Arc restrike determination method and apparatus, computer device, and storage medium

Abstract

The application relates to an arc reignition determination method and device, computer equipment and a storage medium. After a vacuum circuit breaker cuts off a circuit of a power system, a target space-time coordinate point of the vacuum circuit breaker is acquired; the target space-time coordinate point comprises a time coordinate and a space coordinate, the time coordinate corresponds to different time points in an arc-after medium recovery process, and the space coordinate corresponds to different positions in a vacuum circuit breaker electrode gap; the target space-time coordinate point is input into a target simulation model to obtain a target simulation physical quantity; the target simulation physical quantity comprises at least one of electric potential, electron number density, ion number density and electron temperature; and then, according to the target simulation physical quantity, a reignition evaluation result of the arc is determined. According to the above scheme, the target simulation physical quantity can be quickly obtained by inputting the target space-time coordinate point into the target simulation model, a large number of particles do not need to be simulated, and therefore, the simulation efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a method, apparatus, computer device, and storage medium for determining arc reignition. Background Technology

[0002] Vacuum interruption technology is a critical component of power systems, used to rapidly disconnect circuits when the current crosses zero, preventing the arc from reigniting. In a vacuum circuit breaker, a vacuum arc is generated when the electrodes separate. After the current crosses zero, the arc extinguishes, but the dielectric in the electrode gap needs to recover from plasma to an insulating state. This process is called post-arc dielectric recovery.

[0003] If the dielectric recovery rate cannot keep up with the rapidly rising transient recovery voltage (TRV) between the electrodes, dielectric breakdown will occur, leading to arc reignition and potentially causing equipment damage or even a large-scale power outage. Therefore, accurate and rapid simulation and prediction of this process are crucial for guiding the design and optimization of vacuum circuit breakers.

[0004] Traditional techniques typically employ high-fidelity methods such as Particle-In-Cell with Monte Carlo Collisions (PIC-MCC) to simulate the post-arc dielectric recovery process of vacuum circuit breakers. This requires simulating a massive number of particles, and a single simulation can take weeks or even months, resulting in low simulation efficiency. Summary of the Invention

[0005] Therefore, it is necessary to provide a method, apparatus, computer equipment, and storage medium for determining arc reignition, which can improve the simulation efficiency of the post-arc dielectric recovery process of vacuum circuit breakers, in order to address the above-mentioned technical problems.

[0006] In a first aspect, this application provides a method for determining arc reignition, comprising:

[0007] After the vacuum circuit breaker disconnects the power system circuit, the target spatiotemporal coordinates of the vacuum circuit breaker are obtained; the target spatiotemporal coordinates include time coordinates and spatial coordinates, the time coordinates correspond to different moments in the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker.

[0008] Input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities; among them, the target simulation physical quantities include at least one of electric potential, electron number density, ion number density and electron temperature;

[0009] Based on the target simulated physical quantities, the reignition assessment results of the electric arc are determined.

[0010] In one embodiment, the target simulation model includes a sequentially connected input layer, a hidden layer, and an output layer. The hidden layer includes multiple sequentially connected hidden units, with adjacent hidden units connected by a self-attention module. Inputting the target spatiotemporal coordinates into the target simulation model yields the target simulation physical quantities, including:

[0011] The target spatiotemporal coordinates are input into the input layer for feature extraction to obtain the basic feature vector;

[0012] The basic feature vector is input into the hidden layer for processing to obtain the weight feature vector output by the hidden layer; the input data of the next hidden unit is the feature vector obtained by processing the output data of the previous hidden unit through the self-attention module; the input data of the first hidden unit is the basic feature vector, and the output data of the last hidden unit is the weight feature vector.

[0013] The weighted feature vector is input into the output layer for feature recognition to obtain the target simulated physical quantity;

[0014] Each self-attention module is used to perform a weighted summation of each of the multi-dimensional features in the output data of the previous hidden unit, based on the correlation between the dimensional feature and each of the other dimensional features, to obtain the updated feature corresponding to the dimensional feature; and based on the updated features corresponding to each dimensional feature, to obtain the fused feature vector for the input data; the fused feature vector of the input data is the input data of the next hidden unit.

[0015] In one embodiment, the method further includes:

[0016] Based on the target simulated physical quantities, potential cloud map, particle density distribution map, and sheath development dynamic map are generated. Among them, the potential cloud map is used to show the spatial distribution of potential in the electrode gap at different times, the particle density distribution map is used to show the spatiotemporal evolution of electron number density and ion number density, and the sheath development dynamic map is used to locate the position and thickness changes of the sheath by the dynamic changes of particle density gradient.

[0017] The potential cloud map, particle density distribution map, and sheath development dynamic map are displayed.

[0018] In one embodiment, the reignition assessment result of the electric arc is determined based on the target simulated physical quantities, including:

[0019] Determine the current value based on the target simulated physical quantity;

[0020] If the current value is greater than the preset threshold, then arc reignition is determined.

[0021] In one embodiment, after the vacuum circuit breaker disconnects the power system circuit, the method further includes:

[0022] Obtain the target operating parameters of the vacuum circuit breaker; among which, the target operating parameters include the transient recovery voltage (TRV) waveform parameters and the initial plasma density;

[0023] Input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities, including:

[0024] The target operating condition parameters are matched with the sample operating condition parameters used to train the target simulation model to obtain the matching results;

[0025] If the matching result is a match, the target spatiotemporal coordinates are input into the target simulation model to obtain the target simulation physical quantity.

[0026] In one embodiment, the target simulation model is trained in the following manner:

[0027] Input the sample operating parameters and sample spatiotemporal coordinates into the initial simulation model to obtain the predicted simulation physical quantities;

[0028] Based on the predicted physical quantities and constraints, determine the model loss of the initial simulation model;

[0029] The model parameters of the initial simulation model are optimized based on the model loss to obtain the target simulation model;

[0030] The constraints include physical equation constraints, boundary condition constraints, initial condition constraints, and data fitting condition constraints. Physical equation constraints are used to constrain the electromagnetic state, particle number, and thermodynamic state. Boundary condition constraints are used to constrain the boundary parameters of the vacuum circuit breaker. Initial condition constraints are used to constrain the initial state of the vacuum circuit breaker. Data fitting condition constraints are used to constrain the consistency between simulated physical quantities and actual physical quantities.

