Double-fracture high-voltage vacuum circuit breaker fault diagnosis method and system, medium and product

Abstract

A double-fracture high-voltage vacuum circuit breaker fault diagnosis method and system, a medium and a product relate to the field of electrical variable measurement, and the method comprises the following steps: obtaining contact action time sequence data of two fractures, pressure data in a vacuum chamber, environment temperature, environment humidity and a system load index; calculating a mechanical characteristic cooperation vector; calculating a heating imbalance coefficient of the two fractures; collecting voltage sharing ratio and transition recovery voltage characteristics in the breaking process, and determining a voltage endurance capability difference index; inputting the mechanical characteristic cooperation vector and the voltage endurance capability difference index to a fault diagnosis model to obtain a fault diagnosis result; calculating an environmental influence factor of the high-voltage vacuum circuit breaker under the current working condition, generating a working condition feature vector, and determining a fault early warning threshold value; and when the early warning parameter index of the fault diagnosis result exceeds a fault early warning threshold, generating fault early warning information. By implementing the application, the fault of the double-break high-voltage vacuum circuit breaker can be found in time.

Description

Technical Field

[0001] The present application relates to the field of measuring electrical variables, and in particular to a double-break high-voltage vacuum circuit breaker fault diagnosis method, system, medium and product. Background Art

[0002] High-voltage vacuum circuit breakers are important switchgear used to control and protect circuits in power systems. Their main functions are to switch on and off load currents during normal operation and to quickly cut off fault currents in the event of a fault. Compared with the single-break structure, the double-break high-voltage vacuum circuit breaker has two vacuum interrupters in series, each of which bears half of the recovery voltage. It has higher breaking capacity and stronger insulation performance, and is particularly suitable for power transmission and distribution systems with voltage levels of 110kV and above.

[0003] In related technologies, fault diagnosis of double-break high-voltage vacuum circuit breakers mainly relies on regular maintenance and simple online monitoring. Maintenance personnel usually use mechanical property testers to measure parameters such as the circuit breaker's opening and closing time and travel characteristics, and use infrared thermometers to detect contact temperature distribution. The online monitoring system mainly collects basic electrical parameters of the circuit breaker, such as operating current waveforms, switch position signals, etc., and sets fixed thresholds for fault alarms.

[0004] However, due to the different environments in which the equipment is located and the different probabilities of failure, the threshold alarm mechanism in related technologies is often difficult to perceive the corresponding environment. There is a delay in the identification of faults in high-risk environments. Generally, early warnings are only issued after serious problems occur in the equipment, which often leads to increased maintenance costs. Summary of the invention

[0005] The present application provides a double-break high-voltage vacuum circuit breaker fault diagnosis method, system, medium and product, which are used to timely detect the fault of the double-break high-voltage vacuum circuit breaker.

[0006] In the first aspect, the present application provides a double-break high-voltage vacuum circuit breaker fault diagnosis method, which is applied to a fault detection system. The method includes: obtaining the contact action timing data of the high-voltage vacuum circuit breaker at two breaks, the pressure data in the vacuum chamber, and the ambient temperature, ambient humidity and system load index; calculating the contact action speed difference and bounce characteristic parameters of the two breaks according to the contact action timing data, and generating a mechanical characteristic coordination vector in combination with the opening and closing energy consumption data of the two breaks; obtaining the temperature distribution data of the conductive circuit of the two breaks, and calculating the heat imbalance coefficient of the two breaks based on the temperature distribution data; collecting The voltage sharing ratio and transition recovery voltage characteristics of the two breakpoints during the breaking process are combined with the pressure data in the vacuum chamber and the heat imbalance coefficient to determine the voltage tolerance difference index; the mechanical characteristic coordination vector and the voltage tolerance difference index are input into the fault diagnosis model to obtain the fault diagnosis result; based on the ambient temperature, ambient humidity and system load index, the environmental impact factor of the high-voltage vacuum circuit breaker under the current working condition is calculated, the working condition characteristic vector is generated, and the fault warning threshold of the corresponding working condition characteristic vector is determined; when the warning parameter index of the fault diagnosis result exceeds the fault warning threshold, a fault warning information is generated.

[0007] In the above embodiment, the fault detection system obtains the contact action timing, pressure and other data of the two breaks and combines them with environmental factors to calculate the mechanical characteristics and voltage tolerance difference indicators, and then determines the fault warning threshold based on environmental influences, thereby achieving accurate diagnosis and timely warning of double-break high-voltage vacuum circuit breaker faults and reducing maintenance costs.

[0008] In combination with some embodiments of the first aspect, in some embodiments, when the warning parameter index of the fault diagnosis result exceeds the fault warning threshold, after the step of generating fault warning information, the method also includes: obtaining multiple historical fault data including equipment number, fault type, fault degree and maintenance record, and building a historical fault database; performing model training based on the historical fault database to obtain a fault development prediction model; inputting the fault warning information into the fault development prediction model to obtain fault timing characteristics; generating a fault analysis report based on the fault timing characteristics; and pushing the fault analysis report to the mobile terminal of the target personnel.

[0009] In the above embodiment, the fault detection system constructs a historical fault database for model training, obtains a fault development prediction model, and generates an analysis report based on the fault timing characteristics, thereby achieving accurate prediction of the development trend of circuit breaker faults and reasonable planning of maintenance costs.

[0010] In combination with some embodiments of the first aspect, in some embodiments, the step of generating a fault analysis report based on the fault timing characteristics specifically includes: determining the expected service life of the high-voltage vacuum circuit breaker according to the fault timing characteristics; determining the maintenance cost budget according to the expected service life and maintenance records in the historical fault database; and generating a fault analysis report based on the expected service life and the maintenance cost budget.

[0011] In the above embodiment, the fault detection system determines the expected life by analyzing the fault timing characteristics and determines the maintenance cost budget in combination with the historical maintenance records, thereby achieving accurate prediction of the management and maintenance cost of the circuit breaker throughout its life cycle.

[0012] In combination with some embodiments of the first aspect, in some embodiments, after the steps of obtaining the contact action timing data of the high-voltage vacuum circuit breaker at two breaks, the pressure data in the vacuum chamber, and the ambient temperature, ambient humidity and system load index, the method also includes: determining the edge computing node closest to the location of the high-voltage vacuum circuit breaker; integrating the contact action timing data, the pressure data in the vacuum chamber, the ambient temperature, ambient humidity and the system load index into an original data packet; sending the original data packet to the edge computing node; and receiving the fault calculation result data packet returned by the edge computing node.

[0013] In the above embodiment, the fault detection system processes the original data through the edge computing node, realizes rapid data processing and real-time response, and improves the efficiency and accuracy of fault diagnosis.

[0014] In combination with some embodiments of the first aspect, in some embodiments, after the step of receiving the fault diagnosis results and fault warning thresholds returned by the edge computing node, the method also includes: extracting the computing node mark and data timestamp in the fault calculation result data packet; when the data timestamp is inconsistent with the original timestamp of the original data packet, sending a data acquisition request containing the original timestamp to the edge computing node.

[0015] In the above embodiment, the fault detection system ensures the accuracy of data timing and the reliability of the diagnosis result by verifying the consistency of data timestamps.

[0016] In combination with some embodiments of the first aspect, in some embodiments, when the warning parameter index of the fault diagnosis result exceeds the fault warning threshold, after the step of generating fault warning information, the method also includes: extracting the spatiotemporal distribution characteristics of the fault diagnosis result from the historical fault database; determining the correlation influence coefficient of the circuit breakers in each area based on the circuit breaker location data and spatiotemporal distribution characteristics of the geographic information system; when a preset power outage occurs, determining the fault spread range based on the correlation influence coefficient, and generating a partition maintenance priority list.

[0017] In the above embodiment, the fault detection system determines the correlation influence coefficient of the circuit breaker by analyzing the temporal and spatial distribution characteristics of the fault diagnosis results, thereby achieving accurate early warning of the preset power outage event and accurate prediction of the fault propagation range.

[0018] In combination with some embodiments of the first aspect, in some embodiments, when a preset power outage event occurs, after the step of determining the fault propagation range based on the associated impact coefficient and generating a partition maintenance priority list, the method also includes: determining a path planning table based on the partition maintenance priority list and real-time road conditions information; determining a maintenance scheduling plan based on maintenance personnel information, equipment supply information and the path planning table; determining a partition maintenance task sequence based on the maintenance scheduling plan, and pushing multiple maintenance tasks in the partition maintenance task sequence to the corresponding maintenance terminal.

[0019] In the above embodiment, the fault detection system formulates a maintenance scheduling plan by combining real-time road condition information, thereby achieving efficient deployment of maintenance resources and allocation of maintenance tasks.

[0020] In a second aspect, an embodiment of the present application provides a fault detection system, which comprises: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code comprises computer instructions, and the one or more processors call the computer instructions to enable the fault detection system to execute the method described in the first aspect and any possible implementation method of the first aspect.

