Arc energy dissipation prediction system for molded case circuit breaker based on field-circuit coupling

Abstract

The application discloses a field-circuit coupling-based molded case circuit breaker arc energy dissipation prediction system and relates to the technical field of circuit breaker arc energy prediction.The system comprises a multi-modal data acquisition module, a cross-modal feature fusion module, a prediction model construction module, a working condition self-adaptive adjustment module and a prediction output module.The application realizes accurate prediction of double-break molded case circuit breaker arc energy dissipation through multi-modal data acquisition and cross-modal feature fusion.Synchronous acquisition of multi-dimensional feature data is realized through arrayed optical fiber electric field sensors, miniature Hall sensor arrays, hyperspectral imaging units, infrared thermographs and other equipment, covering the full-space field of the arc-extinguishing chamber, and through dynamic weight distribution and physical correlation degree calculation, data heterogeneity among modes is eliminated, a fusion feature vector is generated, the representation capability of arc dynamic behavior is improved, the error between the prediction value and the actual dissipated energy is low, and the problem of poor working condition adaptability of traditional models is solved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of circuit breaker arc energy prediction, in particular to a molded case circuit breaker arc energy dissipation prediction system based on field-circuit coupling. BACKGROUND

[0002] With the rapid development of power electronic technology, the breaking capacity and reliability of the double-break molded case circuit breaker as the core protection device of the low-voltage distribution system are directly related to the safe operation of the power grid. In the breaking process of the circuit breaker, the generation and extinction of the arc are the key links to determine the success or failure of the breaking. If the energy released by the arc during combustion cannot be effectively dissipated, the temperature in the arc chamber will be too high, the contact ablation will be aggravated, and even the circuit breaker will fail. Therefore, accurate prediction of arc energy dissipation is an important prerequisite for optimizing the design of the circuit breaker and improving its breaking performance.

[0003] However, the existing arc energy dissipation prediction technology has some deficiencies. The data acquisition mode is single, and most methods only rely on circuit parameters such as current and voltage, ignoring the direct influence of physical fields such as electric field, magnetic field and light intensity on arc combustion. For example, when the traditional model collects the time-domain waveform of the current through a Rogowski coil, it cannot reflect the modulation effect of the electric field distortion in the arc chamber on the arc column voltage drop, resulting in a high deviation between the predicted value and the actual dissipated energy. There is no cross-modal data fusion mechanism, and the time base and dimension of different physical field data differ significantly. The traditional method directly splices the features or simply weights them, without establishing a model for calculating the physical correlation between modes, so that the fused features are insufficient to represent the dynamic behavior of the arc, and the working condition adaptability is poor. The parameters of the existing model are fixed, and when the load current fluctuation or environmental temperature change exceeds the corresponding threshold, the model prediction error rate increases, which cannot meet the application requirements under complex working conditions. SUMMARY

[0004] The purpose of the present application is to overcome the deficiencies of the prior art and provide a molded case circuit breaker arc energy dissipation prediction system based on field-circuit coupling. The present application realizes accurate prediction of the arc energy dissipation of the double-break molded case circuit breaker through multi-modal data acquisition and cross-modal feature fusion. By using devices such as arrayed optical fiber electric field sensors, micro Hall sensor arrays, hyperspectral imaging units and infrared thermometers, multi-dimensional feature data is synchronously acquired, covering the full spatial field of the arc chamber. Through dynamic weight distribution and physical correlation calculation, the data heterogeneity between modes is eliminated, a fused feature vector is generated, the representation ability of the arc dynamic behavior is improved, the error between the predicted value and the actual dissipated energy is low, and the problem of poor working condition adaptability of the traditional model is solved, providing data support for the design optimization of the circuit breaker.

