A multi-dimensional coupling-based fault automatic detection method for an electric energy metering box

Abstract

The application relates to the technical field of power equipment fault diagnosis, and discloses a kind of electric energy metering box fault automatic detection method based on multidimensional coupling;The method comprises the following steps: establishing the thermal network model of each monitoring point based on the steady-state electrical parameters and ambient temperature collected, and screening abnormal points based on the thermal network model;Calculate the equivalent resistance of each abnormal point, and perform overload fault detection on the abnormal points based on the equivalent resistance;For the abnormal points that are not overload faults, arc pulse detection is performed to determine whether it is a poor contact fault;If it is not a poor contact fault, then perform leakage path inversion, and determine whether it is a leakage heating fault according to the results of leakage path inversion;If it is not a leakage heating fault and the abnormal point is a switch terminal, then perform switch aging assessment to determine whether the corresponding abnormal point is a heating fault caused by switch contact point aging;The application realizes automatic detection of electric energy metering box faults, reduces the misjudgment rate and improves the operation and maintenance efficiency.

Description

Technical Field

[0001] This invention relates to the field of power equipment fault diagnosis technology, and specifically to an automatic fault detection method for power metering boxes based on multidimensional coupling. Background Technology

[0002] As electricity metering boxes integrate more and more communication, fee control, and other functional modules, their equipment structure and operating conditions become more complex, and fault types exhibit diverse and multi-dimensional coupling characteristics. In actual operation, electricity metering boxes commonly experience faults such as poor terminal contact, overload overheating, insulation deterioration and leakage, and mechanical aging of switches. These faults are interconnected and mutually exacerbate each other. Poor contact can cause localized overheating, accelerating the aging of switch contacts; leakage current passing through poor contact points further aggravates abnormal temperature rise; and overload conditions also increase the temperature rise of terminals and components, forming a multi-factor coupled fault evolution mechanism that can easily lead to continuous degradation of equipment performance and affect power supply reliability.

[0003] Current fault detection methods for metering boxes mostly rely on traditional approaches such as single temperature threshold alarms, conventional overcurrent protection, and leakage protection. These methods offer limited detection dimensions and analytical capabilities, failing to effectively differentiate between various fault types and making it difficult to identify subtle, latent faults such as low-level leakage or early-stage poor contact. Furthermore, traditional detection methods can only provide simple warnings for extreme faults, resulting in low fault identification accuracy, frequent false alarms, and the need for maintenance personnel to conduct comprehensive manual investigations after receiving alarms. This leads to time-consuming and labor-intensive fault location, low maintenance efficiency, and the potential for early-stage problems to escalate. Consequently, these methods fail to meet the actual needs of smart grids for accurate, efficient, and early-stage fault detection in electricity metering boxes.

[0004] The information disclosed in this background section is intended only to enhance the understanding of the overall background of this application and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the purpose of this invention is to provide an automatic fault detection method for electricity metering boxes based on multi-dimensional coupling, which solves the problems mentioned in the background technology to a certain extent, realizes automatic detection of faults in electricity metering boxes, reduces the false judgment rate and improves operation and maintenance efficiency.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides an automatic fault detection method for electricity metering boxes based on multidimensional coupling, comprising the following steps: Real-time acquisition of ambient temperature of the power metering box, steady-state electrical parameters of each monitoring point, and transient characteristic waveforms of each monitoring circuit and switch; A thermal network model for each monitoring point is established based on the steady-state electrical parameters and ambient temperature. Abnormal points are then screened based on the thermal network model, and the equivalent resistance of each abnormal point is calculated. Based on the equivalent resistance, overload fault detection is performed on the abnormal point. If it is not an overload fault, arc pulse detection is performed on the corresponding abnormal point to determine whether the abnormal point is a poor contact fault. If it is not a poor contact fault, then the leakage path is inverted based on the transient characteristic waveform of the corresponding monitoring circuit, and the leakage path inversion result is used to determine whether it is a leakage heating fault. If the fault is not due to leakage current and the abnormal point is a switch terminal, then the switch aging assessment is performed based on the transient characteristic waveform of the switch to determine whether the corresponding abnormal point is a heat fault caused by the aging of the switch contacts.

[0007] As a preferred embodiment of the automatic fault detection method for power metering boxes based on multidimensional coupling described in this application, the specific method for screening abnormal points based on the thermal network model is as follows: Calculate the initial value of the conductor resistance at each monitoring point at the reference temperature, and combine the effective value of the current with the ambient temperature to calculate the estimated value of the conductor temperature at the corresponding monitoring point; The initial value of the conductor resistance is corrected based on the temperature coefficient of resistance of the conductor material to obtain the theoretical conductor resistance at the current temperature at the corresponding monitoring point, and the heating power at the corresponding monitoring point is calculated. The heat generation power is used as the input excitation of the thermal network model, and the ambient temperature is used as the external boundary condition of the thermal network model. The differential equation of the thermal network model for the corresponding monitoring point is established, and the real-time predicted temperature is calculated iteratively with a set time step as the real-time temperature of the corresponding monitoring point. The temperature rise rate of the monitoring points is calculated based on the thermal network model. When the real-time temperature of any monitoring point is greater than the preset safety threshold, or when the temperature rise rate of any monitoring point is greater than the temperature rise rate threshold for a preset duration, the corresponding monitoring point is determined to be an abnormal point.

[0008] As a preferred embodiment of the automatic fault detection method for an energy metering box based on multidimensional coupling described in this application, the equivalent resistance of the abnormal point is compared with the reference resistance of the corresponding abnormal point; if the equivalent resistance is less than or equal to the product of the reference resistance and a preset first coefficient, and the effective value of the current is greater than the product of the rated current and a preset second coefficient, then it is determined to be an overload fault. If the equivalent resistance is greater than or equal to the product of the reference resistance and the preset third coefficient, then the transient characteristic waveform of the corresponding monitoring circuit is subjected to arc pulse detection. If the equivalent resistance is greater than the product of the reference resistance and the first coefficient, but less than the product of the reference resistance and the third coefficient, then leakage path inversion is performed based on the transient characteristic waveform of the corresponding monitoring circuit.

[0009] As a preferred embodiment of the automatic fault detection method for power metering boxes based on multidimensional coupling described in this application, the specific method of arc pulse detection is as follows: The time of occurrence of thermal anomaly events at abnormal locations is obtained, and the high-frequency current instantaneous waveform within a preset time window before and after the occurrence of thermal anomaly events is extracted as the signal to be detected. In the preprocessed signal to be detected, it is detected whether there is a pulse signal that meets the preset pulse characteristic conditions; if there is, it is determined whether the pulse signal is an arc pulse based on the pulse repetition frequency of the pulse signal.