[0021] In a third aspect, an embodiment of the present application provides a computer program product comprising instructions, and when the above-mentioned computer program product runs on a fault detection system, the above-mentioned fault detection system executes the method described in the first aspect and any possible implementation method of the first aspect.

[0022] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, comprising instructions. When the instructions are executed on a fault detection system, the fault detection system executes the method described in the first aspect and any possible implementation method of the first aspect.

[0023] It is understandable that the fault detection system provided in the second aspect, the computer program product provided in the third aspect, and the computer storage medium provided in the fourth aspect are all used to execute the method provided in the embodiment of the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.

[0024] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Due to the coordinated analysis of the double-break contact action timing data and the vacuum chamber pressure data, as well as the comprehensive evaluation of the ambient temperature, humidity and load index, combined with the dual diagnosis mechanism of the mechanical characteristic coordination vector and the voltage tolerance difference index, and the introduction of environmental impact factors to dynamically adjust the fault warning threshold, the operating status of the circuit breaker can be comprehensively and accurately evaluated, which effectively solves the problem that related technologies cannot adapt to different environmental conditions by relying solely on fixed thresholds, thereby realizing the intelligent fault diagnosis and precise warning, and improving the operation and maintenance efficiency of the circuit breaker.

[0025] 2. Due to the use of a distributed architecture in which the nearest edge computing node processes data, the original data is integrated for localized computing and processing, and the calculation results are quickly returned through the data packet transmission mechanism. This can greatly reduce data transmission delays and computing loads, and effectively solve the response lag and resource waste problems caused by centralized processing in related technologies, thereby achieving real-time fault diagnosis and improved computing efficiency, ensuring the system's rapid response capabilities.

[0026] 3. Due to the use of spatiotemporal distribution feature analysis based on historical fault data, the construction of an associated impact model combined with the circuit breaker location data of the geographic information system, and the prediction of fault spread when a preset power outage occurs, the risk of fault spread can be accurately assessed, effectively solving the problem of unpredictable chain reactions of faults in related technologies, thereby achieving accurate prediction of the fault range and reasonable division of maintenance priorities, thereby improving the overall reliability of the power grid system. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 It is a flow chart of a double-break high-voltage vacuum circuit breaker fault diagnosis method in an embodiment of the present application; Figure 2 It is another flowchart of the double-break high-voltage vacuum circuit breaker fault diagnosis method in the embodiment of the present application; Figure 3 It is a schematic diagram of the structure of a physical device of a fault detection system in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification of the present application, the singular expressions "one", "a kind of", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more of the listed items.

[0029] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0030] It should be noted that double-break (two-break) refers to two vacuum interrupter chambers (vacuum chambers) connected in series in the circuit breaker, and each break contains a pair of contacts (moving contact and static contact). The core feature of the double-break high-voltage vacuum circuit breaker is its two vacuum interrupter chambers connected in series, each of which contains a pair of conductive contacts (moving contact and static contact). When the circuit breaker is working normally, the contacts of the two interrupter chambers are closed at the same time to form a complete conductive path; when the circuit needs to be disconnected, the moving contacts of the two interrupter chambers need to leave the corresponding static contacts synchronously. At this time, the recovery voltage of the system will be evenly shared by the two breaks. This design significantly improves the insulation strength and breaking capacity of the circuit breaker. Based on this structural feature, the reliable operation of the circuit breaker requires that the two breaks maintain a high degree of synchronization in mechanical action and a balanced sharing in electrical performance. Therefore, when performing fault diagnosis, it is necessary to focus on the coordination of the contact action timing and the balance of voltage sharing.

[0031] For ease of understanding, the application scenarios of the embodiments of the present application are introduced below.

[0032] A large number of double-break high-voltage vacuum circuit breakers are used as core switchgear in the 110kV substation operated and maintained by a provincial power company. This type of circuit breaker shares the voltage through two series-connected vacuum interrupters, which improves the breaking capacity. However, due to the complex and changeable operating environment of the equipment, the temperature ranges from -20℃ to 40℃, the humidity ranges from 20% to 95%, and the load level fluctuates with the seasons. In the summer of a certain year, the area experienced continuous high temperatures of around 40℃, causing multiple circuit breakers to trip due to excessive environmental stress, resulting in large-scale power outages. The traditional fixed threshold warning mechanism cannot adapt to environmental changes, and often triggers an alarm only after the equipment is damaged, which not only increases maintenance costs but also affects the reliability of the power grid.

[0033] In the related art, basic detection of circuit breaker faults can be achieved by using a mechanical property tester and a simple online monitoring system. Maintenance personnel regularly measure parameters such as the circuit breaker opening and closing time and travel characteristics, and trigger fault alarms through fixed thresholds. The following describes the scenario of using the double-break high-voltage vacuum circuit breaker fault diagnosis method in the related art.

[0034] A certain substation uses a traditional circuit breaker fault diagnosis system, which mainly relies on regular maintenance and simple online monitoring. Maintenance personnel use mechanical property testers to measure parameters such as opening and closing time and travel characteristics every quarter, and use infrared thermometers to detect contact temperature distribution. The online monitoring system collects basic electrical parameters of the circuit breaker such as operating current waveform, switch position signal, etc., and sets fixed alarm thresholds. At the beginning of a certain year, a circuit breaker in operation had abnormal heating, but the warning was not triggered because the temperature did not reach the fixed alarm threshold of 85°C. Subsequently, in humid weather, the circuit breaker suffered a phase-to-phase short circuit due to reduced insulation performance, causing the equipment to burn. Subsequent analysis found that this accident could have been avoided if the warning threshold could be dynamically adjusted according to the ambient humidity.

[0035] The double-break high-voltage vacuum circuit breaker fault diagnosis method in the embodiment of the present application is adopted, by real-time collection of multi-dimensional operation data and analysis combined with environmental factors, to achieve accurate diagnosis and timely warning of faults, and to dynamically adjust the warning threshold according to environmental conditions to achieve accurate judgment of faults. The following introduces the scenario of using the double-break high-voltage vacuum circuit breaker fault diagnosis method in the present application.

[0036] A certain smart substation introduced the fault diagnosis method of this application, which collects multi-dimensional data such as the contact action timing, vacuum degree, temperature distribution, etc. of the circuit breaker in real time, and monitors the ambient temperature, humidity and load level at the same time. The system quickly processes data through edge computing nodes to calculate the mechanical characteristic coordination vector and voltage tolerance difference index. In July of a certain year, the system detected that the contact temperature of a circuit breaker reached 65°C. Although it did not exceed the normal threshold, based on the extreme conditions of the ambient temperature of 38°C and humidity of 85% at the time, the system dynamically lowered the warning threshold to 60°C and issued a warning in time. The maintenance personnel carried out timely inspections and found that the contacts were micro-ablated. After replacement, more serious failures were avoided.

[0037] It can be seen that the double-break high-voltage vacuum circuit breaker fault diagnosis method in the embodiment of the present application can not only achieve accurate fault detection, but also effectively solve the problems of poor environmental adaptability and delayed warning, thereby achieving optimization of fault prevention and maintenance costs.

[0038] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 , which is a flow chart of a double-break high-voltage vacuum circuit breaker fault diagnosis method in an embodiment of the present application.

[0039] S101, obtaining contact action timing data of the high-voltage vacuum circuit breaker at two breaks, pressure data in the vacuum chamber, and ambient temperature, ambient humidity, and system load index.

[0040] Among them, the contact action timing data represents the sampling data of the displacement, velocity and acceleration of the circuit breaker contacts changing with time during the opening and closing process; the vacuum chamber pressure data refers to the real-time monitoring value of the vacuum degree inside the vacuum arc extinguishing chamber, which is used to characterize the sealing performance of the vacuum chamber; the ambient temperature and ambient humidity respectively represent the temperature and relative humidity values ​​of the circuit breaker operating place; the system load index is used to indicate the load level of the power system where the circuit breaker is located, which is usually calculated comprehensively from parameters such as current and power factor.

[0041] The fault detection system needs to comprehensively collect various operating parameters when the circuit breaker is operating normally or undergoing routine inspections. Specifically, the fault detection system first collects the contact movement trajectory through the displacement sensor, and records the timestamp of the sampling moment to form a high-precision time series data stream; it also continuously monitors the internal pressure changes of the vacuum chamber through the vacuum gauge to ensure that the sampling frequency is not less than 100Hz; at the same time, it collects environmental data through temperature and humidity sensors, and obtains real-time load data from the system dispatch end. All data need to be time synchronized and calibrated to ensure the accuracy of subsequent analysis.