[0031] In one embodiment, the model loss of the initial simulation model is determined based on the predicted simulation physical quantities and constraints, including:

[0032] Determine the residuals between the predicted simulated physical quantities and each theoretical simulated physical quantity; wherein each theoretical simulated physical quantity is determined based on the predicted simulated physical quantity and each constraint condition.

[0033] The model loss of the initial simulation model is obtained by weighted summation of the residuals.

[0034] Secondly, this application also provides an arc reignition determination device, comprising:

[0035] The acquisition module is used to acquire the target spatiotemporal coordinates of the vacuum circuit breaker after the vacuum circuit breaker disconnects the power system circuit; wherein, the target spatiotemporal coordinates include time coordinates and spatial coordinates, the time coordinates correspond to different moments in the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker.

[0036] The simulation module is used to input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities; wherein, the target simulation physical quantities include at least one of electric potential, electron number density, ion number density and electron temperature;

[0037] The determination module is used to determine the reignition assessment result of the electric arc based on the target simulated physical quantities.

[0038] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0039] After the vacuum circuit breaker disconnects the power system circuit, the target spatiotemporal coordinates of the vacuum circuit breaker are obtained; the target spatiotemporal coordinates include time coordinates and spatial coordinates, the time coordinates correspond to different moments in the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker.

[0040] Input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities; among them, the target simulation physical quantities include at least one of electric potential, electron number density, ion number density and electron temperature;

[0041] Based on the target simulated physical quantities, the reignition assessment results of the electric arc are determined.

[0042] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0043] After the vacuum circuit breaker disconnects the power system circuit, the target spatiotemporal coordinates of the vacuum circuit breaker are obtained; the target spatiotemporal coordinates include time coordinates and spatial coordinates, the time coordinates correspond to different moments in the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker.

[0044] Input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities; among them, the target simulation physical quantities include at least one of electric potential, electron number density, ion number density and electron temperature;

[0045] Based on the target simulated physical quantities, the reignition assessment results of the electric arc are determined.

[0046] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0047] After the vacuum circuit breaker disconnects the power system circuit, the target spatiotemporal coordinates of the vacuum circuit breaker are obtained; the target spatiotemporal coordinates include time coordinates and spatial coordinates, the time coordinates correspond to different moments in the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker.

[0048] Input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities; among them, the target simulation physical quantities include at least one of electric potential, electron number density, ion number density and electron temperature;

[0049] Based on the target simulated physical quantities, the reignition assessment results of the electric arc are determined.

[0050] The aforementioned method, apparatus, computer equipment, and storage medium for determining arc reignition, after the vacuum circuit breaker disconnects the power system circuit, acquires the target spatiotemporal coordinates of the vacuum circuit breaker. These target spatiotemporal coordinates include both temporal and spatial coordinates; the temporal coordinates correspond to different moments during the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker. The target spatiotemporal coordinates are then input into a target simulation model to obtain target simulation physical quantities. These target simulation physical quantities include at least one of electric potential, electron number density, ion number density, and electron temperature. Based on these target simulation physical quantities, the arc reignition assessment result is determined. This scheme, by inputting the target spatiotemporal coordinates into the target simulation model, can quickly obtain the target simulation physical quantities without simulating a massive number of particles, thereby improving simulation efficiency. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 This is a flowchart illustrating the method for determining arc reignition in one embodiment;

[0053] Figure 2 This is a schematic diagram of the target simulation model in one embodiment;

[0054] Figure 3 This is a flowchart illustrating the process of obtaining the target simulated physical quantity in one embodiment;

[0055] Figure 4 This is a flowchart illustrating the process of visually displaying the target simulated physical quantities in one embodiment.

[0056] Figure 5 This is a flowchart illustrating the process of determining the reignition assessment result of an electric arc in one embodiment;

[0057] Figure 6 This is a schematic diagram of the process of training the target simulation model in one embodiment;

[0058] Figure 7 This is a schematic diagram of the arc reignition determination system in one embodiment;

[0059] Figure 8 This is a flowchart illustrating the arc reignition determination method in another embodiment;

[0060] Figure 9 This is a structural block diagram of the arc reignition determination device in one embodiment;

[0061] Figure 10 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0063] The arc reignition determination method provided in this application embodiment can be applied to the post-arc dielectric recovery process of a simulated vacuum circuit breaker to guide the design and optimization of vacuum circuit breakers in application scenarios.

[0064] This method can be executed by a server or a terminal. The server can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server providing cloud computing services. The terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle systems, and projection devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted displays. Head-mounted displays can be virtual reality (VR) devices, augmented reality (AR) devices, and smart glasses.

[0065] In one exemplary embodiment, such as Figure 1As shown, a method for determining arc reignition is provided. Taking the application of this method to a server as an example, the method includes the following steps:

[0066] S101: After the vacuum circuit breaker disconnects the power system circuit, obtain the target spatiotemporal coordinates of the vacuum circuit breaker.

[0067] The target spatiotemporal coordinate point (t, r, z0) can be understood as the time and space coordinates for the arc reignition assessment. The time coordinate t corresponds to different moments in the post-arc dielectric recovery process, and the space coordinates correspond to different positions within the vacuum circuit breaker electrode gap. It can be represented by a two-dimensional axisymmetric cylindrical coordinate system (r, z0), where r is the radial coordinate, belonging to the radial dimension of the two-dimensional axisymmetric cylindrical coordinate system, describing the position within the electrode gap along the radial direction (perpendicular to the electrode axis), such as the spatial position from the electrode axis (r=0) to the electrode edge (r=R, where R is the electrode radius), used to characterize the radial distribution difference of physical quantities; z0 is the axial coordinate, belonging to the axial dimension of the two-dimensional axisymmetric cylindrical coordinate system, describing the position within the electrode gap along the axial direction (parallel to the electrode axis, from the cathode to the anode), such as the spatial position of the cathode surface (z0=0) and the anode surface (z0=d, where d is the electrode gap distance), used to characterize the axial distribution difference of physical quantities.

[0068] After a vacuum circuit breaker disconnects the power system's circuit, the target spatiotemporal coordinates of the vacuum circuit breaker can be obtained. The target spatiotemporal coordinates can be determined according to the actual simulation requirements; it can be a single discrete coordinate point or a continuous spatiotemporal coordinate grid.

[0069] S102, input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities.