[0010] As a preferred embodiment of the automatic fault detection method for power metering boxes based on multidimensional coupling described in this application, if the abnormal point has a real-time temperature greater than a preset safety threshold, then the moment when the real-time temperature of the corresponding abnormal point first exceeds the safety threshold is taken as the moment when the corresponding thermal anomaly event occurs. If the abnormal point is a temperature rise rate that is greater than the temperature rise rate threshold for a preset duration, then the end time of the duration is taken as the time when the thermal anomaly event occurs. The pulse characteristic conditions include rising edge steepness condition, pulse width condition, and spectral centroid condition; For a pulse signal that meets all pulse characteristic conditions, if the pulse repetition frequency is greater than or equal to the preset repetition frequency threshold, then the corresponding pulse signal is determined to be an electric arc pulse.

[0011] As a preferred embodiment of the automatic fault detection method for an energy metering box based on multidimensional coupling described in this application, the method involves: counting the total number of arc pulses detected within the time window; if the total number of arc pulses is greater than a preset pulse number threshold, it is determined to be a dynamic contact failure fault. Otherwise, it is judged as a static high-resistance poor contact fault.

[0012] As a preferred embodiment of the automatic fault detection method for energy metering boxes based on multidimensional coupling described in this application, the specific method of leakage path inversion is as follows: The leakage current contribution of each monitoring circuit is calculated based on the transient characteristic waveform; the monitoring circuit with the largest leakage current contribution is identified as the abnormal circuit, and the abnormal circuit is associated with the abnormal point. If the location association is successful, it is determined whether the abnormal circuit meets the leakage heating condition. If the leakage heating condition is met, the abnormal circuit is determined to be a leakage heating fault caused by insulation deterioration, and the product of the leakage contribution and the effective value of the total residual current waveform is used as the leakage current output. If the leakage current heating condition is not met or the location association fails, it is determined that the leakage current is not directly related to the current thermal anomaly.

[0013] As a preferred embodiment of the automatic fault detection method for energy metering boxes based on multidimensional coupling described in this application, the leakage current contribution of each monitoring circuit is calculated as follows: The effective value and vector angle of the total residual current waveform at the moment of thermal anomaly are obtained to obtain the total residual current phasor. The effective value and vector angle of the self-generated zero-sequence current waveform of each monitoring loop are also obtained to obtain the self-generated zero-sequence current phasor of each monitoring loop. Using the self-generated zero-sequence current phasor of each monitoring circuit as the basis vector and the total residual current phasor as the target vector, the leakage current contribution of each monitoring circuit is solved by the non-negative least squares method. Calculate the cumulative sum vector of the products of the leakage current contribution of all monitoring loops and the self-generated zero-sequence current phasor of the corresponding monitoring loop, and calculate the sum of the cumulative sum vector and the fitting residual. The leakage current contribution of each monitoring loop is greater than or equal to 0, and the sum of the accumulated sum vector and the fitting residual is equal to the total residual current phasor.

[0014] As a preferred embodiment of the automatic fault detection method for power metering boxes based on multidimensional coupling described in this application, the location association method is as follows: determine whether the wiring terminal of the outgoing circuit breaker in the abnormal circuit contains the abnormal point; if it contains it, the location association is successful; if it does not contain it, the location association fails. The leakage heating condition is as follows: the leakage contribution of the abnormal circuit is greater than or equal to the preset leakage contribution threshold, and the effective value of the total residual current waveform is greater than or equal to the preset leakage current threshold.

[0015] As a preferred embodiment of the automatic fault detection method for power metering boxes based on multidimensional coupling described in this application, if the leakage path inversion determines that the leakage is not directly related to the current thermal anomaly, and the anomaly point is the wiring terminal of the incoming main switch or the wiring terminal of the outgoing circuit breaker, then a switch aging assessment is performed. The cumulative value of the coil current waveform collected during the most recent M opening or closing actions of the corresponding switch and the number of actions of the corresponding switch; Feature parameters are extracted from the coil current waveform of each opening or closing action, and the feature parameters are combined into a one-dimensional vector as the feature vector of the corresponding opening or closing action. Calculate the Euclidean distance between the health template feature vector of the corresponding switch at the time of manufacture and the feature vector of each opening or closing action, and use it as the aging index of the corresponding opening or closing action. Calculate the arithmetic mean of the aging index corresponding to M opening or closing actions, and calculate the mean and standard deviation of the aging index corresponding to the previous L opening or closing actions of the corresponding switch; M and L are both positive integers.

[0016] As a preferred embodiment of the automatic fault detection method for power metering boxes based on multidimensional coupling described in this application, the aging trend slope is calculated based on the cumulative value of the corresponding number of actions and the aging index corresponding to M opening or closing actions. If the arithmetic mean of the aging index corresponding to the M opening or closing actions is greater than the sum of the mean of the aging index corresponding to the previous L opening or closing actions of the corresponding switch and 3 times the standard deviation, or if the aging trend slope is greater than or equal to the preset slope threshold, then the corresponding abnormal point is determined to be a heat-generating fault caused by the aging of the switch contacts. Otherwise, the corresponding abnormal point switch is determined to be normal.

[0017] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: By synchronously acquiring ambient temperature, steady-state electrical parameters, and transient characteristic waveforms from multiple dimensions, a complete data foundation is provided for subsequent thermal network modeling, fault diagnosis, and aging assessment, enabling comprehensive perception of the operating status of the power metering box. By establishing an independent thermal network model for each monitoring point, the temperature and temperature rise rate of each monitoring point are calculated in real time using the effective current value, ambient temperature, and equipment parameters. This achieves full-coverage temperature monitoring of all key terminals inside the power metering box, eliminating the need to install temperature sensors at each point and significantly reducing hardware costs. Furthermore, by using measured active power to deduce the equivalent resistance, the influence of theoretical parameter errors on fault determination is eliminated. By comparing the equivalent resistance with the reference resistance, accurate differentiation between overload faults and poor contact faults is achieved, solving the problem that traditional single temperature thresholds cannot distinguish between faults. The system addresses various fault types and further differentiates between dynamic and static high-resistance contact failures through arc pulse detection, enhancing the precision of fault diagnosis. By decomposing the total residual current and the self-generated zero-sequence current of each branch into phasor values, macroscopic leakage current is broken down to specific branches, enabling the location and tracing of minute leakage currents below the leakage protection threshold. Furthermore, the system correlates leakage current contribution with thermal anomalies at corresponding terminals, facilitating automatic reasoning of the causal chain from branch leakage to localized heating, thus identifying potential early insulation degradation. By analyzing the coil current waveform characteristics during switch opening and closing, and extracting peak values, holding time, and other characteristic parameters to calculate the aging index and aging trend slope, the system achieves online assessment and trend prediction of switch contact wear, providing early warning of mechanical aging faults and preventing the loss of cost control functions due to switch failure. Attached Figure Description

[0018] Figure 1 A flowchart of an automatic fault detection method for an energy metering box based on multidimensional coupling provided in this application; Figure 2 A flowchart for fault diagnosis provided in this application. Detailed Implementation

[0019] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof.

[0020] like Figure 1 As shown in the figure, this embodiment introduces an automatic fault detection method for electricity metering boxes based on multi-dimensional coupling, including the following steps: The ambient temperature of the power metering box, the steady-state electrical parameters of each monitoring point, and the transient characteristic waveforms of each monitoring circuit and switch are collected in real time. The ambient temperature is collected by installing an ambient temperature sensor in a location that is not directly affected by the heat radiation of the internal heating elements, such as the side wall of the incoming line compartment or the bottom of the metering compartment of the power metering box, and collecting the ambient temperature through the ambient temperature sensor.

