Casing condition monitoring and fault diagnosis system and method based on digital twin model

Abstract

The present invention belongs to the technical field of state monitoring of power transmission and transformation equipment, and discloses a bushing state monitoring and fault diagnosis system and method based on a digital twin model. Oil temperature and oil pressure data of multiple positions of a high-voltage bushing under different operating conditions are obtained through oil temperature and oil pressure sensors, and an operating state data vector is formed. The data vector is input into a trained neural network as a test set, and the state of the high-voltage bushing is monitored and the fault type is diagnosed according to the output results, so as to solve the problem that the state monitoring effect of the current high-voltage bushing online monitoring sensing method is poor.

Description

Technical Field

[0001] The present invention belongs to the technical field of state monitoring of power transmission and transformation equipment, and specifically relates to a transformer high-voltage bushing state monitoring and fault diagnosis system and method based on a digital twin model. Background Art

[0002] Low-oil power equipment represented by high-voltage bushings plays an important role in the safe and stable operation of power systems. With the increase of voltage levels, the scope of high-voltage bushing failures or damages is getting wider and wider, and the personal and property safety issues caused are becoming more and more prominent. For the insulation status diagnosis of bushings, there are still some problems with traditional bushing status assessment and diagnosis methods, including dielectric loss and capacitance tests, such as reflecting less insulation information and requiring power outages. Oil chromatography experiments have been fully applied in partial discharge detection of multi-oil equipment, but the high-voltage bushing contains less oil, and it is not realistic to frequently take oil from the oil valve. Oil chromatography is not suitable for partial discharge detection of high-voltage bushings. Oil temperature is more sensitive to thermal defects of low-oil equipment, and oil pressure is more sensitive to total dissolved gas in oil. The sensing methods of the two are also relatively mature, but there is a lack of relevant standards to provide targeted guidance on the status monitoring and fault diagnosis of high-voltage bushings, so field applications are not sufficient.

[0003] Digital twins are digital models created for physical equipment in a digital way to simulate their actual operating conditions, thereby providing reliable detection of physical quantities that cannot be directly measured. How to apply digital twin technology to the condition monitoring and fault diagnosis of high-voltage bushings, and realize virtual-reality interaction and data analysis between physical entities and digital twins, so as to perform condition monitoring and fault diagnosis for high-voltage bushings, has become an urgent problem to be solved. Summary of the invention

[0004] The present invention provides a bushing state monitoring and fault diagnosis system and method based on a digital twin model, which solves the problem that the current high-voltage bushing online monitoring sensing method has poor state monitoring effect.

[0005] To achieve the above-mentioned purpose, the casing state monitoring and fault diagnosis system based on the digital twin model described in the present invention includes a physical perception module, a digital twin module and a state monitoring and fault diagnosis module, wherein the physical perception module and the digital twin module are both connected to the state monitoring and fault diagnosis module, and the physical perception module is connected to the digital twin module;

[0006] The physical perception module is used to obtain the physical parameters of the transformer bushing; organize the operating sensor quantity into a historical sensor quantity feature matrix and an operating sensor quantity feature matrix, input the historical sensor quantity feature matrix into the digital twin module and the state monitoring and fault diagnosis module, and input the operating sensor quantity matrix into the state monitoring and fault diagnosis module;

[0007] The digital twin module is used to: establish a three-dimensional model of the high-voltage bushing according to the physical parameters of the transformer bushing; establish a parameter model according to the historical operating status data, factory test report and historical maintenance report of the transformer; establish a multi-physics field model by combining the Maxwell equations, heat transfer formulas and Navier-Stokes formulas, and output a simulated sensor quantity characteristic matrix using the multi-physics field model;

[0008] The state monitoring and fault diagnosis module is used to: transform the simulated sensor quantity characteristic matrix into the simulated sensor quantity characteristic vector to construct a neural network, and use the neural network to perform fault diagnosis.

[0009] Furthermore, the physical sensing module includes a sensing device and a storage device, the sensing device includes an oil temperature sensor, an oil pressure sensor, a current transformer, a voltage transformer and a memory, the oil temperature sensor, the oil pressure sensor, the current transformer and the voltage transformer are all connected to the memory, and the memory is used to store data collected by the oil temperature sensor, the oil pressure sensor, the current transformer and the voltage transformer; the data includes oil temperature, oil pressure, current and voltage.

[0010] Furthermore, the oil temperature sensor, oil pressure sensor, current transformer, voltage transformer and memory are installed at the same position of the high-voltage bushing, and the voltage transformer and current transformer are installed on the outlet line of the high-voltage bushing.