[0070] The target simulated physical quantities include at least one of electric potential, electron number density, ion number density, and electron temperature.

[0071] For example, see Figure 2 , Figure 2 A schematic diagram of the structure of a target simulation model is provided. The target simulation model includes an input layer 201, a hidden layer 202 and an output layer 203 connected in sequence. The hidden layer 202 includes a plurality of hidden units 2021 connected in sequence, and adjacent hidden units 2021 are connected by a self-attention module 2022.

[0072] The input layer receives the target operating condition parameters and target spatiotemporal coordinates, converting them into vector form. The hidden layer performs deep feature extraction and optimization through multiple hidden units and a self-attention module. The self-attention module calculates the correlation weights between feature vectors, allowing the target simulation model to adaptively focus on regions with drastic changes in the physical field gradient, such as the electrode sheath, thus improving the model's ability to capture multi-scale physical phenomena. The output layer outputs the corresponding target simulation physical quantity (φ, n) based on the feature vectors output by the hidden layer. e n i T e ), where φ is the electric potential, n e n is the electron number density. i T is the ion number density. e It represents the electron temperature.

[0073] S103, based on the target simulated physical quantities, determine the reignition assessment results of the electric arc.

[0074] For example, based on the obtained electric potential, electron number density, ion number density, and electron temperature, the characteristics of arc reignition can be analyzed to determine whether arc reignition has occurred. For instance, a potential cloud map can be generated based on the electric potential, and a particle density distribution map can be generated based on the electron and ion number densities, so that the arc reignition assessment result can be determined based on the potential cloud map and the particle density distribution map. Alternatively, a current value can be determined based on the obtained electric potential, electron number density, ion number density, and electron temperature, and then the arc reignition assessment result can be determined based on the relationship between the current value and a preset threshold.

[0075] The aforementioned method for determining arc reignition involves obtaining the target spatiotemporal coordinates of the vacuum circuit breaker after it disconnects the power system circuit. These coordinates include both temporal and spatial coordinates; the temporal coordinates correspond to different moments in the post-arc dielectric recovery process, while the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker. The target spatiotemporal coordinates are then input into a target simulation model to obtain target simulation physical quantities. These quantities include at least one of electric potential, electron number density, ion number density, and electron temperature. Based on these physical quantities, the arc reignition assessment result is determined. This approach, by inputting the target spatiotemporal coordinates into the target simulation model, allows for the rapid acquisition of target simulation physical quantities without simulating a massive number of particles, thus improving simulation efficiency.

[0076] In some alternative implementations, after the vacuum circuit breaker disconnects the power system circuit, the target operating parameters of the vacuum circuit breaker can also be obtained; among them, the target operating parameters include TRV waveform parameters and initial plasma density.

[0077] The target operating parameters include TRV waveform parameters and initial plasma density. Transient recovery voltage (TRV) refers to the voltage that rapidly rises across the electrode gap after the current crosses zero and the arc is initially extinguished. It is the most critical external excitation condition during the post-arc dielectric recovery process of the vacuum circuit breaker, directly determining whether "dielectric breakdown," i.e., arc reignition, will occur in the electrode gap. Initial plasma density refers to the initial concentration of plasma in the electrode gap at the instant the current crosses zero and the arc is initially extinguished (t=0), typically expressed as the number of particles per unit volume (e.g., particles / m³), such as electron number density and ion number density. For example, TRV waveform parameters can be acquired by a voltage sensor, and the initial plasma density can be determined based on the operating parameters of the vacuum circuit breaker and an empirical model.

[0078] Inputting the target spatiotemporal coordinates into the target simulation model yields the target simulation physical quantity. This can be achieved by matching the target operating condition parameters with the sample operating condition parameters used to train the target simulation model, and obtaining the matching result. If the matching result is a match, the target spatiotemporal coordinates are input into the target simulation model to obtain the target simulation physical quantity.

[0079] In this way, the target operating condition parameters are matched with the sample operating condition parameters of the target simulation model to determine whether the target operating condition parameters are within the range of the sample operating condition parameters used for model training. When the target operating condition parameters match the sample operating condition parameters used to train the target simulation model, the target spatiotemporal coordinate points are input into the target simulation model to obtain the target simulation physical quantities, making the obtained target simulation physical quantities more reliable.

[0080] In some alternative implementations, see [link to relevant documentation]. Figure 3 , Figure 3 A flowchart for obtaining target simulation physical quantities is provided, which specifically includes the following steps:

[0081] S301, input the target spatiotemporal coordinates into the input layer for feature extraction to obtain the basic feature vector.

[0082] For example, the input layer can normalize the received target spatiotemporal coordinates to obtain a normalized feature vector, thereby eliminating the dimensional differences between different parameters; then, the normalized feature vector is converted into a basic feature vector of a preset dimension through a linear transformation.

[0083] S302, the basic feature vector is input into the hidden layer for processing, and the weight feature vector output by the hidden layer is obtained.

[0084] The input data of the next hidden unit is the feature vector obtained by processing the output data of the previous hidden unit through the self-attention module; the input data of the first hidden unit is the basic feature vector, and the output data of the last hidden unit is the weight feature vector.

[0085] Each self-attention module is used to perform a weighted summation of each of the multi-dimensional features in the output data of the previous hidden unit, based on the correlation between the dimensional feature and each of the other dimensional features, to obtain the updated feature corresponding to the dimensional feature; and based on the updated features corresponding to each dimensional feature, to obtain the fused feature vector for the input data; the fused feature vector of the input data is the input data of the next hidden unit.

[0086] For example, the self-attention module receives each dimension of the multi-dimensional features from the output data of the previous hidden unit, calculates the correlation score between this dimension and each other dimension, converts the correlation score into attention weights using a softmax function, and then performs a weighted summation of the attention weights for each dimension to obtain a fused feature vector that incorporates global information. This fused feature vector sequence serves as the input data for the next hidden unit. After processing by all hidden units and the self-attention module, the final hidden unit outputs a weighted feature vector. The activation function of the hidden unit can be the tanh function.

[0087] For example, feature extraction using a self-attention module enables the model to dynamically allocate computational resources, assigning higher weights to physically important regions (e.g., high-gradient sheath regions), thereby improving the simulation accuracy of the model without increasing network complexity.

[0088] S303 inputs the weighted feature vector to the output layer for feature recognition to obtain the target simulated physical quantity.