[0021] In this embodiment, the ambient temperature sensor is a PT100 platinum resistance temperature sensor or a thermocouple-type temperature sensor. The sensing element of the ambient temperature sensor senses the air temperature inside the box and converts the temperature signal into a resistance value or a millivolt-level voltage signal. After amplification and analog-to-digital conversion by the signal conditioning circuit, the ambient temperature is obtained.

[0022] The steady-state electrical parameters include the effective value of current, the effective value of voltage, and active power. The acquisition method is as follows: Current transformers or current sensors are installed at the incoming and outgoing terminals of the main switch in the power metering box, the incoming and outgoing terminals of each outgoing circuit breaker, and at each phase connection point of the busbar. The primary current at each monitoring point is converted into a secondary current signal by the current transformers or current sensors. The secondary current signal is filtered and amplified by a signal conditioning circuit and then sent to an analog-to-digital converter (ADC). The ADC synchronously samples the instantaneous current values ​​of each phase and calculates the effective current value of each phase at each monitoring point based on the sampled instantaneous current values. Voltage sampling circuits or voltage sensors are installed between the incoming and outgoing terminals of the main switch and between the incoming and outgoing terminals of each outgoing circuit breaker. The device converts the phase voltage or line voltage of each monitoring point into a low-voltage analog signal. The low-voltage analog signal is filtered and amplified by the signal conditioning circuit and then sent to the analog-to-digital converter. The digital-to-analog converter synchronously samples the instantaneous voltage values ​​of each phase and calculates the effective voltage value of each phase at each monitoring point based on the sampled instantaneous voltage values. The instantaneous current value and the instantaneous voltage value sampled at the same monitoring point at the same time are multiplied, and the average value of the multiplication result is calculated within one cycle as the active power of the corresponding monitoring point. For a three-phase monitoring circuit, each phase is independently treated as a monitoring point, that is, the phase voltage and phase current of phases A, B, and C are collected respectively, and the effective current value, effective voltage value, and active power of each phase are calculated respectively. For a single-phase monitoring circuit, the phase voltage and phase current of a single phase are collected, and the effective current value, effective voltage value, and active power of the single phase are calculated.

[0023] The transient characteristic waveforms of each monitoring circuit and switch are acquired as follows: a high-frequency current transformer or a high-frequency current sensor is installed at the incoming end of the outgoing circuit breaker of each monitoring circuit to acquire the instantaneous high-frequency current waveform of each monitoring circuit; a high-frequency current transformer is installed at the outgoing end of the main incoming switch of the power metering box to acquire the total residual current waveform; a zero-sequence current transformer is installed on the outgoing side of each monitoring circuit or the self-generated zero-sequence current waveform of each monitoring circuit is obtained by using the secondary current signal of the three-phase current transformer; a sampling resistor is connected in series or a Hall effect current sensor is installed in the control circuit of the main incoming switch and each outgoing circuit breaker to acquire the coil current waveform when each switch is opened or closed.

[0024] In this embodiment, the primary-side current signal of each monitoring circuit is converted into a secondary-side current signal by the high-frequency current transformer or high-frequency current sensor. The secondary-side current signal is filtered and amplified by the signal conditioning circuit and then sent to a high-speed analog-to-digital converter. The high-speed digital-to-analog converter synchronously samples the instantaneous current waveform of each monitoring circuit to obtain the high-frequency instantaneous current waveform of each monitoring circuit. The self-generated zero-sequence current signal of each monitoring circuit is directly acquired by the zero-sequence current transformer, or by vector summation after synchronous sampling of the three-phase current to synthesize the self-generated zero-sequence current waveform of each monitoring circuit. The high-frequency current installed at the output terminal of the main incoming line switch is used to... The current transformer collects the total residual current signal of the entire tank, which is then filtered and amplified by the signal conditioning circuit before being sent to a high-speed analog-to-digital converter (ADC). The ADC collects the total residual current waveform and simultaneously extracts the effective value and vector angle of the waveform. The coil current signal is converted into a voltage signal by a sampling resistor connected in series in the switch control circuit, or by a Hall effect current sensor. This signal is then filtered and amplified by the signal conditioning circuit before being sent to the ADC. The ADC collects the coil current waveforms of each switch during opening or closing and records the number of switch actions corresponding to the coil current waveforms.

[0025] A thermal network model is established for each monitoring point based on the steady-state electrical parameters, and the real-time temperature and temperature rise rate of each monitoring point are calculated based on the thermal network model. The thermal network model is established as follows: The topology of the thermal network model is determined, and the thermal characteristics of each monitoring point are equivalent to a parallel network composed of thermal resistance and thermal capacity. The thermal resistance represents the ability of the corresponding monitoring point to conduct heat to the surrounding environment, and the thermal capacity represents the ability of the corresponding monitoring point to store heat. The initial values ​​of thermal resistance and thermal capacity are preset according to the equipment parameters of the corresponding monitoring point. The equipment parameters include the conductor material, conductor cross-sectional area, terminal size, and insulation layer thickness corresponding to each monitoring point. The equipment parameters are obtained before the system is put into operation by consulting the technical manual, factory inspection report, and nameplate parameters provided by the equipment manufacturer of the power metering box, or by on-site measurement. These are conventional technical methods for those skilled in the art and will not be elaborated further here.

[0026] In this embodiment, the initial value of the conductor resistance at each monitoring point at the reference temperature is calculated using the conductor material and conductor cross-sectional area in the device parameters. ;in, Reference temperature The resistivity of the corresponding conductor material. The effective length of the conductive path corresponding to the monitoring point. The conductor cross-sectional area corresponding to the monitoring point; the reference temperature The temperature is set to 20°C, according to the relevant regulations in the temperature correction factor table provided by the International Electrotechnical Commission (IEC) standard, which corrects the measured value of conductor resistance to 20°C. The effective length of the conductive path refers to the equivalent path length of the current actually flowing along the current direction in the conductive part of the monitoring point. Specifically, for terminal block monitoring points, the effective length refers to the equivalent path length of the current flowing from the wire input end of the terminal block along the conductive body of the terminal block to its switch or busbar connection end; for busbar connection point monitoring points, the effective length refers to the equivalent path length of the current flowing longitudinally along the busbar in the busbar section corresponding to the busbar connection point; for switch terminal block monitoring points, the effective length includes the equivalent length of the current path from the input end to the output end in the internal conductive circuit of the switch.

[0027] The estimated conductor temperature at the corresponding monitoring point is calculated based on the ambient temperature and the effective value of the current. The initial value of the conductor resistance is then corrected by combining the temperature coefficient of resistance of the corresponding conductor material to obtain the theoretical conductor resistance at the corresponding monitoring point at the current temperature. The temperature coefficient of resistance is a physical quantity that characterizes the degree of change of conductor resistance with temperature. It is specified by the International Electrotechnical Commission standard IEC 60228 and the national standard GB / T 3956. Its physical meaning is: the relative change of conductor resistance when the temperature changes by 1°C, with the unit being 1 / °C.

