Method for evaluating internal fault bearing time of large-scale oil-filled main equipment

Abstract

The invention discloses a method for evaluating the bearing time of an internal fault of large oil-filled main equipment, and belongs to the technical field of relay protection of oil-filled main equipment of a power system. Obtaining a plurality of fault currents in the internal fault current range of the oil-filled main equipment; establishing a three-dimensional geometric simulation model and a grid model of the oil charge main equipment; calling a transient pressure-based solver, simulating a turbulence phenomenon and an oil pressure rise phenomenon of the oil-filled main equipment under a fault current condition, judging whether the oil-filled main equipment is broken or not, and obtaining allowable fault bearing time of the oil-filled main equipment under different fault currents; and establishing a neural network model for mapping the fault current and the allowable fault withstanding time through a convolutional layer, a pooling layer and a full connection layer in the convolutional neural network, so as to evaluate the allowable fault withstanding time of the oil-filled main equipment. And the fracture and failure phenomena of the oil-filled main equipment in a high-stress state are accurately captured, so that a quantitative method which is more accurate than a traditional empirical method is provided for evaluating the fault bearing time of the oil-filled main equipment.

Description

Technical Field

[0001] The invention belongs to the technical field of relay protection of oil-filled main equipment in an electric power system, and in particular relates to a method for evaluating the internal fault tolerance time of large-scale oil-filled main equipment. Background Art

[0002] In modern power systems, UHV and EHV oil-filled main equipment (such as oil-immersed transformers, oil-immersed reactors and oil-filled circuit breakers) play a vital role in maintaining the stable operation of the power grid. This type of equipment usually operates under high load and high voltage environment. When a high-energy arc fault occurs inside the equipment, it is very easy to cause equipment damage or explosion accidents. In recent years, there have been many explosion accidents of UHV oil-filled main equipment caused by internal high-energy arc discharge in China, which have produced adverse social impacts and caused serious economic losses. This highlights the importance of a comprehensive assessment of the equipment's fault tolerance. The allowable fault tolerance time is a critical indicator for measuring the catastrophic damage of the equipment when no protection measures are intervened under fault conditions. A deep understanding of the allowable tolerance time of oil-filled main equipment under fault conditions is of great significance for accurately determining the equipment protection action time, optimizing protection strategies, and designing preventive measures. However, there is still a lack of research on the assessment of the allowable fault tolerance time of oil-filled main equipment. Summary of the invention

[0003] The present invention provides a method for evaluating the internal fault tolerance time of large-scale oil-filled main equipment to solve the problem that there is a lack of research on the evaluation of the allowable fault tolerance time of oil-filled main equipment in the prior art, and the problem that the allowable fault tolerance time of the oil-filled main equipment cannot be accurately evaluated within the full fault current range.

[0004] In order to achieve the above object, the present invention adopts the following technical solution: A method for evaluating the internal fault tolerance time of a large oil-filled main equipment comprises the following steps: Step 1: Obtain multiple fault currents within the fault current range of the arc fault inside the oil-filled main equipment ; Step 2: Establish a three-dimensional geometric simulation model of the oil filling main equipment, and establish a grid model of the oil filling main equipment based on the three-dimensional geometric simulation model of the oil filling main equipment; Step 3: Call the transient pressure-based solver to perform fluid-solid coupling calculations on the main oil-filled equipment; S1, based on multiple fault currents , the oil pressure increase process inside the oil-filled main equipment is simulated in the grid model of the oil-filled main equipment to reflect different fault currents Oil pressure changes under certain conditions; S2, simulate the turbulence phenomenon caused by the dynamic changes of the fluid inside the oil-filled main equipment, and set the turbulence phenomenon of the oil-filled main equipment to Model; S3, according to the oil pressure changes and The model obtains the oil pressure load; S4. Input the oil pressure load into the solid calculation domain of the oil-filled main equipment to obtain the displacement vectors of the grid points of the grid model of the oil-filled main equipment in the x, y and z directions. , through the displacement vector , obtain the principal strains of the mesh model of the oil-filled main equipment in the x, y and z directions , and ; S5. According to the main strain of the mesh model of the oil-filled main equipment in the x, y and z directions , and Determine whether the main oil-filled equipment is broken. If broken, record the main oil-filled equipment at different fault currents. If no rupture occurs, the system will re-enter S1. Step 4: According to the different fault currents of the oil-filled main equipment Based on the convolutional neural network, a neural network model of fault current and allowable fault tolerance time is established to output the allowable fault tolerance time of the oil-filled main equipment under different fault currents. The allowable fault tolerance time result.

[0005] In step 2, the method for establishing the grid model of the oil-filling main equipment is specifically: using a progressive size function to discretize the three-dimensional geometric simulation model of the oil-filling main equipment, gradually reducing the grid diameter near the boundaries, corners and curves, and gradually increasing the grid diameter as the distance increases.