[0005] To solve the above technical problems, the present application provides the following technical scheme: a molded case circuit breaker arc energy dissipation prediction system based on field-circuit coupling, which comprises:

[0006] Multi-modal data acquisition module: synchronous acquisition of electric field field distribution characteristics, magnetic field field distribution characteristics, circuit current time domain waveform characteristics, circuit voltage time domain waveform characteristics, arc light intensity spectrum characteristics and arc chamber temperature distribution characteristics during the breaking process of double-break plastic shell circuit breaker;

[0007] Cross-modal feature fusion module: standardization, outlier rejection, time reference alignment and key feature extraction are performed on the collected various types of feature data, cross-modal feature fusion is completed, and a fusion feature vector is obtained;

[0008] Prediction model construction module: based on the fusion feature vector, an arc energy dissipation prediction model is constructed, the dynamic weights of each modality are calculated through a modality feature dynamic weight distribution formula, and prediction is performed through an arc energy dissipation prediction formula;

[0009] Working condition self-adaptive adjustment module: real-time monitoring of circuit breaker operating condition parameters, when the working condition parameter change amplitude exceeds the working condition change threshold, the arc energy dissipation prediction model parameters are updated through a working condition self-adaptive model update formula;

[0010] Prediction output module: input the fusion feature vector into the updated arc energy dissipation prediction model, output the arc energy dissipation prediction value, and visually display the prediction value change curve, weight distribution, temperature and field strength thermal map.

[0011] Further, in the multi-modal data acquisition module, the electric field field distribution characteristics are collected by an array optical fiber electric field sensor; the magnetic field field distribution characteristics are collected by a micro Hall sensor array; the circuit current time domain waveform characteristics are collected by a wideband Rogowski coil; the circuit voltage time domain waveform characteristics are collected by a high-voltage capacitor voltage divider; the arc light intensity spectrum characteristics are collected by a hyperspectral imaging unit; and the arc chamber temperature distribution characteristics are collected by an infrared thermal imager.

[0012] Further, in the cross-modal feature fusion module, the cross-modal feature fusion process is realized through a cross-modal feature fusion formula, and the cross-modal feature fusion formula is: wherein, is the mth modality original feature, is a multi-modal feature extraction operator, is an initial weight determined by historical data, is a multi-modal physical correlation degree calculation operator, is a correlation degree weight coefficient determined by historical data, is a normalization function, is a fusion feature vector, is an electric field field distribution characteristic, is a magnetic field field distribution characteristic, is a circuit current time domain waveform characteristic, is an arc voltage time domain waveform feature, is an arc light intensity spectrum feature, is an arc chamber temperature distribution feature.

[0013] Further, in the prediction model construction module, the arc energy dissipation prediction model comprises an input layer, a feature encoding layer, a weight distribution layer, and a prediction output layer; the input layer inputs the fusion feature vector The feature encoding layer performs nonlinear mapping on the fusion feature vector through a two-layer fully connected network, the weight distribution layer calculates the dynamic weight of each mode through a modal feature dynamic weight distribution formula, and the prediction output layer outputs the prediction value through an arc energy dissipation prediction formula.

[0014] Further, in the prediction model construction module, the modal feature dynamic weight distribution formula is: wherein, is the dynamic weight of the mth mode in the kth breaking process, is a feature importance evaluation operator, is a working condition parameter vector of the kth breaking, is a working condition influence factor function, is a weight adjustment sensitivity coefficient, is a fusion feature vector.

[0015] Further, in the prediction model construction module, the arc energy dissipation prediction formula is: wherein, is an arc energy dissipation prediction value, is the dynamic weight of the mth mode in the kth breaking process, is an element-level multiplication, , is a feature encoding layer parameter, , is a prediction output layer parameter, is an activation function, is a hyperbolic tangent function.

[0016] Further, in the working condition adaptive adjustment module, the real-time monitored circuit breaker operating condition parameters include load current, ambient temperature, breaking speed, and arc chamber contact resistance; wherein the load current is monitored by a Rogowski coil sensor; the ambient temperature is monitored by a platinum resistance temperature sensor; the breaking speed is monitored by a laser displacement sensor; the arc chamber contact resistance is monitored by a micro-ohmmeter; and the working condition change threshold is: load current ± 15%, ambient temperature ± 8℃, breaking speed ± 20%, and arc chamber contact resistance ± 25%.