[0011] A casing condition monitoring and fault diagnosis method based on a digital twin model comprises the following steps:

[0012] Step 1, obtaining the historical sensor quantity characteristic matrix and the historical state parameter matrix of the high voltage bushing;

[0013] Step 2, obtaining physical parameters of the high-voltage bushing, and performing three-dimensional modeling on the high-voltage bushing according to the physical parameters to obtain a three-dimensional model of the high-voltage bushing;

[0014] Step 3: Establish a digital twin module parameter model based on the historical operating status data, factory test report and historical maintenance report records of the high-voltage bushing;

[0015] Step 4, perform electric field simulation calculation, temperature field simulation calculation and flow field simulation calculation on the high-voltage bushing one by one, obtain the electric field part of the simulation quantity characteristic matrix, the temperature field part of the simulation quantity characteristic matrix and the flow field part of the simulation quantity characteristic matrix, and integrate the electric field part of the simulation quantity characteristic matrix, the temperature field part of the simulation quantity characteristic matrix and the flow field part of the simulation quantity characteristic matrix into a simulation sensor quantity characteristic matrix;

[0016] Step 5: Difference the simulated sensor characteristic matrix and the historical sensor characteristic matrix to obtain the relative error, and compare the relative error with the set threshold:

[0017] When the relative error is less than or equal to the set threshold, the simulated sensor characteristic matrix obtained in step 4 is output;

[0018] When the relative error is greater than the set threshold, retrieve the curves of the electric field material parameters, temperature field material parameters and flow field material parameters changing with temperature, query the curves according to the temperature in the simulated sensing quantity characteristic matrix to obtain new electric field material parameters, temperature field material parameters and flow field material parameters, return to step 4, update the electric field material parameters, temperature field material parameters and flow field material parameters, and re-perform the electric field, temperature field and flow field simulation calculations until the relative error is less than or equal to the set threshold, and output the simulated sensing quantity characteristic matrix obtained in step 4;

[0019] Step 6: Use the historical sensor feature vector and the simulated sensor feature vector to train the reverse neural network, determine the number of hidden layer nodes of the reverse neural network, perform repeated iterations, and obtain a neural network for fault diagnosis;

[0020] Step 7: Obtain the operating status data of the high-voltage bushing, input it into the neural network for fault diagnosis, and perform status monitoring and fault diagnosis on the high-voltage bushing.

[0021] Furthermore, in step 4, when performing electric field simulation calculations, the boundary conditions are the voltage parameters in the historical state parameter matrix, and the calculation results are the electric field part of the simulation quantity characteristic matrix and the Joule heat matrix; when performing temperature field simulation calculations, the boundary conditions are the heating power and the Joule heat matrix, and the output results are the temperature field part of the simulation quantity characteristic matrix and the fluid temperature field matrix; when performing flow field simulation calculations, the boundary conditions are the fluid temperature field matrix, and the output results are the flow field part of the simulation quantity characteristic matrix.

[0022] Further, in step 4, the process of performing electric field calculation simulation calculation includes the following steps:

[0023] The electric field is calculated using the following formula:

[0024]

[0025] Calculate the Joule heating using the following formula:

[0026]

[0027] Where D is the electric displacement vector, dS is the area element, q0 is the charge, E is the electric field intensity, dl is the distance element, p is the heating power, dV is the volume element, and J is the current element;

[0028] The electric field material parameters of different components of the high-voltage bushing three-dimensional model are set to the electric field material parameters corresponding to the ambient temperature. Then, the electric field boundary conditions are set according to the voltage parameters in the historical state parameter matrix. The solution domain is set to the entire high-voltage bushing, the solution domain is meshed, and the solver is used to solve the electric field distribution. After the solution is completed, the electric field part of the simulation characteristic matrix and the Joule heat matrix are obtained.

[0029] Further, in step 4, the process of performing temperature calculation simulation calculation includes the following steps:

[0030] The temperature field is calculated according to the following formula:

[0031]

[0032]

[0033] Where: z is the axial coordinate; r is the radial coordinate; ρ is the transformer oil density; u is the transformer oil axial velocity; v is the transformer oil radial velocity; c p is the specific heat; T is the temperature; S E is the heat of the heat source; k is the thermal conductivity of the solid; λ is the thermal conductivity of the transformer oil;

[0034] The thermal field material parameters of different components of the high-voltage bushing three-dimensional model are set to the thermal field material parameters corresponding to the ambient temperature. The Joule heat matrix obtained by electric field calculation and the heating power in the historical state parameter matrix are set as boundary conditions. The solution domain is set to the entire high-voltage bushing. The interior of the high-voltage bushing is divided into solid area and fluid area. The solution domain is meshed, and the temperature field distribution is solved using a solver. After the solution is completed, the temperature field part of the simulation characteristic matrix and the fluid temperature field matrix are obtained.

[0035] Further, in step 4, the process of performing flow field calculation simulation calculation includes the following steps:

[0036] The flow field is calculated according to the following formula:

[0037]

[0038]

[0039] Where: μ is the dynamic viscosity; F z is the axial volume force of the microelement; F r is the radial volume force of the microelement;

[0040] The flow field material parameters of different components of the high-voltage bushing three-dimensional model are set to the flow field material parameters corresponding to the ambient temperature. The fluid temperature field matrix obtained by temperature field calculation is set as the boundary condition. The solution domain is set to the fluid area inside the high-voltage bushing. The solution domain is meshed, and the flow field is solved using a solver. After the solution is completed, the flow field part of the simulation characteristic matrix is ​​obtained.

