Active testing method and system for circuit breaker

Abstract

The present invention relates to the technical field of active testing of circuit breakers, and discloses an active testing method and system for circuit breakers, which are used to solve the problem that existing active testing methods are difficult to achieve adaptive threshold adjustment. The method extracts time-domain and frequency-domain features by performing time-frequency analysis on contact velocity signals, and obtains a thermomechanical coupling feature vector by combining heat flow analysis of a thermopile array. A thermomechanical coupling dynamic graph is constructed with sensors and components as nodes and inverse thermal conductances as edge weights, and the structure is updated in real time. Node drift is predicted using a graph convolutional network. The threshold is adaptively adjusted by integrating environmental derating and predicted drift. The thermoelectric coupling component of the fast-changing contact velocity component and the rate of change of the heat flow refraction angle is extracted to generate a reverse polarity pulse compensation signal to suppress interference. Time synchronization is then achieved through secondary alignment with IEEE1588 and the cloud, closed-loop feedback of data and threshold adjustment results, and dynamic updating of graph model node weights and threshold rules to ensure synchronization with equipment health and achieve accurate early warning and protection.

Description

Technical Field

[0001] The present invention relates to the technical field of active testing of circuit breakers, and in particular to an active testing method and system for circuit breakers. Background Art

[0002] Circuit breakers are crucial protective devices in power systems. They have overload protection, short-circuit protection, undervoltage protection, overvoltage protection, and isolation functions. They can quickly cut off abnormal currents through thermal tripping, electromagnetic tripping, or electronic protection mechanisms. In existing active testing and online monitoring solutions for circuit breakers, the industry's common practice generally relies on pre-set fixed alarm or evaluation thresholds to trigger tests or judge status. The thresholds are often given once in the laboratory or at the initial stage of operation based on experience, standard provisions, or a small number of samples. They are difficult to adjust in real time as the working conditions and environment change throughout the life cycle of the equipment, resulting in a long-standing problem of difficulty in adapting the real-time thresholds. For example, a method and system for online monitoring of intelligent circuit breakers with the publication number (CN118604601A) proposes to achieve intelligent monitoring by collecting circuit breaker working data and generating action characteristics and risk assessment results. However, its monitoring triggering and status determination still use preset thresholds as the core control quantity; at the same time, the operating environment and structural status of the circuit breaker show significant fluctuations throughout its life cycle. Changes in ambient temperature, control cabinet temperature, contact and oil tank temperature will change the measurement baseline and characteristic curve. If these external factors and historical operating condition data are not considered at the same time, a single fixed threshold will deviate from the actual health level and it will be difficult to guide refined active test triggering. Because the threshold setting lacks the ability to dynamically respond to multi-factor coupling, long-term degradation laws and operating mode switching, coupled with uncertainties such as manufacturing differences, inconsistent maintenance quality and sensor drift, existing active test systems often experience trigger lags, frequent interruptions to normal power supply, or generate a large number of invalid test records, which in turn affects the accuracy of maintenance plans and the economy and safety of power grid operations.

[0003] In summary, due to the static threshold setting and the lack of real-time compensation for multiple factors such as the environment, load, and mechanical fatigue, existing active testing methods are difficult to strike a balance between sensitivity and reliability, which not only leads to false triggering, power disturbances, and increased operation and maintenance costs, but also conceals major hidden dangers and increases the risk of system power outages. An active testing method and system that can achieve adaptive threshold adjustment is urgently needed to solve the above-mentioned chain of problems. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides an active testing method and system for a circuit breaker to solve the problem that the existing active testing method is difficult to achieve threshold adaptive adjustment.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An active testing method for a circuit breaker, comprising:

[0007] S1: Without power outage, a high-precision sensor array is used to synchronously collect the multi-dimensional operating parameters of the circuit breaker and perform derating normalization processing on the temperature and humidity signals;

[0008] S2: Based on the standardized input data, a time-frequency joint analysis is performed on the contact opening and closing speed signal to extract time and frequency domain features. The heat flux refraction angle is calculated based on the sensor thermal conductivity distribution and then correlated with the coil current time series to obtain the thermal-mechanical coupling eigenvector.

[0009] S3: Construct a thermo-mechanical coupling dynamic graph with sensor locations and key components of the circuit breaker as nodes and the inverse of the equivalent thermal conductance between nodes as edge weights. Use online graph reasoning to output nodes and predict critical drift.

[0010] S4: The environmental derating curve and the predicted critical drift are weighted and integrated. The weights are adaptively adjusted based on the historical trends of contact-related parameters. The new threshold is calculated in real time and updated with the health of the circuit breaker.

[0011] S5: During the threshold interference period, the thermoelectric coupling component that is linearly related to the rate of change of the heat flux refraction angle is extracted from the fast-changing component of the contact opening and closing speed signal, and a small voltage pulse of equal amplitude and opposite polarity is generated and synchronously injected into the signal circuit;

[0012] S6: Local sub-microsecond synchronization is achieved through IEEE1588 precision clocks. Cross-device timestamps are corrected with cloud-based secondary alignment. The closed-loop compensated operating data and threshold adjustment results are continuously fed back to the coupled dynamic graph model to update node weights and threshold generation rules.

[0013] Furthermore, step S2 includes:

[0014] The synchronously collected contact opening and closing speed, coil current, and various temperature and humidity parameters are zero-referenced and amplitude normalized according to the preset range to eliminate the dimensional differences between different sensing channels;

[0015] The standardized contact opening and closing speed signal is first segmented into time windows. Then, time domain envelope extraction and frequency domain amplitude-phase statistics are synchronously performed on each signal segment to obtain the time characteristic quantity reflecting the travel smoothness and the frequency characteristic quantity reflecting the mechanical impact intensity.