[0017] Further, in the working condition adaptive adjustment module, the working condition adaptive model updating formula is: wherein, is the arc energy dissipation prediction model parameter set at the first is the arc energy dissipation prediction model parameter set at the first is the parameter gradient operator, is the arc energy dissipation prediction value, is the arc energy dissipation prediction model parameter set at the first is the arc energy dissipation prediction model parameter set at the first is the hybrid loss function, is the working condition change amount, is the adaptive update switch function, is the incremental learning rate, is the arc energy dissipation prediction model parameter set at the first is the arc energy dissipation prediction model parameter set at the first

[0018] Compared with the prior art, the arc energy dissipation prediction system based on field-circuit coupling of the molded case circuit breaker has the following beneficial effects:

[0019] I. The present application realizes accurate prediction of arc energy dissipation of a double-break molded case circuit breaker through multi-modal data acquisition and cross-modal feature fusion, synchronously acquires multi-dimensional feature data through array optical fiber electric field sensors, micro Hall sensor arrays, hyperspectral imaging units, infrared thermographs and other equipment, covers the full space field of the arc-extinguishing chamber, eliminates data heterogeneity between modes through dynamic weight distribution and physical correlation calculation, generates a fusion feature vector, improves the representation ability of the arc dynamic behavior, ensures that the error between the prediction value and the actual dissipated energy is low, solves the problem of poor working condition adaptability of traditional models, and provides data support for circuit breaker design optimization.

[0020] II. The present application builds a working condition adaptive adjustment mechanism, so that the arc energy dissipation prediction model can respond to changes in the operating environment in real time, and through devices such as Rogowski coils and platinum resistance temperature sensors, key working condition parameters such as load current, environmental temperature and breaking speed are monitored in real time, when the parameter changes exceed the threshold value, the parameter incremental learning is triggered through the working condition adaptive model updating formula, the parameter update of the arc energy dissipation prediction model is completed, the prediction error rate is reduced, in addition, the prediction output module integrates a visual thermal map and a weight distribution map, which can intuitively display the temperature field and electric field intensity distribution of the arc-extinguishing chamber, and quickly locate design defects.

[0021] Other advantages, objects, and features of the present application will be apparent to those skilled in the art upon reading the following specification, and will be learned from the practice of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0023] Figure 1 The flowchart shows a system for predicting arc energy dissipation in molded case circuit breakers based on field-circuit coupling.

[0024] Figure 2 This is a framework diagram of a field-circuit coupling-based system for predicting arc energy dissipation in molded case circuit breakers. Detailed Implementation

[0025] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0026] Example 1:

[0027] Multimodal data acquisition module: In the application scenario of double-break molded case circuit breakers in low-voltage power distribution rooms, considering the characteristic of circuit breakers being installed in enclosed cabinets, array-type fiber optic electric field sensors are arranged in a matrix along the upper and lower end covers and side walls of the arc-extinguishing chamber of the circuit breaker to collect the electric field distribution characteristics; a miniature Hall sensor array is arranged along the two pole axes of the circuit breaker and the central axis of the arc-extinguishing chamber to obtain the magnetic field distribution characteristics; a broadband Rogowski coil is connected in series in the main circuit of the circuit breaker near the moving contact to collect the time-domain waveform characteristics of the circuit current; a high-voltage capacitor divider is connected in parallel between the two breaks of the double-break circuit breaker to capture the time-domain waveform characteristics of the circuit voltage; the lens of the hyperspectral imaging unit is horizontally aligned with the center area of ​​the arc column channel through the high-temperature resistant quartz window on the side wall of the arc-extinguishing chamber to collect the arc light intensity spectrum characteristics; an infrared thermal imager is arranged vertically downward through the high-temperature resistant observation window on the top of the arc-extinguishing chamber, covering the entire internal grid area and arc column channel of the arc-extinguishing chamber to obtain the temperature distribution characteristics of the arc-extinguishing chamber, and all sensors complete the acquisition simultaneously.

[0028] The cross-modal feature fusion module: It processes the collected original features from six modes—electric field distribution features, magnetic field distribution features, circuit current time-domain waveform features, circuit voltage time-domain waveform features, arc light intensity spectrum features, and arc-extinguishing chamber temperature distribution features—in sequence, performing standardization, outlier removal, and time-base alignment. Then, it extracts key features to complete the cross-modal feature fusion, obtaining a fused feature vector, such as... Figure 1 As shown, the cross-modal feature fusion process is implemented through a cross-modal feature fusion formula, which is: wherein, is the mth modal original feature, is the multi-modal feature extraction operator, is the initial weight determined by historical data, is the multi-modal physical correlation degree calculation operator, is the correlation degree weight coefficient determined by historical data, is the normalization function, is the fusion feature vector, is the electric field field domain distribution feature, is the magnetic field field domain distribution feature, is the circuit current time domain waveform feature, is the circuit voltage time domain waveform feature, is the arc light intensity spectrum feature, is the arc chamber temperature distribution feature.