[0041] Further, step 6 includes the following steps:

[0042] S6.1. Extract the oil pressure vector and oil temperature vector at the sensor installation location from the historical sensor quantity feature matrix to form a historical sensor quantity feature vector; extract the oil pressure vector and oil temperature vector from the historical sensor quantity feature matrix at different faults to form a historical sensor quantity feature vector;

[0043] S6.2. According to the possible operating conditions of the high-voltage bushing, the ambient temperature, the voltage of the central conductive rod of the high-voltage bushing, the current of the central conductive rod of the high-voltage bushing and the heating power are sorted into a simulation state parameter matrix in the format of a historical state parameter matrix; the simulation state parameter matrix is ​​input into the multi-physics field model of the digital twin module, and the simulation sensor quantity feature matrix is ​​calculated according to the method described in steps S4 and S5; the oil temperature vector and the oil pressure vector are extracted from the simulation sensor quantity feature matrix to form a simulation sensor quantity feature vector;

[0044] S6.3, merge the historical sensor quantity feature vector and the historical state parameter matrix, the simulation sensor quantity feature vector and the simulation state parameter matrix into a data set and make labels; wherein the simulation sensor quantity feature vector and the simulation state parameter matrix correspond to the high-voltage bushing being in a normal state, and the label is recorded as 0; the label of the high-voltage bushing in the historical sensor quantity feature vector and the historical state parameter matrix is ​​recorded as 0, the label of the overheating defect is recorded as 1, and the label of the oil pressure is too high is recorded as 2; randomly shuffle the order of the data set and the labels, and divide them into a training set and a test set;

[0045] S6.4. Building a Neural Network

[0046] Build a three-layer neural network. The number of nodes in the neural network input layer is equal to the total number of elements in the historical sensor feature vector and the historical state parameter matrix or the simulated sensor feature vector and the simulated state parameter matrix. The number of nodes in the output layer is 3, which corresponds to the three fault diagnosis results of normal state, overheating state and excessive oil pressure. Input the training set and test set obtained in step S6.3 into the neural network, train the neural network, and obtain a neural network for fault diagnosis.

[0047] Compared with the prior art, the present invention has at least the following beneficial technical effects:

[0048] Based on the digital twin model, the present invention designs a system and method for effectively monitoring the state and diagnosing faults of transformer high-voltage bushings under the action of multiple physical fields including electric field, flow field and temperature field. The oil temperature and oil pressure data of multiple positions of the high-voltage bushing under different operating conditions are obtained through oil temperature and oil pressure sensors, and the operating state data vector is formed. The vector is input as a test set into the trained neural network, and the state of the high-voltage bushing is monitored and the fault type is diagnosed according to the output results.

[0049] Furthermore, due to the lack of commonality of high-voltage bushings used in different working conditions, the installation position of the oil temperature and oil pressure sensors will also largely affect the measurement results, and it is impossible to determine an appropriate oil temperature and oil pressure threshold for high-voltage bushing status monitoring. The present invention fully learns the historical sensor quantity feature vector and the simulated sensor quantity feature vector through a neural network to obtain a neural network that meets certain accuracy requirements, so that the running sensor quantity feature vector can be tested in real time to achieve high-voltage bushing status monitoring and fault diagnosis.

[0050] Furthermore, the present invention realizes the virtual-real interaction between the physical entity and the digital twin. The oil temperature and oil pressure calculated by the digital twin are compared with the sensor values ​​of the physical entity to determine whether the high-voltage bushing has failed, thus realizing the virtual-real interaction between the physical entity and the digital twin, and using the calculation results to guide the state monitoring and fault diagnosis of the physical entity.

[0051] Furthermore, the present invention overcomes the error caused by the nonlinear change of material properties with temperature through an iterative method, and improves the accuracy of multi-physics field simulation calculation; the coupled calculation of electric field, temperature field and flow field is adopted to make the simulation calculation closer to engineering practice. The multi-physics field simulation results of the present invention will have a strong guiding significance for the state monitoring and fault diagnosis of high-voltage bushings. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0053] Figure 2 Schematic diagram of the process flow for building a digital twin multiphysics model. DETAILED DESCRIPTION

[0054] In order to make the purpose and technical solution of the present invention clearer and easier to understand, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. The specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0055] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, unless otherwise specified, "multiple" means two or more. In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0056] Example 1

[0057] Reference Figure 1 ,The transformer high voltage bushing condition monitoring and fault diagnosis system based on ,digital twin model 1) consists of three modules: physical perception module, digital twin module, and condition ,monitoring and fault diagnosis module. The digital twin module is connected with the physical perception module; ,both the digital twin module and the physical perception module are connected with the condition ,monitoring and fault diagnosis module.

[0058] The physical perception module includes the physical entity of the high-voltage bushing, the sensing device and the storage. The sensing device includes the oil temperature sensor, the oil pressure sensor, the current transformer, the voltage transformer and the storage. The oil temperature sensor, the oil pressure sensor, the current transformer and the voltage transformer are all connected to the storage. The storage is used to store the data collected by the oil temperature sensor, the oil pressure sensor, the current transformer and the voltage transformer; the data includes oil temperature, oil pressure, current and voltage. The physical perception module covers all the physical entities of the system and all the data returned by the sensors.

[0059] The physical perception module is used to complete the following two tasks:

[0060] 1) Obtain the physical parameters of the transformer bushing and save them in the memory: obtain its geometric parameters and material parameters according to the transformer bushing manual, and fully and accurately describe the interaction relationship between the various components in the bushing;

[0061] 2) Data cleaning and output: The memory cleans and organizes the operating sensor quantities sent back by the sensor to form a historical sensor quantity feature matrix and an operating sensor quantity feature matrix. The operating sensor quantities from 3-6 months before the current moment to the current moment are organized into a historical sensor quantity feature matrix, and the operating sensor quantities at the current moment are organized into an operating sensor quantity feature matrix. The historical sensor quantity feature matrix is ​​input into the digital twin module and the state monitoring and fault diagnosis module, and the operating sensor quantity matrix is ​​input into the state monitoring and fault diagnosis module.