[0016] Based on the thermal conductivity distribution of the sensor at each installation point inside the circuit breaker, the refraction angle of the heat flow in the XY plane within the same time window is calculated as the heat transfer direction feature;

[0017] The heat flux refraction angle is mapped one-to-one with the corresponding coil current time series according to the timestamp, and is combined with the time characteristic quantity and frequency characteristic quantity to form a multidimensional vector to form a characteristic direction star set that characterizes the thermal-mechanical coupling state.

[0018] Furthermore, step S3 includes:

[0019] The various sensors installed inside the circuit breaker body and the key mechanical, electrical and heat dissipation components of the circuit breaker are arranged according to their physical locations - corresponding to the nodes of the dynamic graph;

[0020] The inverse of the equivalent thermal conductivity value of the heat transfer channel between nodes is calculated and used as the weight of the edge;

[0021] The node sensor data is mapped into a time series snapshot with variable weights at a preset sampling period, and the real-time operation feature vector of the node is synchronously recorded in each snapshot;

[0022] The dynamic graph inference process deployed on the edge processor is called to perform sliding window calculations on continuous snapshots and output the predicted critical drift of each node.

[0023] Furthermore, step S4 includes:

[0024] Real-time reading of the ambient temperature, humidity, air pressure and busbar load current at the circuit breaker location, mapping the data to the pre-stored environmental derating curve to obtain the compensation coefficient;

[0025] Call the online graph inference results to obtain the predicted critical drift for each node;

[0026] Retrieve the time series of historical operating parameters related to the contacts and calculate the recent change trends of each parameter.

[0027] Furthermore, according to the magnitude and direction of the change trend, weights are assigned to the environmental compensation coefficient and the predicted critical drift respectively, and the weight values ​​are automatically adjusted up or down along with the trend of the contact parameters;

[0028] The environmental compensation coefficient and the predicted critical drift amount are weightedly integrated according to the weight to obtain a threshold increment;

[0029] The threshold increment is added to the base threshold to generate a new real-time threshold, and the threshold is stored in the local database along with the health score.

[0030] Furthermore, step S5 includes:

[0031] Taking the threshold interference period as the boundary, a synchronous time window is selected for the contact opening and closing speed signal, and the fast-changing component is extracted.

[0032] The fast-changing component is time-aligned with the rate of change of the heat flux refraction angle in the same time window. The signal section where the two are linearly correlated is identified by comparing them with the pre-calibrated sensitive section, and the thermoelectric coupling component to be compensated is determined.

[0033] Furthermore, the amplitude and polarity of the thermoelectric coupling component are measured, and a small voltage pulse of equal amplitude, opposite polarity and duration 1 to 3 times the thermal expansion relaxation time is generated;

[0034] A small voltage pulse is injected into the contact signal circuit in phase synchronization with the thermoelectric coupling component through a precision voltage source, so that the two cancel each other out in amplitude.

[0035] The difference in threshold fluctuation before and after compensation is recorded, and the compensation result is fed back to the threshold adaptive module to correct the pulse amplitude and phase of the next cycle.

[0036] Furthermore, step S6 includes:

[0037] Step 1: Configure the circuit breaker monitoring terminal as an IEEE1588 precision time protocol slave clock and synchronize it with the master clock in the substation using hardware timestamps to achieve sub-microsecond time synchronization.

[0038] Step 2: Obtain a reference time scale from the Global Positioning System (GPS) or an equivalent high-stability reference in the cloud-side data center, and perform secondary correction on the timestamps of the data packets uploaded by each monitoring terminal to eliminate cross-device and cross-network segment propagation delays.

[0039] Step 3: Feedback the contact stroke, coil current, threshold adjustment results and other operating data after secondary correction to the local edge processor in real time;

[0040] Step 4: Based on the feedback data, the node weights of the thermal-mechanical-electrical coupling dynamic graph are adjusted incrementally or incrementally, and the threshold generation rules are simultaneously refreshed so that the model and thresholds evolve dynamically with the health of the circuit breaker.

[0041] Step 5: In subsequent sampling cycles, the timing errors of the old and new threshold judgment results and the actual action timing are compared. When the error exceeds the preset limit, the resynchronization process of steps 1 and 2 is automatically triggered until the time scales are consistent.

[0042] In another aspect, the present invention provides an active testing system for a circuit breaker, comprising:

[0043] The sensing and acquisition module is used to synchronously obtain multi-dimensional signals of coil current, contact stroke and speed, shell and contact temperature, ambient temperature and humidity, and bus current to form unified time-based raw health data;

[0044] The feature analysis module is used to perform time-domain envelope and frequency-domain amplitude-phase statistics on the contact opening and closing speeds, separating high-frequency impact, low-frequency inertia, and steady-state displacement characteristics;

[0045] The coupled modeling module is used to construct a thermo-mechanical dynamic diagram with variable weights as the temperature field, using sensor points and key components such as coils, contacts, and arc extinguishing chambers as nodes and taking the inverse of the thermal conductance of the two nodes as the edge weights.