[0089] For example, the output layer uses a linear transformation to map the weighted feature vectors to the corresponding target simulation physical quantity space, and outputs specific values ​​of physical quantities such as electric potential, electron number density, ion number density, and electron temperature.

[0090] In the above embodiments, by setting a self-attention module between two adjacent hidden units, more weights can be assigned to important features, thereby improving the accuracy of the target simulation physical quantity output by the model.

[0091] In some alternative implementations, see [link to relevant documentation]. Figure 4 , Figure 4 A flowchart for visually displaying target simulation physical quantities is provided, which includes the following steps:

[0092] S401 generates a potential cloud map, a particle density distribution map, and a dynamic map of sheath development based on the target simulated physical quantities.

[0093] Among them, the potential cloud map is used to display the spatial distribution of potential in the electrode gap at different times. Different colors or gray levels represent the potential level, which can intuitively reflect the propagation and distribution of transient recovery voltage in the electrode gap.

[0094] Particle density distribution maps are used to show the spatiotemporal evolution of electron number density and ion number density, and can present the variation of carrier concentration with time and space.

[0095] The dynamic map of sheath development is used to locate the position and thickness changes of the sheath by the dynamic changes of the particle density gradient. As a high gradient region near the electrode, the development state of the sheath directly affects the medium recovery process.

[0096] S402 displays the potential cloud map, particle density distribution map, and sheath development dynamic map.

[0097] For example, the generated potential cloud map, particle density distribution map, and sheath development dynamic map can be displayed through the terminal's display device or the server's visualization interface, allowing users to intuitively understand the changes in the physical field during the post-arc dielectric recovery process. Users can analyze the key stages and weak points of dielectric recovery based on these charts, providing an intuitive reference for the structural optimization and parameter adjustment of the vacuum circuit breaker.

[0098] In the above embodiments, by visually displaying the potential cloud map, particle density distribution map, and sheath development dynamic map, users can intuitively understand the changes in the physical field during the post-arc dielectric recovery process, so as to efficiently optimize the structure and adjust the parameters of the vacuum circuit breaker.

[0099] In some alternative implementations, see [link to relevant documentation]. Figure 5 , Figure 5 A flowchart for determining the reignition assessment result of an electric arc is provided, which specifically includes the following steps:

[0100] S501, determine the current value based on the target simulated physical quantity.

[0101] The target simulated physical quantities include electric potential, electron number density, ion number density, and electron temperature. The current value can be calculated based on physical quantities such as electron number density, ion number density, and electric field strength. According to plasma physics theory, the current density is related to the carrier concentration (electron number density, ion number density) and carrier drift velocity. The carrier drift velocity can be determined by the electric field strength and mobility, and the electric field strength can be calculated by combining the potential distribution. Furthermore, the total current value can be obtained by integrating the current density over the entire electrode gap region.

[0102] For example, the electric field strength can first be calculated based on the electric potential, specifically using the following formula:

[0103]

[0104] in, The electric field intensity vector, For gradient operators, For electric potential, , , , , and , respectively, are the partial derivatives of the electric potential in the x, y, and z directions. , , These are unit vectors in the x, y, and z directions, respectively.

[0105] Furthermore, carrier mobility can be corrected based on electron temperature using the following formula:

[0106]

[0107] Among them, carrier mobility is a physical quantity that describes the ease with which carriers (electrons / ions) move in an electric field. Its value is directly related to the electron temperature (increasing temperature increases the thermal motion of carriers and reduces mobility).

[0108] Where e is the electron charge, q i m is the ionic charge. e For electron mass, m i For ion mass, For the collision time of electrons, The collision time of the ions. For electron mobility, This represents ion mobility.

[0109] Furthermore, the carrier drift velocity can be calculated based on the electric field strength and mobility using the following formula:

[0110]

[0111] in, For electron drift velocity, This represents the ion drift velocity.

[0112] Furthermore, the total current density can be calculated based on the carrier concentration and drift velocity using the following formula:

[0113]

[0114] in, The total current density, For electron current density, This represents the ion current density.

[0115] Finally, the total current value can be solved by integrating the current density using the following formula:

[0116]

[0117] Where I is the total current value, and S is the effective electrode area. This represents the electrode area element.

[0118] S502, if the current value is greater than the preset threshold, then arc reignition is determined.

[0119] For example, the preset threshold can be the critical current value for determining whether the arc will reignite, which can be determined based on the rated parameters, insulation requirements and experimental data of the vacuum circuit breaker.

[0120] When the calculated total current value is greater than the preset threshold, it indicates that the carrier concentration in the electrode gap is high enough, the dielectric fails to effectively restore its insulating properties, and dielectric breakdown will occur, thus confirming arc reignition; otherwise, it is confirmed that the arc has not reignited.

[0121] In the above embodiments, the current value is determined based on the target simulated physical quantity to facilitate the determination of arc reignition, providing a reliable theoretical basis for judging whether the arc has reignited and improving the reliability of the judgment result on whether the arc has reignited.

[0122] In some alternative implementations, see [link to relevant documentation]. Figure 6 , Figure 6 A flowchart illustrating the process of training a target simulation model is provided, specifically including the following steps:

[0123] S601: Input the sample operating parameters and sample spatiotemporal coordinates into the initial simulation model to obtain the predicted simulation physical quantities.

[0124] The sample operating parameters include multiple sets of different TRV waveform parameters and initial plasma density, covering the common operating conditions of vacuum circuit breakers. The sample spatiotemporal coordinates are selected from typical time periods during the post-arc dielectric recovery process and key spatial regions of the electrode gap, and can be obtained through uniform or random sampling.

[0125] The initial simulation model is an untrained neural network model with the same structure as the target simulation model, and its parameters (such as weights and biases) are randomly initialized. The sample operating parameters and sample spatiotemporal coordinates are input into the initial simulation model, and after processing by the input layer, hidden layer and output layer, the predicted simulated physical quantities are obtained.

[0126] S602, determine the model loss of the initial simulation model based on the predicted simulation physical quantities and constraints.

[0127] The constraints include physical equation constraints, boundary condition constraints, initial condition constraints, and data fitting constraints. Physical equation constraints are used to constrain the electromagnetic state, particle number, and thermodynamic state; boundary condition constraints are used to constrain the boundary parameters of the vacuum circuit breaker; initial condition constraints are used to constrain the initial state of the vacuum circuit breaker; and data fitting constraints are used to constrain the consistency between simulated and actual physical quantities.