[0006] The oil pressure increase process inside the oil-filled main equipment is simulated in the grid model of the oil-filled main equipment. The bubble dynamics is used to simulate the oil pressure increase process inside the oil-filled main equipment. The bubble dynamics equation is:

[0007] In the formula, represents the radius of the bubble, is the normal velocity of the bubble surface, represents the normal acceleration of the bubble surface; Indicates the density of insulating oil, represents the pressure at the boundary of the fluid domain, represents the heat transfer coefficient, represents the specific heat ratio, represents the fluid dynamic viscosity, represents the surface tension of the fluid, represents the external work done by the bubble surface during the bubble expansion process, It means that the surface tension does work on the outside during the bubble expansion process. It means that the viscous force does work externally during the bubble expansion process. represents the initial internal energy of the bubble, Indicates the arc energy.

[0008] Arc energy The calculation formula is:

[0009] In the formula, Indicates the fault current, Indicates the fault current The corresponding arc voltage, Indicates the fault duration.

[0010] The method for simulating the turbulence phenomenon caused by the dynamic changes of the fluid inside the oil-filled main equipment is as follows: using the dynamic grid to update the instantaneous fluid field inside the oil-filled main equipment, and simulating the turbulence phenomenon caused by the dynamic changes of the fluid caused by the expansion of bubbles inside the oil-filled main equipment by capturing the flow characteristics and transient changes of physical quantities of the fluid during the structural changes of the oil-filled main equipment. The calculation formula is:

[0011] In the formula, represents the time derivative, is the density of the fluid, u is the velocity vector, is the area vector, is the diffusion coefficient, express The source term of is the turbulence characteristic quantity, is the mesh speed of the moving mesh, is the control volume, is the vector differential operator, Turbulence characteristic quantity The gradient of is the control volume of the border.

[0012] The dynamic mesh update uses the Laplace smoothing model, and the calculation formula of the Laplace smoothing model is:

[0013] In the formula, Representation Node i The new location, Representation Node The number of adjacent nodes of Is an adjacent node location.

[0014] Set the turbulence phenomenon of the oil-filled main equipment to The model is set up with the equation:

[0015]

[0016] In the formula, represents the turbulent kinetic energy, Turbulent kinetic energy The differential change of represents the time differential, Represents spatial coordinates The differential of represents the symbol of partial derivative, represents the turbulent kinetic energy, The turbulent Prandtl number, represents the turbulence specific dissipation rate, represents the turbulence specific dissipation rate The turbulent Prandtl number, represents the average velocity of the fluid, represents the specific dissipation rate, represents the molecular viscosity, represents the turbulent viscosity, The Prandtl number representing the kinetic energy equation, represents the generation term of turbulent kinetic energy, which is generated by shear stress, represents the magnitude of the strain rate, and represents the model constant, Is an adjacent node location.

[0017] According to the oil pressure changes and The model obtains the oil pressure load, which needs to be further calculated through the momentum equation. The calculation formula of the momentum equation is:

[0018] In the formula, is the dynamic viscosity, represents the external force vector per unit volume acting on the fluid, It's pressure. is the density of the fluid, is the flow rate of the fluid inside the oil-filled main equipment, Represents the dot product operation of the vector field on the vector micromolecule; the pressure is solved by iteratively solving the momentum equation and the flow rate of the fluid inside the oil-filled main equipment , through the pressure and the flow rate of the fluid inside the oil-filled main equipment The pressure distribution is obtained by coupling and compared with the oil pressure change and The models are combined to obtain the oil pressure load.

[0019] Displacement vectors of the grid points of the oil-filled main equipment grid model in the x, y and z directions It is obtained through explicit dynamics calculation. The specific calculation method is: use the central difference method to perform discrete integration of acceleration, velocity and displacement in time steps to obtain the displacement vector of the grid point of the grid model of the oil-filled main equipment , The displacement of the grid points of the oil-filled main equipment grid model in the x direction, The displacement of the grid points of the oil-filled main equipment grid model in the y direction, The displacement of the grid points of the oil-filled main equipment grid model in the z direction, and the principal strains of the oil-filled main equipment grid model in the x, y and z directions , and The strain calculation formula is: .

[0020] The method for judging whether the main oil-filled device is broken is as follows: judging whether the main oil-filled device is broken according to the constant strain failure criterion, and the calculation formula for the judgment is:

[0021] In the formula, represents the failure strain, represents the equivalent plastic strain. If the equivalent plastic strain >Failure strain , then the oil-filled main equipment breaks. If the equivalent plastic strain ≤Failure strain , the main oil-filled equipment has not been broken.