[0029] The prediction model construction module: based on the fusion feature vector, an arc energy dissipation prediction model suitable for the working condition of the power distribution room is constructed, and the arc energy dissipation prediction model includes an input layer, a feature encoding layer, a weight distribution layer and a prediction output layer; the fusion feature vector is input through the input layer, the feature encoding layer performs nonlinear mapping on the fusion feature vector through a 2-layer fully connected network, and especially for the gradual change scene of the contact resistance caused by the aging of the equipment in the power distribution room, the processing of the related coding of the arc chamber temperature distribution feature is strengthened; the weight distribution layer calculates the dynamic weight of each modal through a modal feature dynamic weight distribution formula, for example, in the stable load period, the weight of the circuit current and voltage time domain waveform feature is increased, in the temperature sudden change period, the weight of the arc chamber temperature distribution feature is increased, and the weight distribution is ensured to be in line with the actual working condition requirements, and the modal feature dynamic weight distribution formula is: wherein, is the dynamic weight of the mth modal in the kth breaking process, is the feature importance evaluation operator, is the working condition parameter vector of the kth breaking, is the working condition influence factor function, is the weight adjustment sensitivity coefficient, is the fusion feature vector; the prediction output layer uses an arc energy dissipation prediction formula to combine the related parameters of the feature encoding layer and the prediction output layer, and outputs an arc energy dissipation prediction value after processing by an activation function, thereby providing data support for the operation and maintenance of the power distribution room, and the arc energy dissipation prediction formula is: ; wherein, is the arc energy dissipation prediction value, is the dynamic weight of the mth modal in the kth breaking process, is the element-level multiplication, , encoding layer parameters, , prediction output layer parameters, activation function, hyperbolic tangent function.

[0030] The working condition adaptive adjustment module: real-time monitoring of the load current, environmental temperature, breaking speed and arc chamber contact resistance of the circuit breaker, among which the load current is monitored by a Rogowski coil sensor, focusing on capturing the instantaneous large current during industrial motor starting; the environmental temperature is monitored by a platinum resistance temperature sensor, which is installed near the ventilation of the arc chamber to avoid temperature misjudgment caused by direct sunlight or equipment heat dissipation; the breaking speed is monitored by a laser displacement sensor, which records the displacement change in the breaking process in real time; the arc chamber contact resistance is monitored by a micro-ohmmeter, and is calibrated regularly every month to ensure accuracy; when the working condition parameter changes by more than the set threshold, for example, the original temperature of the power distribution room is 28℃ in summer noon, and the indoor temperature rises to 40℃ due to air conditioner failure, exceeding the environmental temperature threshold, the arc energy dissipation prediction model parameters are updated through the working condition adaptive model update formula to ensure that the arc energy dissipation prediction model can adapt to the dynamic changes of the power distribution room working condition, and the working condition adaptive model update formula is: wherein, is the arc energy dissipation prediction model parameter set at the nth breaking, is the arc energy dissipation prediction value, is the parameter gradient operator, is the arc energy measured value of the nth breaking, is the hybrid loss function, is the working condition change amount, is the adaptive update switch function, is the incremental learning rate, is the arc energy dissipation prediction model parameter set after updating at the nth breaking; wherein the working condition change threshold is: load current ± 15%, environmental temperature ± 8℃, breaking speed ± 20%, arc chamber contact resistance ± 25%.

[0031] ​​​The prediction output module inputs the fusion feature vector into the updated arc energy dissipation prediction model, obtains the arc energy dissipation prediction value through the arc energy dissipation prediction formula, and simultaneously visually displays the prediction value change curve, the weight distribution of each mode, the temperature thermal map of the arc chamber, and the thermal maps of the electric field and the magnetic field on the terminal screen of the power distribution room monitoring center. The terminal can be used to monitor the operating state of the circuit breaker in real time. For example, when the current fluctuates due to the wear of the motor bearing of a production line, the prediction value change curve will synchronously display the abnormal rising trend of the arc energy, and the temperature thermal map will present the hot spot of the local temperature rise of the arc chamber, so that the fault circuit can be quickly located, the maintenance can be timely carried out, and the expansion of the fault to affect the power supply of the entire power distribution room can be avoided.