[0128] For example, the physical equation constraints include a set of macroscopic fluid control equations describing the evolution of the back-arc plasma, specifically including the Poisson equation, the electronic continuity equation, the ion continuity equation, and the electronic energy conservation equation, to ensure that the model output conforms to physical laws.

[0129] Boundary condition constraints include potential conditions and particle flux conditions at boundaries such as electrodes and shields, such as the TRV waveform of cathode potential changing with time, anode grounding, and particle flux constraints.

[0130] For example, the cathode potential satisfies the following condition:

[0131]

[0132] The cathode potential satisfies the following condition:

[0133]

[0134] in, Let be the cathode potential at time t. Peak voltage, representing the maximum voltage value of the TRV waveform. For time index, This is the anode potential. The potential of the shielding cover.

[0135] The particle flux constraint satisfies the following conditions:

[0136]

[0137] in, Let n be the flux of particles along the boundary normal, and n be the particle number density. For particle drift velocity, The normal vector of the boundary.

[0138] Initial condition constraints refer to the initial state constraints such as plasma density and temperature distribution at time t=0.

[0139] Data fitting constraints refer to the fitting constraints between the model output and high-fidelity simulation data (such as PIC-MCC simulation data), which are used to improve the simulation accuracy of the model.

[0140] For example, a loss function can be constructed to quantify the deviation between the model output and the high-fidelity simulation data. The loss function is as follows:

[0141]

[0142] in, For loss function, For the i-th sampled value of the simulated physical quantity, N represents the corresponding sampled value of the actual physical quantity, and N is the number of sampling points.

[0143] If it is necessary to fit multiple physical quantities simultaneously, a weighted loss function can be used, for example:

[0144]

[0145] Where L is the total loss function and M is the number of physical quantities. The weight of the j-th physical quantity Let j be the j-th sampled value of the simulated physical quantity. These are the corresponding sampled values ​​of the actual physical quantities.

[0146] Furthermore, by minimizing the total loss function L through optimization algorithms (such as gradient descent and genetic algorithms), the parameters of the initial simulation model are adjusted so that the deviation between the simulated physical quantities and the actual physical quantities gradually decreases.

[0147] Repeat the iterative process of "simulation output → loss calculation → parameter optimization" until the total loss function L meets the accuracy requirements. At this point, the consistency between the simulated physical quantity and the actual physical quantity reaches the expected level, and the model accuracy is effectively improved.

[0148] For each type of constraint, it is necessary to derive the theoretical simulation physical quantity under the corresponding constraint based on the predicted simulation physical quantity (such as electric potential, electron number density, etc.) output by the initial simulation model, and then calculate the difference between the two (i.e., residual). For example, the physical equation residual is the difference between the calculated value and the theoretical value after substituting the predicted physical quantity into the equation, and the boundary residual is the difference between the predicted physical quantity and the boundary theoretical value, etc.

[0149] For example, the residuals between the predicted simulated physical quantities and each theoretical simulated physical quantity can be determined, and then the model loss can be determined based on the residuals between the predicted simulated physical quantities and each theoretical simulated physical quantity. Each theoretical simulated physical quantity is determined based on the predicted simulated physical quantity and each constraint condition.

[0150] For example, for physical equation constraints, the predicted simulated physical quantities can be substituted into the corresponding physical equations, and the residuals of the equations, i.e., the differences between the left and right sides of the equations, can be calculated using automatic differentiation techniques; for boundary condition constraints, the difference between the predicted simulated physical quantities and the theoretical values ​​specified by the boundary conditions can be calculated as the boundary residuals; for initial condition constraints, the difference between the predicted simulated physical quantities and the theoretical values ​​of the initial state can be calculated as the initial residuals; for data fitting condition constraints, the difference between the predicted simulated physical quantities and the high-fidelity simulation data can be calculated as the data residuals.

[0151] Furthermore, the model loss of the initial simulation model can be obtained by weighted summation of the residuals.

[0152] For example, to balance the importance of different residuals, a corresponding weighting coefficient is assigned to each residual. The weighting coefficient can be determined based on engineering experience or cross-validation. After summing the squares of the physical equation residuals, boundary residuals, initial residuals, and data fitting residuals, multiplying them by the corresponding weighting coefficients, and then summing all the weighted residuals, the model loss is obtained.

[0153] S603 optimizes the model parameters of the initial simulation model based on the model loss to obtain the target simulation model.

[0154] For example, the model parameters of the initial simulation model can be optimized based on the model loss. For instance, a gradient descent optimization algorithm (such as the Adaptive Moment Estimation (Adam) optimizer) can be used to calculate the gradient of the model loss relative to each parameter (weights, biases) of the initial simulation model. The parameter values ​​are then adjusted according to the gradient direction to reduce the model loss. During the optimization process, the loss signal is transmitted from the output layer to the input layer via backpropagation, updating the parameters of each layer sequentially.

[0155] If the model loss meets the iteration stopping condition, the model parameters are output, resulting in the target simulation model. The iteration stopping condition can be set to either the model loss being less than a preset loss threshold or the number of iterations reaching a preset maximum number of iterations. When the model loss is less than the preset loss threshold, it indicates that the model's prediction accuracy meets the requirements; when the number of iterations reaches the preset maximum number of iterations, iteration stops regardless of whether the model loss reaches the threshold. After the iteration stopping condition is met, the parameters of the current model are output, resulting in the trained target simulation model.

[0156] In the above embodiments, by setting various constraints to train the initial simulation model, the output accuracy of the target simulation model can be improved.

[0157] In some alternative implementations, see [link to relevant documentation]. Figure 7 , Figure 7A schematic diagram of an arc reignition determination system is provided, including:

[0158] The data preparation module provides all the necessary input data for model training, including:

[0159] The physical model configuration unit is used to define the set of macroscopic physical governing equations to be solved, mainly including the Poisson equation, the particle continuity equation and the energy conservation equation, and to configure simplified chemical reaction model parameters, such as the ionization reaction rate coefficient.

[0160] The initial and boundary condition acquisition unit is used to set the initial state of the simulation (plasma density, temperature distribution, etc. at t=0, which can be derived from a short PIC-MCC simulation or empirical model) and boundary conditions (such as time-varying TRV applied to the cathode, anode grounded, etc.).