[0046] The threshold generation module is used to adaptively weight the environmental derating curve and the predicted drift according to historical trends, output a real-time threshold, and update it in real time with the health level;

[0047] The compensation output module is used to identify the thermoelectric coupling component within the interference time window and inject voltage pulses of equal amplitude and opposite polarity to offset thermal drift and record the compensation effect;

[0048] The clock calibration module is used to monitor the terminal's local time synchronization using the IEEE1588PTP slave clock, solving sub-microsecond synchronization of equipment within the station and performing secondary alignment using a high-stability reference on the cloud side.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] After the sampling units, protection and measurement devices, mechanical sensors and edge controllers in the station are locked to the same sub-microsecond time base through the IEEE1588 precision clock, each local data frame has an absolute timestamp. Subsequently, the cloud dispatch center regularly collects the "master clock-slave clock drift" logs of each station, and uses the timestamp mapping-playback and congestion ranging methods to perform secondary alignment on the cross-station data, converge the residual error, and solve the causal dislocation caused by asynchronous reporting of multiple circuit breakers. Then, the edge end combines the contact speed, thermal refraction angle and dynamic threshold after closed-loop compensation into a multi-mode digital The data package is pushed to the cloud side, and the cloud side then periodically writes this batch of data back to the coupled dynamic graph model. On the one hand, the current feature vector of the node is updated to , and on the other hand, the thermal resistance weight Wᵢⱼ between nodes is refreshed in real time based on the temperature and humidity distribution and power flow changes of the entire network as calculated by the cloud side. The model weights evolve accordingly, making the graph reasoning network sensitive to the "real thermal-mechanical-electrical channel" without overfitting the old weights. After such closed-loop iterations, false alarms caused by data misalignment are significantly reduced. Compared with traditional monitoring, the false alarm rate is reduced. At the same time, the subsequent threshold can be adaptively adjusted and remain stable.

[0051] The local IEEE1588 precise clock synchronizes sensors, switches, and protection IEDs at the sub-microsecond level, significantly reducing the millisecond error compared to traditional NTP, ensuring accurate sequencing of mechanical, electrical, and thermal events. Secondary alignment in the cloud uses the sliding window maximum correlation method to correct cross-site RTU time drift, resolving deviations introduced by multi-level forwarding and avoiding asynchrony between threshold triggering and actual actions due to timestamp misalignment.

[0052] By writing the "threshold tightening range, environmental coefficient, and drift coefficient" back into the graph model node weights, the graph neural network can be retrained instantly, and the health index curve scrolls in real time with the equipment status to avoid the static asset health index being slow to respond to sudden failures. Engineering practice shows that the fused H curve can show an inflection point 2 to 3 weeks before a failure, which is at least one maintenance cycle faster than regular offline testing, supporting proactive maintenance and precise resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 This is a flow chart of an active testing method for a circuit breaker according to the present invention;

[0054] Figure 2 The figure is a structural diagram of an active testing system for a circuit breaker according to the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] Example 1: Figure 1 The present invention provides an active testing method for a circuit breaker, comprising:

[0057] S1: Under the condition of power outage, a high-precision sensor array is used to synchronously collect the multi-dimensional operating parameters of the circuit breaker and perform derating normalization processing on the temperature and humidity signals. The specific implementation is as follows:

[0058] Through current sensors, displacement sensors and temperature and humidity sensors, the current signal, displacement signal and temperature and humidity signal of the target device are synchronously collected as data support. The current sensor should be installed at the input and output ends of the circuit breaker, that is, in the wires of the power input end and the load output end, to ensure that the sensor can accurately measure the current passing through the circuit breaker. The displacement sensor should be installed on the moving parts of the circuit breaker contacts, usually installed at the connection between the contacts and the operating mechanism. During installation, it is necessary to ensure that the measuring axis of the sensor is consistent with the movement direction of the contacts to accurately measure the displacement of the contacts. The temperature and humidity sensor is installed in a representative position of the environment where the circuit breaker is located, usually on the top or side of the circuit breaker cabinet, away from heat and humidity sources. During installation, ensure that the air inlet of the sensor is unobstructed to ensure that it can accurately sense the temperature and humidity changes in the environment. The installation position should avoid being affected by external airflow and direct sunlight to ensure the stability and accuracy of the measurement.

[0059] During the data collection process, the IEEE1588 precision time protocol is used to synchronize the time of all sensors to ensure that the data of each sensor is collected under the same time reference. For example, the synchronization period of the master clock source is set to 1 second, and the clock offset compensation threshold is set to ±10 milliseconds. When the threshold is exceeded, the clock resynchronization is triggered. The collected multi-dimensional operating parameter data is uploaded to the cloud monitoring platform through the communication protocol.

[0060] Before data is uploaded, the temperature and humidity signals are derated and normalized. First, the digital temperature and humidity probes installed on-site inside the switchgear continuously output real-time environmental values. These raw points undergo rapid quality screening. Out-of-range readings, readings with rapid transitions, and readings with large deviations from adjacent redundant probes are marked as abnormal and discarded or replaced with the average of the most recent credible window. This is used to filter out false signals caused by sensor drift, poor contact, or transient airflow shock, thereby preventing environmental glitches from directly affecting subsequent threshold calculations. Furthermore, based on industry maintenance experience and literature on the effects of high temperature, humidity, and condensation on insulation and conductive joint temperature rise, as well as switch reliability, performance reduction ratios are set for different temperature and humidity ranges. For example, for every 10°C increase in temperature or 20%RH increase in humidity, the current carrying capacity and diagnostic thresholds are adjusted downward by a certain percentage based on experience.

[0061] During operation, the controller receives real-time temperature and humidity, looks up the corresponding derating factor in the database, and scales or rescales the original measured values ​​of coil current, contact stroke, opening and closing speed, and contact resistance temperature rise collected synchronously in the same time slice according to the factor to the equivalent value under the nominal base point. In this way, the environmental differences between seasons and sites are devalued, and the online trend and threshold comparison return to the same benchmark. To maintain long-term accuracy, the system continuously writes the "environmental value, performance indicator of the same period, and whether to trigger an alarm" triplet into the local database and uploads it to the cloud regularly. The cloud makes small step corrections to the derating table nodes based on the accumulated samples. When the offset exceeds the set tolerance, the new version of the derating coefficient is issued, which automatically takes effect after the on-site hot update without power outage.