[0062] The digital twin module and the condition monitoring and fault diagnosis module establish and store the entire mathematical model of the system, among which the digital twin module consists of a geometric model, a parameter model and a multi-physical field model.

[0063] The geometric model refers to the three-dimensional model of the high-voltage bushing; the geometric model is a three-dimensional model that meets certain accuracy requirements and is established based on the actual physical shape of the high-voltage bushing, including all components such as conductive rods, capacitor cores and flanges.

[0064] The parameter model refers to the engineering parameters of the high-voltage bushing, including historical operating status data, factory test reports, historical maintenance report records and other information; as well as the curves of the electric field material parameters (dielectric constant and conductivity) and temperature field material parameters (heat transfer coefficient) of transformer oil, insulating paperboard and ceramics, and the curves of the flow field material parameters (dynamic viscosity, surface tension and fluid density) of transformer oil changing with temperature.

[0065] The multi-physics model includes Maxwell's equations, heat transfer formulas and Navier-Stokes formulas. The multi-physics model is a physical model that couples the electric field, temperature field and flow field one by one. The multi-physics model calculates Maxwell's equations, heat transfer formulas and Navier-Stokes formulas respectively to obtain the characteristic matrix of the simulated sensor quantity;

[0066] The digital twin module consists of a geometric model, a parametric model, and a multi-physics model. The specific implementation method is as follows:

[0067] 1) Establishing a geometric model: Establishing a three-dimensional model of the high-voltage bushing based on the physical parameters of the transformer bushing obtained by the physical perception module. The modeling process should meet the accuracy requirements;

[0068] 2) Establish parameter model: obtain and store basic information about the transformer and bushing, such as historical operating status data of the transformer, factory test reports, and historical maintenance reports, and establish a parameter model.

[0069] 3) Establish a multi-physics field model: Use the historical sensor characteristic matrix sent back by the physical perception module as the boundary condition, perform multi-physics field simulation calculations based on the Maxwell equations, heat transfer formulas and Navier-Stokes formulas, and obtain the simulated sensor characteristic matrix.

[0070] The condition monitoring and fault diagnosis module consists of a neural network and its training and testing parts.

[0071] The condition monitoring and fault diagnosis module is used to train and test the neural network and store the resulting neural network. Its main operating steps include:

[0072] 1) Data cleaning and network construction: The simulated sensor feature matrix and historical sensor feature matrix transmitted by the digital twin module are cleaned and converted into simulated sensor feature vectors and historical sensor feature vectors. The number of hidden layer nodes is given using an empirical formula to build a neural network framework.

[0073] 2) Training the neural network: Input the simulated sensor feature vector and the historical sensor feature vector into the neural network, use the back propagation algorithm to train the neural network, and obtain a neural network that meets the accuracy requirements.

[0074] 3) Use neural network for fault diagnosis: Input the characteristic matrix of the operating sensor quantity into the neural network, diagnose the fault type with the highest probability, and perform condition monitoring and fault diagnosis on the high-voltage bushing.

[0075] Example 2

[0076] Reference Figure 2 A transformer high-voltage bushing state monitoring and fault diagnosis method based on a specific digital twin model uses a high-voltage bushing physical perception module and a constructed digital twin module for virtual-real interaction, and inputs the output results into a state monitoring and fault diagnosis module for corresponding monitoring and diagnosis, specifically including the following steps:

[0077] S1. Build a physical perception module to obtain the historical state parameter matrix and the historical sensor feature matrix;

[0078] An oil temperature sensor and an oil pressure sensor are installed at the oil inlet of the high-voltage bushing, the upper part, the middle part and the lower part of the capacitor core. A thermometer is installed in the environment where the high-voltage bushing is running to obtain the ambient temperature. A voltage transformer and a current transformer are installed on the outlet line of the high-voltage bushing to measure the voltage and current of the central conductive rod of the high-voltage bushing. A coordinate system is established with the axial direction and radial direction of the high-voltage bushing as the coordinate axes to obtain the coordinates of the above four oil temperature sensors and four oil pressure sensors. After the high-voltage bushing has been running for three to six months, the above sensors collect a number of historical sensing quantities. The historical sensing quantities include the oil temperature at the oil inlet of the high-voltage bushing, the upper part, the middle part and the lower part of the capacitor core, the oil pressure at the oil inlet of the high-voltage bushing, the upper part, the middle part and the lower part of the capacitor core, the ambient temperature, the voltage of the central conductive rod of the high-voltage bushing and the current of the central conductive rod of the high-voltage bushing measured during the running period. According to formula (1), the heating power Q is calculated using the voltage and current historical sensing quantities collected at the same time.

[0079] Q=IU (1)

[0080] Wherein, I is the current of the central conductive rod of the high-voltage bushing, and U is the voltage of the central conductive rod of the high-voltage bushing.