[0161] A high-fidelity data generation unit (optional) is used to run one or more short-duration, high-fidelity PIC-MCC simulations and sample sparse spatiotemporal data points from them as supervision data for subsequent network training, thereby improving training accuracy and speed.

[0162] The model training module is used to build and train Physical Information Neural Networks (PINN) surrogate models, including:

[0163] Network architecture building blocks are used to construct a network that takes spatiotemporal coordinates (t, r, z0) as input and macroscopic physical quantities (φ, n) as input. e n i T e The output is a deep neural network. The key is that, within this unit, a self-attention module needs to be embedded between the fully connected layers of the network to improve the network's ability to resolve high-gradient regions.

[0164] The loss function definition unit is used to define a comprehensive loss function based on the physical model. This function must include at least: physical equation residuals (ensuring the network output satisfies the governing equations), boundary residuals, and initial residuals. Optionally, it may also include data fitting residuals.

[0165] The model optimization training unit is used to iteratively minimize the comprehensive loss function using a gradient descent algorithm (such as the Adam optimizer), thereby training the weights and bias parameters of the neural network until the loss function converges, resulting in a surrogate model that can accurately reproduce the physical process.

[0166] The simulation prediction module is used for rapid simulations using a pre-trained model. Its functions include:

[0167] The input request processing unit is used to receive any spatiotemporal coordinate point (t, r, z0) or a spatiotemporal coordinate grid specified by the user.

[0168] The fast solver unit is used to input the requested coordinates into the pre-trained PINN model, perform a single forward propagation calculation, and instantly obtain the physical field solution (φ, n) corresponding to that point. e n i T e ).

[0169] The results visualization and analysis unit is used to process the solved data and generate potential cloud maps, particle density distribution maps, sheath development dynamics, etc. It can also determine whether reignition has occurred based on the calculated total current, thus completing a complete simulation prediction.

[0170] See Figure 8 , Figure 8 A flowchart illustrating another method for determining arc reignition is provided, which includes the following steps:

[0171] Step S801: Construct the physics model and collision model.

[0172] First, the microscopic dynamics describing the evolution of the post-arc plasma are simplified and abstracted into a set of macroscopic fluid control equations. This set of equations includes at least: a Poisson equation describing the spatial potential φ, and an equation describing the electron number density n. e and ion number density n i The particle continuity equation for spacetime evolution, and the equation used to describe electron temperature T e The energy conservation equation for evolution.

[0173] Then, the complex microscopic collision process is characterized using a simplified chemical reaction model: the ionization source term S... iz Modeling as In the form of, where the ionization rate coefficient k iz Processed to electron temperature T e The function, n e n is the electron number density. n This represents the neutral particle number density. This functional relationship can be pre-calculated by fitting experimental data or using specialized software. In this way, complex physical collisions are encoded into a mathematical term in the governing equations.

[0174] Step S802: Prepare initial and boundary condition data.

[0175] To solve the above system of equations, the boundary value conditions for the simulation need to be defined. This step requires setting the initial plasma field distribution (n) at the simulation start time (t=0). e (0, r, z0), n i(0, r, z0), T e (0, r, z0)), and define the physical conditions on each boundary of the computational domain (such as cathode, anode, shield) throughout the simulation time, such as the function V_TRV(t) for the change of cathode potential over time.

[0176] Step S803: Construct the PINN architecture with attention mechanism.

[0177] This step constructs a special neural network. This network takes time t and spatial coordinates (r, z0) as input. The network output is the macroscopic physical quantity (φ, n) corresponding to that spatiotemporal point. e n i T e The main body of the network consists of multiple hidden layers (such as fully connected layers).

[0178] In this embodiment, a self-attention module is inserted between at least two hidden layers. This module works by receiving a batch of feature vectors output from the previous layer and calculating the correlation weights between each pair of these feature vectors. For regions with physically drastic gradient changes (such as the sheath near electrodes), the network learns during training to assign higher attention weights to sampling points in these regions. By weighted summing of the feature vectors, this module allows the network to "focus" on physically more important regions, thereby accurately capturing high-gradient phenomena such as those in the sheath with a limited number of parameters, avoiding the difficulties faced by traditional neural networks in such problems.

[0179] Step S804: Define and combine the physical information loss function.

[0180] To ensure that the output of the neural network conforms to physical laws, this step defines a comprehensive loss function L. total The function consists of a weighted sum of the following parts:

[0181] Physical equation residual L pde : The output of the network (φ, n) e n i T e Substitute all the governing equations defined in step S801. Calculate all derivative terms using the automatic differentiation function of the deep learning framework to obtain the "residuals" of the equations. This loss term is the mean square error of the residuals of all equations at a large number of randomly sampled points within the computational domain. Minimizing it forces the network to learn and satisfy the physical laws.

[0182] Boundary residual L bc Sample points on the boundary of the computational domain and calculate the mean square error between the network output and the true boundary value defined in step S802.

[0183] Initial residual L icAt the sampling point at t=0, calculate the mean square error between the network output and the initial state defined in step S802.

[0184] Data fitting residual L data (Optional): If true data is prepared, calculate the mean squared error of the network output relative to the "true data" at these specific data points.

[0185] Step S805: Perform model training.

[0186] In this step, the system employs a gradient descent optimization algorithm (such as Adam) to iteratively adjust all weights and bias parameters of the neural network to minimize the comprehensive loss function L defined in step S804. total The training process continues until the loss function converges to a sufficiently small value, indicating that the neural network's output can simultaneously satisfy all constraints. At this point, the target simulation model is obtained.

[0187] Step S806: Perform rapid simulation prediction.

[0188] For any given operating condition (within the parameter space covered by training), the user only needs to provide the desired spatiotemporal coordinates (t, r, z0), input these coordinates into the target simulation model, and obtain accurate physical field results through a single forward computation without iteration. Because neural network computation is highly efficient, the entire simulation process (e.g., obtaining solutions for tens of thousands of spatiotemporal points and plotting contour maps) can be completed within minutes, achieving a speedup of several orders of magnitude compared to the traditional PIC-MCC method.