[0028] The estimated conductor temperature is calculated as follows: based on the effective current value and the initial conductor resistance value, the initial heating power at the corresponding monitoring point is calculated, which is the product of the square of the effective current value and the initial conductor resistance value; the product of the initial heating power and the initial thermal resistance value at the corresponding monitoring point is calculated to obtain the temperature rise of the estimated conductor temperature at the corresponding monitoring point relative to the ambient temperature, and the temperature rise is added to the ambient temperature to obtain the initial value of the estimated conductor temperature; it should be noted that the estimated conductor temperature is subsequently updated based on the real-time temperature output by the thermal network model.

[0029] In this embodiment, the initial value of the thermal resistance is preset according to the geometric dimensions and heat dissipation conditions of the corresponding monitoring point. The method of obtaining the value is as follows: before the system is put into operation, the heat dissipation area, heat dissipation coefficient and heat conduction path length of the corresponding monitoring point are obtained by consulting the technical manual provided by the equipment manufacturer, and the corresponding thermal resistance value is calculated according to the definition of thermal resistance.

[0030] The specific method for correcting the initial value of the conductor resistance by combining the temperature coefficient of resistance of the corresponding conductor material is as follows: calculate the difference between the estimated value of the conductor temperature and the reference temperature, and multiply the difference by the temperature coefficient of resistance of the corresponding conductor material to obtain the relative rate of change of the conductor resistance at the corresponding monitoring point, that is, the percentage change of unit reference resistance with temperature; multiply the relative rate of change by the initial value of the conductor resistance at the reference temperature to obtain the amount of resistance change; add the amount of resistance change to the initial value of the conductor resistance to obtain the theoretical conductor resistance at the corresponding monitoring point at the current temperature.

[0031] Based on the effective value of the current and the theoretical conductor resistance of the corresponding monitoring point at the current temperature, the heating power of the corresponding monitoring point is calculated. That is, the product of the square of the effective current value and the theoretical conductor resistance at the corresponding monitoring point at the current temperature; the heating power is used as the input excitation of the thermal network model, and the ambient temperature is used as the input excitation. As the external boundary conditions of the aforementioned thermal network model, the differential equations of the thermal network model corresponding to the monitoring points are established as follows: ;in, To determine the heat capacity of the corresponding monitoring point, To provide real-time predicted temperatures for the corresponding monitoring points, For time, The thermal resistance at the corresponding monitoring point is used; the differential equation of the thermal network model is discretized and solved, and the real-time estimated temperature is iteratively calculated at a set time step, which is then used as the real-time temperature of the corresponding monitoring point. This represents the rate of temperature rise at the corresponding monitoring point.

[0032] In this embodiment, the time step setting must simultaneously meet the stability and accuracy requirements of numerical calculation. From the perspective of numerical stability, when using the backward Euler method to discretize and solve the differential equation of the thermal network model, in order to ensure the convergence of the recursive calculation, the time step must be less than or equal to the product of heat capacity and thermal resistance, i.e., the thermal time constant. From the perspective of accuracy requirements, in order to accurately capture the dynamic changes in the temperature rise rate, the time step should be much smaller than the thermal time constant, which is usually taken as one-tenth of the thermal time constant in engineering. Therefore, the time step of each monitoring point can be determined independently based on its own thermal time constant, and a larger value should be taken as much as possible to reduce the computational cost while meeting the stability and accuracy requirements.

[0033] It should be further explained that for monitoring points equipped with temperature sensors, the thermal network model continuously uses the measured temperature collected by the temperature sensors as the target value during operation, and uses the recursive least squares method to correct the heat capacity and thermal resistance online, so that the real-time predicted temperature continuously approaches the measured temperature.

[0034] When the real-time temperature of any monitoring point is greater than the preset safety threshold, or when the temperature rise rate of any monitoring point is greater than the temperature rise rate threshold for a preset duration, the monitoring point is identified as an abnormal point, and the equivalent resistance of the abnormal point is calculated; the equivalent resistance is the ratio of the active power of the corresponding monitoring point to the square of the effective value of the corresponding current.

[0035] In this embodiment, the safety threshold is the absolute upper limit of the real-time estimated temperature of each monitoring point. It is set according to the temperature rise limit specified in the relevant technical standards for electricity metering boxes, and the safety threshold is set as the sum of the highest ambient temperature and the temperature rise limit. The highest ambient temperature is the expected highest ambient temperature at the installation location of the electricity metering box, which is determined according to the usage conditions specified in GB / T20641 and Q / GDW 11008.

[0036] The temperature rise rate characterizes the drastic change in temperature at the monitoring point over time. The temperature rise rate threshold and duration are used to identify sudden thermal faults, such as instantaneous deterioration of poor contact or short-circuit arcing. The temperature rise rate of these faults is much higher than the slow temperature rise caused by normal load fluctuations or changes in ambient temperature. The temperature rise rate threshold is set as follows: during the initial operation of the system, the temperature rise rate of each monitoring point under normal operating conditions (i.e., no fault alarms and stable load) is collected, and the statistical distribution is calculated. The 95th percentile is used as the upper limit of the normal operating temperature rise rate, and the temperature rise rate threshold is set as the product of the upper limit of the normal operating temperature rise rate and a preset safety margin coefficient. The safety margin coefficient is set to a value of 2 to 3, depending on the type of monitoring point: for points with a small thermal time constant, such as busbar connection points, the normal temperature rise rate fluctuates more, and a larger safety margin coefficient of 3 is used to avoid false alarms; for points with a large thermal time constant, such as switch terminals, the normal temperature rise rate fluctuates less, and a smaller safety margin coefficient of 2 can be used to improve sensitivity.

[0037] The duration is used to filter out transient disturbances, such as brief temperature rise rate spikes caused by instantaneous switching of large loads, to avoid misjudging them as persistent thermal faults. Its specific value is determined based on the thermal time constant of the monitoring point: for points with a small thermal time constant, the temperature rise rate responds quickly to load changes, the disturbance duration is short, and the duration can be set to a shorter value, such as 10 seconds; for points with a large thermal time constant, the thermal inertia is large, the duration of transient disturbances is relatively long, and the duration can be extended accordingly, such as 30 seconds; in this embodiment, the duration is set to a uniform engineering value of 30 seconds, that is, a thermal anomaly event is only triggered when the temperature rise rate exceeds the temperature rise rate threshold for 30 consecutive seconds.

[0038] like Figure 2As shown, the equivalent resistance of the abnormal point is compared with the reference resistance of the corresponding abnormal point; if the equivalent resistance is less than or equal to the product of the reference resistance and the preset first coefficient, and the effective value of the current is greater than the product of the rated current and the preset second coefficient, then it is determined to be an overload fault; if the equivalent resistance is greater than or equal to the product of the reference resistance and the preset third coefficient, then the transient characteristic waveform of the corresponding monitoring circuit is subjected to arc pulse detection.