[0081] The ambient temperature, voltage of the central conductive rod of the high-voltage bushing, current and heating power of the central conductive rod of the high-voltage bushing at the same time are combined into a historical state parameter matrix; the oil temperature, oil pressure and the corresponding oil temperature sensor and oil pressure sensor coordinates are combined into a historical sensor quantity feature matrix. Therefore, the historical state parameter matrix has a total of 4 column vectors of ambient temperature, voltage, current, and heating power, and the historical sensor quantity feature matrix has a total of 4 column vectors of x-coordinate, y-coordinate, oil temperature, and oil pressure.

[0082] S2. Build the geometric model of the digital twin module

[0083] Obtain the geometric dimensions of the high-voltage bushing, adopt the coordinate system established in step S1, and use 3D modeling software such as UG or 3Ds max to perform 3D modeling on the high-voltage bushing to obtain a 3D model of the high-voltage bushing, which is the geometric model of the digital twin module, and save the 3D model of the high-voltage bushing in stl format or in a format supported by other multi-physics field simulation software;

[0084] S3. Build the digital twin module parameter model

[0085] The parameter model of the digital twin module stores the engineering information of the high-voltage bushing. The establishment steps include: storing the historical operating status data, factory test reports and historical maintenance report records of the high-voltage bushing; storing the electric field material parameters of transformer oil, insulation cardboard and ceramics, the temperature field material parameter change curve with temperature, storing the flow field material parameter change curve of transformer oil with temperature, and classifying the three types of faults, namely normal state, overheating defect and excessive oil pressure, and storing the historical sensor quantity characteristic matrix when the same type of high-voltage bushing fails into the storage. The electric field material parameters include dielectric constant and conductivity, the temperature field material parameters are heat transfer coefficient, and the flow field material parameters include dynamic viscosity, surface tension and fluid density.

[0086] S4. Build a digital twin multi-physics model and use the multi-physics model to perform multi-physics simulation calculations on high-voltage bushings

[0087] Use commercial finite element simulation software to perform multi-physics field simulation calculations on high-voltage bushings. Use the historical state parameter matrix as the boundary condition to calculate the electric field, temperature field and flow field one by one. The electric field calculation obtains the distribution of the electric field, voltage and Joule heat of the high-voltage bushing; the temperature field calculation obtains the distribution of the temperature field of the high-voltage bushing; the flow field calculation obtains the temperature distribution and flow field distribution of the high-voltage bushing. The Joule heat calculated by the electric field analysis will be used as the boundary condition for the temperature field calculation, and the temperature field distribution calculated by the temperature field will also be used as the boundary condition for the flow field calculation. Finally, the simulation quantity characteristic matrix is ​​calculated, which refers to the coordinates of all grid points in the solution domain and the corresponding simulation calculation values, including the electric field, temperature field and flow field parts. The specific steps are described as follows:

[0088] S4.1. First, the electric field is calculated to obtain the electric field part of the simulation characteristic matrix and the Joule heat matrix. The calculation formula of the electric field satisfies formula (2):

[0089]

[0090] The calculation of Joule heat satisfies formula (3):

[0091]

[0092] Where D is the electric displacement vector, dS is the area element, q0 is the charge, E is the electric field intensity, dl is the distance element, p is the Joule heat, dV is the volume element, and J is the current element.

[0093] The electric field material parameters of different components of the three-dimensional model of the high-voltage bushing are set to the electric field material parameters corresponding to the ambient temperature. Then, the electric field boundary conditions are set according to the voltage parameters in the historical state parameter matrix, the solution domain is set to the entire high-voltage bushing, the solution domain is meshed, and the electric field distribution is solved using a commonly used solver. After the solution is completed, the electric field part of the simulation characteristic matrix and the Joule heat matrix are obtained. The electric field part of the simulation characteristic matrix consists of three column vectors, the first two column vectors are the x-coordinate vectors and y-coordinate vectors of all grid points in the solution domain, and the third column is the electric field intensity of the corresponding grid point. The Joule heat matrix consists of three column vectors, the first two columns are the x-coordinate vectors and y-coordinate vectors of the grid points at the boundary of the conductive rod in the center of the high-voltage bushing, and the third column is the corresponding thermal power value.

[0094] S4.2. Then, the temperature field is calculated to obtain the temperature field part of the simulation characteristic matrix and the fluid temperature field matrix. The temperature field calculation satisfies the energy conservation equation and the heat conduction formula, that is, formula (4).

[0095]

[0096] Where: z is the axial coordinate; r is the radial coordinate; ρ is the transformer oil density; u is the transformer oil axial velocity; v is the transformer oil radial velocity; c p is the specific heat; T is the temperature; S E is the heat of the heat source; k is the thermal conductivity of the solid; λ is the thermal conductivity of the transformer oil.

[0097] The thermal field material parameters (heat transfer coefficient) of different components of the high-voltage bushing three-dimensional model are set to the thermal field material parameters corresponding to the ambient temperature. The Joule heat matrix obtained by the electric field calculation and the heating power in the historical state parameter matrix are set as boundary conditions, and the solution domain is set to the entire high-voltage bushing. The interior of the high-voltage bushing is divided into a solid area and a fluid area, and the solution domain is meshed. The temperature field distribution is solved using a commonly used solver, and the temperature field part of the simulation characteristic matrix and the fluid temperature field matrix are obtained after the solution. The temperature field part of the simulation characteristic matrix consists of three column vectors. The first two column vectors are the x-coordinate vectors and y-coordinate vectors of all grid points in the solid area in the solution domain, and the third column is the temperature vector of the corresponding grid point. The fluid temperature field matrix consists of three column vectors. The first two column vectors are the x-coordinate vectors and y-coordinate vectors of the grid points in the fluid part of the solution domain, and the third column is the temperature vector of the corresponding grid point.

