A method for temperature prediction of digital cable and related products

Abstract

The application discloses a method for temperature prediction of a digital cable and related products, and the method comprises the following steps: acquiring a digital cable temperature data set, wherein the digital cable temperature data set comprises multi-dimensional temperature data and various single-dimensional temperature data; constructing a plurality of physical information neural networks based on the various single-dimensional temperature data; fusing the features of the plurality of physical information neural networks to obtain a fusion result, wherein the fusion result is used for initializing the parameters of a preset initial physical information neural network; and training the initial physical information neural network based on the multi-dimensional temperature data to obtain a target physical information neural network, wherein the target physical information neural network is used for predicting the temperature distribution of the digital cable under the conditions of a given carrying capacity and a radial position. In this way, the stability and accuracy of temperature prediction of the digital cable under multiple working conditions are effectively improved.

Description

Technical Field

[0001] This application relates to the fields of power systems and cyber-physical systems, and in particular to a method for temperature prediction of digital cables and related products. Background Technology

[0002] With the rapid development of smart grids and cyber-physical systems (CPS), power cables, as a critical infrastructure for power transmission, are facing higher operational requirements and increasingly complex electromagnetic environments. As a representative of the next generation of smart cables, digital cables not only need to meet traditional power transmission requirements but also undertake communication and monitoring tasks, involving the coupling of multiple physical fields such as electricity, magnetism, and heat during their operation. Therefore, conducting accurate multi-physics analysis and simulation of digital cables is particularly important.

[0003] Traditional digital cable analysis methods are typically based on single-physics models. For example, electric field analysis primarily studies electrical properties such as voltage, current, and insulation performance, often employing numerical methods like the finite element method and boundary element method. Thermal field analysis, on the other hand, mainly studies the heat generated during cable operation and the resulting heat transfer processes such as conduction, convection, and radiation, commonly using methods like the thermal network method and the finite volume method. However, these single-physics model analysis methods cannot effectively characterize the coupling effects between multiple physical fields (electric, magnetic, and thermal), making it difficult to accurately predict the temperature of digital cables. Consequently, they fail to fully reflect the true operating state of digital cables under complex conditions. Summary of the Invention

[0004] This application provides a method and related products for predicting the temperature of digital cables, which can achieve accurate temperature prediction of digital cables.

[0005] In a first aspect, embodiments of this application provide a method for predicting the temperature of a digital cable, the method comprising:

[0006] The electro-magnetic-thermal coupling control equations related to the digital cable are obtained, and the digital cable temperature dataset is simulated based on the electro-magnetic-thermal coupling control equations. The digital cable temperature dataset includes multi-dimensional temperature data and various single-dimensional temperature data. The multi-dimensional temperature data includes temperature data under different radial positions and different current carrying combinations. The single-dimensional temperature data includes axial temperature data under different current carrying and temperature data under different radial positions.

[0007] Based on the various single-dimensional temperature data, multiple physical information neural networks are constructed respectively;

[0008] The features of the multiple physical information neural networks are fused to obtain a fusion result, which is used to initialize the parameters of a preset initial physical information neural network.

[0009] The initial physical information neural network is trained based on the multi-dimensional temperature data to obtain a target physical information neural network, which is used to predict the temperature distribution of the digital cable under given current carrying capacity and radial position conditions.

[0010] One feasible implementation involves constructing multiple physical information neural networks based on the various single-dimensional temperature data, including:

[0011] A first physical information neural network is constructed based on the first training data and the first objective loss function. The first training data consists of shaft core temperature data under different current carrying capacities. The first objective loss function includes a first data loss function and a physical loss function. The first data loss function is used to measure the error between the shaft core temperature predicted by the first physical information neural network and the corresponding shaft core temperature in the first training data. The physical loss function is used to constrain the shaft core temperature predicted by the first physical information neural network to satisfy the electro-magnetic-thermal coupling control equation related to the digital cable.

[0012] One feasible implementation involves constructing multiple physical information neural networks based on the various single-dimensional temperature data, including:

[0013] A second physical information neural network is constructed based on the second training data and the second objective loss function. The second training data consists of temperature data at different radial positions. The second objective loss function includes a second data loss function and a boundary loss function. The second data loss function measures the error between the temperature predicted by the second physical information neural network at the radial position and the temperature at the corresponding radial position in the second training data. The boundary loss function constrains the predicted temperature of the second physical information neural network at the radial boundary position to meet a preset radial boundary condition. The radial boundary position includes the axis position and the outermost position of the digital cable.

[0014] One feasible implementation, wherein training the initial physical information neural network based on the multi-dimensional temperature data to obtain the target physical information neural network, includes:

[0015] The initial physical information neural network is trained based on the multi-dimensional temperature data and the third objective loss function to obtain the target physical information neural network. The third objective loss function includes a third data loss function, a third physical loss function, and a third boundary loss function. The third data loss function is used to measure the error between the temperature predicted by the initial physical information neural network under different radial positions and different current carrying capacity combinations and the corresponding temperature in the multi-dimensional temperature data. The third physical loss function is used to constrain the predicted temperature of the initial physical information neural network under multi-dimensional temperature data conditions to satisfy the electro-magnetic-thermal coupling control equation. The third boundary loss function is used to constrain the predicted temperature of the initial physical information neural network at the radial boundary position to satisfy the preset radial boundary condition.