[0062] S2: Based on the standardized input data, a time-frequency joint analysis is performed on the contact opening and closing speed signal to extract the time and frequency domain features. The heat flux refraction angle is calculated based on the sensor thermal conductivity distribution and then correlated with the coil current time series to obtain the thermal-mechanical coupling eigenvector. The specific implementation is as follows:

[0063] The displacement of the circuit breaker drive or contact moving guide rod is recorded continuously over time. The first-order difference of this displacement curve is performed in the instrument to obtain the velocity curve, that is, the contact opening and closing velocity signal, and the standardized velocity waveform is obtained. After that, short-time Fourier transform is performed in combination with envelope extraction within the 32ms sliding window;

[0064] The wide window focuses on the low frequency band of 0-200Hz, outputs the time domain characteristics of peak velocity and zero crossing number, which are used to judge whether the spring energy storage is full. The narrow window locks the high frequency band of 500-1500Hz, outputs the energy center of mass and pulse coefficient frequency domain characteristics, which are used to capture hinge collision, jamming or multiple rebounds. In the same time window, in the circuit breaker cabinet without power outage, the micro thermopile array evenly pasted on the inner wall of the shell samples and outputs the temperature point cloud at 5kHz; first, the discrete temperature field is collected in real time using the thermopile array pasted on the surface of the circuit breaker shell , and then calculate the temperature gradient at each sensing point according to Fourier's law Finally, the inner wall of the shell or the plane where the sensor array is located is regarded as the XY reference plane, and the longitudinal direction of the shell or the design reference line is defined as the baseline direction , then the instantaneous heat flux density vector can be obtained by Fourier's law: in: is the instantaneous temperature field at any point on the plane, is the instantaneous temperature field at any point on the plane, is the instantaneous temperature field at any point on the plane;

[0065] And through Fourier's law we get: in is the heat flux vector, is the thermal conductivity of the material, is the temperature gradient;

[0066] Then After projecting onto the plane, we get the components ), heat flow refraction angle Defined as Deflection angle relative to the tangent direction: = when =0°, the heat flow is strictly conducted along the designed heat dissipation channel, indicating that the thermal resistance of each component of the mechanism is balanced. When the temperature is greater than 0° and continues to increase, heat is refracted in the horizontal direction, indicating a sudden increase in thermal resistance or migration of heat sources. The first-order rate of change = / t, can map the thermal disturbance propagation speed, for subsequent linear pairing with the contact high-frequency velocity center of mass, extract the thermal-mechanical coupling component, and then With the aforementioned high-frequency energy center of mass When the machine is in good condition, the two are highly positively correlated. However, once the lubrication fails or there is uneven wear, the refraction angle will increase suddenly. The change is not big, the correlation drops suddenly into an abnormal scatter band, and then the coil current The differential peak is used as the time anchor, and the time within 5ms after the peak is The point marks the electromagnetic thrust section, and the peak front or empty window marks the inertia / rebound section, and then we get the three-element coupling vector. Stacking them in chronological order, we get a thermal-mechanical coupling characteristic vector sequence, which can be directly sent to the subsequent dynamic diagram diagnosis module.

[0067] S3: A thermo-mechanical coupling dynamic graph is constructed with sensor locations and key components of the circuit breaker as nodes and the inverse of the equivalent thermal conductance between nodes as edge weights. The critical drift is predicted by calling the online graph inference output node. The specific implementation is as follows:

[0068] The controller first distributes the temperature, displacement, vibration, and current sensors and treats them as nodes together with the key structural parts of the circuit breaker. Then, the equivalent thermal conductivity obtained by finite element thermal analysis Establish heat flow channels between node pairs and then take their inverses As the edge weight, the larger the value, the more difficult it is for heat flow to pass through this edge. Then the system reads the temperature point cloud every 100ms. , use Kriging interpolation to reconstruct the continuous temperature surface and update it in real time For example, if the contact area of ​​the connecting rod and the contact pair is reduced by 20% due to wear, the effective thermal conductivity will be automatically reduced in proportion. The potential heat dissipation bottleneck is mapped to the graph structure in real time by raising it accordingly. According to the 3D assembly drawing of the circuit breaker, the moving contact, connecting rod, coil, arc extinguishing chamber shell and the position of the attached temperature probe are discretized into several rectangular or cylindrical units, and the geometric dimensions of each unit are marked. , cross-sectional area and thermal conductivity of the material , and then according to Fourier's law, the series path is written as And use the contact thermal resistance of experience or bench calibration for parallel surface flow, bolt contact surface, etc. For the air gap between the moving contact and the shell, it is also necessary to convert the gas heat conduction, natural convection and radiation into an equivalent thermal resistance branch, and then summarize all series and parallel thermal resistances into a single thermal resistance between nodes i and j according to the circuit superposition method. , and the adjacency matrix is ​​written by the controller when it is first powered on During operation, after sampling the temperature field every 100ms, the system will use Kriging interpolation to calculate the local temperature gradient. If the temperature difference and heat flow direction of a certain section of material show that the thermal bottleneck is increasing, the corresponding , so that the coupling dynamic diagram can be updated in real time with the structure’s heat dissipation capacity or contact aging;