[0039] The reference resistance is the equivalent resistance reference value of the corresponding monitoring point under healthy conditions. It is obtained as follows: In the initial stage of system commissioning, the theoretical conductor resistance of the corresponding monitoring point is used as the initial value of the reference resistance. After the system has been running for 30 days and has accumulated enough normal operation data, the reference resistance is switched to a sliding statistical value based on the normal operation data. Specifically, the equivalent resistance of the corresponding monitoring point is calculated every 10 minutes every day as an equivalent resistance time series. The median value of the equivalent resistance time series is taken as the daily equivalent resistance. The arithmetic mean of the daily equivalent resistance during the period without abnormalities in the last 30 days is taken as the reference resistance of the corresponding monitoring point.

[0040] In this embodiment, the first coefficient, the second coefficient, and the third coefficient are preset fault judgment threshold coefficients, and their values ​​are set according to the allowable deviation range and engineering experience in the relevant technical standards of the power metering box. The first coefficient is the equivalent resistance overload judgment threshold coefficient, used to determine whether the current equivalent resistance is within the healthy range while the current is overloaded. Its value is 1.2, that is, when the equivalent resistance does not exceed 1.2 times the reference resistance and the current exceeds the product of the rated current and the second coefficient, it is judged as a pure overload fault. The second coefficient is the current overload judgment threshold coefficient, used to determine whether the current significantly exceeds the rated current. Its value is 1.1, that is, when the effective value of the current exceeds 1.1 times the rated current, it is considered to be in an overload state. This value is determined according to the lower limit of the overload protection action current specified in GB / T 7251.1 to ensure reliable identification of overload conditions when the rated current is 1.1 times or more. The third coefficient is the equivalent resistance poor contact judgment threshold coefficient, used to determine whether the equivalent resistance has increased significantly to the point that it is sufficient to judge poor contact. Its value is 2.0, that is, when the equivalent resistance reaches 2.0 times or more of the reference resistance, it is considered that there is a significant abnormality in the contact interface resistance, triggering the poor contact judgment process.

[0041] The specific method for arc pulse detection is as follows: If the abnormal point has a real-time temperature greater than a safety threshold, the moment when the real-time temperature of the abnormal point first exceeds the safety threshold is taken as the occurrence time of the corresponding thermal anomaly event; if the abnormal point has a temperature rise rate greater than a temperature rise rate threshold for a preset duration, the end time of that duration is taken as the occurrence time of the thermal anomaly event; the high-frequency current instantaneous waveform within a preset time window before and after the occurrence time of the thermal anomaly event is extracted as the signal to be detected, and the signal to be detected is preprocessed; in the preprocessed signal to be detected, it is detected whether there is a pulse signal that meets the preset pulse characteristic conditions; the pulse characteristic conditions are used to identify from the high-frequency current instantaneous waveform. The arc pulse generated by poor contact is distinguished from normal load current waveforms, transient pulses generated by switching operations, and other high-frequency signals such as electromagnetic interference. This distinction is based on three conditions: rise edge steepness, pulse width, and spectral centroid. The rise edge steepness condition is defined as the rise edge rate of the pulse signal being greater than a preset rise edge steepness threshold. The pulse width condition is defined as the duration of the pulse signal being within a preset pulse width range. The spectral centroid condition is defined as the spectral centroid of the pulse signal being greater than a preset spectral centroid threshold. If a pulse signal satisfying all pulse characteristic conditions exists in the preprocessed detection signal, then the pulse repetition frequency of the pulse signal is used to determine whether the pulse signal forms a pulse cluster.

[0042] Specifically, the preset time window is the time range for capturing the instantaneous waveform of high-frequency current for arc pulse detection. It is set as follows: taking the moment of the thermal anomaly as a reference point, extending forward by a first duration and backward by a second duration together constitutes the time window. The first duration is used to capture intermittent arc pulses that may have already existed before the thermal anomaly occurred, and the second duration is used to capture the continuation of the arc pulses after the thermal anomaly occurred. In this embodiment, the first duration is 10 seconds, and the second duration is 2 seconds. Arc pulses generated by poor contact at the wiring terminals typically exhibit intermittent characteristics, with a single arc discharge duration being extremely short, usually on the order of microseconds to hundreds of microseconds. However, arc activity may repeatedly occur in cycles of several seconds before the thermal anomaly occurs, extending forward by 10 seconds... A two-second window can cover the intermittent periods of multiple arc pulses under typical poor contact conditions, effectively capturing arc activity traces before thermal anomalies occur, establishing the time series characteristics of arc pulses, and thus confirming the causal relationship between arc activity and thermal anomaly events. After poor contact causes thermal expansion of materials or mechanical damage to the oxide layer due to heat generation, arc activity may persist or show an intensifying trend. Meanwhile, the tripping action time of the outgoing circuit breaker in the power metering box after detecting a fault is usually in the range of tens to hundreds of milliseconds. Extending the window by 2 seconds can not only completely record the transition process of arc pulses before and after the occurrence of thermal anomalies, verifying the temporal correlation between arc pulses and thermal anomalies, but also avoid introducing a large amount of invalid data due to an excessively long time window, thus balancing the computational resource consumption of data storage and real-time processing.

[0043] In this embodiment, the specific method for preprocessing the signal to be detected is as follows: a bandpass filter is used to filter out the power frequency and its integer multiple harmonic components, retaining the signal components in the frequency band from 1kHz to 100kHz, so as to remove the power frequency components and harmonic interference; the filtered signal to be detected is subjected to amplitude normalization processing to map the signal amplitude to a uniform range, so as to eliminate the influence of the difference in signal amplitude under different load currents on the detection.

[0044] The rise edge steepness threshold is used to distinguish between arc pulses, normal load current changes, and common electromagnetic interference signals. An arc is essentially a gas breakdown discharge phenomenon; the time it takes for the current to rise from zero to the arc current during the breakdown is extremely short. For low-pressure air gaps, the arc breakdown time is typically on the order of nanoseconds to microseconds, and the rise rate of the arc current can reach tens to hundreds of amperes per microsecond. Normal load current changes are limited by line impedance and load characteristics, and their rise rate is much lower than that during arc breakdown. Although transient pulses generated by switching operations have a relatively fast rise edge, their rise edge steepness is usually lower than that of contact pulses. The steepness of a good arc; although electromagnetic interference signals may have steep rising edges, their pulse width and spectral characteristics are significantly different from those of arc pulses; therefore, setting a rising edge steepness threshold can initially separate arc pulses from the aforementioned interference signals in the time domain; the rising edge steepness threshold is preset before the system is put into operation based on the statistical distribution of the rising edges of the arc waveform measured by typical terminals in the power metering box under simulated poor contact conditions in the laboratory, and the 10th percentile of the measured arc pulse rising edge steepness distribution is used as the rising edge steepness threshold to ensure that the vast majority of real arc pulses can be effectively detected.

