors undergoes the complete processing of steps S01 to S06 independently, and the feature vectors and diagnostic conclusions of each connector are written into a shared database. After all connectors have been scanned, an aging status matrix reflecting the overall health status of the box is generated, with rows corresponding to circuit numbers and columns corresponding to... , , The system includes fields such as rating. For connectors classified as hazardous, the system immediately sends a priority maintenance alarm, notifying maintenance personnel to handle them in the next maintenance window to prevent metering anomalies or equipment accidents caused by contact failures.

Claims

1. A method for monitoring the aging of connectors in an energy metering box based on loop impedance identification, characterized in that, Includes the following steps: Step S01: Inject a wideband composite excitation signal with timing coding characteristics into the circuit where the connector under test is located; Step S02: For the broadband composite excitation signal injected in step S01, synchronously capture the reflected and transmitted signals generated at the loop interface and collect the loop load current in real time, extract the amplitude and phase of each frequency component, and construct the original impedance response dataset. Step S03: The original impedance response dataset constructed in step S02 is processed using a dielectric polarization spectrum decoupling analysis algorithm to identify the chemical composition and growth thickness of the oxide layer at the contact interface. Step S04: Based on the chemical composition and growth thickness of the oxide layer identified in step S03, calculate the electron penetration probability and the effective contact point area density of the contact interface using the interface quantum tunneling effect mapping model. Step S05: Using the effective contact point area density obtained in step S04 and the loop load current collected in step S02 as inputs, the bolt tightening torque retention rate under load fluctuation thermal stress is dynamically calculated through the thermomechanical deformation evolution model coupled with the thermoelastic equation of the material. Step S06: Match the oxide layer growth thickness identified in step S03, the effective contact point area density obtained in step S04, and the bolt tightening torque retention rate calculated in step S05 with the preset health fingerprint grading standard, and output a diagnostic conclusion that includes aging degree rating, remaining life prediction, and maintenance suggestions.

2. The method for monitoring the aging of connectors in an energy metering box based on loop impedance identification according to claim 1, characterized in that: In step S01, the wideband composite excitation signal is generated by a combination of multi-frequency sine wave superposition and pseudo-random binary sequence modulation, covering a frequency band of 1 Hz to 1 MHz. The amplitude of each frequency component is weighted according to a predetermined amplitude spectrum distribution to ensure a uniform power spectral density across the entire frequency band.

3. The method for monitoring the aging of connectors in an energy metering box based on loop impedance identification according to claim 1, characterized in that: In step S02, reflected and transmitted signals are independently acquired under at least three different load conditions to eliminate background noise introduced by load fluctuations; the impedance response under each load condition is organized in matrix form to form a multi-load state impedance matrix, which serves as the input for subsequent decoupling analysis.

4. The method for monitoring the aging of connectors in an energy metering box based on loop impedance identification according to claim 1, characterized in that: In step S03, the dielectric polarization spectrum decoupling analysis algorithm uses an improved Cole-Cole model to perform layered fitting of the relaxation arcs of each frequency band, extracts the relaxation time constant and arc width parameters corresponding to each polarization mechanism, and performs similarity matching with the pre-stored standard spectrum feature library of oxide components such as copper oxide, cuprous oxide, and copper sulfide to determine the mole fraction of each component, and then calculates the comprehensive equivalent thickness of the oxide layer at the contact interface.

5. The method for monitoring the aging of connectors in an energy metering box based on loop impedance identification according to claim 1, characterized in that: In step S04, the interface quantum tunneling effect mapping model uses the Wenzel-Clemers-Brillouin approximation method to calculate the probability of electrons penetrating the oxide barrier. The equivalent thickness of the oxide layer is used as the barrier width parameter, and the weighted average value of the band gap energy of each oxide component is used as the barrier height parameter. A monotonic mapping function between the penetration probability and the effective contact point area density is established.

6. The method for monitoring the aging of connectors in an energy metering box based on loop impedance identification according to claim 1, characterized in that: In step S05, the thermomechanical deformation evolution model calculates the temperature rise of the contact area based on Joule's law using the current density within the effective contact area, uses the thermal expansion coefficient and elastic modulus of the contact material as thermoelastic parameters, and uses the Prandtl-Royce elastoplastic constitutive equation to calculate the cumulative plastic strain caused by periodic temperature changes, and then uses the strain-preload conversion relationship to deduce the staged relaxation amount of the bolt tightening torque.

7. The method for monitoring the aging of connectors in an energy metering box based on loop impedance identification according to claim 1, characterized in that: In step S06, the health fingerprint grading standard includes three levels: normal, warning, and dangerous. The grading judgment is based on the equivalent thickness threshold of the oxide layer, the effective contact point area density threshold, and the bolt tightening torque retention rate threshold as joint criteria. If all three indicators are within the normal threshold range, it is judged as normal level; if any one reaches the warning threshold, it is judged as warning level; and if any one reaches the dangerous threshold, it is directly judged as dangerous level.

8. The method for monitoring the aging of connectors in an energy metering box based on loop impedance identification according to claim 1, characterized in that: Step S01 also includes adaptive frequency band adjustment, which dynamically adjusts the weight of each frequency band component based on the equivalent thickness of the oxide layer identified in the previous measurement in step S03; when the equivalent thickness of the oxide layer is lower than the preset thin film threshold, the weight of the high frequency band component is increased to improve the spectral resolution of the thin oxide layer; when the equivalent thickness of the oxide layer exceeds the preset thick film threshold, the weight of the low frequency band component is increased to ensure the penetration depth and signal-to-noise ratio of deep polarization information.

Images