[0022] The method is based on a convolutional neural network to establish a neural network model of fault current and allowable fault tolerance time. Specifically, the obtained allowable fault tolerance time data under different fault current conditions are used as input values, and the data are trained by a convolutional neural network to generate a neural network model of the relationship between the fault current and the allowable fault tolerance time. The convolutional neural network includes a convolutional layer, a pooling layer, and a fully connected layer. The convolutional layer is used for different fault currents. The corresponding allowable fault tolerance time data is used for feature extraction. The pooling layer optimizes the extracted features. The fully connected layer maps the optimized feature data between the fault current and the allowable fault tolerance time. The calculation formula for feature extraction by the convolution layer is:

[0023] In the formula, Indicates k The feature quantity is at the position ( i , j ), Indicates that the input data is at position The value at Indicates that the convolution kernel is k The weight matrix of the channel, Represents the bias term; the calculation of the pooling layer to optimize the extracted features includes average pooling and maximum pooling. The calculation formulas of average pooling and maximum pooling are as follows:

[0024]

[0025] in represents average pooling, represents the maximum pooling, N represents the dimension of the convolution kernel, It represents the data points in the pooled area. The fully connected layer maps the optimized feature data between the fault current and the allowable fault tolerance time, using the Softmax function, and its calculation formula is as follows:

[0026]

[0027] In the formula, It is the fully connected layer j The output of a neuron, and b Represent the weight and bias respectively, represents the i-th optimized feature data output by the pooling layer, and Represents the j The output probability corresponding to a neuron in Softmax is, is the probability distribution in the Softmax function.

[0028] Compared with the prior art, the present invention has the following beneficial effects: The present invention discloses a method for evaluating the internal fault tolerance time of a large oil-filled main equipment. The present application establishes a fluid-solid coupling calculation model of the oil-filled main equipment, comprehensively considers the bubble dynamics of the fluid domain, the oil pressure change and the structural mechanical response of the solid domain during an arc fault, and fully reflects the interaction between multiple physical fields. By accurately simulating the physical process inside the oil-filled main equipment, the allowable fault tolerance time of the oil-filled main equipment under different fault currents can be more accurately determined, thereby improving the accuracy of the evaluation results. The relationship between the fault current and the fault tolerance time within the full fault current range is modeled using machine learning methods such as convolutional neural networks. The machine learning model can handle complex nonlinear relationships, train and learn through a large amount of sample data, and mine the potential laws in the data, thereby achieving accurate prediction of the fault tolerance time under any fault current, making up for the shortcomings of traditional methods in handling complex relationships. Combined with the constant strain failure criterion, the present invention can accurately capture the fracture and failure process of the structural material of the oil-filled main equipment under high stress, thereby providing a more accurate evaluation method than the traditional empirical data method for determining the allowable fault tolerance time of the oil-filled main equipment. In addition, the present invention applies deep neural networks to model the relationship between fault current and fault tolerance time for the first time, which can achieve high-precision evaluation within the full fault current range and successfully solve the problem of fitting complex nonlinear relationships. This machine learning-based method significantly improves the efficiency of fault tolerance time evaluation and provides efficient and intelligent technical support for the design and optimization of oil-filled main equipment protection systems.

[0029] Furthermore, the present application not only analyzes the allowable fault tolerance time under the sampled fault current conditions, but also expands the evaluation range to the full fault current range through a machine learning model. This allows the fault tolerance capacity of the oil-filled main equipment to be accurately evaluated in practical applications, regardless of the level of the fault current, providing a more comprehensive and reliable basis for equipment protection. During the evaluation process, the effects of multiple factors on the fault tolerance time, such as arc energy, bubble expansion, fluid turbulence, and mechanical properties of structural materials, are comprehensively considered. This comprehensive consideration method can more realistically reflect the performance changes of the oil-filled main equipment under actual fault conditions, and avoid evaluation errors caused by ignoring certain important factors.

[0030] Furthermore, based on accurate fault tolerance time assessment results, the protection action time of the oil-filled main equipment can be set more accurately. This avoids aggravated equipment damage due to too long protection action time, or unnecessary power outages and equipment malfunctions due to too short protection action time, thereby improving the safety and reliability of equipment operation. The evaluation method of the present application provides a scientific basis for the design and optimization of the oil-filled main equipment protection system. By deeply understanding the relationship between fault current and fault tolerance time, the parameters and structure of the protection system can be improved in a targeted manner, improving the performance and adaptability of the protection system and reducing the losses caused by equipment failures. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 : Geometric model diagram of the oil filling equipment riser; Figure 2 : Grid division diagram of oil filling equipment riser; Figure 3 : Flowchart of evaluation of allowed fault tolerance time in an embodiment of the present invention; Figure 4 : A diagram showing the relationship between fault current and allowable fault tolerance time in an embodiment of the present invention. DETAILED DESCRIPTION

[0032] To further understand the content of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the embodiments are merely to explain the present invention and are not intended to limit it, and the present invention can be implemented in a variety of different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thoroughly understood.