[0016] One feasible implementation, prior to obtaining the electro-magnetic-thermal coupling control equations associated with the digital cable, further includes:

[0017] Based on Maxwell's equations, the electromagnetic control equations of the digital cable are established.

[0018] Based on Fourier's heat transfer law and the law of conservation of energy, the heat transfer control equation of the digital cable is established.

[0019] Based on the conductivity-temperature relationship of each layer of the digital cable, the electromagnetic control equation and the heat transfer control equation are coupled to obtain the electro-magnetic-thermal coupling control equation of the digital cable.

[0020] One feasible implementation, wherein the digital cable temperature dataset obtained by simulation based on the electro-magnetic-thermal coupling control equations includes:

[0021] A simulation model of the digital cable is built based on the structural parameters related to the digital cable.

[0022] Based on the electro-magnetic-thermal coupling control equations and the digital cable simulation model, simulation boundary conditions and initial parameters are set.

[0023] Based on the simulation boundary conditions and initial parameters, the digital cable simulation model is subjected to electro-magnetic-thermal coupling simulation under various current carrying capacities. After the simulation, the temperature dataset of the digital cable is obtained.

[0024] One feasible implementation involves fusing the features of the plurality of physical information neural networks to obtain a fusion result, the fusion result being used to initialize the parameters of a preset initial physical information neural network, including:

[0025] Based on a preset splicing order, the hidden layer output feature vectors of each of the multiple physical information neural networks are spliced ​​together to obtain a fusion result.

[0026] Secondly, embodiments of this application provide an apparatus for predicting the temperature of a digital cable, comprising:

[0027] The data acquisition module is used to acquire the electro-magnetic-thermal coupling control equations related to the digital cable, and to simulate the digital cable temperature dataset based on the electro-magnetic-thermal coupling control equations. The digital cable temperature dataset includes multi-dimensional temperature data and various single-dimensional temperature data. The multi-dimensional temperature data includes temperature data under different radial positions and different current carrying combinations. The single-dimensional temperature data includes axial temperature data under different current carrying capacities and temperature data under different radial positions.

[0028] The neural network construction module is used to construct multiple physical information neural networks based on the various single-dimensional temperature data.

[0029] The neural network fusion module is used to fuse the features of the multiple physical information neural networks to obtain a fusion result, which is used to initialize the parameters of a preset initial physical information neural network.

[0030] The target network training module is used to train the initial physical information neural network based on the multi-dimensional temperature data to obtain the target physical information neural network, which is used to predict the temperature distribution of the digital cable under given current carrying capacity and radial position conditions.

[0031] Thirdly, embodiments of this application provide an electronic device, the device including: a processor, a memory, and a system bus;

[0032] The processor and the memory are connected via the system bus;

[0033] The memory is used to store a program, the program including instructions that, when executed by the processor, cause the processor to perform any of the implementation steps of the method for predicting the temperature of a digital cable described above.

[0034] Fourthly, embodiments of this application provide a computer-readable storage medium for storing a computer program, which, when executed by a terminal device, implements any of the implementation steps of the method for predicting the temperature of a digital cable described above.

[0035] As can be seen from the above technical solutions, the embodiments of this application have the following advantages:

[0036] In this embodiment, firstly, the electro-magnetic-thermal coupling control equations related to the digital cable are obtained, and a digital cable temperature dataset is simulated based on the electro-magnetic-thermal coupling control equations. The digital cable temperature dataset includes multi-dimensional temperature data and various single-dimensional temperature data. The multi-dimensional temperature data includes temperature data under different radial positions and different current-carrying combinations, while the single-dimensional temperature data includes axial temperature data under different current-carrying conditions and temperature data under different radial positions. Next, multiple physical information neural networks are constructed based on the various single-dimensional temperature data. Subsequently, the features of the multiple physical information neural networks are fused, and the fusion result is used to initialize the parameters of a preset initial physical information neural network. Finally, the initial physical information neural network is trained based on the multi-dimensional temperature data to obtain a target physical information neural network, which is used to predict the temperature distribution of the digital cable under given current-carrying and radial position conditions.

[0037] As can be seen, this scheme constructs multiple physical information neural networks (PINs) based on single-dimensional temperature data obtained from simulations of the electro-magnetic-thermal coupling control equations. Each PIN learns independently for axial temperature data under different current-carrying capacities or temperature data at different radial positions. Then, the features of the multiple PINs are fused, and the fusion result is used to initialize the parameters of a pre-set initial PIN. This allows the initial PIN to directly utilize the temperature change features learned from the multiple PINs during the initial training phase. Subsequently, the initial PIN is trained based on multi-dimensional temperature data to obtain a target PIN. This target PIN can predict the temperature distribution of digital cables under given current-carrying capacities and radial positions, thereby improving the stability and accuracy of temperature prediction for digital cables under various operating conditions. Compared to existing technologies, this scheme, by constructing multi-dimensional PINs and performing feature fusion and initialization training, effectively overcomes the problem of insufficient accuracy in predicting digital cable temperature under complex operating conditions using traditional single-physics methods. Attached Figure Description

[0038] Figure 1 A flowchart illustrating a method for predicting the temperature of a digital cable, provided as an embodiment of this application;

[0039] Figure 2 This is a schematic diagram of a device for predicting the temperature of a digital cable, provided in an embodiment of this application. Detailed Implementation

[0040] As mentioned earlier, traditional digital cable analysis methods are typically based on single-physics models. For example, electric field analysis mainly studies electrical properties such as voltage, current, and insulation performance, often employing numerical methods like the finite element method and boundary element method. Thermal field analysis, on the other hand, primarily studies the heat generated during cable operation and the resulting heat transfer processes such as conduction, convection, and radiation, commonly using methods like the thermal network method and the finite volume method. However, these single-physics model analysis methods cannot effectively characterize the coupling effects between multiple physical fields (electric, magnetic, and thermal), making it difficult to accurately predict the temperature of digital cables. Consequently, they fail to fully reflect the true operating state of digital cables under complex conditions.