[0032] In summary, for the working conditions such as centralized start-stop of equipment in the power distribution room, load fluctuation, and seasonal temperature change, six types of feature data are collected by the multi-modal data acquisition module, and electromagnetic interference is avoided. The cross-modal feature fusion module processes the features in combination with the historical operation data of the power distribution room, excavates the physical correlation, and obtains a fusion feature vector that is adapted to complex working conditions. The prediction model construction module dynamically allocates the weight of each mode according to the working condition, and outputs an accurate prediction value. The working condition self-adaptive adjustment module monitors the parameters in real time, responds to temperature sudden rise, load sudden increase and other abnormalities, and updates the model. The prediction output module visually displays the data through the monitoring terminal, helps the operation and maintenance personnel to timely troubleshoot faults, and ensures the stable power supply of the power distribution room.

[0033] Embodiment two:

[0034] Multi-modal data acquisition module: In the application scenario of DC power distribution system double-break molded case circuit breaker in new energy vehicles, considering the compact vehicle space and the influence of environmental factors such as jolt, waterproof, temperature change, etc., the array type optical fiber electric field sensor is arranged in a compact matrix along the upper and lower end covers and the two side walls of the arc chamber of the circuit breaker, packaged in a waterproof shockproof shell, and the shell is fixed with the circuit breaker shell through an insulating support to avoid displacement of the sensor caused by vehicle body vibration; The array of miniature Hall sensors is miniaturized along the two-pole axis and the arc chamber axis of the circuit breaker, and the miniature Hall sensor lead is made of silica gel cable resistant to high and low temperature; The wideband Rogowski coil is connected in series in the form of a flat structure on the side of the main circuit of the circuit breaker close to the moving contact, mounted on the circuit breaker shell, reducing the occupation of the vehicle space, and wrapped in an electromagnetic shielding layer to prevent the strong magnetic field generated by the motor controller from interfering with current signal acquisition; The high-voltage capacitor voltage divider is designed in a small size and connected in parallel between the two breakpoints of the double-break circuit breaker, and its interface uses a waterproof aviation plug to prevent water vapor from entering and affecting the voltage division accuracy when driving in rainy weather; The hyperspectral imaging unit uses a miniature lens, which is horizontally aligned with the center area of the arc column channel through a high-temperature-resistant quartz window on the side wall of the arc chamber, and a anti-glare coating is added outside the lens to reduce the interference of vehicle light on spectrum acquisition; The miniature infrared thermal imager is arranged vertically downward through a high-temperature-resistant observation window on the top of the arc chamber, and the lens view angle covers the entire arc chamber internal grid area and arc column channel. All acquisition devices realize synchronous data acquisition through the vehicle CAN bus, which is suitable for the dynamic driving scene of the vehicle.

[0035] Cross-modal feature fusion module: When processing the collected 6 kinds of modal original features, the focus is on typical scenes such as vehicle acceleration overtaking, sudden increase of motor power, brake kinetic energy recovery, reverse current flow, low temperature start, and battery output characteristic change, etc. For example, in the acceleration scene, the instantaneous peak segment in the circuit current time domain waveform feature and the instantaneous voltage drop segment in the circuit voltage time domain waveform feature are preferentially removed, to ensure that the data can reflect the true arc state under dynamic load; Then, the collection time of each modal feature is unified to avoid time deviation caused by sensor collection delay due to vehicle jolt; After extracting the key features, cross-modal feature fusion is completed to obtain a fusion feature vector. The cross-modal feature fusion process is realized through a cross-modal feature fusion formula, which is: .

[0036] The prediction model construction module: based on the fusion feature vector, an arc energy dissipation prediction model suitable for the vehicle dynamic working condition is constructed, which includes an input layer, a feature encoding layer, a weight distribution layer and a prediction output layer; the fusion feature vector is input through the input layer, the feature encoding layer performs nonlinear mapping on the fusion feature vector through a 2-layer fully connected network, especially for the reverse current scene during the dynamic energy recovery of automobile braking, the processing of the relevant coding of the circuit current, voltage time domain waveform features is strengthened to ensure that the arc characteristics under the reverse current can be reflected; the dynamic weight of each mode is calculated by the modal feature dynamic weight distribution formula in the weight distribution layer, for example, in the transient high current scene of acceleration and overtaking, the weights of the circuit current, voltage time domain waveform features are increased, in the low temperature starting scene, the weights of the temperature distribution features of the arc chamber and the magnetic field domain distribution features are increased, to ensure that the weight distribution is consistent with the real-time changes of the vehicle working condition, the modal feature dynamic weight distribution formula is: ; the prediction output layer uses the arc energy dissipation prediction formula to output the arc energy dissipation prediction value, which provides a basis for decision-making for the vehicle control system, and the arc energy dissipation prediction formula is: .