[0045] The pulse width range is used to distinguish arc pulses from high-frequency noise with excessively short durations and load transient changes with excessively long durations. The duration of a single discharge of a low-pressure air gap arc is affected by factors such as air dielectric recovery time, electrode material, circuit parameters, and arc extinguishing conditions. According to typical measured data of low-pressure electrical contact arcs, the duration of a single arc pulse caused by poor contact is usually in the range of several microseconds to hundreds of microseconds. Pulses with excessively short durations, such as those in the sub-microsecond range, are usually caused by high-frequency oscillation noise in the circuit or sensor parasitic effects. Pulses with excessively long durations, such as those in the millisecond range or longer, may be normal transient changes in the load, such as motor starting current surges or surge currents in switching power supplies, rather than arc discharges. Therefore, setting the pulse width range can effectively separate arc pulses from the above two types of non-arc signals. The pulse width range is preset before the system is put into operation based on the statistical distribution of the measured arc pulse duration of typical terminals in the power metering box under simulated poor contact conditions in the laboratory. The 5th percentile of the measured arc pulse duration is used as the lower limit of the pulse width and the 95th percentile is used as the upper limit of the pulse width to ensure that the duration distribution of most real arc pulses is covered. In this embodiment, the pulse width range is 10μs to 100μs.

[0046] The spectral centroid threshold is used to distinguish arc pulses from the 50Hz power frequency and its integer multiples of harmonics. Arc pulses have wide bandwidth characteristics, with their energy distribution spanning a wide frequency range, typically extending from several kHz to several MHz. When the power frequency current crosses zero, the arc undergoes an extinction and reignition process. The current surge generated during this process excites high-frequency components, while the spectral energy of normal load current is mainly concentrated in the power frequency and low-order harmonic frequency bands, with very few high-frequency components. Therefore, setting a spectral centroid threshold can effectively separate the arc pulse from the power frequency components and harmonic interference. The spectral centroid threshold is preset before system commissioning based on the measured spectral centroid distribution of arc pulses under simulated poor contact conditions at typical terminals in the power metering box in the laboratory. The 5th percentile of the measured arc pulse spectral centroid distribution is used as the spectral centroid threshold, ensuring that this threshold is significantly higher than the power frequency and its main harmonics. In this embodiment, the spectral centroid threshold is 10kHz.

[0047] For a pulse signal that meets all pulse characteristic conditions, if the pulse repetition frequency is greater than or equal to a preset repetition frequency threshold, the pulse signal is determined to be an arc pulse. Arc discharge caused by poor contact of the terminal block usually manifests as multiple consecutive discharges rather than a single isolated event. When there is a microscopic gap at the contact interface, breakdown discharge may occur in each half-wave of the power frequency voltage, forming a pulse cluster that repeats at the power frequency or its multiples. A single isolated pulse may be a transient pulse generated by a switching operation or external electromagnetic interference. Only when multiple pulses that meet all pulse characteristic conditions appear continuously at a certain repetition frequency can a continuous arc discharge phenomenon be confirmed. The repetition frequency threshold is preset before the system is put into operation based on the statistical distribution of the repetition frequency of the arc pulse cluster measured by typical terminal blocks in the power metering box under simulated poor contact conditions in the laboratory. The 5th percentile of the measured arc pulse repetition frequency distribution is used as the repetition frequency threshold. In this embodiment, the preset repetition frequency threshold is 1 time / second.

[0048] The total number of arc pulses detected within the time window is counted. If the total number of arc pulses is greater than a preset pulse count threshold, it is determined to be a dynamic contact failure; otherwise, it is determined to be a static high-resistance contact failure. The pulse count threshold is set before the system is put into operation based on the statistical distribution of the total number of arc pulses measured within the time window under simulated contact failure conditions in the laboratory for typical terminals in the power metering box. The 5th percentile of the measured arc pulse count distribution is used as the pulse count threshold to ensure that the arc pulses generated by the vast majority of real contact failures can be effectively detected, while effectively excluding a small number of isolated transient pulses generated by non-arc interference such as load start-up and switching operation. In this embodiment, the pulse count threshold is preferably 5 times. That is, if the total number of arc pulses detected within the 12-second time window is greater than 5 times, it is determined to be a dynamic contact failure; if the total number of arc pulses is less than or equal to 5 times, it is determined to be a static high-resistance contact failure.

[0049] If the equivalent resistance is greater than the product of the reference resistance and the first coefficient, but less than the product of the reference resistance and the third coefficient, then leakage path inversion is performed based on the transient characteristic waveform of the corresponding monitoring circuit; the specific method of leakage path inversion is as follows: The effective value and vector angle of the total residual current waveform at the moment of thermal anomaly are obtained to obtain the total residual current phasor. The effective value and vector angle of the self-generated zero-sequence current waveform of each monitoring loop are also obtained to obtain the self-generated zero-sequence current phasor of each monitoring loop. Using the self-generated zero-sequence current phasor of each monitoring loop as the basis vector and the total residual current phasor as the target vector, the leakage current contribution of each monitoring loop is solved using the non-negative least squares method. The cumulative sum vector of the products of the leakage current contribution of all monitoring loops and the self-generated zero-sequence current phasor of the corresponding monitoring loop is calculated, and the sum of the cumulative sum vector and the fitting residual is calculated. The leakage current contribution satisfies the following conditions: the leakage current contribution of each monitoring loop is greater than or equal to 0, and the sum of the cumulative sum vector and the fitting residual is equal to the total residual current phasor.

[0050] The monitoring circuit with the largest leakage current contribution is designated as the abnormal circuit, and the abnormal circuit is associated with the abnormal point. If the location association is successful, it is further determined whether the abnormal circuit meets the leakage current heating condition. If the abnormal circuit meets the leakage current heating condition, it is determined that the abnormal circuit is a leakage current heating fault caused by insulation degradation, and the leakage current is output as approximately the product of the leakage current contribution and the effective value of the total residual current waveform. If the leakage current heating condition is not met or the location association fails, it is determined that the leakage current is not directly related to the current thermal anomaly.

[0051] Specifically, the location association method is as follows: determine whether the outgoing circuit breaker terminal in the abnormal circuit contains the abnormal point; if it does, the location association is successful; if it does not, the location association fails.

[0052] The leakage heating condition is as follows: the leakage contribution of the abnormal circuit is greater than or equal to the preset leakage contribution threshold, and the effective value of the total residual current waveform is greater than or equal to the preset leakage current threshold.

[0053] When the leakage current contribution of a certain monitoring circuit is small, it indicates that the proportion of the monitoring circuit in the total residual current is low. Even if the monitoring circuit has leakage, its leakage current value is low, and the additional Joule heat generated is limited, which is insufficient to be the main cause of thermal anomalies. Only when the leakage current contribution of a certain monitoring circuit reaches a certain proportion can the leakage current of the monitoring circuit generate enough heat to cause an abnormal temperature rise in the contact resistance or line resistance. The leakage current contribution threshold is set before the system is put into operation based on the measured data of typical terminals in the power metering box under simulated leakage heating conditions in the laboratory. The minimum leakage current contribution that can cause the terminal temperature to rise is used as the leakage current contribution threshold. In this embodiment, the leakage current contribution threshold is preferably 0.3. That is, when the leakage current contribution of the abnormal circuit is greater than or equal to 0.3, the proportion of the leakage current of the monitoring circuit to the total residual current is considered sufficient to be a potential cause of thermal anomalies.