[0189] This application employs the PINN (Peripheral-Induced Neural Network) technology to achieve rapid simulation of the post-arc breaking process in a vacuum. Traditional high-fidelity simulation methods such as PIC-MCC typically require weeks or even months to complete a single calculation. In contrast, this application constructs and trains a PINN surrogate model, transforming the complex physical process into a single, efficient neural network forward computation. Once the model is trained, obtaining solutions for tens of thousands of spatiotemporal points and plotting contour maps takes only minutes, achieving an acceleration of several orders of magnitude compared to existing technologies. This enables large-scale parameter optimization and design iteration.

[0190] Furthermore, this application embodiment effectively addresses the technical challenge of insufficient accuracy in processing high-gradient regions such as the sheath near electrodes by embedding a self-attention mechanism into the PINN network. The self-attention module allows the network to automatically "focus" on physically more important regions during training, accurately capturing multi-scale physical phenomena with a limited number of parameters. Simultaneously, this application embodiment simplifies complex microscopic collision physics (such as electron collision ionization) into a function of macroscopic parameters (electron temperature) and encodes it into the PINN loss function. This ensures the physical accuracy of the simulation while avoiding the cumbersome microscopic calculations of traditional methods, thus providing an end-to-end, complete, and efficient solution for the design and optimization of vacuum circuit breakers.

[0191] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0192] Based on the same inventive concept, this application also provides an arc reignition determination apparatus for implementing the arc reignition determination method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the arc reignition determination apparatus provided below can be found in the limitations of the arc reignition determination method described above, and will not be repeated here.

[0193] In one exemplary embodiment, such as Figure 9 As shown, an arc reignition determination device is provided, comprising:

[0194] The acquisition module 10 is used to acquire the target spatiotemporal coordinates of the vacuum circuit breaker after the vacuum circuit breaker disconnects the power system circuit; wherein, the target spatiotemporal coordinates include time coordinates and spatial coordinates, the time coordinates correspond to different times in the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions in the electrode gap of the vacuum circuit breaker.

[0195] The simulation module 20 is used to input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities; wherein, the target simulation physical quantities include at least one of electric potential, electron number density, ion number density and electron temperature;

[0196] The determination module 30 is used to determine the reignition assessment result of the electric arc based on the target simulated physical quantities.

[0197] The aforementioned method for determining arc reignition involves obtaining the target spatiotemporal coordinates of the vacuum circuit breaker after it disconnects the power system circuit. These coordinates include both temporal and spatial coordinates; the temporal coordinates correspond to different moments in the post-arc dielectric recovery process, while the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker. The target spatiotemporal coordinates are then input into a target simulation model to obtain target simulation physical quantities. These quantities include at least one of electric potential, electron number density, ion number density, and electron temperature. Based on these physical quantities, the arc reignition assessment result is determined. This approach, by inputting the target spatiotemporal coordinates into the target simulation model, allows for the rapid acquisition of target simulation physical quantities without simulating a massive number of particles, thus improving simulation efficiency.

[0198] In one embodiment, the target simulation model includes a sequentially connected input layer, a hidden layer, and an output layer. The hidden layer includes multiple sequentially connected hidden units, with adjacent hidden units connected by a self-attention module. The simulation module 20 is specifically used for:

[0199] The target spatiotemporal coordinates are input into the input layer for feature extraction, yielding a basic feature vector. This basic feature vector is then input into the hidden layer for processing, resulting in a weighted feature vector output by the hidden layer. The input data for the next hidden unit is the feature vector obtained after processing the output data of the previous hidden unit by a self-attention module. The input data for the first hidden unit is the basic feature vector, and the output data for the last hidden unit is the weighted feature vector. The weighted feature vector is then input into the output layer for feature recognition, yielding the target simulated physical quantity. Each self-attention module is used to perform weighted summation on each of the multi-dimensional features included in the output data of the previous hidden unit, based on the correlation between the dimensional feature and each other dimensional feature, to obtain the updated feature corresponding to the dimensional feature. Based on the updated features corresponding to each dimensional feature, a fused feature vector for the input data is obtained. This fused feature vector of the input data becomes the input data for the next hidden unit.

[0200] In one embodiment, the device further includes a display module for:

[0201] Based on the target simulated physical quantities, a potential cloud map, a particle density distribution map, and a sheath development dynamic map are generated. Among them, the potential cloud map is used to show the spatial distribution of potential in the electrode gap at different times, the particle density distribution map is used to show the spatiotemporal evolution of electron number density and ion number density, and the sheath development dynamic map is used to locate the position and thickness changes of the sheath through the dynamic changes of particle density gradient. The potential cloud map, particle density distribution map, and sheath development dynamic map are then displayed.

[0202] In one embodiment, the determining module 30 is specifically used for:

[0203] The current value is determined based on the target simulated physical quantity; if the current value is greater than the preset threshold, the arc reignition is determined.

[0204] In one embodiment, the acquisition module 10 is further configured to acquire target operating condition parameters of the vacuum circuit breaker after the vacuum circuit breaker disconnects the power system circuit; wherein the target operating condition parameters include transient recovery voltage (TRV) waveform parameters and initial plasma density.

[0205] The simulation module 20 is specifically used to match the target operating condition parameters with the sample operating condition parameters used to train the target simulation model to obtain the matching result; if the matching result is a match, the target spatiotemporal coordinate points are input into the target simulation model to obtain the target simulation physical quantity.

[0206] In one embodiment, the device further includes a training module for:

[0207] The sample operating parameters and sample spatiotemporal coordinates are input into the initial simulation model to obtain the predicted simulation physical quantities. Based on the predicted simulation physical quantities and constraints, the model loss of the initial simulation model is determined. The model parameters of the initial simulation model are optimized based on the model loss to obtain the target simulation model. The constraints include physical equation constraints, boundary condition constraints, initial condition constraints, and data fitting condition constraints. Physical equation constraints are used to constrain the electromagnetic state, particle number, and thermodynamic state. Boundary condition constraints are used to constrain the boundary parameters of the vacuum circuit breaker. Initial condition constraints are used to constrain the initial state of the vacuum circuit breaker. Data fitting condition constraints are used to constrain the consistency between the simulated physical quantities and the actual physical quantities.

[0208] In one embodiment, the training module is specifically used for:

[0209] Determine the residuals between the predicted simulated physical quantities and each theoretical simulated physical quantity; wherein each theoretical simulated physical quantity is determined based on the predicted simulated physical quantity and each constraint condition; and perform a weighted summation of each residual to obtain the model loss of the initial simulation model.