[0033] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0034] Example 1 This embodiment proposes a method for evaluating the internal fault tolerance time of a large oil-filled main device, including the following steps: Step 1: Obtain multiple fault currents within the fault current range of the arc fault inside the oil-filled main equipment ; Step 2: Establish a three-dimensional geometric simulation model of the oil filling main equipment, and establish a grid model of the oil filling main equipment based on the three-dimensional geometric simulation model of the oil filling main equipment; Step 3: Call the transient pressure-based solver to perform fluid-solid coupling calculations on the main oil-filled equipment; S1, based on multiple fault currents , the oil pressure increase process inside the oil-filled main equipment is simulated in the grid model of the oil-filled main equipment to reflect different fault currents Oil pressure changes under certain conditions; S2, simulate the turbulence phenomenon caused by the dynamic changes of the fluid inside the oil-filled main equipment, and set the turbulence phenomenon of the oil-filled main equipment to Model; S3, according to the oil pressure changes and The model obtains the oil pressure load; S4. Input the oil pressure load into the solid calculation domain of the oil-filled main equipment to obtain the displacement vectors of the grid points of the grid model of the oil-filled main equipment in the x, y and z directions. , through the displacement vector , obtain the principal strains of the mesh model of the oil-filled main equipment in the x, y and z directions , and ; S5. According to the main strain of the mesh model of the oil-filled main equipment in the x, y and z directions , and Determine whether the main oil-filled equipment is broken. If broken, record the main oil-filled equipment at different fault currents. If no rupture occurs, go back to step 3. Step 4: According to the different fault currents of the oil-filled main equipment Based on the convolutional neural network, a neural network model of fault current and allowable fault tolerance time is established, and the output of the main oil-filled equipment under different fault currents is The allowable fault tolerance time result.

[0035] Example 2 In this embodiment, the main oil-filled equipment is taken as an example of an oil-filled equipment riser. The oil-filled equipment riser is taken as a research object with a diameter of 600 mm and a height of 1000 mm. The cylinder wall material of the oil-filled equipment riser is carbon steel, and an insulator structure with a height of 2040 mm is installed thereon. Figure 1 This implementation targets Figure 1 A method for evaluating the internal fault tolerance time of a large oil-filled main equipment is carried out on the oil-filled equipment lifting seat, and the specific implementation method includes the following steps: Step 1: Determine the fault current range of the arc fault in the oil-filled equipment lifting seat. After determination, the internal arc fault current range of the oil-filled equipment lifting seat is: 50 kA-80 kA.

[0036] Step 2: Set uniform sampling points within the determined fault current range of 50 kA to 80 kA, and collect several fault currents at the sampling points. As the input value of the subsequent fluid-structure interaction calculation of the oil-filled equipment riser, and according to the fault current at the sampling point , and obtain the corresponding arc voltage .

[0037] Step 3: Use SolidWorks software to create a 1:1 equivalent three-dimensional geometric simulation model of the oil-filled equipment riser.

[0038] Step 4: Import the three-dimensional geometric simulation model of the oil filling equipment riser into the ANSYS software.

[0039] Step 5: In ANSYS software, use the progressive size function to discretize the spatial area defined by the three-dimensional geometric simulation model of the oil filling equipment lifting seat, establish the mesh model of the oil filling equipment lifting seat, generate finer meshes near the boundaries, corners, and curves, and gradually generate coarser meshes as the distance increases. The mesh model of the oil filling equipment lifting seat is shown in the figure below. Figure 2 as shown in .

[0040] Step 6: Call the transient pressure-based solver in the ANSYS software, apply the transient pressure-based solver to the grid model of the oil-filling equipment lifting seat, and perform fluid-solid coupling calculations on the oil-filling equipment lifting seat.

[0041] The first step is to simulate the oil pressure increase caused by the bubbles induced by the arc fault inside the oil-filled equipment riser by applying the bubble dynamics equation that takes into account the continuous injection of arc energy, and obtain the oil pressure change. The bubble dynamics equation is:

[0042] In the formula, represents the radius of the bubble, is the normal velocity of the bubble surface, represents the normal acceleration of the bubble surface; Indicates the density of insulating oil, represents the pressure at the boundary of the fluid domain, represents the heat transfer coefficient, represents the specific heat ratio, represents the fluid dynamic viscosity, represents the surface tension of the fluid, represents the external work done by the bubble surface during the bubble expansion process, It means that the surface tension does work on the outside during the bubble expansion process. It means that the viscous force does work externally during the bubble expansion process. represents the initial internal energy of the bubble, Represents the arc energy, and its calculation formula is:

[0043] In the formula, represents the arc voltage, Indicates the fault current, Indicates the fault duration.