[0054] During normal operation, a small leakage current to ground is unavoidable in the electricity metering box. This leakage current usually originates from the distributed capacitance of the line, surface leakage of the insulation material, etc., and its value is usually below a few milliamperes, which will not cause significant heating and temperature rise. Only when the leakage current amplitude exceeds a certain level can Joule heating sufficient to cause abnormal temperature rise be generated on the contact resistance or line resistance. In addition, the rated operating current of the residual current device (RCD) installed in the electricity metering box is usually 30mA. When the leakage current is greater than this value, it will trigger the protection to trip, and the system can be known by the operation and maintenance personnel without performing leakage path inversion. Therefore, the setting of the leakage current threshold must simultaneously meet two constraints: higher than the background leakage current level during normal operation, and lower than the rated operating current of the residual current device, so as to achieve effective early warning in the range where the leakage current has formed a heat source but the leakage protection has not yet tripped. In this embodiment, the leakage current threshold is 10mA, that is, when the effective value of the total residual current waveform is greater than or equal to 10mA, it is considered that there is a leakage level that requires early warning.

[0055] If the leakage path inversion determines that the leakage is not directly related to the current thermal anomaly, then a switch aging assessment is performed. The triggering condition for the switch aging assessment is that the abnormal point where the current thermal anomaly event occurs is a terminal of the incoming main switch or an outgoing circuit breaker, and this abnormal point has not yet been identified as having any fault. The switch aging assessment method is as follows: When the switch aging assessment is triggered, the coil current waveforms collected and stored during the most recent M opening or closing actions of the corresponding switch are retrieved, along with the cumulative number of actions of the switch. Feature parameters are extracted from the coil current waveforms of each opening or closing action, including the peak value, the first holding time from the peak value down to 90% of the peak value, and the second holding time from the peak value down to 10% of the peak value. These feature parameters are combined into a one-dimensional vector as the feature vector of the corresponding opening or closing action. The Euclidean distance between the health template feature vector stored at the time of the switch's manufacture and the feature vector of the corresponding opening or closing action is calculated as the aging index of the corresponding opening or closing action. The arithmetic mean of the aging indices corresponding to the M opening or closing actions is calculated, and the mean and standard deviation of the aging indices corresponding to the previous L opening or closing actions of the corresponding switch are calculated. The mean and standard deviation are calculated after the system has run a full L opening or closing actions and are continuously updated. The aging trend slope is calculated based on the cumulative number of actions, i.e., the linear regression slope of the aging index corresponding to the M opening or closing actions on N.

[0056] Where M is a positive integer, its setting must simultaneously meet the requirements of statistical stability and real-time performance. From the perspective of statistical stability, the single calculation result of the aging index is affected by factors such as random noise of coil current waveform sampling, voltage fluctuation during switch operation, and random friction of operating mechanism, resulting in certain random fluctuations. The aging index of a single operation is difficult to reliably reflect the true wear state of the switch contacts. Taking the arithmetic mean of the most recent operations can effectively smooth out random fluctuations, making the aging index closer to the true state of the switch contacts. In this embodiment, the arithmetic mean of the most recent 10 operations is taken as the characterization of the degree of aging. Statistically, it can meet the stability requirements of simple statistics, while avoiding feature lag due to excessively large sample size. From the perspective of real-time performance, if M is too small, the statistical fluctuations will be large and the reliability will be insufficient. If M is too large, the update cycle of the average value will be too long, and it will be unable to reflect the rapid deterioration trend of the aging degree of the switch in a timely manner. In particular, when the aging enters the accelerated stage, an excessively large M value will lead to a delay in the diagnostic response. Therefore, after comprehensively considering statistical stability and real-time response speed, this embodiment sets M to 10.

[0057] The health template feature vector is obtained by consulting the technical manual provided by the equipment manufacturer of the switch before the system is put into operation, or by measuring the coil current waveform of the switch during three consecutive opening or closing operations after the initial installation and taking the arithmetic mean of each characteristic parameter as the health template feature vector.

[0058] L is a positive integer, and its setting must meet the requirements of statistical robustness and engineering feasibility. From the perspective of statistical robustness, the aging index has certain random fluctuations during the healthy operation of the switch due to factors such as measurement noise, load fluctuations, and changes in ambient temperature. The larger the sample size used to calculate the mean and standard deviation, the more accurately the statistical parameters obtained can reflect the aging index distribution characteristics of the switch in a healthy state. However, an excessively large sample size will cause the system to wait longer before it can start aging judgment, affecting the availability of diagnostic functions in the early stage of commissioning. Moreover, the switch may have already undergone slow aging during long-term operation, and an excessively large sample size will mix the aged data into the healthy baseline, causing baseline shift and reducing the sensitivity of subsequent aging judgment. From the engineering practice experience of statistics, a sample size greater than or equal to 30 can meet the requirement of the central limit theorem that the sample mean distribution approaches a normal distribution, providing a statistical basis for the subsequent use of the 3σ criterion. At the same time, considering the actual operating frequency of the switch in the power metering box, setting the value of L to 100 can complete the baseline establishment within 1 to 3 months after the system is put into operation, taking into account both the timely availability of diagnostic functions and statistical robustness. Therefore, in this embodiment, L is set to 100 times.

[0059] If the arithmetic mean of the aging index corresponding to the M opening or closing actions is greater than the sum of the mean of the aging index corresponding to the previous L opening or closing actions of the corresponding switch and 3 times the standard deviation, or if the slope of the aging trend is greater than or equal to a preset slope threshold, then the corresponding abnormal point is determined to be a heat fault caused by the aging of the switch contacts, i.e., a switch aging fault, and a warning message is output that the switch is nearing the end of its life and a planned replacement is recommended; otherwise, the corresponding abnormal point switch is determined to be normal, the current thermal abnormality cannot be attributed to switch aging, and a warning message is output that the thermal abnormality is of unknown cause and an on-site inspection is recommended.

[0060] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0061] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of protection of this application, and these forms are all within the protection scope of this application.