[0210] Each module in the aforementioned arc reignition determination device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0211] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 10 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data about vacuum circuit breakers. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When executed by the processor, the computer program implements a method for determining arc reignition.

[0212] Those skilled in the art will understand that Figure 10 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0213] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the arc reignition determination method described in any of the above embodiments.

[0214] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the arc reignition determination method described in any of the above embodiments.

[0215] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the arc reignition determination method described in any of the above embodiments.

[0216] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0217] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0218] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An arc restrike determination method, characterized by, The method includes: After the vacuum circuit breaker disconnects the power system circuit, the target spatiotemporal coordinates of the vacuum circuit breaker are obtained; wherein, the target spatiotemporal coordinates include time coordinates and spatial coordinates, the time coordinates correspond to different moments in the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker, which are used to characterize the radial and axial distribution differences of physical quantities. The target spatiotemporal coordinates are input into the target simulation model to obtain the target simulation physical quantities; wherein, the target simulation physical quantities include at least one of electric potential, electron number density, ion number density and electron temperature; Based on the target simulated physical quantities, determine the reignition assessment results of the electric arc; The target simulation model is trained in the following way: Input the sample operating parameters and sample spatiotemporal coordinates into the initial simulation model to obtain the predicted simulation physical quantities; Based on the predicted simulation physical quantities and constraints, determine the model loss of the initial simulation model; The model parameters of the initial simulation model are optimized based on the model loss to obtain the target simulation model; The constraints include physical equation constraints, boundary condition constraints, initial condition constraints, and data fitting condition constraints. The physical equation constraints are used to constrain the electromagnetic state, particle number, and thermodynamic state. The boundary condition constraints are used to constrain the boundary parameters of the vacuum circuit breaker. The initial condition constraints are used to constrain the initial state of the vacuum circuit breaker. The data fitting condition constraints are used to constrain the consistency between the simulated physical quantities and the actual physical quantities.

2. The method of claim 1, wherein, The target simulation model includes an input layer, a hidden layer and an output layer connected in sequence. The hidden layer includes multiple hidden units connected in sequence, and adjacent hidden units are connected by a self-attention module. The step of inputting the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities includes: The target spatiotemporal coordinates are input into the input layer for feature extraction to obtain the basic feature vector. The basic feature vector is input into the hidden layer for processing to obtain the weight feature vector output by the hidden layer; wherein, the input data of the next hidden unit is the feature vector obtained by processing the output data of the previous hidden unit through the self-attention module; the input data of the first hidden unit is the basic feature vector, and the output data of the last hidden unit is the weight feature vector; The weighted feature vector is input into the output layer for feature recognition to obtain the target simulated physical quantity; Each self-attention module is used to perform a weighted summation of each of the multi-dimensional features included in the output data of the previous hidden unit, based on the correlation between the multi-dimensional feature and each of the other multi-dimensional features, to obtain the updated feature corresponding to the multi-dimensional feature; and based on the updated features corresponding to each multi-dimensional feature, to obtain the fused feature vector for the input data; the fused feature vector of the input data is the input data of the next hidden unit.

3. The method of claim 1, wherein, The method further includes: Based on the target simulated physical quantities, an electric potential cloud map, a particle density distribution map, and a sheath development dynamic map are generated. The electric potential cloud map is used to display the spatial distribution of electric potential in the electrode gap at different times. The particle density distribution map is used to display the spatiotemporal evolution of electron number density and ion number density. The sheath development dynamic map is used to locate the position and thickness changes of the sheath by the dynamic changes of the particle density gradient. The potential cloud map, the particle density distribution map, and the sheath development dynamic map are displayed.

4. The method of claim 1, wherein, The step of determining the reignition assessment result of the electric arc based on the target simulated physical quantity includes: Determine the current value based on the target simulated physical quantity; If the current value is greater than a preset threshold, then arc reignition is determined.

5. The method of claim 1, wherein, After the vacuum circuit breaker disconnects the power system circuit, the method further includes: Obtain the target operating condition parameters of the vacuum circuit breaker; wherein, the target operating condition parameters include the transient recovery voltage (TRV) waveform parameters and the initial plasma density; The step of inputting the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantities includes: The target operating condition parameters are matched with the sample operating condition parameters used to train the target simulation model to obtain the matching result; If the matching result is a match, the target spatiotemporal coordinates are input into the target simulation model to obtain the target simulation physical quantity.

6. The method of claim 1, wherein, The step of determining the model loss of the initial simulation model based on the predicted simulation physical quantities and constraints includes: Determine the residual between the predicted simulated physical quantity and each theoretical simulated physical quantity; wherein each theoretical simulated physical quantity is determined based on the predicted simulated physical quantity and each constraint condition; The model loss of the initial simulation model is obtained by weighted summation of the residuals.

7. An arc restriking determination device characterized by comprising: The device includes: The acquisition module is used to acquire the target spatiotemporal coordinates of the vacuum circuit breaker after the vacuum circuit breaker disconnects the power system circuit; wherein, the target spatiotemporal coordinates include time coordinates and spatial coordinates, the time coordinates correspond to different moments in the post-arc dielectric recovery process, and the spatial coordinates correspond to different positions within the electrode gap of the vacuum circuit breaker, which are used to characterize the radial and axial distribution differences of physical quantities. The simulation module is used to input the target spatiotemporal coordinates into the target simulation model to obtain the target simulation physical quantity; wherein, the target simulation physical quantity includes at least one of electric potential, electron number density, ion number density and electron temperature; The determination module is used to determine the reignition assessment result of the electric arc based on the target simulated physical quantity; The training module is used to input sample operating parameters and sample spatiotemporal coordinates into the initial simulation model to obtain predicted simulation physical quantities; based on the predicted simulation physical quantities and constraints, the model loss of the initial simulation model is determined; the model parameters of the initial simulation model are optimized based on the model loss to obtain the target simulation model; wherein, the constraints include physical equation constraints, boundary condition constraints, initial condition constraints, and data fitting condition constraints; the physical equation constraints are used to constrain the electromagnetic state, particle number, and thermodynamic state; the boundary condition constraints are used to constrain the boundary parameters of the vacuum circuit breaker; the initial condition constraints are used to constrain the initial state of the vacuum circuit breaker; and the data fitting condition constraints are used to constrain the consistency between the simulated physical quantities and the actual physical quantities.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images