[0041] To address the aforementioned issues, this application provides a method and related products for temperature prediction of digital cables. First, the electro-magnetic-thermal coupling control equations related to the digital cable are obtained, and a digital cable temperature dataset is simulated based on these equations. This dataset includes multi-dimensional temperature data and various single-dimensional temperature data. The multi-dimensional data includes temperature data under different radial positions and current-carrying combinations, while the single-dimensional data includes axial temperature data under different current-carrying conditions and temperature data under different radial positions. Next, multiple physical information neural networks are constructed based on the various single-dimensional temperature data. Then, the features of these multiple physical information neural networks are fused, and the fusion result is used to initialize the parameters of a preset initial physical information neural network. Finally, the initial physical information neural network is trained based on the multi-dimensional temperature data to obtain a target physical information neural network, which is used to predict the temperature distribution of the digital cable under given current-carrying and radial position conditions.

[0042] As can be seen, this scheme constructs multiple physical information neural networks (PINs) based on single-dimensional temperature data obtained from simulations of the electro-magnetic-thermal coupling control equations. Each PIN learns independently for axial temperature data under different current-carrying capacities or temperature data at different radial positions. Then, the features of the multiple PINs are fused, and the fusion result is used to initialize the parameters of a pre-set initial PIN. This allows the initial PIN to directly utilize the temperature change features learned from the multiple PINs during the initial training phase. Subsequently, the initial PIN is trained based on multi-dimensional temperature data to obtain a target PIN. This target PIN can predict the temperature distribution of digital cables under given current-carrying capacities and radial positions, thereby improving the stability and accuracy of temperature prediction for digital cables under various operating conditions. Compared to existing technologies, this scheme, by constructing multi-dimensional PINs and performing feature fusion and initialization training, effectively overcomes the problem of insufficient accuracy in predicting digital cable temperature under complex operating conditions using traditional single-physics methods.

[0043] It should be noted that the implementation of the temperature prediction method for digital cables in this application is not limited to a single entity. For example, the temperature prediction method for digital cables in this application can be applied to information processing devices such as servers or terminal devices. The server can be a standalone server, a cluster server, or a cloud server. The terminal device can be an electronic device such as a smartphone, computer, personal digital assistant (PDA), or tablet computer.

[0044] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0045] Figure 1 This is a flowchart illustrating a method for temperature prediction of a digital cable, provided as an embodiment of this application. (In conjunction with...) Figure 1 As shown, it may include steps S101-S104.

[0046] S101: Obtain the electro-magnetic-thermal coupling control equations related to the digital cable, and simulate the digital cable temperature dataset based on the electro-magnetic-thermal coupling control equations. The digital cable temperature dataset includes multi-dimensional temperature data and various single-dimensional temperature data. The multi-dimensional temperature data includes temperature data under different radial positions and different current carrying combinations. The single-dimensional temperature data includes axial temperature data under different current carrying capacities and temperature data under different radial positions.

[0047] In this embodiment, before obtaining the electro-magnetic-thermal coupling control equations related to the digital cable, it is first necessary to establish the electromagnetic control equations of the digital cable based on Maxwell's equations. Specifically, the vector magnetic potential equations for each region in the digital cable are obtained by deriving Maxwell's equations:

[0048] ;

[0049] in, For the Laplace operator, For magnetic vector position, The magnetic permeability of the material, For source current density, This represents the conductor region where the source current resides. By solving this vector magnetic potential equation, the magnetic vector potential under a given current distribution can be obtained. And further calculate the electric field strength caused by the source current. :

[0050] ;

[0051] Subsequently, the current distribution of the digital cable in each layer of the material structure can be obtained:

[0052] ;

[0053] in, For digital cables in the first Current density of the layer material For digital cables in the first The electrical conductivity of the layer material. Based on the current distribution in each layer, the power loss per unit length of each layer can be further calculated:

[0054] ;

[0055] in, For digital cables in the first Power loss per unit length of the layer material For digital cables in the first The volumetric elements of the layered material are represented by the above equations. Under different current conditions, the current distribution and power loss of each layer of the digital cable can be accurately calculated, providing an electromagnetic field basis for subsequent temperature prediction.

[0056] The above derivation shows that the electromagnetic control equation of digital cable consists of vector magnetic potential equation, electric field and magnetic field relationship, and current distribution equation of each layer of material. It can accurately describe the current distribution and loss in digital cable under given current conditions, and provides a mathematical basis for establishing electro-magnetic-thermal coupling control equation.