[0098] S4.3. Then, the flow field is calculated to obtain the flow field part of the simulation characteristic matrix. The calculation formula satisfies the Navier-Stokes formula, that is, formula (5).

[0099]

[0100] Where: μ is the dynamic viscosity; F z is the axial volume force of the microelement; F r is the radial volume force of the microelement.

[0101] The flow field material parameters of different components of the high-voltage bushing three-dimensional model are set to the flow field material parameters corresponding to the ambient temperature. The fluid temperature field matrix obtained by temperature field calculation is set as the boundary condition, the solution domain is set to the fluid area in the high-voltage bushing, the solution domain is meshed, and the flow field is solved using a commonly used solver. After the solution is completed, the flow field part of the simulation characteristic matrix is ​​obtained. The flow field part of the simulation characteristic matrix consists of four column vectors. The first two column vectors are the x-coordinate vectors and y-coordinate vectors of all grid points in the solution domain, the third column vector is the oil temperature of the corresponding grid point, and the fourth column vector is the oil pressure of the corresponding grid point.

[0102] S4.4, finally, the electric field part, the temperature field part and the flow field part of the simulation quantity characteristic matrix are synthesized to obtain the simulation quantity characteristic matrix. The electric field part, the temperature field part and the flow field part corresponding to the sensor coordinate in step S1 are read out from the simulation quantity characteristic matrix to synthesize the simulation sensor quantity characteristic matrix.

[0103] S5. Establishing the digital twin module of high-voltage bushing

[0104] Subtract the simulated sensing quantity characteristic matrix obtained in S4.4 from the historical sensing quantity characteristic matrix obtained in step S1 and calculate the relative error. When the relative error is greater than 5%, retrieve the curves of the change of electric field material parameters, temperature field material parameters and flow field material parameters with temperature from the digital twin module parameter model established in step S3, and read the material parameters corresponding to the temperature vector in the simulated sensing quantity characteristic matrix from the curves. Substitute the read new material parameters into step S4 and iterate again until the above relative error is less than 5%, and output the simulated sensing quantity characteristic matrix obtained in S4.4.

[0105] S6. Create neural network training and test sets

[0106] The amount of data returned by the sensor equipment during the actual operation of the high-voltage bushing cannot meet the requirements of neural network training. In order to make up for the lack of data in the historical sensor feature matrix, the digital twin module generates a large number of simulated sensor feature matrices to supplement it. The specific steps are as follows:

[0107] S6.1. First, extract the oil pressure vector and oil temperature vector at the sensor installation location from the historical sensor feature matrix to form the historical sensor feature vector. Extract the historical sensor feature matrix at different faults from the digital twin module parameter model, and extract the oil pressure vector and oil temperature vector to form the historical sensor feature vector.

[0108] S6.2. Then, according to the possible operating conditions of the high-voltage bushing, the ambient temperature, the voltage of the central conductive rod of the high-voltage bushing, the current of the central conductive rod of the high-voltage bushing and the heating power are sorted into a simulation state parameter matrix in the format of the historical state parameter matrix. The simulation state parameter matrix is ​​input into the multi-physics field model of the digital twin module, and the simulation sensor quantity feature matrix is ​​calculated according to steps S4 and S5. The oil temperature vector and the oil pressure vector are extracted from the simulation sensor quantity feature matrix to form a simulation sensor quantity feature vector.

[0109] S6.3, then, the historical sensor feature vector and the historical state parameter matrix, the simulated sensor feature vector and the simulated state parameter matrix are combined into a data set and labeled. The simulated sensor feature vector and the simulated state parameter matrix correspond to the high-voltage bushing in a normal state, and the label is recorded as 0; the label of the high-voltage bushing in the historical sensor feature vector and the historical state parameter matrix is ​​recorded as 0, the label of the overheating defect is recorded as 1, and the label of the oil pressure is too high is recorded as 2; the order of the data set and the label is randomly shuffled, and 75% of the vectors are divided as training sets and 25% of the vectors are divided as test sets.

[0110] S7. Build a neural network

[0111] A three-layer neural network is constructed, which is the input layer, hidden layer and output layer. The number of nodes in the input layer of the neural network is equal to the total number of elements of the historical sensor feature vector and the historical state parameter matrix or the simulated sensor feature vector and the simulated state parameter matrix. The number of nodes in the output layer is 3, which is the three fault diagnosis results corresponding to the normal state, overheating state and high oil pressure. The number of nodes in the hidden layer is determined by referring to the empirical formula (6)

[0112]

[0113] Where m is the number of hidden layer nodes, n is the number of input layer nodes, l is the number of output layer nodes, and α is a constant between 1 and 10. The training set and test set obtained in S6.3 are input into the neural network for repeated iterations, and the parameters of the neural network are repeatedly corrected and calibrated until a neural network that meets the error requirements is obtained through training.