[0037] The working condition self-adaptive adjustment module: four running working condition parameters of the load current, the environmental temperature, the breaking speed and the contact resistance of the arc chamber of the circuit breaker are monitored in real time, among which the load current is monitored by a Rogowski coil sensor, focusing on capturing the transient high current during acceleration and overtaking and the reverse current during braking; the environmental temperature is monitored by a platinum resistance temperature sensor, which is installed on the outside of the circuit breaker shell and avoids the heat dissipation area of the battery pack to ensure that the real vehicle environmental temperature can be reflected; the breaking speed is monitored by a laser displacement sensor; the contact resistance of the arc chamber is monitored by a micro-ohmmeter, and resistance calibration is automatically completed once before each charging; when the working condition parameter changes by more than a set threshold value, for example, the low temperature in winter exceeds the environmental temperature threshold value, the breaking speed fluctuation exceeds the threshold value due to the bumpy road section, or the load current exceeds the threshold value due to the unstable output current of the charging pile during charging, the working condition self-adaptive model update formula is used to update the arc energy dissipation prediction model parameters, as shown in Figure 2 , to adapt to the dynamic changes of the vehicle working condition, the working condition self-adaptive model update formula is: ; and the working condition change threshold value is: load current ± 15%, environmental temperature ± 8℃, breaking speed ± 20%, and contact resistance of arc chamber ± 25%.

[0038] The prediction output module inputs the fusion feature vector into the updated arc energy dissipation prediction model, outputs the arc energy dissipation prediction value, and visualizes the prediction value change curve, the weight distribution of each mode, and the temperature thermal map of the arc extinguishing chamber through the vehicle-mounted screen. At the same time, detailed data is uploaded to the vehicle-mounted T-BOX, and core monitoring information can be viewed through the vehicle-mounted screen. For example, if the output current of the charging pile is unstable during charging, the prediction value change curve will show abnormal fluctuations in arc energy. If the current suddenly increases due to motor failure during driving, the prediction value change curve will also reflect the abnormal rise of arc energy, ensuring driving safety.

[0039] In summary, in the direct current distribution system of new energy vehicles, the multi-modal data acquisition module uses waterproof and shockproof, small-sized sensors and synchronously acquires data through the CAN bus, adapting to dynamic scenes such as compact vehicle-mounted space, jolting and vibration, and high and low temperature. The cross-modal feature fusion module processes features for scenes such as acceleration and overtaking, and low-temperature starting in combination with vehicle driving history data. The prediction model construction module dynamically adjusts weights according to vehicle-mounted working conditions, adapting to reverse current, instantaneous large current, and other situations. The working condition self-adaptive adjustment module monitors working condition parameters and quickly updates the model to cope with changes in working conditions caused by low temperature and jolting. The prediction output module displays data through the central control screen and issues warnings to ensure the safety of vehicle driving and charging, and adapt to vehicle-mounted power demand.

[0040] The above is only a preferred embodiment of the present application, and does not limit the present application in any form. Although the present application has been disclosed as above, it is not intended to limit the present application. Any person skilled in the art can make some changes or modifications to the above disclosed technical content without departing from the scope of the present application, and any equivalent embodiments with equivalent changes are equivalent to the above embodiments. Any modification, change, and modification of the above embodiments according to the technical essence of the present application are still within the scope of the present application.