[0057] Next, the heat transfer control equations for the digital cable are established based on Fourier's law of heat transfer and the law of conservation of energy. Specifically, the local heat transfer process of the digital cable is analyzed using the infinitesimal element method. According to the law of conservation of energy, the heat flowing into the infinitesimal element... and the heat generated inside the micro-element The sum of these equals the heat released by the infinitesimal element. The increase in internal energy of a infinitesimal element per unit time ,Right now:

[0058] ;

[0059] Combining Fourier's law of heat transfer, the heat transfer control equation for digital cables can be obtained:

[0060] ;

[0061] in, For the density of digital cable materials, The constant-pressure specific heat capacity of digital cable materials. For the fluid velocity field, For the temperature gradient of the digital cable, For time, The heat flux density vector, This refers to the heat generation power density per unit volume of a digital cable.

[0062] According to Fourier's law, the heat flux density vector satisfies:

[0063] ;

[0064] in, Let be the thermal conductivity of the digital cable material. For the surface of a digital cable, heat dissipation mainly includes two parts: heat convection and heat radiation, with corresponding heat flux densities of convective heat flux density and radiative heat flux density, respectively. and radiation convection density The calculation formula is as follows:

[0065] ;

[0066] ;

[0067] in, The heat transfer coefficient is... For emission rate, The temperature of the environment in which the digital cable is located. The surface temperature of the digital cable. This represents the Stefan-Boltzmann constant. Based on the above formula, the heat transfer on the surface of the digital cable and the total heat flux density on the surface of the digital cable can be calculated. :

[0068] ;

[0069] Therefore, based on the law of conservation of energy and Fourier's law of heat transfer, the heat transfer control equation for the temperature variation within the digital cable over time and space can be obtained. Simultaneously, the convective heat flux density can be used to... and radiative heat flux density Calculate the total heat flux density on the surface of the digital cable. This allows for a complete depiction of the internal heat transfer and surface heat dissipation processes of digital cables.

[0070] Furthermore, based on the conductivity-temperature relationship of each layer of the digital cable material, the electromagnetic control equation and the heat transfer control equation can be coupled to obtain the electro-magnetic-thermal coupled control equation of the digital cable. Specifically, for digital cables, this means the heat generation power density... With the square of the current density Related:

[0071] ;

[0072] The conductivity of digital cable materials changes with temperature; therefore, when heating causes temperature changes, the digital cable... Electrical conductivity of layer material It will also change accordingly, and the relationship can be expressed as:

[0073] ;

[0074] in, This is the reference temperature for digital cables. For the first The electrical conductivity of the layer material at 20°C The temperature coefficient of electrical conductivity is denoted as . .

[0075] Based on this conductivity-temperature relationship The electromagnetic control equations and the heat transfer control equations can be coupled to obtain the electro-magnetic-thermal coupled control equations for digital cables:

[0076] ;

[0077] After obtaining the electro-magnetic-thermal coupling control equations for the digital cable, the temperature of the digital cable can be simulated based on these equations, thereby generating a digital cable temperature dataset. Specifically, firstly, a digital cable simulation model is constructed based on the relevant structural parameters of the digital cable. Digital cables typically consist of a cable core, a semiconductor shielding layer, a cross-linked polyethylene insulation layer, a metallic shielding layer, and an outer sheath. A simulation model of the digital cable is built based on these structures, and the corresponding material properties are filled into each layer.

[0078] Subsequently, based on the electromagnetic-magnetic-thermal coupling control equations and the constructed digital cable simulation model, simulation boundary conditions and initial parameters were set. Specifically, according to the electromagnetic control equations, the main heat sources of the digital cable were determined to be the cable core and the metal shielding layer. The cable core was modeled as a coil model excited by the load current, while the metal shielding layer was modeled as a coil model excited by the circuit current. The heat transfer mode of the digital cable was set according to the heat transfer control equations, with solid-state conduction heat transfer between the internal layers of the digital cable and convection and thermal radiation heat transfer between the outer sheath and the air. Initial parameters such as the initial temperature of the digital cable and the operating environment temperature also needed to be specified. Mesh generation parameters (such as average mesh size and curvature factor) were set according to the thickness and accuracy requirements of each layer of the digital cable to mesh the simulation model of the digital cable.

[0079] Next, under the premise of setting simulation boundary conditions and initial parameters, an electro-magnetic-thermal coupled simulation of the digital cable simulation model is performed under different current-carrying conditions. After the simulation, the temperature dataset of the digital cable is obtained. The current-carrying capacity is set as a scannable physical parameter for parametric scanning, i.e., electro-magnetic-thermal coupled analysis of the digital cable under different current-carrying conditions is performed, and the convergence of the digital cable simulation model is determined by finite element analysis. If the digital cable simulation model does not converge, the simulation boundary conditions and initial parameters need to be readjusted, and the relevant structural parameters of the digital cable need to be modified. If the digital cable simulation model converges, the temperature distribution of the digital cable under different current-carrying conditions is further calculated, and the temperature dataset of the digital cable is exported as the raw data for subsequent training of the physical information neural network.

[0080] When simulating and solving the model, it should be noted that the embodiments of this application are based on COMSOL Multiphysics software to carry out electro-magnetic-thermal coupling analysis, and solve the temperature field under different current carrying conditions through parametric scanning method, and finally obtain digital cable temperature dataset.