[0114] S8. Condition monitoring and fault diagnosis of high voltage bushings

[0115] In the working state, the sensors in the physical perception module collect the operating sensor quantities in real time to generate the operating sensor quantity feature matrix and the real-time state parameter matrix. The real-time state parameter matrix is ​​input into the multi-physical field model of the digital twin module, and the simulation sensor quantity feature matrix is ​​obtained by simulation calculation. The operating sensor quantity feature matrix and the simulation sensor quantity feature matrix are subtracted and the relative error is taken. If the relative error is less than or equal to 5%, it is determined that the high-voltage bushing is in a normal operating state; if the relative error is greater than 5%, the real-time sensor quantity feature vector is extracted from the operating sensor quantity feature matrix, and the real-time sensor quantity feature vector and the real-time state parameter matrix are input into the neural network, and the output label of the neural network corresponds to the fault diagnosis result.

[0116] The above contents are only for explaining 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 in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A casing condition monitoring and fault diagnosis method based on a digital twin model, characterized in that: The specific steps include: Step 1, obtaining the historical sensor quantity characteristic matrix and the historical state parameter matrix of the high voltage bushing; Step 2, obtaining physical parameters of the high-voltage bushing, and performing three-dimensional modeling on the high-voltage bushing according to the physical parameters to obtain a three-dimensional model of the high-voltage bushing; Step 3: Establish a digital twin module parameter model based on the historical operating status data, factory test report and historical maintenance report records of the high-voltage bushing; Step 4, perform electric field simulation calculation, temperature field simulation calculation and flow field simulation calculation on the high-voltage bushing one by one, obtain the electric field part of the simulation quantity characteristic matrix, the temperature field part of the simulation quantity characteristic matrix and the flow field part of the simulation quantity characteristic matrix, and integrate the electric field part of the simulation quantity characteristic matrix, the temperature field part of the simulation quantity characteristic matrix and the flow field part of the simulation quantity characteristic matrix into a simulation sensor quantity characteristic matrix; Step 5: Difference the simulated sensor characteristic matrix and the historical sensor characteristic matrix to obtain the relative error, and compare the relative error with the set threshold: When the relative error is less than or equal to the set threshold, the simulated sensor characteristic matrix obtained in step 4 is output; When the relative error is greater than the set threshold, retrieve the curves of the electric field material parameters, temperature field material parameters and flow field material parameters changing with temperature, query the curves according to the temperature in the simulated sensing quantity characteristic matrix to obtain new electric field material parameters, temperature field material parameters and flow field material parameters, return to step 4, update the electric field material parameters, temperature field material parameters and flow field material parameters, and re-perform the electric field, temperature field and flow field simulation calculations until the relative error is less than or equal to the set threshold, and output the simulated sensing quantity characteristic matrix obtained in step 4; Step 6: Use the historical sensor feature vector and the simulated sensor feature vector to train the reverse neural network, determine the number of hidden layer nodes of the reverse neural network, perform repeated iterations, and obtain a neural network for fault diagnosis; Step 7, obtaining the operating status data of the high-voltage bushing, inputting the data into the neural network for fault diagnosis, and performing status monitoring and fault diagnosis on the high-voltage bushing; The step 6 comprises the following steps: S6.

1. Extract the oil pressure vector and oil temperature vector at the sensor installation location from the historical sensor quantity feature matrix to form a historical sensor quantity feature vector; extract the oil pressure vector and oil temperature vector from the historical sensor quantity feature matrix at different faults to form a historical sensor quantity feature vector; S6.

2. According to the possible operating conditions of the high-voltage bushing, the ambient temperature, the voltage of the central conductive rod of the high-voltage bushing, the current of the central conductive rod of the high-voltage bushing and the heating power are sorted into a simulation state parameter matrix in the format of a historical state parameter matrix; the simulation state parameter matrix is ​​input into the multi-physics field model of the digital twin module, and the simulation sensor quantity feature matrix is ​​calculated according to the method described in steps S4 and S5; the oil temperature vector and the oil pressure vector are extracted from the simulation sensor quantity feature matrix to form a simulation sensor quantity feature vector; S6.3, merge the historical sensor quantity feature vector and the historical state parameter matrix, the simulation sensor quantity feature vector and the simulation state parameter matrix into a data set and make labels; wherein the simulation sensor quantity feature vector and the simulation state parameter matrix correspond to the high-voltage bushing being in a normal state, and the label is recorded as 0; the label of the high-voltage bushing in the historical sensor quantity feature vector and the historical state parameter matrix is ​​recorded as 0, the label of the overheating defect is recorded as 1, and the label of the oil pressure is too high is recorded as 2; randomly shuffle the order of the data set and the labels, and divide them into a training set and a test set; S6.

4. Building a Neural Network Build a three-layer neural network. The number of nodes in the neural network input layer is equal to the total number of elements in the historical sensor feature vector and the historical state parameter matrix or the simulated sensor feature vector and the simulated state parameter matrix. The number of nodes in the output layer is 3, which corresponds to the three fault diagnosis results of normal state, overheating state and excessive oil pressure. Input the training set and test set obtained in step S6.3 into the neural network, train the neural network, and obtain a neural network for fault diagnosis.