[0042] In some embodiments, high-quality data collection and preprocessing can be achieved in a variety of ways: Optionally, the fault detection system uses an adaptive sampling algorithm to dynamically adjust the sampling frequency, preprocesses the original data through piecewise linear interpolation and wavelet denoising, and then uses the Kalman filter algorithm for data fusion, and finally obtains time series data with a high signal-to-noise ratio; Optionally, the fault detection system is based on a distributed sensor network architecture, adopts a multi-source heterogeneous data collaborative collection solution, performs real-time data cleaning and feature extraction through edge computing nodes, and then realizes continuous data acquisition and updating through time window sliding. It is understandable that other data collection and preprocessing methods can also be used to obtain the required types of data, which are not limited here.

[0043] In some embodiments, the fault detection system uses multi-source sensors to collect circuit breaker operation data in real time. The contact action timing data collection uses a high-precision displacement sensor to achieve continuous tracking of the contact position through the magnetic encoding principle. After the displacement signal x(t) is digitally filtered, Calculate the velocity using the central difference method: v(t)=(x(t+Δt)-x(t-Δt)) / (2Δt); Acceleration: a(t)=(v(t+Δt)-v(t-Δt)) / (2Δt); where x(t) is the position of the contact at time t, in millimeters (mm); Δt is the sampling time interval, the recommended value is 1ms, and the value range is 0.5ms~2ms; v(t) is the speed of the contact at time t, in meters per second (m / s); a(t) is the acceleration of the contact at time t, in meters per square second (m / s²).

[0044] The vacuum chamber pressure is measured by an ionization gauge with a sensitivity of 10⁻ 6 Pa, the pressure signal is filtered through exponential smoothing to eliminate high-frequency noise. Among the environmental parameters, temperature and humidity data are collected using a composite sensor, and the system load index F is calculated using the following formula: F=α(I / I_n)²+β(cosφ)+γ(T / T_n); where I is the real-time current, I_n is the rated current, cosφ is the power factor, T is the ambient temperature, T_n is the rated operating temperature, and α, β, and γ are weight coefficients obtained through historical data regression.

[0045] For example, during the opening process of a circuit breaker, the contact position recorded by the displacement sensor moves uniformly from x1=0mm to x2=200mm, which takes t=50ms. The average speed is calculated to be 4m / s, and the peak acceleration is 80m / s². At the same time, the vacuum chamber pressure is recorded to be 2×10⁻ 5 Pa, ambient temperature 35℃, humidity 75%RH, load current 0.8 times rated value, substitute into the load index formula to calculate F=0.85.

[0046] S102. Calculate the contact action speed difference and bounce characteristic parameters of the two breaks according to the contact action timing data, and generate a mechanical characteristic coordination vector in combination with the opening and closing energy consumption data of the two breaks.

[0047] Among them, the difference in contact action speed indicates the degree of inconsistency in the speed of the two break contacts during the opening and closing process; the bounce characteristic parameters include the number of bounces, bounce time and bounce amplitude; the opening and closing energy consumption data refers to the energy loss of the operating mechanism during the opening and closing process; the mechanical characteristic coordination vector is used to characterize the coordination of the mechanical actions of the two breaks.

[0048] After obtaining the contact action timing data, the fault detection system needs to perform feature extraction and vector construction. Specifically, the fault detection system first performs differential operation on the timing data to obtain the velocity curve, calculates the correlation coefficient and root mean square error of the two fracture velocity curves to quantify the velocity difference; then identifies the key time points in the bounce process through the peak detection algorithm, extracts the bounce characteristic parameters; at the same time, calculates the energy consumption through the current and voltage waveforms of the operating mechanism; finally, combines these features into a feature vector that describes the synergy of mechanical characteristics.

[0049] In some embodiments, mechanical characteristic analysis and vector construction can be achieved in a variety of ways: Optionally, the fault detection system uses wavelet transform to perform time-frequency analysis on the speed curve, extracts the energy characteristics of each frequency band, combines the fuzzy clustering algorithm to calculate the speed coordination index, and then reduces the dimension through principal component analysis to obtain the final feature vector; Optionally, the fault detection system is based on a deep learning method, uses a long short-term memory network to extract timing features, captures the interaction relationship between the two fractures through an attention mechanism, and finally obtains a fixed-dimensional coordination vector through full-connection layer mapping. It is understandable that other feature extraction and vector construction methods can also be used to achieve quantitative characterization of mechanical characteristics, which is not limited here.

[0050] In some embodiments, the synergy of the mechanical properties of the two fractures is quantitatively analyzed through three dimensions: contact action speed difference, bounce characteristics, and energy consumption.

[0051] For the velocity curve v_k(t) of fracture k (k=1, 2), define the velocity difference function: ΔV(t)=|v1(t)-v2(t)|; The feature extraction is performed through the speed difference curve after wavelet denoising. The speed coordination index S_v is defined as: S_v=exp(-λ∫ΔV(t)²dt / T); where T is the total action time and λ is the scale factor.

[0052] The bounce characteristics are analyzed by continuous wavelet transform, using Morlet wavelet with a scale range of a∈[1,64]. The ridges of the wavelet coefficients W(a, b) of the bounce signal x(t) are extracted to obtain the number of bounces n, the average bounce time τ and the bounce amplitude sequence {A_i}. The bounce characteristic parameter B is calculated as: B=w1n+w2τ+w3σ(A); where σ(A) is the standard deviation of the bounce amplitude, and w1, w2, and w3 are weight coefficients.

[0053] The energy consumption is calculated from the current i(t) and voltage u(t) of the actuator: E = ∫u(t)i(t)dt; The final mechanical property coordination vector M is composed of the above parameters: M=[S_v, B, E].

[0054] For example, during the closing process of a circuit breaker, the maximum speed difference between the two fractures was recorded to be 0.5m / s, with a duration of 8ms, and S_v=0.92 was calculated. Bounce analysis showed 3 bounces, with an average duration of 2.5ms and an amplitude standard deviation of 0.3mm, and B=0.45 was obtained. The energy consumption of the actuator was 850J. The synthetic mechanical characteristic coordination vector M=[0.92, 0.45, 0.85].

[0055] S103, obtaining temperature distribution data of the conductive loops of the two breaks, and calculating the heat generation imbalance coefficient of the two breaks based on the temperature distribution data.

[0056] Among them, the temperature distribution data of the conductive circuit represents the spatial distribution of the temperature field of the circuit breaker contacts and their connecting conductors; the heat imbalance coefficient refers to the degree of difference in the temperature distribution of the two fractures, which is used to characterize the heating uniformity of the conductive circuit; the temperature distribution data includes the temperature values ​​of multiple measuring points such as the contact surface temperature, the connecting conductor temperature and the ambient reference temperature.

[0057] The fault detection system monitors the temperature when the circuit breaker is running under power. Specifically, the fault detection system uses an infrared thermal imager to perform a full-scale scan of the conductive circuits of the two breaks of the circuit breaker to obtain a high-resolution temperature field image; then the temperature field image is meshed and feature points are extracted to establish a three-dimensional temperature distribution model; then the temperature difference between the corresponding positions of the two breaks is calculated, and combined with the spatial gradient information of the temperature field, the heat imbalance coefficient is obtained by weighted averaging.

[0058] In some embodiments, temperature field analysis and imbalance assessment can be achieved in a variety of ways: Optionally, the fault detection system uses a thermal field finite element analysis method to construct a thermal conduction model of the conductive circuit, obtains the steady-state temperature distribution by solving the temperature field distribution equation, and then uses statistical methods to calculate the discreteness index and unevenness coefficient of the temperature field; Optionally, the fault detection system uses computer vision technology to perform image segmentation and region recognition on infrared images, extract temperature features of key areas, establish a temperature distribution model through a deep learning model, and finally calculate the similarity of the temperature patterns of the two fractures to obtain the imbalance coefficient. It is understandable that other temperature field analysis methods can also be used to evaluate heat imbalance, which is not limited here.

[0059] In some embodiments, the calculation of the heat imbalance coefficient is first based on the temperature field data collected by the infrared thermal imager. The system uses a high-precision infrared thermal imager with a spatial resolution of 0.1mm to scan the conductive circuits of the two fractures of the circuit breaker in real time at a frequency of 10Hz. The temperature field data is pre-processed by grid division, and the temperature distribution of each fracture is converted into a two-dimensional matrix T (i, j), where i and j represent grid coordinates. For each fracture k (k=1, 2), the spatial gradient of its temperature field is calculated: G_k(i,j)=√[(∂T / ∂x)²+(∂T / ∂y)²]; The heat dissipation coefficient η of the two fractures is defined as: η=∑∑w(i, j)|T1(i, j)-T2(i, j)| / √[(G1(i, j)+G2(i, j)) / 2]; where w(i, j) is the position weight factor, which reflects the contribution of different positions to the imbalance. The weight factor is determined according to the importance of the component. The contact area has a larger weight (typical value 0.6-0.8), and the connecting conductor area has a smaller weight (typical value 0.2-0.4). The final heat imbalance coefficient η is normalized and has a value range of [0, 1]. The larger the value, the more serious the heat imbalance.