[0069] In the same time window, the contact high-frequency velocity centroid, coil current peak, shell vibration RMS, and electromechanical domain characteristics are written into the respective node attributes to form a frame time mark The thermo-mechanical coupling dynamic graph, the vertex set contains sensors and structural parts, the edge set is Weighted, next frame After the update and new features are written, the previous frame is automatically replaced. The sliding window stack becomes a time series graph stream. The back-end online graph reasoning module inputs the latest 10 frames into the convolution-autoregressive unit and first uses the adjacency matrix Perform weighted aggregation to extract the thermal-mechanical coupling propagation path, and then predict the node migration. For example, the temperature rise rate of the moving contact node is predicted to exceed Or its thermal engine correlation coefficient is expected to fall below 0.5, in which case the warning light will be turned on in the diagram and the shortest thermal resistance path will be output for operation and maintenance positioning;

[0070] The system samples the ten frames of dynamic images in the last second. After stacking in chronological order and feeding into the lightweight spatiotemporal graph convolutional network, the model first uses the adjacency matrix at that moment Perform weighted aggregation in the spatial dimension to strengthen the heat engine propagation path with the minimum thermal resistance path. Then, use gated loops or one-dimensional convolution in the time dimension to capture the short-term gradients and high-order trends of node features to achieve modeling of temporal evolution under dynamic topology. Finally, the fully connected output only produces a row of vectors. , where each component represents how much temperature rise, vibration RMS increase, or travel distance each node is expected to have in the next frame. This quantified result can be directly compared with the design margin to determine whether to set the node to a warning;

[0071] After the graph inference engine receives the coupled dynamic graph stream of the last 10 frames, it first takes the updated thermal resistance matrix As a convolution kernel, the graph convolution layer uses The temperature, refraction angle, and velocity centroid characteristics of each node are weighted and summed along the path with minimum thermal resistance;

[0072] Next, the temporal convolution / gated recurrent unit captures the slope, peak interval, and drift trend of node features over time within a 1-second window. The final fully connected layer no longer outputs the entire graph, but instead gives a concise vector. Each component corresponds to the node in the next frame. For example, the current temperature rise rate of the moving contact node is 10K / min, and the concise vector given by the model is +7K / min, while the design limit only allows 15K / min. Therefore, the system immediately lights the node in red in the graph and returns the shortest thermal resistance chain (moving contact → connecting rod → coil) discovered during the reasoning process, allowing the on-duty personnel to understand the actual diffusion path of the thermal anomaly.

[0073] S4: The environmental derating curve and the predicted critical drift are weighted and integrated. The weights are adaptively adjusted based on the historical trends of contact-related parameters. The new threshold is calculated in real time and updated with the health of the circuit breaker. The specific implementation is as follows:

[0074] When the latest environmental value is obtained from the temperature and humidity probe, the controller will immediately refer to the pre-stored two-dimensional derating table based on IEC standards and manufacturer tests to determine the environmental derating factor. For example, 0.88 means that the current threshold should be reduced by 12% compared to the nominal value to cope with thermal stress. After each action cycle, the system will predict how much the thermal / force state of the next window node will change based on the characteristics of the current node. The system will compare the predicted value with the design limit or the historical 95% quantile to obtain a drift coefficient between 0 and 1. , used to indicate the intensity of the trend that the contact, connecting rod, and coil equipment may approach the critical value in the next time period. For example, if the temperature rise is predicted to increase by another 8K / min, and the allowable limit is 15K / min, then ;

[0075] The system introduces adaptive weight adjustment based on historical trends. It automatically monitors the slope changes of the contact operation behavior in the past 180 days, and accumulates the key features of the contact peak speed, rebound amplitude, and stroke end point every day. It uses a sliding window to calculate its short-term slope, and then compares the short-term slope with the long-term slope to obtain the change rate of the past 180 days. ,like , indicating that the rate of change is stable, indicating that the current state of the contact is stable, the system tends to refer to more environmental influences, if , indicating accelerated equipment aging, increases the drift weight. The controller then integrates the environmental and drift effects, linearly integrating the weights to calculate a new real-time threshold. This value replaces the original design threshold and is directly written into the circuit breaker logic processing module. For example, if the static contact travel threshold is designed to be 16mm, and the contact drift is evident in a high-temperature, high-humidity environment, the real-time threshold may be adjusted to approximately 9.5mm, thereby preemptively amplifying potential fault signs and triggering a warning as required.

[0076] S5: During the threshold interference period, the thermoelectric coupling component that is linearly related to the rate of change of the heat flux refraction angle is extracted from the fast-changing component of the contact opening and closing speed signal, and a small voltage pulse of equal amplitude and opposite polarity is generated and synchronously injected into the signal circuit. The specific implementation is as follows:

[0077] On a static test bench, the contact and its surrounding structure are placed in a vacuum insulation cover, and a 50W chip heater is installed outside. The heater generates a constant heat flow through a constant current source, and two miniature K-type thermocouples are used to measure the temperature difference between the contact and the reference end. T, and use a high-resistance digital multimeter to measure the thermopile output voltage in real time V, recorded when the system reaches steady state ( T, V) point, linear fit for all data, slope = is the thermoelectric coupling coefficient;

[0078] Contact opening and closing speed signal First decompose and extract the fast-changing component, capture the high-frequency impact section and the contact rebound section, and then simultaneously calculate the first-order derivative of the heat flow refraction angle to obtain the refraction angle change rate. The controller uses the envelope peak and high-frequency energy centroid within the sliding window as quantitative indicators representing the fast-changing segment of the signal, and aligns them with the sliding window time to generate a time-varying Sequence, and perform linear correlation analysis within the same time window:

[0079] If the fast variation Heat flow refraction angle If the Pearson correlation coefficient is ≥ 0.7, the component is considered to be a thermoelectric coupling component. ,Right now in is the thermoelectric coupling coefficient;

[0080] Generates a compensation voltage of equal magnitude but opposite polarity in the signal loop ,Right now in is the system impedance calibration value;

[0081] The duration of the compensation pulse is consistent with that of the fast-changing component to filter out the interference of the coupling component on the trigger threshold to the greatest extent possible, thus ensuring a balance between accuracy and compensation effect;

[0082] pulse When injected into the contact speed signal loop or the signal amplifier input at the same time position as the original signal, the system will measure the change in threshold fluctuation before and after compensation of the contact trigger threshold fluctuation before and after the injection, that is, observe whether the alarm or logic judgment level is suppressed by the interference component. If the change in threshold fluctuation before and after compensation is reduced by ≥30%, the compensation is considered effective. Otherwise, the system will automatically adjust the system impedance calibration value or thermoelectric coupling coefficient and try again until the compensation effect is stable.