2. The casing condition monitoring and fault diagnosis method based on digital twin model according to claim 1 is characterized in that: In step 4, when performing electric field simulation calculation, the boundary conditions are the voltage parameters in the historical state parameter matrix, and the calculation results are the electric field part of the simulation quantity characteristic matrix and the Joule heat matrix; when performing temperature field simulation calculation, the boundary conditions are the heating power and the Joule heat matrix, and the output results are the temperature field part of the simulation quantity characteristic matrix and the fluid temperature field matrix; when performing flow field simulation calculation, the boundary conditions are the fluid temperature field matrix, and the output results are the flow field part of the simulation quantity characteristic matrix.

3. The casing condition monitoring and fault diagnosis method based on digital twin model according to claim 1 or 2, characterized in that: In step 4, the process of performing electric field calculation simulation calculation comprises the following steps: The electric field is calculated using the following formula: Calculate the Joule heating using the following formula: in, D is the electric displacement vector, d S is the area element, q 0 is the charge, E is the electric field strength, d l is the distance element, p is the heating power, d V is the volume element, J is the current element; The electric field material parameters of different components of the high-voltage bushing three-dimensional model are set to the electric field material parameters corresponding to the ambient temperature. Then, the electric field boundary conditions are set according to the voltage parameters in the historical state parameter matrix. The solution domain is set to the entire high-voltage bushing, the solution domain is meshed, and the solver is used to solve the electric field distribution. After the solution is completed, the electric field part of the simulation characteristic matrix and the Joule heat matrix are obtained.

4. The casing condition monitoring and fault diagnosis method based on digital twin model according to claim 1 or 2, characterized in that: In step 4, the process of performing temperature calculation simulation calculation comprises the following steps: The temperature field is calculated according to the following formula: in: is the axial coordinate; is the radial coordinate; is the transformer oil density; is the transformer oil axial velocity; is the radial velocity of transformer oil; is the specific heat capacity; is the temperature; is the heat of the heat source; is the thermal conductivity of the solid; is the thermal conductivity of transformer oil; The thermal field material parameters of different components of the high-voltage bushing three-dimensional model are set to the thermal field material parameters corresponding to the ambient temperature. The Joule heat matrix obtained by electric field calculation and the heating power in the historical state parameter matrix are set as boundary conditions. The solution domain is set to the entire high-voltage bushing. The interior of the high-voltage bushing is divided into solid area and fluid area. The solution domain is meshed, and the temperature field distribution is solved using a solver. After the solution is completed, the temperature field part of the simulation characteristic matrix and the fluid temperature field matrix are obtained.

5. The casing condition monitoring and fault diagnosis method based on digital twin model according to claim 1, characterized in that: In step 4, the process of performing flow field calculation simulation calculation comprises the following steps: The flow field is calculated according to the following formula: in: is the dynamic viscosity; is the axial volume force of the microelement; is the radial volume force of the microelement; The flow field material parameters of different components of the high-voltage bushing three-dimensional model are set to the flow field material parameters corresponding to the ambient temperature. The fluid temperature field matrix obtained by temperature field calculation is set as the boundary condition. The solution domain is set to the fluid area inside the high-voltage bushing. The solution domain is meshed, and the flow field is solved using a solver. After the solution is completed, the flow field part of the simulation characteristic matrix is ​​obtained.

6. A casing condition monitoring and fault diagnosis system based on the method of claim 1, characterized in that: It includes a physical perception module, a digital twin module, and a state monitoring and fault diagnosis module, wherein the physical perception module and the digital twin module are both connected to the state monitoring and fault diagnosis module, and the physical perception module is connected to the digital twin module; The physical perception module is used to obtain the physical parameters of the transformer bushing; organize the operating sensor quantity into a historical sensor quantity feature matrix and an operating sensor quantity feature matrix, input the historical sensor quantity feature matrix into the digital twin module and the state monitoring and fault diagnosis module, and input the operating sensor quantity matrix into the state monitoring and fault diagnosis module; The digital twin module is used to: establish a three-dimensional model of the high-voltage bushing according to the physical parameters of the transformer bushing; establish a parameter model according to the historical operating status data, factory test report and historical maintenance report of the transformer; establish a multi-physics field model by combining the Maxwell equations, heat transfer formulas and Navier-Stokes formulas, and output a simulated sensor quantity characteristic matrix using the multi-physics field model; The state monitoring and fault diagnosis module is used to: transform the simulated sensor quantity characteristic matrix into the simulated sensor quantity characteristic vector to construct a neural network, and use the neural network to perform fault diagnosis.

7. The casing condition monitoring and fault diagnosis system based on digital twin model according to claim 6 is characterized in that: The physical sensing module includes a sensing device and a storage device. The sensing device includes an oil temperature sensor, an oil pressure sensor, a current transformer, a voltage transformer and a memory. The oil temperature sensor, the oil pressure sensor, the current transformer and the voltage transformer are all connected to the memory. The memory is used to store data collected by the oil temperature sensor, the oil pressure sensor, the current transformer and the voltage transformer; the data includes oil temperature, oil pressure, current and voltage.

8. The casing condition monitoring and fault diagnosis system based on digital twin model according to claim 7, characterized in that: The oil temperature sensor, oil pressure sensor, current transformer, voltage transformer and memory are installed at the same position of the high-voltage bushing, and the voltage transformer and current transformer are installed on the outlet line of the high-voltage bushing.

Images