[0081] In this embodiment, the digital cable temperature dataset includes multi-dimensional temperature data and various single-dimensional temperature data. The multi-dimensional temperature data includes temperature data at different radial positions and different current-carrying combinations, while the single-dimensional temperature data includes axial temperature data at different current-carrying ratios and temperature data at different radial positions. It should be noted that the temperature data at different radial positions refers to the temperature values ​​at multiple spatial sampling points along the radial direction of the digital cable, from the center of the cable core to the outer surface of the outer sheath. That is, temperature data is acquired within each structural layer and at the interface of the digital cable, thereby reflecting the temperature variation pattern in the radial direction of the digital cable.

[0082] S102: Based on multiple single-dimensional temperature data, construct multiple physical information neural networks respectively.

[0083] In this embodiment, a first physical information neural network is constructed based on first training data and a first target loss function. The first training data consists of shaft core temperature data under different current carrying capacities. The first target loss function includes a first data loss function and a physical loss function. The first data loss function measures the error between the shaft core temperature predicted by the first physical information neural network and the corresponding shaft core temperature in the first training data, and its specific calculation formula is as follows:

[0084] ;

[0085] in, Denotes the first data loss function. This represents the actual temperature at the m-th sampling point under a current carrying capacity I. This represents the predicted temperature at the m-th sampling point under a current carrying capacity I. This represents the core temperature data under different current carrying capacity I. It should be noted that the sampling points here refer to finite element simulation calculation points collected within the core region of the digital cable according to different current carrying capacity I.

[0086] The physical loss function is used to constrain the axis temperature predicted by the first physical information neural network to satisfy the electro-magnetic-thermal coupling control equations related to the digital cable. Its specific calculation formula is as follows:

[0087] ;

[0088] in, Represents the physical loss function. The current density of the cable core. The cross-sectional area of ​​the cable core can be expressed as:

[0089] ;

[0090] Based on the calculation formulas for the first data loss function and the physical loss function, the first target loss function can be obtained. for:

[0091] ;

[0092] It should be noted that, after sufficient training, the first physical information neural network constructed in this embodiment of the application can calculate the core temperature of the digital cable given a current carrying capacity.

[0093] Furthermore, based on the second training data and the second objective loss function, a second physical information neural network is constructed. The second training data consists of temperature data at different radial positions. The second objective loss function includes a second data loss function and a boundary loss function. The second data loss function measures the error between the temperature predicted by the second physical information neural network at the radial position and the corresponding temperature at the radial position in the second training data. Its specific calculation formula is as follows:

[0094] ;

[0095] in, This represents the second data loss function. This indicates that the temperature at the m-th sampling point in the cable core is... The actual temperature below, This indicates that the temperature at the m-th sampling point in the cable core is... The predicted temperature below, Indicates different core temperatures Temperature data at different radial positions.

[0096] The boundary loss function is used to constrain the predicted temperature of the second physical information neural network at the radial boundary position to meet the preset radial boundary conditions. Its specific calculation formula is as follows:

[0097] ;

[0098] in, This represents the boundary loss function, where the sampling points are only located at... area, This represents the radial distance from the outer surface of the outermost layer of the digital cable to the axis, where x represents the radial distance from the sampling point to the axis.

[0099] Based on the calculation formulas for the second data loss function and the boundary loss function, the second objective loss function can be obtained. for:

[0100] ;

[0101] The boundary loss function in this embodiment is constructed based on the physical law that the radial temperature of a digital cable exhibits a piecewise linear distribution. Specifically, by building a simulation model of the digital cable and setting corresponding parameters according to the electro-magnetic-thermal coupling control equation, the temperature distribution data of the digital cable at various radial positions under different current carrying capacities can be obtained. Simulation results for the digital cable show that the temperature value is mainly affected by the radial distance from the observation point to the axis. The influence of current carrying capacity. When the current carrying capacity remains constant, the temperature within each layer of the digital cable has a linear relationship with the radial distance to the axis. With temperature The relationship can be represented as a piecewise linear function, the number of which is determined by the number of layers in the digital cable structure. This application uses a five-layer structure as an example, resulting in the following relationship:

[0102] ;

[0103] This indicates the current carrying capacity of the digital cable at a given current carrying capacity. Temperature at the interface of the layers Indicates the digital cable number The radial distance from the outer surface of the layered structure to the axis. For digital cables The thermal conductivity of the layer, where i = 1, 2, 3, 4, 5.

[0104] It should be noted that the radial boundary position in the embodiments of this application includes the axis position and the outermost position of the digital cable. The preset radial boundary condition can be that the gradient of the predicted temperature of the digital cable at the axis position with respect to the radial distance is zero, so as to reflect the symmetry of heat flow at the center of the digital cable structure. At the outermost position, its predicted temperature should be equal to the theoretical temperature of the outer surface calculated by the piecewise linear relationship of radial temperature, so as to reflect the heat transfer boundary characteristics between the digital cable and the external environment.

[0105] Therefore, the embodiments of this application construct a first physical information neural network and a second physical information neural network based on the axial temperature data under different current carrying capacities and temperature data at different radial positions, combined with the corresponding target loss functions, thus laying the foundation for the subsequent fusion of multiple physical information neural networks.

[0106] S103: Fuse the features of multiple physical information neural networks to obtain a fusion result, which is used to initialize the parameters of a preset initial physical information neural network.

[0107] This application embodiment can fuse the features of multiple physical information neural networks constructed as described above. Specifically, based on a preset concatenation order, the hidden layer output feature vectors of each physical information neural network can be concatenated to obtain the fusion result.