[0083] S6: Local sub-microsecond synchronization is achieved through IEEE 1588 precision clocks. Cross-device timestamps are corrected using cloud-based secondary alignment. The closed-loop compensated operating data and threshold adjustment results are continuously fed back to the coupled dynamic graph model to update node weights and threshold generation rules. The specific implementation is as follows:

[0084] All monitoring units first complete sub-microsecond local alignment through the master-slave mechanism of IEEE1588 precision clock hardware stamping. The switch operates in transparent clock or boundary clock mode, so that the transmission and reception delays of frames such as temperature, displacement, and current are written into the correction field in real time, locking each sampled message to the nanosecond scale under the same second pulse. Then, the edge gateway packages the local timestamps and original message delay values ​​of the last thousand frames every thirty seconds and uploads them to the cloud-side unified time scale platform. The cloud first aligns according to the principle of homologous power frequency events, and then uses the dynamic time warping algorithm to fine-tune the alignment error, and feeds back the residual component as a correction vector. During downlink, the gateway writes the correction vector returned by the cloud into the local clock offset register, and gradually compensates it within several synchronization cycles by adding phase and microstepping, ultimately achieving absolute synchronization of the nodes within the station and the unified cloud time scale of less than 500ns. The original timestamps reported by each station node are first uniformly coarsely calibrated at the millisecond level by GPS. Then, the cloud scheduler compares the key event IDs reported by adjacent stations or different devices at the same station in the past 5 minutes to calculate the arrival delay difference to the cloud. ;

[0085] The scheduler sends a secondary alignment offset : To the corresponding edge controller, the edge end will uniformly decrement all subsequent local timestamps , eliminate the drift of hundreds of microseconds caused by cross-site routing asymmetry. This step is performed every five minutes; if three consecutive rounds The execution cycle is extended to 30 minutes, reducing cloud-edge communication overhead to achieve "secondary alignment";

[0086] The edge controller collects compensated speed, temperature rise, and vibration vectors every 100ms And the dynamic threshold just calculated , together with the unified timestamp corrected by the cloud Encapsulate them into data frames;

[0087] Then Write the corresponding node characteristics, Writes to the threshold channel and generates a threshold offset ; when <0.9, indicating that the threshold is tightened, and the system weights the node thermal resistance edge , making the graph convolution more sensitive to the change of heat flux at this moment. >1.05, indicating that the threshold is relaxed, then press The node's residual safety margin indicator is amplified to avoid false alarms. Through this threshold-edge weight linkage, the model automatically closes the network when temperature and humidity shock or aging is accelerated, and releases the network when stable conditions occur, keeping pace with the healthy evolution of assets.

[0088] All measurement frames, along with this precise timestamp, are then re-aligned across all devices every 10 minutes. A distributed alignment algorithm is used to cross-correlate the same physical events recorded by different acquisition sources, generating the residual time difference between devices. A sliding window least squares algorithm is then used to fit the accumulated time difference curves of each device to generate a correction vector. This correction vector is then sent to the corresponding acquisition board's time register via the edge API, ensuring that the local clock does not drift by more than 2 microseconds on any other node even after long periods of operation.

[0089] All operating data after closed-loop compensation and real-time thresholds All of these flows are returned to the edge controller with the same "cloud-edge secondary alignment" timestamp. The controller pushes them as incremental samples into the thermo-electromechanical coupling dynamic graph model. Node features are directly replaced with the latest temperature rise and displacement drift, and edge weights are recalculated based on the new temperature difference. The threshold generation rules are simultaneously written into the graph network's loss function weights, allowing the model to prioritize paths that have just drifted in the next round of inference. Node weights and thresholds are dynamically tightened or relaxed based on the actual health status. When the cloud detects that the temperature rise slope of a contact has doubled within 30 days, the corresponding edge weight and drift coefficient are automatically increased. Once lubrication maintenance is completed and the drift trend subsides, the system will lower the weight in the next round of threshold calculation to avoid over-conservatism.

[0090] After completing the precise timing, the controller labels the closed-loop compensated speed, temperature rise and real-time adjusted threshold together and continuously transmits them to the thermo-mechanical coupling dynamic graph model for node feature update. The edge feature continues to use the inverse of thermal conductivity as the weight. The model adopts the graph attention network idea to allow the edge weight to adaptively amplify the channel that is more sensitive to the next drift prediction during the training process, that is, the weight adaptive mechanism. When there is a residual between the latest batch of real sampling values ​​and the previous frame model prediction, the system performs gain correction on the edge that generates the residual according to the size of the residual. If the prediction error is continuously higher than 5%, the edge weight is adjusted. Multiply At the same time, the cloud-based asset management platform will synchronize the long-term health index back to the model. After receiving the signal that the health index is decreasing, the model will tighten the overall threshold generation rules. For example, the warning coefficient is adjusted from 0.85 to 0.78, and it is written back to the protection logic chip synchronously, so that the threshold and health level can resonate in real time. The system not only solves the time-scale drift across devices, but also allows the edge-node weight to be updated online adaptively according to the actual drift. At the same time, due to the double-layer correction of timestamps, the cross-cabinet synchronization error is reduced by 91%, so that the parallel waveforms of multiple circuit breakers in the same field can be compared within 0.1ms, ensuring that the threshold generation rules always follow the dynamic adjustment of the health status of the circuit breakers.