[0060] For example, when a circuit breaker is in operation, the temperature distribution data of the two fractures are obtained through infrared imaging. The highest temperature of the contact area of ​​fracture A is 85°C, and the temperature gradient is 8°C / cm; the temperature of the corresponding position of fracture B is 65°C, and the temperature gradient is 5°C / cm. Substituting into the calculation formula, considering the contact area weight of 0.7, the contribution of the imbalance coefficient of this area is 0.82. After similar calculations for other areas, the overall heat imbalance coefficient η=0.76 is finally obtained, indicating that there is obvious heat imbalance.

[0061] S104. Collect the voltage sharing ratio and transition recovery voltage characteristics of the two breakers during the breaking process, combine the pressure data in the vacuum chamber and the heat imbalance coefficient, and determine the voltage tolerance difference index.

[0062] Among them, the voltage sharing ratio indicates the proportion of overvoltage borne by the two breaks during the breaking process; the transition recovery voltage characteristics include parameters such as the rise rate, peak value, and oscillation frequency; the voltage tolerance difference index is used to characterize the degree of difference in the insulation performance of the two breaks; and the heat imbalance coefficient reflects the change in the contact resistance of the contacts.

[0063] The fault detection system measures the voltage characteristics when the circuit breaker performs the breaking operation. Specifically, the fault detection system first collects the instantaneous voltage waveforms of the two breaks synchronously through the high-voltage probe, calculates the dynamic change process of the voltage sharing ratio; then extracts the time domain and frequency domain characteristics of the transition recovery voltage, including the rise time, peak multiple, oscillation period, etc.; then conducts a multi-dimensional correlation analysis of the voltage characteristics with the vacuum chamber pressure and the heat imbalance coefficient, comprehensively evaluates the difference in the voltage tolerance capacity of the two breaks, and finally outputs a standardized difference index.

[0064] In some embodiments, voltage characteristic analysis and difference evaluation can be achieved in a variety of ways: Optionally, the fault detection system uses wavelet multi-resolution analysis method to perform time-frequency decomposition on the voltage waveform, extract the energy distribution characteristics of each scale, and establish a multivariate regression model based on vacuum and temperature factors to calculate the comprehensive score of voltage tolerance; Optionally, the fault detection system is based on artificial neural network technology, takes voltage waveform characteristics, pressure data and temperature coefficient as input, learns their internal correlation through a multi-layer perceptron model, and outputs a normalized voltage tolerance difference index. It is understandable that other analysis methods can also be used to evaluate voltage tolerance differences, which are not limited here.

[0065] In some embodiments, the voltage characteristic analysis during the circuit breaker opening process is based on a high-speed data acquisition system with a sampling rate of 10 MHz and an accuracy of 16 bits. For the instantaneous voltages u1(t) and u2(t) of the two breaks, the voltage sharing ratio K(t) is defined as: K(t)=u1(t) / [u1(t)+u2(t)]; The system analyzes the transition recovery voltage characteristics through wavelet transform. The db4 wavelet basis function is selected to perform a 5-layer decomposition of the voltage signal and extract the energy coefficient E_j (j=1, 2, ..., 5) of each frequency band. Combined with the vacuum chamber pressure p and the heat imbalance coefficient η, the voltage tolerance difference index D is calculated as follows: D=w1|K(t)-0.5|_max+w2∑(E1_j-E2_j)² / p+w3η; Among them, w1, w2, and w3 are weight coefficients, which are set by the user based on experience.

[0066] For example, when a circuit breaker interrupts the short-circuit current, it is recorded that the A breakpoint bears 58% of the recovery voltage and the B breakpoint bears 42%. Wavelet analysis shows that the energy difference in the high-frequency band (>1MHz) is as high as 30%, and the vacuum degree is 3×10⁻ 5 Pa, heat imbalance coefficient 0.76. Substituting into the formula, the voltage tolerance difference index D=0.68 is obtained, indicating that there is a significant difference in the insulation performance of the two fractures.

[0067] S105 , inputting the mechanical characteristic coordination vector and the voltage tolerance difference index into the fault diagnosis model to obtain a fault diagnosis result.

[0068] Among them, the mechanical characteristic coordination vector characterizes the synchronization and stability of the mechanical actions of the two breaks; the voltage tolerance difference index reflects the inconsistency of the insulation performance of the two breaks; the fault diagnosis results include information such as fault type, fault degree and fault development trend.

[0069] After obtaining the feature vector, the fault detection system evaluates the fault status. Specifically, the fault detection system first performs data standardization on the mechanical characteristic coordination vector and the voltage tolerance difference index; then the processed features are input into a pre-trained fault diagnosis model, which uses a multi-layer structure to process mechanical characteristics and electrical characteristics respectively; then the fault probability distribution is obtained through model reasoning, and the fault type is determined in combination with the expert rule system; finally, a diagnostic report containing detailed fault information is generated.

[0070] It should be noted that the fault diagnosis model here adopts a multi-layer neural network structure, which includes three core layers: feature extraction, attention mechanism, and classification output. In the training phase, the mechanical characteristic coordination vector (including contact speed difference, bounce characteristics, energy consumption data) and voltage tolerance difference index (including voltage sharing ratio, transition recovery characteristics, vacuum degree data) in the historical operation data are input, and the output end is the corresponding fault type and degree label. The cross entropy loss function is used to optimize the model parameters, and the model performance is evaluated through K-fold cross validation. In actual application, the model receives the real-time collected operating parameters, dynamically calculates the fault probability distribution in combination with environmental influencing factors (temperature, humidity, load index), and determines whether to trigger the early warning mechanism based on the confidence level. The model can adaptively adjust the weight importance of different features to improve the accuracy of diagnosis.

[0071] In some embodiments, fault diagnosis uses a deep learning model structure, which includes three modules: feature extraction, attention mechanism, and classification output. The input layer receives the mechanical characteristic coordination vector M and the voltage tolerance difference index D, and obtains a high-dimensional representation through feature mapping: H = ReLU(W1[M;D]+b1); The attention layer calculates the importance weight α of each feature: e_i=v^Ttanh(W2h_i+b2); α_i = softmax(e_i); The weighted features are passed through the fully connected layer to obtain the fault type probability distribution p and the fault severity score s: p=softmax(W3∑α_iH_i+b3); s=σ(W4∑α_iH_i+b4); where σ is the sigmoid function.

[0072] The model is optimized using the cross entropy loss function: L=-∑y_ilog(p_i)+λ|ss*|²; where y_i is the true label, s* is the true fault degree, and λ is the balance coefficient.

[0073] For example, the characteristic vector of a circuit breaker [0.92, 0.45, 0.85, 0.68] is input, and the fault type probability distribution [0.05, 0.82, 0.10, 0.03] is obtained through model calculation, indicating that the most likely fault is contact ablation fault (Type 2), and the fault severity score is 0.65, corresponding to medium severity.

[0074] S106. Based on the ambient temperature, ambient humidity and system load index, calculate the environmental impact factor of the high-voltage vacuum circuit breaker under the current working condition, generate a working condition characteristic vector, and determine a fault warning threshold corresponding to the working condition characteristic vector.

[0075] Among them, the environmental impact factor represents the comprehensive impact of environmental conditions on the performance of the circuit breaker; the operating condition feature vector refers to the feature set that describes the current operating state; the fault warning threshold is used to determine the dynamic boundary value of the fault severity; and the system load index reflects the actual load level of the circuit breaker.

[0076] The fault detection system needs to dynamically adjust the warning standard according to environmental conditions. Specifically, the fault detection system first uses multivariate regression analysis to calculate the weight of the impact of ambient temperature and humidity on circuit breaker performance; then combines the system load index to build a working condition evaluation model to generate the working condition feature vector of the current state; then based on historical data statistical analysis, establish a mapping relationship between the working condition feature vector and the probability of fault occurrence; finally, use an adaptive algorithm to calculate the fault warning threshold corresponding to the current working condition.

[0077] In some embodiments, environmental impact assessment and threshold calculation can be achieved in a variety of ways: Optionally, the fault detection system uses a fuzzy reasoning system, takes environmental parameters and load indicators as input variables, calculates environmental impact factors through fuzzy rule reasoning, and dynamically optimizes the warning threshold in combination with historical fault data; Optionally, the fault detection system is based on a reinforcement learning method, takes environmental conditions as state space, and warning thresholds as action space, and gradually optimizes the threshold adjustment strategy through a value function to achieve adaptive optimization of the warning mechanism. It is understandable that other methods can also be used to achieve working condition assessment and threshold calculation, which are not limited here.