Claims

1. A field-circuit coupled two-break molded case circuit breaker arc energy dissipation prediction system, characterized in that, The system comprises: A multi-modal data acquisition module: synchronously acquires electric field field distribution characteristics, magnetic field field distribution characteristics, circuit current time domain waveform characteristics, circuit voltage time domain waveform characteristics, arc light intensity spectral characteristics, and arc chamber temperature distribution characteristics during the breaking process of a double-break plastic shell circuit breaker; The cross-modal feature fusion module: standardize, remove outliers, align time reference and extract key features of the collected various types of feature data, complete cross-modal feature fusion, get the fusion feature vector, the cross-modal feature fusion process is realized through the cross-modal feature fusion formula, the cross-modal feature fusion formula is: Wherein, is the mth modal original feature, is a multi-modal feature extraction operator, is the initial weight, determined by historical data, is a multi-modal physical correlation degree calculation operator, is the correlation degree weight coefficient, determined by historical data, is a normalization function, is a fusion feature vector, is an electric field field distribution feature, is a magnetic field field distribution feature, is a circuit current time domain waveform feature, is a circuit voltage time domain waveform feature, is an arc light intensity spectrum feature, is an arc chamber temperature distribution feature; A prediction model construction module: based on the fusion feature vector, an arc energy dissipation prediction model is constructed, the dynamic weight of each mode is calculated through a modal feature dynamic weight distribution formula, and prediction is performed through an arc energy dissipation prediction formula, the modal feature dynamic weight distribution formula is: wherein, is the dynamic weight of the mth mode in the kth breaking process, is a feature importance evaluation operator, is the working condition parameter vector of the kth breaking, is a working condition influence factor function, is a weight adjustment sensitivity coefficient, is a fusion feature vector; the arc energy dissipation prediction formula is: ; wherein, is an arc energy dissipation prediction value, is the dynamic weight of the mth mode in the kth breaking process, is an element-level multiplication, is a feature encoding layer parameter, is a prediction output layer parameter, is an activation function; A working condition self-adaptive adjustment module: real-time monitoring of circuit breaker operating condition parameters, when the working condition parameter change amplitude exceeds the working condition change threshold, the arc energy dissipation prediction model parameters are updated through the working condition self-adaptive model update formula; A prediction output module: inputs the fusion feature vector into the updated arc energy dissipation prediction model, outputs the arc energy dissipation prediction value, and visualizes the prediction value change curve, weight distribution, temperature and field strength thermal map.

2. The field-linkage based double-break molded case circuit breaker arc energy dissipation prediction system of claim 1, wherein, In the multi-modal data acquisition module, the electric field field distribution characteristics are acquired by an array optical fiber electric field sensor; the magnetic field field distribution characteristics are acquired by a micro Hall sensor array; the circuit current time domain waveform characteristics are acquired by a wideband Rogowski coil; the circuit voltage time domain waveform characteristics are acquired by a high-voltage capacitor voltage divider; the arc light intensity spectral characteristics are acquired by a hyperspectral imaging unit; and the arc chamber temperature distribution characteristics are acquired by an infrared thermal imager.

3. The field-circuit coupled based double-break plastic housed circuit breaker electric arc energy dissipation prediction system of claim 1, wherein, In the prediction model construction module, the arc energy dissipation prediction model comprises an input layer, a feature coding layer, a weight distribution layer and a prediction output layer; the input layer inputs the fusion feature vector The feature coding layer performs nonlinear mapping on the fusion feature vector through a two-layer fully connected network, the weight distribution layer calculates the dynamic weight of each mode through a modal feature dynamic weight distribution formula, and the prediction output layer outputs the prediction value through an arc energy dissipation prediction formula.

4. The field-circuit coupled based double-break plastic housed circuit breaker electric arc energy dissipation prediction system of claim 1, wherein, In the working condition self-adaptive adjustment module, the real-time monitored circuit breaker operating condition parameters include load current, ambient temperature, breaking speed, and arc chamber contact resistance; wherein the load current is monitored by a Rogowski coil sensor; the ambient temperature is monitored by a platinum resistance temperature sensor; the breaking speed is monitored by a laser displacement sensor; the arc chamber contact resistance is monitored by a micro-ohmmeter; and the working condition change threshold is: load current ± 15%, ambient temperature ± 8℃, breaking speed ± 20%, and arc chamber contact resistance ± 25%.

5. The field-circuit coupled based double-break plastic housed circuit breaker electric arc energy dissipation prediction system of claim 1, wherein, The working condition adaptive adjustment module, the working condition adaptive model update formula is: Wherein, The arc energy dissipation prediction model parameter set at the kth breaking, The arc energy dissipation prediction value, The arc energy measured value of the kth breaking, The hybrid loss function, The working condition change amount, The adaptive update switch function, The incremental learning rate, The updated arc energy dissipation prediction model parameter set at the k+1th breaking.

Images