[0108] It should be noted that the parameters of a pre-defined initial physical neural network can be initialized based on the fusion results. This initial physical information neural network refers to the target physical information neural network to be trained, whose parameters have not yet been trained. By mapping the fusion results to the parameter space of this initial physical information neural network through linear mapping or projection, its parameters can be reasonably initialized. Using this initialization method, the initial physical information neural network can begin training in a more suitable parameter space, thereby accelerating convergence and improving prediction accuracy.

[0109] S104: The initial physical information neural network is trained based on multi-dimensional temperature data to obtain the target physical information neural network, which is used to predict the temperature distribution of digital cables under given current carrying capacity and radial position conditions.

[0110] In this embodiment, an initial physical information neural network is trained based on multi-dimensional temperature data and a third objective loss function to obtain a target physical information neural network. The third objective loss function includes a third data loss function, a third physical loss function, and a third boundary loss function. The third data loss function measures the error between the temperature predicted by the initial physical information neural network under different combinations of current carrying capacity and radial position and the corresponding temperature in the multi-dimensional temperature data. Its specific calculation formula is as follows:

[0111] ;

[0112] in, This represents the third data loss function. This represents temperature data at different radial positions under different current carrying capacity I.

[0113] The third physical loss function is used to constrain the neural network's prediction of temperature under multidimensional temperature data conditions to satisfy the electro-magnetic-thermal coupling control equation. The specific calculation formula is as follows:

[0114] ;

[0115] The third boundary loss function is used to constrain the neural network's prediction of the temperature at the radial boundary position based on the initial physical information to meet preset radial boundary conditions. The specific calculation formula is as follows:

[0116] ;

[0117] The radial boundary conditions are consistent with those used in the aforementioned second objective loss function, and will not be repeated here.

[0118] Based on the calculation formulas for the third data loss function, the third physical loss function, and the third boundary loss function, the third target loss function can be obtained. for:

[0119] ;

[0120] Therefore, it can be seen that the embodiments of this application construct a six-layer fully connected physical information neural network, with a hyperbolic tangent function used as the activation function after each of the first five layers. This target physical information neural network uses radial position... and carrying capacity As input, the corresponding temperature is output. This enables the prediction of temperature distribution in digital cables under given current carrying capacity and radial position conditions.

[0121] Based on the relevant content of steps S101-S104 above, in this embodiment, firstly, the electro-magnetic-thermal coupling control equation related to the digital cable is obtained, and a digital cable temperature dataset is obtained by simulation based on the electro-magnetic-thermal coupling control equation. The digital cable temperature dataset includes multi-dimensional temperature data and multiple single-dimensional temperature data. The multi-dimensional temperature data includes temperature data under different radial positions and different current-carrying combinations, and the single-dimensional temperature data includes axial temperature data under different current-carrying conditions and temperature data under different radial positions. Next, multiple physical information neural networks are constructed based on the multiple single-dimensional temperature data. Subsequently, the features of the multiple physical information neural networks are fused, and the fusion result is used to initialize the parameters of the preset initial physical information neural network. Finally, the initial physical information neural network is trained based on the multi-dimensional temperature data to obtain the target physical information neural network, which is used to predict the temperature distribution of the digital cable under given current-carrying and radial position conditions. As can be seen, this scheme constructs multiple physical information neural networks (PINs) based on single-dimensional temperature data obtained from simulations of the electro-magnetic-thermal coupling control equations. Each PIN learns independently for axial temperature data under different current-carrying capacities or temperature data at different radial positions. Then, the features of the multiple PINs are fused, and the fusion result is used to initialize the parameters of a pre-set initial PIN. This allows the initial PIN to directly utilize the temperature change features learned from the multiple PINs during the initial training phase. Subsequently, the initial PIN is trained based on multi-dimensional temperature data to obtain a target PIN. This target PIN can predict the temperature distribution of digital cables under given current-carrying capacities and radial positions, thereby improving the stability and accuracy of temperature prediction for digital cables under various operating conditions. Compared to existing technologies, this scheme, by constructing multi-dimensional PINs and performing feature fusion and initialization training, effectively overcomes the problem of insufficient accuracy in predicting digital cable temperature under complex operating conditions using traditional single-physics methods.

[0122] Furthermore, Figure 2 This is a schematic diagram of a device for predicting the temperature of a digital cable, provided as an embodiment of this application. (Combined with...) Figure 2 As shown, the device 200 for predicting the temperature of digital cables provided in this application embodiment may include:

[0123] The data acquisition module 201 is used to acquire the electro-magnetic-thermal coupling control equations related to the digital cable, and to simulate the digital cable temperature dataset based on the electro-magnetic-thermal coupling control equations. The digital cable temperature dataset includes multi-dimensional temperature data and various single-dimensional temperature data. The multi-dimensional temperature data includes temperature data under different radial positions and different current carrying combinations. The single-dimensional temperature data includes axial temperature data under different current carrying capacities and temperature data under different radial positions.

[0124] The neural network construction module 202 is used to construct multiple physical information neural networks based on the multiple single-dimensional temperature data.

[0125] The neural network fusion module 203 is used to fuse the features of the multiple physical information neural networks to obtain a fusion result, which is used to initialize the parameters of a preset initial physical information neural network.

[0126] The target network training module 204 is used to train the initial physical information neural network based on the multi-dimensional temperature data to obtain the target physical information neural network, which is used to predict the temperature distribution of the digital cable under given current carrying capacity and radial position conditions.