[0078] In some embodiments, the environmental impact factor is calculated by a multivariate nonlinear mapping model. Assuming that the reference values ​​of the ambient temperature T, relative humidity H, and load index F are T0, H0, and F0 respectively, the environmental impact factor E is defined as: E=α·exp[k1(T-T0) / T0]+β·exp[k2(H-H0) / H0]+γ·(F / F0)^n; where α, β, and γ are weight coefficients and satisfy α+β+γ=1, k1 and k2 are temperature and humidity sensitivity coefficients, and n is the nonlinear exponent of the load index.

[0079] The operating condition characteristic vector C is constructed through polynomial kernel transformation: C=[E, E², T·H, H·F, T·F, φ(T, H, F)]; where φ(T, H, F) is the three-variable interaction term.

[0080] An adaptive function for establishing the fault warning threshold η based on historical data: η=η0·[1+μ1tanh(w1C1)+μ2tanh(w2C2)+…+μ_ntanh(w_nC_n)]; where η0 is the baseline threshold, and μᵢ and wᵢ are adaptive coefficients.

[0081] For example, the operating environment temperature of a circuit breaker is 38°C (baseline 20°C), the humidity is 85% (baseline 65%), and the load index is 0.9 (baseline 0.8). The calculated environmental impact factor E=1.35. The generated operating condition feature vector is threshold mapped, and the baseline warning threshold of 0.8 is adjusted to 1.08.

[0082] S107: When the warning parameter index of the fault diagnosis result exceeds the fault warning threshold, generate fault warning information.

[0083] Among them, the warning parameter index refers to the quantitative index used to evaluate the severity of the fault; the fault warning threshold refers to the critical value that triggers the warning; the fault warning information includes the fault type, location, severity and handling suggestions.

[0084] The fault detection system needs to issue a warning in time when an abnormality is found. Specifically, the fault detection system first monitors the changing trends of various warning parameter indicators in real time; when a certain indicator is detected to exceed the corresponding warning threshold, the warning mechanism is immediately triggered; then the urgency of the fault is evaluated according to the degree of over-limit and duration; then a warning report containing detailed fault information and handling suggestions is generated; finally, the warning information is pushed to relevant maintenance personnel through multiple communication channels.

[0085] In some embodiments, early warning triggering and information push can be achieved in a variety of ways: Optionally, the fault detection system adopts a multi-level early warning mechanism, sets different early warning levels according to the degree of parameter overrun, determines the comprehensive early warning level through weight calculation, and selects the corresponding push strategy and processing flow according to the early warning level; Optionally, the fault detection system establishes an equipment-fault-measure association network based on knowledge graph technology, matches the most suitable processing solution in real time, and arranges maintenance tasks through an intelligent scheduling system. It is understandable that other methods can also be used to achieve early warning management, which is not limited here.

[0086] In practical applications, multiple edge nodes can be deployed according to the distribution of substations to achieve reasonable allocation of computing resources. The following is a supplement to the scenario of this embodiment.

[0087] Based on the application of this solution, a regional power grid company further integrated the historical fault database and geographic information system. During a thunderstorm in August of a certain year, the system detected abnormalities in the circuit breakers of multiple substations. Based on the analysis of spatiotemporal distribution characteristics and the assessment of associated impacts, the system quickly located the substation at the source of the fault and predicted the possible path of fault propagation. Through the zoning maintenance priority list and intelligent scheduling scheme, the system planned the optimal maintenance route for maintenance personnel and assigned tasks according to professional expertise. Finally, all maintenance was completed within 4 hours, and the scope of power outage was kept to a minimum.

[0088] After combining the above scenarios, the following is a more detailed description of the process of the method provided by this implementation. Figure 2 , is another flow chart of the double-break high-voltage vacuum circuit breaker fault diagnosis method in the embodiment of the present application.

[0089] S201, obtaining contact action timing data of the high-voltage vacuum circuit breaker at two breaks, pressure data in the vacuum chamber, and ambient temperature, ambient humidity and system load index.

[0090] Referring to step S101 , the fault detection system acquires various operating data and environmental data.

[0091] In some embodiments, after acquiring the data, the fault detection system may not refer to the above steps S102-S106 for data processing. The fault detection system may send the data to the edge device for calculation to improve computing efficiency; specifically, the fault detection system will determine the edge computing node closest to the high-voltage vacuum circuit breaker, and then integrate the contact action timing data, vacuum chamber pressure data, ambient temperature, ambient humidity and system load index into the original data packet; then send the original data packet to the edge computing node, and receive the fault calculation result data packet returned by the edge computing node.

[0092] Among them, edge devices refer to computing devices deployed at the edge of the network; raw data packets represent unprocessed multi-source data sets; edge computing nodes refer to local computing units with data processing capabilities; and fault calculation result data packets contain processed feature vectors and diagnostic results.

[0093] The fault detection system needs to make full use of distributed computing resources. Specifically, the fault detection system first obtains available node information through the edge computing node registry, selects the nearest computing node based on geographic location information, then encapsulates the collected multi-source heterogeneous data, generates a standard data packet containing data identification, timestamp and original data, and then sends the data packet to the edge node through a secure channel for parallel computing, thus achieving distributed processing of computing load.

[0094] In some embodiments, the selection of edge computing nodes adopts a comprehensive evaluation mechanism based on physical distance, network latency and computing load. The system uses a weighted distance function: D(i)=w1d(p0,p_i)+w2L(i)+w3C(i) calculates the optimal node, where the physical distance d, network delay L and computing load C are assigned different weights respectively.

[0095] Data transmission adopts a reliability guarantee mechanism, which is dynamically adjusted through adaptive retransmission timeout RTO. The RTO value is calculated based on the smoothed round-trip delay RTT, specifically: RTO(i)=RTT(i)+4σ(RTT).

[0096] The original data packet is designed with a three-segment structure, which includes a header part with device identification, timestamp, and priority, a data payload part including contact action timing, vacuum chamber pressure, and environmental parameters, and a checksum part based on the entire packet data.

[0097] To ensure transmission reliability, the system monitors the round-trip time (RTT) in real time and dynamically updates the retransmission strategy. The RTT is calculated using the exponentially weighted moving average method, namely: RTT(i)=α·RTT(i-1)+(1-α)·(t_ack-t_send).

[0098] For example, after a circuit breaker triggered a fault alarm, the system selected the optimal edge node (physical distance 2 kilometers, network delay 15 milliseconds) within 50 milliseconds through the above mechanism. The total time for data transmission and calculation was 530 milliseconds. The entire process was completed automatically and data consistency was guaranteed.

[0099] In some embodiments, after receiving the data packet returned by the edge computing node, the fault detection system needs to perform data verification; specifically, the fault detection system will extract the computing node mark and data timestamp in the fault calculation result data packet, and when the data timestamp is inconsistent with the original timestamp of the original data packet, send a data acquisition request containing the original timestamp to the edge computing node.

[0100] Among them, the computing node tag is used to uniquely identify the edge node that processes the data; the data timestamp represents the time information of data processing; the original timestamp refers to the time stamp when the data is collected; and the data acquisition request is used to re-acquire the calculation results.

[0101] The fault detection system needs to ensure the timing consistency of data processing. Specifically, the fault detection system first parses the metadata information in the returned data packet, extracts the computing node tag and processing timestamp, and then compares it with the timestamp of the original data packet to determine whether there is a timing deviation. When the time inconsistency is found, it immediately generates a data re-acquisition request containing the original timestamp to ensure the timing accuracy of the data.

[0102] In some embodiments, the data verification mechanism implements a timestamp consistency check based on a distributed clock synchronization protocol. The system calculates the relationship between the edge node timestamp, the original timestamp and the transmission delay through a time deviation function Δt, that is, Δt=|T_edge-T_origin-T_trans|.

[0103] Taking into account the randomness of network transmission, the system uses a dynamic fault tolerance threshold ε(t)=ε0+k·log(1+σ²(RTT)) for judgment, where the basic fault tolerance threshold ε0 is set according to the system stability, k is the adjustment coefficient, and σ²(RTT) is the variance of the round-trip delay.

[0104] When it is detected that the time deviation exceeds the fault tolerance threshold, the system automatically generates a re-retrieval request containing the original timestamp, data range and verification code. The verification code is generated based on the session key through the HMAC algorithm to ensure the authenticity of the request.

[0105] For example, during a data verification, the system discovered that the timestamp of the data packet returned by the edge node was delayed by 2.5 seconds, significantly exceeding the fault tolerance threshold of 200 milliseconds. It immediately triggered the data retrieval mechanism to ensure the timing consistency of the data.

[0106] S202. Calculate the contact action speed difference and bounce characteristic parameters of the two breaks according to the contact action timing data, and generate a mechanical characteristic coordination vector in combination with the opening and closing energy consumption data of the two breaks.

[0107] Referring to step S102 , the fault detection system calculates a mechanical characteristic coordination vector.

[0108] S203, obtaining temperature distribution data of the conductive loops of the two breaks, and calculating the heat generation imbalance coefficient of the two breaks based on the temperature distribution data.