[0044] The second step is that due to the complex fluid state inside the oil-filling equipment lifting seat, the dynamic changes of the fluid caused by the expansion of bubbles in the fluid trigger turbulence. The dynamic grid is used to update the instantaneous fluid field inside the oil-filling equipment lifting seat. The turbulence phenomenon inside the oil-filling equipment lifting seat is simulated by capturing the flow characteristics of the fluid during the fluid change process inside the oil-filling equipment lifting seat and the transient changes of physical quantities. The calculation formula is:

[0045] In the formula, represents the time derivative, is the density of the fluid, u is the velocity vector, is the area vector, is the diffusion coefficient, express The source term of is the turbulence characteristic quantity, is the mesh speed of the moving mesh, is the control volume, is the vector differential operator, Turbulence characteristic quantity The gradient of is the control volume By solving this equation, the instantaneous fluid field can be updated according to the bubble expansion speed.

[0046] Step 3: During the dynamic mesh update process, due to factors such as bubble expansion, the position of the mesh nodes will change, which may cause the mesh quality to deteriorate, which will affect the accuracy and stability of the numerical calculation. By using the Laplace smoothing model to redistribute the mesh nodes, reduce the distortion and deformity of the mesh, improve the quality of the mesh, and keep the mesh in a good shape and topological structure, it is possible to accurately simulate the dynamic changes of complex fluids in the lifting seat of the oil-filled equipment. The dynamic mesh update uses the Laplace smoothing model, and the calculation formula of the Laplace smoothing model is:

[0047] In the formula, Representation Node i The new location, Representation Node The number of adjacent nodes of Is an adjacent node location.

[0048] Step 4: In the fault arc simulation analysis of the oil-filled equipment riser, the fluid inside it will show complex turbulence phenomenon under the condition of arc fault. Therefore, the turbulence phenomenon in the fluid domain is set as Model, in order to simulate turbulence phenomena more accurately, set the fluid domain turbulence model to Model equation:

[0049]

[0050] In the formula, represents the turbulent kinetic energy, Turbulent kinetic energy The differential change of represents the time differential, Represents spatial coordinates The differential of represents the symbol of partial derivative, represents the turbulent kinetic energy, The turbulent Prandtl number, represents the turbulence specific dissipation rate, represents the turbulence specific dissipation rate The turbulent Prandtl number, represents the average velocity of the fluid, represents the specific dissipation rate, represents the molecular viscosity, represents the turbulent viscosity, The Prandtl number representing the kinetic energy equation, represents the generation term of turbulent kinetic energy, which is generated by shear stress, represents the magnitude of the strain rate, and represents the model constant, Is an adjacent node location.

[0051] Step 5: Based on the oil pressure changes and Model, the oil pressure load is calculated by the momentum equation, and the calculation formula of the momentum equation is:

[0052] In the formula, is the dynamic viscosity, represents gravity, It's pressure. is the density of the fluid, To increase the flow rate of the fluid inside the oil-filled equipment seat, Represents the dot product operation of the vector field on the vector micromolecule; the pressure is solved by iteratively solving the momentum equation and the flow rate of the fluid inside the oil-filled equipment riser , through the pressure and the flow rate of the fluid inside the oil-filled equipment riser Coupling is performed to obtain pressure distribution and compare it with oil pressure changes and The models are combined to obtain the oil pressure load.

[0053] Step 6: Input the oil pressure load into the solid calculation domain of the oil-filled equipment lifting seat, and calculate the structural response of the lifting seat through explicit dynamics. The explicit dynamics uses the central difference method to perform discrete integration of the acceleration, velocity and displacement in time steps. The displacement update formula is:

[0054] In the formula, Represents the time step n +1 moment displacement, Represents the time step n The displacement of time, Represents the time step n The speed of time, Represents the time step n The acceleration of time, Represents the time step.

[0055] The speed update formula is:

[0056] In the formula, Represents the time step n +1 / 2 time speed, Represents the time step n -1 / 2 time speed.

[0057] The acceleration update formula is:

[0058] Where M represents the mass matrix, Indicates external force, Represents internal forces.

[0059] By solving the above displacement update formula, velocity update formula and acceleration update formula, the displacement vector of the grid point of the grid model of the oil filling equipment lifting seat is obtained when the oil filling equipment lifting seat receives the oil pressure load caused by the internal arc fault. , The displacement of the grid points of the mesh model of the oil filling equipment riser in the x direction, The displacement of the grid points of the mesh model of the oil filling equipment riser in the y direction, The displacement of the grid points of the mesh model of the oil-filled equipment riser in the z direction is expressed by the displacement vector , the principal strains in the x, y and z directions are obtained, and the strain calculation formula is as follows:

[0060] Step 7: Determine whether the structural material of the oil-filled equipment riser is broken according to the constant strain failure criterion. The calculation formula for judgment is:

[0061] In the formula, represents the failure strain, represents the equivalent plastic strain, , , represents the principal strains in three directions. If the equivalent plastic strain Failure strain , then the structural material of the oil-filled equipment riser will rupture. If the equivalent plastic strain Failure strain If the structural material of the oil-filled equipment lifting seat is not broken, the fluid-solid coupling calculation of the fault arc at the test point is performed again, that is, the first step of step six is ​​re-entered until it is determined that the structural material of the oil-filled equipment lifting seat is broken, and the time at this time is recorded, which is the allowable fault tolerance time of the oil-filled equipment lifting seat under the fault arc current.