[0127] Optionally, the neural network construction module 202 is specifically used for:

[0128] A first physical information neural network is constructed based on the first training data and the first objective loss function. The first training data consists of shaft core temperature data under different current carrying capacities. The first objective loss function includes a first data loss function and a physical loss function. The first data loss function is used to measure the error between the shaft core temperature predicted by the first physical information neural network and the corresponding shaft core temperature in the first training data. The physical loss function is used to constrain the shaft core temperature predicted by the first physical information neural network to satisfy the electro-magnetic-thermal coupling control equation related to the digital cable.

[0129] Optionally, the neural network construction module 202 is specifically used for:

[0130] A second physical information neural network is constructed based on the second training data and the second objective loss function. The second training data consists of temperature data at different radial positions. The second objective loss function includes a second data loss function and a boundary loss function. The second data loss function measures the error between the temperature predicted by the second physical information neural network at the radial position and the temperature at the corresponding radial position in the second training data. The boundary loss function constrains the predicted temperature of the second physical information neural network at the radial boundary position to meet a preset radial boundary condition. The radial boundary position includes the axis position and the outermost position of the digital cable.

[0131] Optionally, the target network training module 204 is specifically used for:

[0132] The initial physical information neural network is trained based on the multi-dimensional temperature data and the third objective loss function to obtain the target physical information neural network. The third objective loss function includes a third data loss function, a third physical loss function, and a third boundary loss function. The third data loss function is used to measure the error between the temperature predicted by the initial physical information neural network under different radial positions and different current carrying capacity combinations and the corresponding temperature in the multi-dimensional temperature data. The third physical loss function is used to constrain the predicted temperature of the initial physical information neural network under multi-dimensional temperature data conditions to satisfy the electro-magnetic-thermal coupling control equation. The third boundary loss function is used to constrain the predicted temperature of the initial physical information neural network at the radial boundary position to satisfy the preset radial boundary condition.

[0133] Optionally, before obtaining the electro-magnetic-thermal coupling control equations associated with the digital cable, the device 200 for predicting the temperature of the digital cable may include:

[0134] The electromagnetic equation establishment module is used to establish the electromagnetic control equations of the digital cable based on Maxwell's equations.

[0135] The heat transfer equation establishment module is used to establish the heat transfer control equation of the digital cable based on Fourier's heat transfer law and the law of conservation of energy.

[0136] The equation coupling module is used to couple the electromagnetic control equation and the heat transfer control equation based on the conductivity-temperature relationship of each layer of the digital cable to obtain the electro-magnetic-thermal coupling control equation of the digital cable.

[0137] Optionally, the data acquisition module 201 is specifically used for:

[0138] A simulation model of the digital cable is built based on the structural parameters related to the digital cable.

[0139] Based on the electro-magnetic-thermal coupling control equations and the digital cable simulation model, simulation boundary conditions and initial parameters are set.

[0140] Based on the simulation boundary conditions and initial parameters, the digital cable simulation model is subjected to electro-magnetic-thermal coupling simulation under various current carrying capacities. After the simulation, the temperature dataset of the digital cable is obtained.

[0141] Optionally, the neural network fusion module 203 is specifically used for:

[0142] Based on a preset splicing order, the hidden layer output feature vectors of each of the multiple physical information neural networks are spliced ​​together to obtain a fusion result.

[0143] Furthermore, embodiments of this application also provide an electronic device, including: a processor, a memory, and a system bus;

[0144] The processor and the memory are connected via the system bus;

[0145] The memory is used to store one or more programs, the one or more programs including instructions that, when executed by the processor, cause the processor to perform any of the implementation steps of the method for predicting the temperature of a digital cable described above.

[0146] Furthermore, embodiments of this application also provide a computer-readable storage medium for storing a computer program, which, when executed by a terminal device, implements any of the implementation steps of the above-described method for predicting the temperature of a digital cable.

[0147] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that all or part of the steps in the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network communication device such as a media gateway, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application. It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on describing the differences from other embodiments. The same or similar parts between the various embodiments can be referred to mutually.

[0148] The system disclosed in the embodiments is described in a relatively simple manner because it corresponds to the method disclosed in the embodiments. For relevant details, please refer to the method section.

[0149] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0150] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method of temperature prediction for a digital cable, characterized by, The method comprises: acquiring an electro-magnetic-thermal coupling control equation related to a digital cable, and simulating a digital cable temperature data set based on the electro-magnetic-thermal coupling control equation, wherein the digital cable temperature data set comprises multi-dimensional temperature data and multiple single-dimensional temperature data, the multi-dimensional temperature data comprises temperature data under different radial position and different load current combinations, and the single-dimensional temperature data comprises axial temperature data under different load currents and temperature data under different radial positions; based on the multiple single-dimensional temperature data, multiple physical information neural networks are respectively constructed; features of the multiple physical information neural networks are fused to obtain a fusion result, and the fusion result is used for initializing parameters of a preset initial physical information neural network; the initial physical information neural network is trained based on the multi-dimensional temperature data to obtain a target physical information neural network, and the target physical information neural network is used for predicting temperature distribution of the digital cable under given load current and radial position conditions; the initial physical information neural network is trained based on the multi-dimensional temperature data and a third target loss function to obtain a target physical information neural network, wherein the third target loss function comprises a third data loss function, a third physical loss function and a third boundary loss function, the third data loss function is used for measuring errors between temperatures predicted by the initial physical information neural network under different radial position and different load current combinations and corresponding temperatures in the multi-dimensional temperature data, the third physical loss function is used for constraining the predicted temperature of the initial physical information neural network under the multi-dimensional temperature data to satisfy the electro-magnetic-thermal coupling control equation, and the third boundary loss function is used for constraining the predicted temperature of the initial physical information neural network at the radial boundary position to satisfy the preset radial boundary condition. the multiple physical information neural networks are respectively constructed based on the multiple single-dimensional temperature data, comprising:

2. The method of claim 1, wherein, a first physical information neural network is constructed based on first training data and a first target loss function, wherein the first training data is the axial temperature data under different load currents, and the first target loss function comprises a first data loss function and a physical loss function, the first data loss function is used for measuring errors between axial temperatures predicted by the first physical information neural network and corresponding axial temperatures in the first training data, and the physical loss function is used for constraining the axial temperature predicted by the first physical information neural network to satisfy the electro-magnetic-thermal coupling control equation related to the digital cable. the multiple physical information neural networks are respectively constructed based on the multiple single-dimensional temperature data, comprising:

3. The method of claim 1, wherein, ​ constructing a second physical information neural network based on second training data and a second target loss function, wherein the second training data is temperature data at different radial positions, the second target loss function comprises a second data loss function and a boundary loss function, the second data loss function is used to measure the error between the temperature at the radial position predicted by the second physical information neural network and the temperature at the corresponding radial position in the second training data, and the boundary loss function is used to constrain the predicted temperature of the second physical information neural network at the radial boundary position to satisfy a preset radial boundary condition, and the radial boundary position includes the axial position and the outermost position of the digital cable.

4. The method of claim 1, wherein, Before obtaining the electro-magnetic-thermal coupling control equation related to the digital cable, the method further comprises: establishing an electromagnetic control equation of the digital cable based on a Maxwell equation set; establishing a heat transfer control equation of the digital cable based on a Fourier heat transfer law and an energy conservation law; coupling the electromagnetic control equation and the heat transfer control equation according to the conductivity-temperature relationship of each layer material of the digital cable to obtain the electro-magnetic-thermal coupling control equation of the digital cable.

5. The method of claim 1, wherein, The digital cable temperature data set obtained by simulation based on the electro-magnetic-thermal coupling control equation comprises: building a digital cable simulation model based on the structural parameters related to the digital cable; setting simulation boundary conditions and initial parameters based on the electro-magnetic-thermal coupling control equation and the digital cable simulation model; performing electro-magnetic-thermal coupling simulation on the digital cable simulation model under multiple current-carrying capacities based on the simulation boundary conditions and the initial parameters, and obtaining the digital cable temperature data set after the simulation ends.

6. The method of claim 1, wherein, The fusion result is used to initialize the parameters of a preset initial physical information neural network, and the method comprises: splicing the hidden layer output feature vectors of each of the plurality of physical information neural networks based on a preset splicing order to obtain a fusion result.

7. An apparatus for temperature prediction of a digital cable, characterized by comprises: a data acquisition module configured to obtain an electro-magnetic-thermal coupling control equation related to a digital cable and simulate a digital cable temperature data set based on the electro-magnetic-thermal coupling control equation, wherein the digital cable temperature data set comprises multi-dimensional temperature data and multiple single-dimensional temperature data, the multi-dimensional temperature data comprises temperature data at different radial positions and under different current-carrying capacities, and the single-dimensional temperature data comprises axial temperature data under different current-carrying capacities and temperature data at different radial positions; a neural network construction module configured to construct a plurality of physical information neural networks based on the multiple single-dimensional temperature data; a neural network fusion module configured to fuse features of the plurality of physical information neural networks to obtain a fusion result, and the fusion result is used to initialize parameters of a preset initial physical information neural network; and a target network training module configured to train the initial physical information neural network based on the multi-dimensional temperature data to obtain a target physical information neural network, the target physical information neural network being configured to predict a temperature distribution of the digital cable under a given load flow and radial position condition; the training of the initial physical information neural network based on the multi-dimensional temperature data to obtain a target physical information neural network comprises: training the initial physical information neural network based on the multi-dimensional temperature data and a third target loss function to obtain a target physical information neural network, wherein the third target loss function comprises a third data loss function, a third physical loss function and a third boundary loss function, the third data loss function being configured to measure an error between a temperature predicted by the initial physical information neural network under a combination of different radial positions and different load flows and a corresponding temperature in the multi-dimensional temperature data, the third physical loss function being configured to constrain the predicted temperature of the initial physical information neural network under the multi-dimensional temperature data to satisfy an electro-magnetic-thermal coupling control equation, and the third boundary loss function being configured to constrain the predicted temperature of the initial physical information neural network at the radial boundary position to satisfy the preset radial boundary condition.

8. An electronic device, comprising: The device comprises a processor, a memory and a system bus; The processor and the memory are connected through the system bus; The memory is configured to store a program, the program comprising instructions which, when executed by the processor, cause the processor to perform the steps of the method for predicting a temperature of a digital cable according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium is configured to store a computer program which, when executed by a terminal device, implements the steps of the method for predicting a temperature of a digital cable according to any one of claims 1 to 6.

Images