[0091] This embodiment uses a high-precision sensor array to synchronously collect multi-dimensional parameters such as circuit breaker current, displacement, temperature, and humidity without power outages. After IEEE1588 time synchronization, the temperature and humidity signals are derated and normalized, and thresholds are dynamically adjusted based on environmental factors to update the de-rating table. Time-frequency analysis is performed on the contact velocity signal to extract time and frequency domain features. This is combined with thermal flow analysis of the thermopile array to obtain a thermo-mechanical coupling feature vector. A dynamic thermo-mechanical coupling graph is constructed with sensors and components as nodes and the inverse of thermal conductance as edge weights. The structure is updated in real time, and a graph convolutional network is used to predict node drift and output warnings. The threshold is adaptively adjusted by integrating environmental derating with predicted drift. The thermoelectric coupling component of the fast-changing contact velocity component and the rate of change of the heat flux refraction angle are extracted to generate a reverse polarity pulse compensation signal to suppress interference. Sub-microsecond time synchronization is then achieved through secondary alignment between IEEE1588 and the cloud. Closed-loop feedback of data and threshold adjustment results is used to dynamically update the graph model node weights and threshold rules, ensuring synchronization with device health and achieving precise warning and protection.

[0092] Example 2: Figure 2 A structural diagram of an active testing system for a circuit breaker according to the present invention is provided. The active testing system for a circuit breaker comprises:

[0093] The sensing and acquisition module is used to synchronously obtain multi-dimensional signals of coil current, contact stroke and speed, shell and contact temperature, ambient temperature and humidity, and bus current to form unified time-based raw health data;

[0094] The feature analysis module is used to perform time-domain envelope and frequency-domain amplitude-phase statistics on the contact opening and closing speeds, separating high-frequency impact, low-frequency inertia, and steady-state displacement characteristics;

[0095] The coupled modeling module is used to construct a thermo-mechanical dynamic diagram with variable weights as the temperature field, using sensor points and key components such as coils, contacts, and arc extinguishing chambers as nodes and taking the inverse of the thermal conductance of the two nodes as the edge weights.

[0096] The threshold generation module is used to adaptively weight the environmental derating curve and the predicted drift according to historical trends, output a real-time threshold, and update it in real time with the health level;

[0097] The compensation output module is used to identify the thermoelectric coupling component within the interference time window and inject voltage pulses of equal amplitude and opposite polarity to offset thermal drift and record the compensation effect;

[0098] The clock calibration module is used to monitor the terminal's local time synchronization using the IEEE1588PTP slave clock, solving sub-microsecond synchronization of equipment within the station and performing secondary alignment using a high-stability reference on the cloud side.

[0099] It should be noted that the present invention can be deployed on the device itself to implement embedded applications, and can also be run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0100] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions or computer programs. When loaded or executed on a computer, the processes or functions described in the embodiments of this application are fully or partially performed. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission. Wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission methods include infrared, microwave, etc. The computer-readable storage medium can be any available medium accessible by a computer, or a data storage device such as a server or data center that contains a collection of one or more available media. The available medium can be magnetic media (e.g., floppy disks, hard disks, tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0101] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and modules described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0102] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0103] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, and may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of this embodiment according to actual needs.

[0104] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0105] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0106] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

[0107] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An active testing method for a circuit breaker, characterized in that: include: S1: Without power outage, a high-precision sensor array is used to synchronously collect the multi-dimensional operating parameters of the circuit breaker and perform derating normalization processing on the temperature and humidity signals; S2: Based on the standardized input data, a time-frequency joint analysis is performed on the contact opening and closing speed signal to extract time and frequency domain features. The heat flux refraction angle is calculated based on the sensor thermal conductivity distribution and then correlated with the coil current time series to obtain the thermal-mechanical coupling eigenvector. S3: Construct a thermo-mechanical coupling dynamic graph with sensor locations and key components of the circuit breaker as nodes and the inverse of the equivalent thermal conductance between nodes as edge weights. Use online graph reasoning to output nodes and predict critical drift. S4: The environmental derating curve and the predicted critical drift are weighted and integrated. The weights are adaptively adjusted based on the historical trends of contact-related parameters. The new threshold is calculated in real time and updated with the health of the circuit breaker. S5: During the threshold interference period, the thermoelectric coupling component that is linearly related to the rate of change of the heat flux refraction angle is extracted from the fast-changing component of the contact opening and closing speed signal, and a small voltage pulse of equal amplitude and opposite polarity is generated and synchronously injected into the signal circuit; S6: Local sub-microsecond synchronization is achieved through IEEE1588 precision clocks. Cross-device timestamps are corrected with cloud-based secondary alignment. The closed-loop compensated operating data and threshold adjustment results are continuously fed back to the coupled dynamic graph model to update node weights and threshold generation rules.