[0062] After the fault currents obtained at different sampling points are calculated in the above step six, the allowable fault tolerance time data under different sampling fault current conditions are obtained.

[0063] Step 7: Using a machine learning algorithm, the obtained data of allowable fault withstand time under different fault current conditions are used as input values, and the data is trained through a convolutional neural network to generate a neural network model of the relationship between fault current and allowable fault withstand time. The data used for model training in the convolutional neural network is shown in Table 1 below.

[0064] Table 1

[0065] The convolutional layer operation formula is:

[0066] In the formula, Indicates k The feature quantity is at the position ( i , j ), Indicates that the input data is at position The value at Indicates that the convolution kernel is k The weight matrix of the channel, In the convolution layer, the above convolution layer operation formula is used to extract the features of the allowable fault withstand time data under different fault current conditions, and obtain the corresponding feature map.

[0067] After the convolutional layer features are extracted, the data is pooled. The pooling of the neural network is mainly divided into average pooling and maximum pooling. Average pooling can retain the information of all features in the feature area, reflecting the average state of the fault current at different sampling points and its corresponding allowable fault tolerance time, providing stable feature input for subsequent accurate evaluation of the allowable fault tolerance time. The calculation formula of average pooling is as follows:

[0068] Where N represents the dimension of the convolution kernel. It represents the data points in the pooling area. The maximum pooling takes the maximum value in the pooling area as the feature output, which reflects the significant information characteristics and captures the maximum strain value when the stress concentration area inside the oil-filled equipment riser fails, so as to more accurately evaluate the allowable fault tolerance time of the oil-filled equipment riser under different fault currents. The calculation formula of the maximum pooling is as follows:

[0069] After the convolution and pooling operations of the convolution layer and the pooling layer, the data of different fault currents and the corresponding allowable fault tolerance time are transformed from the original input form into feature representations with a certain degree of abstraction. M The pooling window size is F , the step length is L , and its length after pooling is . The fully connected layer displays and connects all the feature maps output by the aforementioned convolutional layer and pooling layer, so that the neural network model can comprehensively consider all these feature information, determine the relationship between the fault current and the allowable fault tolerance time within the full fault current range, and establish a mapping relationship between the fault current and the allowable fault tolerance time within the full fault current range. The last layer of the fully connected layer uses the Softmax function to convert the input into a probability distribution with a sum of 1, where the highest probability is the type to which the sample belongs, that is, the allowable fault tolerance time category. The fully connected layer completes the classification prediction of the allowable fault tolerance time through the mapping relationship between the fault current and the allowable fault tolerance time within the full fault current range. The formulas for the fully connected layer and the Softmax function are as follows:

[0070]

[0071] In the formula, It is the fully connected layer j The output of a neuron, and b Represent the weight and bias respectively, represents the i-th optimized feature data output by the pooling layer, and Represents the j The output probability corresponding to a neuron in Softmax is, is the probability distribution in the Softmax function.

[0072] Step 8: Output the allowable fault withstand time results of the oil-filled equipment riser under different fault current conditions.

[0073] By constructing a multi-physics field coupling simulation model, the present invention lays a solid theoretical foundation for the dynamic evaluation of the allowable fault tolerance time of the oil-filled equipment lifting seat. Combined with deep neural network technology, the nonlinear mapping relationship between fault current and allowable fault tolerance time is further optimized, and efficient and accurate evaluation within the full fault current range is achieved. This method provides strong technical support for the scientific setting of the protection action time of the oil-filled equipment lifting seat.

[0074] In a specific embodiment of the present invention, the allowable fault tolerance time of the oil-filled equipment riser under a 60KA fault current is evaluated, and according to the strain condition of the grid model of the oil-filled equipment riser, the strain condition of the oil-filled equipment riser at 5 ms, 10 ms, 20 ms and 34.8 ms under the action of the 60KA fault current is tested. With the increase of the duration of the 60KA fault current, the strain distribution of the oil-filled equipment riser gradually expands, especially after 10 ms, the strain at the bottom of the oil-filled equipment riser and the area near the hand hole increases significantly, reflecting that obvious stress concentration occurs at these locations. According to the stress distribution test of the oil-filled equipment riser under the 60KA fault current condition, at 34.8 ms, the oil pressure load caused by the 60KA fault current causes the hand hole bolts of the oil-filled equipment riser to break, so under this fault current condition, the upper limit of the allowable arc fault duration of the oil-filled equipment riser is 34.8 ms. This result is compared with the actual fault case. The actual fault case uses the same fault current and the same equipment. Under the condition of 60KA fault current, the allowable fault tolerance time in the actual fault test case is: 30~40ms. The allowable fault tolerance time evaluated by the oil-filled equipment riser under the same fault current condition is 34.8ms. Therefore, the evaluation result of the present invention is highly consistent with the actual test case, which proves the correctness and effectiveness of the present invention. Based on the above analysis results and the calculation results under different fault current conditions at other sampling points, the present invention accurately establishes the nonlinear mapping relationship between the fault current within the full fault current range and the allowable fault tolerance time through the convolutional neural network model, such as Figure 3 As shown in the figure, multiple fault currents collected by the sampling points As the input value of the input layer, that is, the multiple fault currents collected Corresponding to X1, X2, X3… X n , the final output layer outputs Represents different fault currents The corresponding different oil-filled equipment risers allow for fault tolerance time. Figure 4 The quantitative relationship between the arc current peak value and the allowable fault withstand time is shown. The larger the value, the higher the arc current peak value. The results show that when the arc current peak value increases from 50 kA to 80 kA, the allowable fault tolerance time of the oil-filled equipment riser decreases from about 39 ms to about 21 ms, indicating that under high current fault conditions, the allowable fault tolerance time of the oil-filled equipment riser is shortened by about 50%. The above quantitative analysis results provide a solid theoretical support for the dynamic protection strategy of the oil-filled equipment riser under arc faults, and provide a theoretical basis for the setting of equipment protection action time in engineering design, which has important engineering application value.