[0109] Referring to step S103 , the fault detection system calculates the heat imbalance coefficient.

[0110] S204. Collect the voltage sharing ratio and transition recovery voltage characteristics of the two breakers during the breaking process, and determine the voltage tolerance difference index by combining the pressure data in the vacuum chamber and the heat imbalance coefficient.

[0111] Referring to step S104 , the fault detection system calculates a voltage tolerance difference index.

[0112] S205 , inputting the mechanical characteristic coordination vector and the voltage tolerance difference index into the fault diagnosis model to obtain a fault diagnosis result.

[0113] Referring to step S105 , the fault detection system determines the fault diagnosis result.

[0114] S206. Based on the ambient temperature, ambient humidity and system load index, calculate the environmental impact factor of the high-voltage vacuum circuit breaker under the current working condition, generate a working condition characteristic vector, and determine a fault warning threshold corresponding to the working condition characteristic vector.

[0115] Referring to step S106 , the fault detection system determines a fault warning threshold.

[0116] S207: When the warning parameter index of the fault diagnosis result exceeds the fault warning threshold, generate fault warning information.

[0117] Referring to step S107 , the fault detection system will generate a warning message when the fault diagnosis result is abnormal.

[0118] S208. Acquire a plurality of historical fault data including equipment numbers, fault types, fault severity and maintenance records, and construct a historical fault database.

[0119] Among them, the historical fault database refers to a structured data set that stores the historical operation fault information of the circuit breaker; the equipment number is used to uniquely identify each circuit breaker; the fault type indicates the specific classification of the fault; the fault degree reflects the severity of the fault; and the maintenance record contains fault handling methods and effect evaluation information.

[0120] The fault detection system needs to establish a complete historical data management system. Specifically, the fault detection system first extracts historical fault records from the equipment management system, including basic information such as the time, type, and location of the fault; then collects the fault handling reports filled out by maintenance personnel, extracts fault cause analysis and treatment plans; then cleans and standardizes the data to establish a unified data format; finally, builds a relational database, establishes a fault type index and equipment association relationship.

[0121] In some embodiments, historical data collection and management can be achieved in a variety of ways: Optionally, the fault detection system adopts a distributed database architecture, designs a multi-layer data model including dimensions such as equipment information, operating status, fault records, and maintenance measures, and analyzes fault modes and evolution laws through data mining technology; Optionally, the fault detection system establishes a decentralized data storage system based on blockchain technology to ensure data traceability and non-tamperability, and achieve secure sharing of fault data. It is understandable that other methods can also be used to achieve historical data management, which is not limited here.

[0122] In some embodiments, the historical fault database adopts a spatiotemporal multidimensional index structure. Each fault record is represented as a vector: F = [ID, t, l, s, {x_i}, {y_j}, {z_k}]; where t is the timestamp, l is the location coordinate, s is the fault severity, {x_i} is the equipment parameter set, {y_j} is the environmental parameter set, and {z_k} is the maintenance record set.

[0123] Build an R-tree spatial index: Node={MBR, childPtr[], entryCount}; MBR=[t_min, t_max, l_min, l_max]; Data retrieval uses multidimensional range query: Q(t1, t2, l1, l2) = {F|t1≤Ft≤t2∧l1≤Fl≤l2}; to achieve spatiotemporal correlation analysis and pattern mining.

[0124] For example, the system recorded 534 circuit breaker failures in a certain area within five years. Through the spatiotemporal index, it can quickly locate 68 contact failures that occurred during the high temperature period of a certain summer, providing data support for failure mode analysis.

[0125] S209: Perform model training based on the historical fault database to obtain a fault development prediction model.

[0126] Among them, the fault development prediction model refers to a mathematical model used to predict the fault evolution trend. The model training process includes steps such as feature engineering, parameter optimization and performance verification; the historical fault database provides the sample data required for model training.

[0127] The fault detection system needs to build a prediction model based on historical data. Specifically, the fault detection system first performs time series analysis on historical data to extract the key features and change patterns of fault development; then designs a deep learning model structure suitable for time series prediction, including a data preprocessing layer, a feature extraction layer, and a prediction output layer; then uses the training data set for model training and parameter optimization, and evaluates model performance through cross-validation; finally, selects the optimal model for fault prediction.

[0128] It should be noted that the fault development prediction model here is built based on the long short-term memory network (LSTM), which includes two functional modules: time series feature extraction and multi-step prediction. The training data comes from the historical fault database, which contains multi-dimensional time series information such as equipment number, fault type, fault degree, maintenance record, etc. The model takes the minimization of mean square error as the optimization goal and adjusts the network parameters through the back propagation algorithm. The attention mechanism is used to capture the feature changes at key time points to improve the prediction accuracy. When applied, the model inputs the current fault warning information and outputs the fault development trend prediction results, including the expected service life estimation and maintenance cost budget, to provide a basis for maintenance decisions. The model can handle long-term dependencies and realize multi-time scale prediction.

[0129] S210 , inputting fault warning information into a fault development prediction model to obtain fault time sequence characteristics.

[0130] Among them, the fault time series characteristics represent the set of time dimension characteristics in the fault development process; the fault warning information contains various indicators of the current fault status; the fault development prediction model is a trained time series prediction model; the time series characteristics include information such as fault development speed, trend direction and key time nodes.

[0131] The fault detection system needs to predict and analyze the fault development trend. Specifically, the fault detection system first converts the fault warning information into a standardized model input format; then performs multi-step time series prediction through the prediction model to obtain the fault state estimation at different time points in the future; then performs confidence analysis on the prediction results to evaluate the reliability of the prediction; finally, extracts the key feature points in the prediction sequence to form a complete time series feature description.

[0132] S211. Generate a fault analysis report based on the fault timing characteristics.

[0133] Among them, the fault analysis report is a systematic description document of the fault status; the fault timing characteristics reflect the development law of the fault; the analysis report includes multiple parts such as fault diagnosis results, development trend prediction, maintenance suggestions, etc.; the report format must meet the standardization requirements.

[0134] The fault detection system needs to generate professional analysis reports. Specifically, the fault detection system first determines the overall framework and content structure of the report based on the preset report template; then converts the fault sequence characteristics into easy-to-understand charts and text descriptions; then combines historical maintenance experience to give targeted processing suggestions and preventive measures; finally, through natural language generation technology, a logically clear and structured analysis report is formed.

[0135] In some embodiments, report generation and content organization can be achieved in a variety of ways: Optionally, the fault detection system uses a rule-based expert system to automatically match processing solutions based on fault characteristics and prediction results, and generate a comprehensive report containing fault analysis, risk assessment, and maintenance recommendations; Optionally, the fault detection system uses natural language processing technology based on deep learning to convert data analysis results into professional technical documents, and generate differentiated report content based on different user roles. It is understandable that other methods can also be used to achieve intelligent generation of analysis reports, which are not limited here.

[0136] In some embodiments, report generation is based on a natural language processing model based on deep learning. First, a professional terminology knowledge graph G is constructed: G = (V, E, R); V={equipment term, fault type, maintenance measures}; E={relationship type}; R: V × V → E; Perform semantic mapping on the fault diagnosis result F: S = Encoder (F); A = Attention (S, G); T = Decoder (A); Where S is the semantic vector, A is the attention weight, and T is the generated text description. The report template is defined as: M = {paragraph structure, keyword position, format specification}; Fill the template by conditionally generating a network: P(w_t|w_{ <t},M,F)=softmax(W_oh_t+b_o); h_t=GRU(embed(w_{t-1}),h_{t-1},c_t); Where w_t is the tth word and c_t is the context vector.

[0137] For example, after the system analyzes the fault data of a circuit breaker, it automatically generates an analysis report containing complete content such as "basic equipment information, fault phenomenon description, cause analysis, development trend prediction, and processing suggestions."

[0138] In some embodiments, the fault detection system will display the service life and maintenance budget in the report; specifically, the fault detection system will determine the expected service life of the high-voltage vacuum circuit breaker based on the fault timing characteristics; then determine the maintenance cost budget based on the expected service life and maintenance records in the historical fault database; and then generate a fault analysis report based on the expected service life and maintenance cost budget.

[0139] Among them, the expected service life refers to the remaining time that the equipment is expected to operate normally; the maintenance cost budget refers to the estimated equipment maintenance cost; and the maintenance record contains historical maintenance costs and maintenance plan information.

[0140] The fault detection system needs to perform life assessment and cost prediction. Specifically, the fault detection system first establishes a life prediction model based on the fault timing characteristics to evaluate the remaining service life of the equipment; then analyzes the cost data in the historical maintenance records to establish a maintenance cost prediction model; then combines the life prediction results and the cost model to generate a detailed budget plan, including cost estimates for regular maintenance and emergency repairs.