2. The active testing method for a circuit breaker according to claim 1, characterized in that S2 include: The synchronously collected contact opening and closing speed, coil current, and various temperature and humidity parameters are zero-referenced and amplitude normalized according to the preset range to eliminate the dimensional differences between different sensing channels; The standardized contact opening and closing speed signal is first segmented into time windows. Then, time domain envelope extraction and frequency domain amplitude-phase statistics are synchronously performed on each signal segment to obtain the time characteristic quantity reflecting the travel smoothness and the frequency characteristic quantity reflecting the mechanical impact intensity. Based on the thermal conductivity distribution of the sensor at each installation point inside the circuit breaker, the refraction angle of the heat flow in the XY plane within the same time window is calculated as the heat transfer direction feature; The heat flux refraction angle is mapped one-to-one with the corresponding coil current time series according to the timestamp, and is combined with the time characteristic quantity and frequency characteristic quantity to form a multidimensional vector to form a characteristic direction star set that characterizes the heat engine coupling state.

3. The active testing method for a circuit breaker according to claim 1, characterized in that: S3 includes: The various sensor points installed inside the circuit breaker body are set as nodes of the dynamic graph according to their physical locations in a one-to-one correspondence with the key mechanical, electrical and heat dissipation components of the circuit breaker; The inverse of the equivalent thermal conductivity value of the heat transfer channel between nodes is calculated and used as the weight of the edge; The node sensor data is mapped into a time series snapshot with variable weights at a preset sampling period, and the real-time operation feature vector of the node is synchronously recorded in each snapshot; The dynamic graph inference process deployed on the edge processor is called to perform sliding window calculations on continuous snapshots and output the predicted critical drift of each node.

4. The active testing method for a circuit breaker according to claim 1, characterized in that: S4 include: Real-time reading of the ambient temperature, humidity, air pressure and busbar load current at the circuit breaker location, mapping the data to the pre-stored environmental derating curve to obtain the compensation coefficient; Call the online graph inference results to obtain the predicted critical drift for each node; Retrieve the time series of historical operating parameters related to the contacts and calculate the recent change trends of each parameter.

5. The active testing method for a circuit breaker according to claim 4, characterized in that: include: According to the magnitude and direction of the change trend, weights are assigned to the environmental compensation coefficient and the predicted critical drift respectively, and the weight values ​​are automatically adjusted up or down along with the trend of the contact parameters; The environmental compensation coefficient and the predicted critical drift amount are weightedly integrated according to the weight to obtain a threshold increment; The threshold increment is added to the base threshold to generate a new real-time threshold, and the threshold is stored in the local database along with the health score.

6. The active testing method for a circuit breaker according to claim 1, wherein S5 include: Taking the threshold interference period as the boundary, a synchronous time window is selected for the contact opening and closing speed signal, and the fast-changing component is extracted. The fast-changing component is time-aligned with the rate of change of the heat flux refraction angle in the same time window. The signal section where the two are linearly correlated is identified by comparing them with the pre-calibrated sensitive section, and the thermoelectric coupling component to be compensated is determined.

7. The active testing method for a circuit breaker according to claim 6, characterized in that: include: Determine the amplitude and polarity of the thermoelectric coupling component and generate a small voltage pulse of equal amplitude, opposite polarity and duration 1 to 3 times the thermal expansion relaxation time; A small voltage pulse is injected into the contact signal circuit in phase synchronization with the thermoelectric coupling component through a precision voltage source, so that the two cancel each other out in amplitude. The difference in threshold fluctuation before and after compensation is recorded, and the compensation result is fed back to the threshold adaptive module to correct the pulse amplitude and phase of the next cycle.

8. The active testing method for a circuit breaker according to claim 1, wherein S6 include: Step 1: Configure the circuit breaker monitoring terminal as an IEEE1588 precision time protocol slave clock and synchronize it with the master clock in the substation using hardware timestamps to achieve sub-microsecond time synchronization. Step 2: Obtain a reference time scale from the Global Positioning System (GPS) or an equivalent high-stability reference in the cloud-side data center, and perform secondary correction on the timestamps of the data packets uploaded by each monitoring terminal to eliminate cross-device and cross-network segment propagation delays. Step 3: Feedback the contact stroke, coil current, threshold adjustment results and other operating data after secondary correction to the local edge processor in real time; Step 4: Based on the feedback data, the node weights of the thermal-mechanical-electrical coupling dynamic graph are adjusted incrementally or incrementally, and the threshold generation rules are simultaneously refreshed so that the model and thresholds evolve dynamically with the health of the circuit breaker. Step 5: In subsequent sampling cycles, the timing errors of the old and new threshold judgment results and the actual action timing are compared. When the error exceeds the preset limit, the resynchronization process of steps 1 and 2 is automatically triggered until the time scales are consistent.

9. An active testing system for a circuit breaker, used to implement the active testing method for a circuit breaker according to any one of claims 1 to 8, characterized in that: include: The sensing and acquisition module is used to synchronously obtain multi-dimensional signals of coil current, contact stroke and speed, shell and contact temperature, ambient temperature and humidity, and bus current to form unified time-based raw health data; The feature analysis module is used to perform time-domain envelope and frequency-domain amplitude-phase statistics on the contact opening and closing speeds, separating high-frequency impact, low-frequency inertia, and steady-state displacement characteristics; The coupled modeling module is used to construct a thermo-mechanical dynamic diagram with variable weights as the temperature field, using sensor points and key components such as coils, contacts, and arc extinguishing chambers as nodes and taking the inverse of the thermal conductance of the two nodes as the edge weights. The threshold generation module is used to adaptively weight the environmental derating curve and the predicted drift according to historical trends, output a real-time threshold, and update it in real time with the health level; The compensation output module is used to identify the thermoelectric coupling component within the interference time window and inject voltage pulses of equal amplitude and opposite polarity to offset thermal drift and record the compensation effect; The clock calibration module is used to monitor the terminal's local time synchronization using the IEEE1588PTP slave clock, solving sub-microsecond synchronization of equipment within the station and performing secondary alignment using a high-stability reference on the cloud side.

Images