[0075] In addition, it should be understood that although this specification is described in accordance with the implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation modes that can be understood by those skilled in the art. The above content is only to illustrate the technical idea of ​​the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution according to the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment, characterized in that: The following steps are involved: Step 1: Obtain multiple fault currents within the fault current range of the arc fault inside the oil-filled main equipment ; Step 2: Establish a three-dimensional geometric simulation model of the oil filling main equipment, and establish a grid model of the oil filling main equipment based on the three-dimensional geometric simulation model of the oil filling main equipment; Step 3: Call the transient pressure-based solver to perform fluid-solid coupling calculations on the main oil-filled equipment; S1, based on multiple fault currents , the oil pressure increase process inside the oil-filled main equipment is simulated in the grid model of the oil-filled main equipment to reflect different fault currents Oil pressure changes under certain conditions; S2, simulate the turbulence phenomenon caused by the dynamic changes of the fluid inside the oil-filled main equipment, and set the turbulence phenomenon of the oil-filled main equipment to Model; S3, according to the oil pressure changes and The model obtains the oil pressure load; S4. Input the oil pressure load into the solid calculation domain of the oil-filled main equipment to obtain the displacement vectors of the grid points of the grid model of the oil-filled main equipment in the x, y and z directions. , through the displacement vector , obtain the principal strains of the mesh model of the oil-filled main equipment in the x, y and z directions , and ; S5. According to the main strain of the mesh model of the oil-filled main equipment in the x, y and z directions , and Determine whether the main oil-filled equipment is broken. If broken, record the main oil-filled equipment at different fault currents. If no rupture occurs, the system will re-enter S1. Step 4: According to the different fault currents of the oil-filled main equipment Based on the convolutional neural network, a neural network model of fault current and allowable fault tolerance time is established to output the allowable fault tolerance time of the oil-filled main equipment under different fault currents. The allowable fault tolerance time result.

2. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment according to claim 1, characterized in that: In the step 2, the method for establishing the grid model of the oil-filling main equipment is specifically: using a progressive size function to discretize the three-dimensional geometric simulation model of the oil-filling main equipment, gradually reducing the grid diameter near the boundaries, corners and curves, and gradually increasing the grid diameter as the distance increases.

3. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment according to claim 1, characterized in that: The oil pressure increase process inside the oil filling main equipment is simulated in the grid model of the oil filling main equipment, and the oil pressure increase process inside the oil filling main equipment is simulated using bubble dynamics. The bubble dynamics equation is: In the formula, represents the radius of the bubble, is the normal velocity of the bubble surface, represents the normal acceleration of the bubble surface; Indicates the density of insulating oil, represents the pressure at the boundary of the fluid domain, represents the heat transfer coefficient, represents the specific heat ratio, represents the fluid dynamic viscosity, represents the surface tension of the fluid, represents the external work done by the bubble surface during the bubble expansion process, It means that the surface tension does work on the outside during the bubble expansion process. It means that the viscous force does work externally during the bubble expansion process. represents the initial internal energy of the bubble, Indicates arc energy; Arc energy The calculation formula is: In the formula, Indicates the fault current, Indicates the fault current The corresponding arc voltage, Indicates the fault duration.

4. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment according to claim 1, characterized in that: The method for simulating the turbulence phenomenon caused by the dynamic change of the fluid inside the oil-filled main device is specifically: using a dynamic grid to update the instantaneous fluid field inside the oil-filled main device, and simulating the turbulence phenomenon caused by the dynamic change of the fluid caused by the expansion of bubbles inside the oil-filled main device by capturing the flow characteristics and transient changes of physical quantities of the fluid during the structural change of the oil-filled main device. The calculation formula is: In the formula, represents the time derivative, is the density of the fluid, u is the flow velocity vector, is the area vector, is the diffusion coefficient, express The source term of is the turbulence characteristic quantity, is the mesh speed of the moving mesh, is the control volume, is the vector differential operator, Turbulence characteristic quantity The gradient of is the control volume of the border.

5. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment according to claim 4, characterized in that: The dynamic mesh update uses the Laplace smoothing model, and the calculation formula of the Laplace smoothing model is: In the formula, Representation Node i The new location, Representation Node The number of adjacent nodes of Is an adjacent node location.

6. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment according to claim 1, characterized in that: Set the turbulence phenomenon of the oil-filled main equipment to Model, which sets the equation: In the formula, represents the turbulent kinetic energy, Turbulent kinetic energy The differential change of represents the time differential, Represents spatial coordinates The differential of represents the symbol of partial derivative, represents the turbulent kinetic energy, The turbulent Prandtl number, represents the turbulence specific dissipation rate, represents the turbulence specific dissipation rate The turbulent Prandtl number, represents the average velocity of the fluid, represents the specific dissipation rate, represents the molecular viscosity, represents the turbulent viscosity, The Prandtl number representing the kinetic energy equation, represents the generation term of turbulent kinetic energy, which is generated by shear stress, represents the magnitude of the strain rate, and represents the model constant, Is an adjacent node location.

7. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment according to claim 1, characterized in that: According to the oil pressure changes and The model obtains the oil pressure load, which needs to be further calculated through the momentum equation. The calculation formula of the momentum equation is: In the formula, is the dynamic viscosity, represents the external force vector per unit volume acting on the fluid, It's pressure. is the density of the fluid, is the flow rate of the fluid inside the oil-filled main equipment, Represents the dot product operation of the vector field on the vector micromolecule; the pressure is solved by iteratively solving the momentum equation and the flow rate of the fluid inside the oil-filled main equipment , through the pressure and the flow rate of the fluid inside the oil-filled main equipment The pressure distribution is obtained by coupling and compared with the oil pressure change and The models are combined to obtain the oil pressure load.

8. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment according to claim 1, characterized in that: The displacement vectors of the grid points of the oil-filled main equipment grid model in the x, y and z directions It is obtained through explicit dynamics calculation. The specific calculation method is: use the central difference method to perform discrete integration of acceleration, velocity and displacement in time steps to obtain the displacement vector of the grid point of the grid model of the oil-filled main equipment , The displacement of the grid points of the oil-filled main equipment grid model in the x direction, The displacement of the grid points of the oil-filled main equipment grid model in the y direction, The displacement of the grid points of the oil-filled main equipment grid model in the z direction, and the principal strains of the oil-filled main equipment grid model in the x, y and z directions , and The strain calculation formula is: .

9. A method for evaluating the internal fault tolerance time of a large oil-filled main equipment according to claim 1, characterized in that: The method for judging whether the oil-filled main device is broken is specifically: judging whether the oil-filled main device is broken according to the constant strain failure criterion, and the calculation formula for the judgment is: In the formula, represents the failure strain, represents the equivalent plastic strain. If the equivalent plastic strain >Failure strain , then the oil-filled main equipment breaks. If the equivalent plastic strain ≤Failure strain , the main oil-filled equipment is not broken.

10. A method for evaluating internal fault tolerance time of large oil-filled main equipment according to claim 1, characterized in that: The method is based on a convolutional neural network to establish a neural network model of fault current and allowable fault tolerance time. Specifically, the obtained allowable fault tolerance time data under different fault current conditions are used as input values, and the data are trained by a convolutional neural network to generate a neural network model of the relationship between the fault current and the allowable fault tolerance time. The convolutional neural network includes a convolutional layer, a pooling layer, and a fully connected layer. The convolutional layer is used for different fault currents. The corresponding allowable fault tolerance time data is used for feature extraction. The pooling layer optimizes the extracted features. The fully connected layer maps the optimized feature data between the fault current and the allowable fault tolerance time. The calculation formula for feature extraction by the convolution layer is: In the formula, Indicates k The feature quantity is at the position ( i , j ), Indicates that the input data is at position The value at Indicates that the convolution kernel is k The weight matrix of the channel, Represents the bias term; the calculation of the pooling layer to optimize the extracted features includes average pooling and maximum pooling. The calculation formulas of average pooling and maximum pooling are as follows: in represents average pooling, represents the maximum pooling, N represents the dimension of the convolution kernel, It represents the data points in the pooled area; The fully connected layer maps the optimized feature data between the fault current and the allowable fault tolerance time, using the Softmax function, and its calculation formula is as follows: In the formula, It is the fully connected layer j The output of a neuron, and b Represent the weight and bias respectively, represents the i-th optimized feature data output by the pooling layer, and Represents the j The output probability corresponding to a neuron in Softmax is, is the probability distribution in the Softmax function.

Images