[0141] In some embodiments, the system uses a Weibull distribution model to estimate the expected life of the circuit breaker through a failure rate function: λ(t)=(β / η)(t / η)^(β-1); t is the operating time in years; β is the shape parameter, reflecting the failure rate trend, with a range of values ​​of [2.0, 3.5] and a recommended value of 2.8; η is the scale parameter, representing the characteristic life in years, with a range of values ​​of [10, 15] and a recommended value of 12; λ(t) is the failure rate function, representing the failure probability per unit time.

[0142] And the reliability function: R(t)=exp[-(t / η)^β]; describes the aging characteristics of the equipment, where β and η are the shape parameter and scale parameter respectively; R(t) is the reliability function, which represents the probability of normal operation before time t.

[0143] The maintenance cost budget adopts a polynomial model that takes time decay into account, namely: C(t)=∑[c_i·exp(-ri·t)]+ε(t), where the basic cost item c_i is affected by the depreciation rate r_i, and the random fluctuation item ε(t) represents the uncertainty caused by market factors. c_i is the basic maintenance cost of the i-th item, in units of 10,000 yuan; r_i is the depreciation rate of the i-th item cost, with a range of [0.05, 0.15], and a recommended value of 0.1; t is a time variable, in units of years; ε(t) is a random fluctuation item, which follows a normal distribution with a mean of 0 and a standard deviation of 5% of the total cost; M(t) is the maintenance cost at time t, in units of 10,000 yuan / year; O(t) is the operating cost at time t, in units of 10,000 yuan / year; LCC is the total life cycle cost, in units of 10,000 yuan, and the integral interval is [0, expected life].

[0144] The total life cycle cost is obtained by integrating the maintenance cost M(t) and the operating cost O(t) over time, that is, LCC=∫[M(t)+O(t)]dt.

[0145] For example, for a circuit breaker that has been in operation for 5 years, the Weibull distribution parameters of β=2.8 and η=12 years are obtained through failure data analysis, and the remaining service life is predicted to be 5.8 years, and the maintenance budget for the next 3 years is estimated to be 1.85 million yuan.

[0146] S212: Push the fault analysis report to the mobile terminal of the target person.

[0147] Among them, the target personnel include equipment maintenance personnel, operation and maintenance managers and other relevant responsible persons; mobile terminals refer to portable devices such as smart phones and tablets; push methods include message notifications, emails, text messages and other channels.

[0148] The fault detection system needs to ensure that the report is delivered to the relevant personnel in a timely manner. Specifically, the fault detection system first determines the list of target personnel to be pushed according to the fault level and processing authority; then selects the appropriate push channel according to the personnel configuration information; then converts the analysis report into a format suitable for display on mobile terminals; finally, pushes the report to the target terminal through a secure communication protocol and tracks and confirms the receipt status.

[0149] In some embodiments, report push and status tracking can be achieved in a variety of ways: Optionally, the fault detection system adopts a multi-level push strategy, sets different push priorities and reminder methods according to the severity of the fault, ensures the reliability of the push through the message queue, and records the reading and processing status of the report; Optionally, the fault detection system is based on a microservice architecture to achieve high availability of the report push service, support flexible configuration of multiple push channels, and provide real-time push status feedback. It is understandable that other methods can also be used to achieve intelligent push of reports, which are not limited here.

[0150] In some embodiments, information push uses a dynamic priority scheduling algorithm. Define the message priority function: P(m)=w1U(m)+w2D(m)+w3T(m); where U(m) is the urgency of the message, D(m) is the scope of influence, and T(m) is the timeliness.

[0151] The push time interval Δt is optimized through a Markov decision process: V(s)=max_a{R(s,a)+γ∑P(s'|s,a)V(s')}; where s is the system state, a is the push action, R is the immediate reward, and γ is the discount factor.

[0152] The receiving confirmation mechanism uses a reliable transmission protocol: ACK(m)=Hash(m||t||key); Retry(m)=min(2^n·T0,T_max); Hash is the hash function, m is the message content, t is the timestamp, key is the session key, n is the number of retries, T0 is the basic timeout, T_max is the maximum timeout, and the recommended value is 300 seconds.

[0153] For example, the system detects that the temperature of a circuit breaker contact is abnormal, calculates the message priority to be 0.85, and chooses to push it to the on-duty personnel via the APP immediately, and sends a copy to the supervisor via SMS. If no confirmation is received within 15 minutes, the push level is automatically upgraded.

[0154] In some embodiments, the fault detection system will quickly locate the maintenance focus when a circuit breaker failure causes a large-scale impact; specifically, the fault detection system will extract the spatiotemporal distribution characteristics of the fault diagnosis results from the historical fault database, and then determine the associated influence coefficients of the circuit breakers in each area based on the circuit breaker location data and spatiotemporal distribution characteristics of the geographic information system; when a preset power outage occurs, the fault detection system will determine the fault spread range based on the associated influence coefficient and generate a partition maintenance priority list.

[0155] Among them, the spatiotemporal distribution characteristics represent the distribution law of faults in time and space dimensions; the correlation influence coefficient is used to quantify the degree of mutual influence between circuit breakers; and the partition maintenance priority list guides the on-site maintenance sequence.

[0156] The fault detection system needs to respond quickly to large-scale faults. Specifically, the fault detection system first analyzes the spatiotemporal pattern of historical fault data and establishes a fault propagation model; then, it combines the equipment location relationship in the geographic information system to calculate the correlation coefficient matrix between the equipment; when a major fault occurs, it immediately starts the fault impact assessment program, determines the fault propagation range based on the correlation coefficient, and generates a zoning maintenance plan.

[0157] In some embodiments, fault propagation analysis can be implemented in a variety of ways: Optionally, the fault detection system uses complex network theory to establish a circuit breaker fault propagation network model, and determines key nodes and propagation paths through network topology analysis; Optionally, the fault detection system is based on a graph neural network to learn the complex associations between devices and achieve accurate prediction of fault impacts. It is understandable that other methods can also be used to implement fault analysis, which is not limited here.

[0158] It should be noted that the fault propagation network model here adopts the graph convolutional network (GCN) architecture, which includes two sub-modules: spatial feature aggregation and time series evolution prediction. The training process uses the spatiotemporal distribution data of historical faults. The input includes the geographic location information of the circuit breaker, the physical connection relationship between the devices, and the historical fault propagation records. The fault association pattern between the devices is learned through graph convolution operations. The model uses a combined loss function of cross entropy and mean square error to optimize the calculation accuracy of the correlation influence coefficient between devices. In the case of large-scale faults, the model receives the initial fault point information, predicts the possible spread path and impact range of the fault, and outputs a partition maintenance priority list based on the severity and urgency of the fault. The advantage of the model is that it can effectively capture the fault propagation law in complex networks and support emergency decision-making.

[0159] The system analyzes the fault propagation effect by constructing a circuit breaker association graph G = (V, E). The edge weight w (i, j) comprehensively considers the topological distance and fault correlation, that is: w(i, j)=α·d(i, j)^(-γ)+β·f(i, j); d(i, j) is the physical distance between devices i and j, in kilometers (km); γ is the distance attenuation index, ranging from [1.5 to 2.5], with a recommended value of 2.0; f(i, j) is the fault correlation, ranging from [0 to 1], which is obtained through historical data statistics; α is the distance impact weight, ranging from [0.4 to 0.6], with a recommended value of 0.5; β is the correlation weight, ranging from [0.4 to 0.6], with a recommended value of 0.5.

[0160] The propagation impact intensity I (v) is obtained by weighted accumulation of the impact of each source point, that is: I(v)=∑w(u,v)·s(u)·p(u,v), where s(u) is the severity of the source fault, ranging from [0,1]; p(u,v) is the probability of fault propagation, ranging from [0,1], calculated based on historical data; w(i,j) is the edge weight, indicating the impact strength between devices; I(v) is the intensity of the propagation impact on node v, ranging from [0,1].

[0161] The maintenance priority is calculated based on the impact intensity, load importance and repair time, namely: Priority(v)=w1I(v)+w2C(v)+w3T(v).

[0162] For example, after a circuit breaker in a substation fails, the system uses the above model analysis to show that it may affect 12 devices in three surrounding areas. Through priority calculation, area 2 with the largest load is listed as the highest maintenance priority.

[0163] In some embodiments, the fault detection system will further generate a personnel maintenance plan; specifically, the fault detection system will determine the path planning table based on the partition maintenance priority list and real-time road condition information, and then determine the maintenance scheduling plan based on the maintenance personnel information, equipment supply information and the path planning table; the fault detection system will determine the partition maintenance task sequence according to the maintenance scheduling plan, and push multiple maintenance tasks in the partition maintenance task sequence to the corresponding maintenance terminal.

[0164] Among them, the maintenance scheduling plan includes personnel allocation, equipment deployment and time arrangement; the path planning table guides maintenance personnel on the optimal travel route; and the maintenance task sequence is a list of maintenance work sorted by priority.