Claims

1. An automatic fault detection method for electricity metering boxes based on multidimensional coupling, characterized in that, Includes the following steps: Real-time acquisition of ambient temperature of the power metering box, steady-state electrical parameters of each monitoring point, and transient characteristic waveforms of each monitoring circuit and switch; A thermal network model for each monitoring point is established based on the steady-state electrical parameters and ambient temperature. Abnormal points are then screened based on the thermal network model, and the equivalent resistance of each abnormal point is calculated. Based on the equivalent resistance, overload fault detection is performed on the abnormal point. If it is not an overload fault, arc pulse detection is performed on the corresponding abnormal point to determine whether the abnormal point is a poor contact fault. If it is not a poor contact fault, then the leakage path is inverted based on the transient characteristic waveform of the corresponding monitoring circuit, and the leakage path inversion result is used to determine whether it is a leakage heating fault. If the fault is not due to leakage current and the abnormal point is a switch terminal, then the switch aging assessment is performed based on the transient characteristic waveform of the switch to determine whether the corresponding abnormal point is a heat fault caused by the aging of the switch contacts. The specific method for screening outlier locations based on the aforementioned thermal network model is as follows: Calculate the initial value of the conductor resistance at each monitoring point at the reference temperature, and combine the effective value of the current with the ambient temperature to calculate the estimated value of the conductor temperature at the corresponding monitoring point; The initial value of the conductor resistance is corrected based on the temperature coefficient of resistance of the conductor material to obtain the theoretical conductor resistance at the current temperature at the corresponding monitoring point, and the heating power at the corresponding monitoring point is calculated. The heat generation power is used as the input excitation of the thermal network model, and the ambient temperature is used as the external boundary condition of the thermal network model. The differential equation of the thermal network model for the corresponding monitoring point is established, and the real-time predicted temperature is calculated iteratively with a set time step as the real-time temperature of the corresponding monitoring point. The temperature rise rate of the monitoring points is calculated based on the thermal network model. When the real-time temperature of any monitoring point is greater than the preset safety threshold, or when the temperature rise rate of any monitoring point is greater than the temperature rise rate threshold for a preset duration, the corresponding monitoring point is determined to be an abnormal point.

2. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 1, characterized in that, The equivalent resistance of the abnormal point is compared with the reference resistance of the corresponding abnormal point. If the equivalent resistance is less than or equal to the product of the reference resistance and the preset first coefficient, and the effective value of the current is greater than the product of the rated current and the preset second coefficient, then it is determined to be an overload fault. If the equivalent resistance is greater than or equal to the product of the reference resistance and the preset third coefficient, then the transient characteristic waveform of the corresponding monitoring circuit is subjected to arc pulse detection. If the equivalent resistance is greater than the product of the reference resistance and the first coefficient, but less than the product of the reference resistance and the third coefficient, then leakage path inversion is performed based on the transient characteristic waveform of the corresponding monitoring circuit.

3. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 2, characterized in that, The specific method for arc pulse detection is as follows: The time of occurrence of thermal anomaly events at abnormal locations is obtained, and the high-frequency current instantaneous waveform within a preset time window before and after the occurrence of thermal anomaly events is extracted as the signal to be detected. The presence of a pulse signal that meets the preset pulse characteristic conditions is detected in the preprocessed signal to be detected. If present, determine whether the pulse signal is an arc pulse based on the pulse repetition frequency of the pulse signal.

4. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 3, characterized in that, If the abnormal point is where the real-time temperature is greater than a preset safety threshold, then the moment when the real-time temperature of the corresponding abnormal point first exceeds the safety threshold shall be taken as the moment when the corresponding thermal anomaly event occurs. If the abnormal point is a temperature rise rate that is greater than the temperature rise rate threshold for a preset duration, then the end time of the duration is taken as the time when the thermal anomaly event occurs. The pulse characteristic conditions include rising edge steepness condition, pulse width condition, and spectral centroid condition; For a pulse signal that meets all pulse characteristic conditions, if the pulse repetition frequency is greater than or equal to the preset repetition frequency threshold, then the corresponding pulse signal is determined to be an electric arc pulse.

5. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 4, characterized in that, The total number of arc pulses detected within the time window is counted. If the total number of arc pulses is greater than the preset pulse number threshold, it is determined to be a dynamic contact failure. Otherwise, it is judged as a static high-resistance poor contact fault.

6. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 5, characterized in that, The specific method for inverting the leakage path is as follows: The leakage current contribution of each monitoring circuit is calculated based on the transient characteristic waveform; the monitoring circuit with the largest leakage current contribution is identified as the abnormal circuit, and the abnormal circuit is associated with the abnormal point. If the location association is successful, it is determined whether the abnormal circuit meets the leakage heating condition. If the leakage heating condition is met, the abnormal circuit is determined to be a leakage heating fault caused by insulation deterioration, and the product of the leakage contribution and the effective value of the total residual current waveform is used as the leakage current output. If the leakage current heating condition is not met or the location association fails, it is determined that the leakage current is not directly related to the current thermal anomaly.

7. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 6, characterized in that, The leakage current contribution of each monitoring circuit is calculated as follows: The effective value and vector angle of the total residual current waveform at the moment of thermal anomaly are obtained to obtain the total residual current phasor. The effective value and vector angle of the self-generated zero-sequence current waveform of each monitoring loop are also obtained to obtain the self-generated zero-sequence current phasor of each monitoring loop. Using the self-generated zero-sequence current phasor of each monitoring circuit as the basis vector and the total residual current phasor as the target vector, the leakage current contribution of each monitoring circuit is solved by the non-negative least squares method. Calculate the cumulative sum vector of the products of the leakage current contribution of all monitoring loops and the self-generated zero-sequence current phasor of the corresponding monitoring loop, and calculate the sum of the cumulative sum vector and the fitting residual. The leakage current contribution is such that: the leakage current contribution of each monitoring circuit is greater than or equal to 0, and the sum of the accumulated sum vector and the fitting residual is equal to the total residual current phasor.

8. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 7, characterized in that, The location association method is as follows: determine whether the outgoing circuit breaker terminal in the abnormal circuit contains the abnormal point; if it does, the location association is successful; if it does not, the location association fails. The leakage heating condition is as follows: the leakage contribution of the abnormal circuit is greater than or equal to the preset leakage contribution threshold, and the effective value of the total residual current waveform is greater than or equal to the preset leakage current threshold.

9. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 6, characterized in that, If the leakage path inversion determines that the leakage is not directly related to the current thermal anomaly, and the anomaly point is the wiring terminal of the incoming main switch or the wiring terminal of the outgoing circuit breaker, then switch aging assessment shall be performed. The cumulative value of the coil current waveform collected during the most recent M opening or closing actions of the corresponding switch and the number of actions of the corresponding switch; Feature parameters are extracted from the coil current waveform of each opening or closing action, and the feature parameters are combined into a one-dimensional vector as the feature vector of the corresponding opening or closing action. Calculate the Euclidean distance between the health template feature vector of the corresponding switch at the time of manufacture and the feature vector of each opening or closing action, and use it as the aging index of the corresponding opening or closing action. Calculate the arithmetic mean of the aging index corresponding to M opening or closing actions, and calculate the mean and standard deviation of the aging index corresponding to the previous L opening or closing actions of the corresponding switch. M and L are both positive integers.

10. The automatic fault detection method for an energy metering box based on multidimensional coupling as described in claim 9, characterized in that, The aging trend slope is calculated based on the cumulative value of the corresponding number of actions and the aging index corresponding to M opening or closing actions. If the arithmetic mean of the aging index corresponding to the M opening or closing actions is greater than the sum of the mean of the aging index corresponding to the previous L opening or closing actions of the corresponding switch and 3 times the standard deviation, or if the aging trend slope is greater than or equal to the preset slope threshold, then the corresponding abnormal point is determined to be a heat-generating fault caused by the aging of the switch contacts. Otherwise, the corresponding abnormal point switch is determined to be normal.

Images