[0165] The fault detection system needs to realize the intelligent scheduling of maintenance resources. Specifically, the fault detection system first plans the optimal maintenance route based on the maintenance priority of the zone and real-time traffic data; then generates a detailed scheduling plan based on the professional skills of the maintenance personnel, equipment inventory status and route planning; finally, the maintenance task is broken down into specific work instructions and pushed to the on-site personnel through mobile terminals.

[0166] In some embodiments, the system uses an improved A* algorithm for maintenance path planning, and the heuristic function h(n) comprehensively considers spatial distance, traffic time and congestion cost, that is: h(n)=w1d(n,goal)+w2t(n)+w3c(n). The maintenance scheduling problem is modeled as a multi-objective optimization problem with the objective function: F(x)=[f1(x), f2(x), ..., f_m(x)] includes multiple dimensions such as total repair time, personnel utilization, and equipment deployment efficiency.

[0167] Task sequence generation uses priority-based queue management. The priority calculation formula is: priority=w1u(task)+w2r(task)+w3s(task) balances task urgency, resource readiness, and system impact.

[0168] For example, after a chain failure occurred in a certain area, the system completed the optimal allocation of 15 maintenance tasks within 10 minutes through the above-mentioned planning mechanism, planned the optimal path for the three maintenance teams, and it is expected that all repair work can be completed within 4 hours.

[0169] In the embodiment of the present application, due to the use of the collaborative analysis technology of the double-break contact action timing data and the vacuum chamber pressure data, combined with the comprehensive evaluation mechanism of the ambient temperature, humidity and load index, and the introduction of the edge computing architecture and the dynamic warning threshold adjustment strategy, it is possible to achieve accurate diagnosis and rapid processing of faults, effectively solving the problems of poor environmental adaptability, low data processing efficiency and delayed fault warning of the fixed threshold warning mechanism in traditional technologies, thereby achieving the improvement of equipment reliability, effective reduction of maintenance costs, and efficient collaborative processing in large-scale fault scenarios. Through historical data mining and spatiotemporal feature analysis, the system can also predict fault development trends and optimize the configuration of maintenance resources, providing a full range of intelligent solutions for power grid operation and maintenance management.

[0170] The following describes the fault detection system in the embodiment of the present invention from the perspective of hardware processing. Figure 3 , which is a schematic diagram of a physical device structure of a fault detection system in an embodiment of the present application.

[0171] It should be noted that Figure 3 The structure of the fault detection system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0172] like Figure 3 As shown, the fault detection system includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part 308 to the random access memory (RAM) 303, such as executing the method described in the above embodiment. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, the ROM 302 and the RAM 303 are connected to each other through the bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0173] The following components are connected to the I / O interface 305: an input section 306 including an audio input device, a button switch, etc.; an output section 307 including a liquid crystal display (LCD) and an audio output device, an indicator light, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed so that a computer program read therefrom is installed into the storage section 308 as needed.

[0174] In particular, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 309, and / or installed from a removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, various functions defined in the present invention are performed.

[0175] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, apparatus, or device.

[0176] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. Each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings.

[0177] Specifically, the fault detection system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, the double-break high-voltage vacuum circuit breaker fault diagnosis method provided in the above embodiment is implemented.

[0178] As another aspect, the present invention further provides a computer-readable storage medium, which may be included in the fault detection system described in the above embodiment; or may exist independently without being assembled into the fault detection system. The above storage medium carries one or more computer programs, and when the above one or more computer programs are executed by a processor of the fault detection system, the fault detection system implements the double-break high-voltage vacuum circuit breaker fault diagnosis method provided in the above embodiment.

[0179] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present application.

[0180] As used in the above embodiments, the term "when..." may be interpreted to mean "if..." or "after..." or "in response to determining..." or "in response to detecting...", depending on the context. Similarly, the phrases "upon determining..." or "if (the stated condition or event) is detected" may be interpreted to mean "if determining..." or "in response to determining..." or "upon detecting (the stated condition or event)" or "in response to detecting (the stated condition or event)", depending on the context.

[0181] Those skilled in the art can understand that to implement all or part of the processes in the above-mentioned embodiments, the processes can be completed by computer programs to instruct related hardware, and the programs can be stored in computer-readable storage media. When the programs are executed, they can include the processes of the above-mentioned method embodiments. The aforementioned storage media include: ROM or random access memory RAM, magnetic disk or optical disk and other media that can store program codes.

Claims

1. A double-break high-voltage vacuum circuit breaker fault diagnosis method, characterized in that: Applied to a fault detection system, the method comprises: Obtain the contact action timing data of the high-voltage vacuum circuit breaker at two breaks, the pressure data in the vacuum chamber, as well as the ambient temperature, ambient humidity and system load index; Calculate the contact action speed difference and bounce characteristic parameters of the two fractures according to the contact action timing data, and generate a mechanical characteristic coordination vector in combination with the opening and closing energy consumption data of the two fractures; Acquiring temperature distribution data of the conductive loops of the two breaks, and calculating the heat dissipation coefficient of the two breaks based on the temperature distribution data; The voltage sharing ratio and transition recovery voltage characteristics of the two break points during the breaking process are collected, and the voltage tolerance difference index is determined by combining the pressure data in the vacuum chamber and the heat imbalance coefficient; Inputting the mechanical characteristic coordination vector and the voltage tolerance capability difference index into a fault diagnosis model to obtain a fault diagnosis result; Based on the ambient temperature, the ambient humidity and the system load index, the environmental impact factor of the high-voltage vacuum circuit breaker under the current working condition is calculated, a working condition characteristic vector is generated, and a fault warning threshold corresponding to the working condition characteristic vector is determined; When the warning parameter index of the fault diagnosis result exceeds the fault warning threshold, fault warning information is generated.

2. The method according to claim 1, characterized in that When the warning parameter index of the fault diagnosis result exceeds the fault warning threshold, after the step of generating fault warning information, the method further includes: Acquire multiple historical fault data including equipment number, fault type, fault degree and maintenance record, and build a historical fault database; Performing model training based on the historical fault database to obtain a fault development prediction model; Inputting the fault warning information into the fault development prediction model to obtain fault timing characteristics; Generate a fault analysis report based on the fault timing characteristics; Push the fault analysis report to the mobile terminal of the target person.

3. The method according to claim 2, characterized in that The step of generating a fault analysis report based on the fault timing characteristics specifically includes: Determining the expected service life of the high-voltage vacuum circuit breaker according to the fault timing characteristics; Determining a maintenance cost budget according to the expected service life and the maintenance records in the historical failure database; A failure analysis report is generated based on the expected service life and the maintenance cost budget.

4. The method according to claim 1, characterized in that After the step of obtaining the contact action timing data of the high-voltage vacuum circuit breaker at two breaks, the pressure data in the vacuum chamber, and the ambient temperature, ambient humidity and system load index, the method further includes: Determine an edge computing node closest to the location of the high-voltage vacuum circuit breaker; Integrating the contact action timing data, the pressure data in the vacuum chamber, the ambient temperature, the ambient humidity and the system load index into an original data packet; Sending the original data packet to the edge computing node; Receive the fault calculation result data packet returned by the edge computing node.

5. The method according to claim 4, characterized in that After the step of receiving the fault diagnosis result and the fault warning threshold returned by the edge computing node, the method further includes: Extracting the computing node mark and data timestamp in the fault computing result data packet; When the data timestamp is inconsistent with the original timestamp of the original data packet, a data acquisition request including the original timestamp is sent to the edge computing node.

6. The method according to claim 1, characterized in that When the warning parameter index of the fault diagnosis result exceeds the fault warning threshold, after the step of generating fault warning information, the method further includes: Extracting the spatiotemporal distribution characteristics of fault diagnosis results from the historical fault database; Determine the correlation influence coefficient of the circuit breakers in each area according to the circuit breaker location data of the geographic information system and the spatiotemporal distribution characteristics; When a preset power outage event occurs, the fault propagation range is determined according to the associated impact coefficient, and a partition maintenance priority list is generated.

7. The method according to claim 6, characterized in that When a preset power outage event occurs, after the step of determining the fault propagation range according to the correlation influence coefficient and generating a partition maintenance priority list, the method further includes: Determine a path planning table based on the partition maintenance priority list and real-time traffic information; Determine a maintenance scheduling plan based on maintenance personnel information, equipment supply information and the path planning table; A partition maintenance task sequence is determined according to the maintenance scheduling plan, and multiple maintenance tasks in the partition maintenance task sequence are pushed to corresponding maintenance terminals.

8. A fault detection system, characterized in that: The fault detection system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the fault detection system to execute the method described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instructions are executed on a fault detection system, the fault detection system is caused to execute the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product runs on a fault detection system, the fault detection system is caused to perform the method according to any one of claims 1 to 7.

Images