Transformer fault prediction method, device, equipment, medium and product
By combining finite element analysis and deep learning, a temperature field prediction sample set is generated and a deep neural network model is trained, which solves the problems of insufficient accuracy and efficiency in transformer fault prediction and achieves high-precision and high-efficiency fault prediction.
Patent Information
- Application Number
- CN202511082177.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-04
- Publication Date
- 2025-11-18
AI Technical Summary
Existing technologies for transformer fault prediction suffer from insufficient prediction accuracy, generalization, and prediction efficiency.
A temperature field prediction sample set is generated through finite element analysis. The sample set is trained based on a deep neural network model. The temperature field prediction data is compared with the temperature field reference data of the transformer when it is not in a fault state to determine the fault prediction result of the transformer.
It improves the accuracy and efficiency of transformer fault prediction, overcomes the limitations of traditional methods in modeling high-dimensional temperature fields, and enhances generalization.
Smart Images
Figure CN120974102A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present invention relate to the field of artificial intelligence technology, and in particular to a method, apparatus, equipment, medium and product for predicting transformer faults. Background Technology
[0002] As a core piece of equipment in the power system, transformers are responsible for the transmission and distribution of electricity in the power grid. A transformer's failure can trigger a series of chain reactions, affecting everything from local power supply to the stability of the entire power grid. Transformer fault prediction allows for the timely detection and handling of potential problems, facilitating the safe, stable, and efficient operation of transformers and ensuring the safety and reliability of the power system.
[0003] In existing technologies, methods such as support vector machines and genetic algorithms can be used for transformer fault prediction, but these methods have shortcomings in terms of prediction accuracy, generalization, and prediction efficiency. Summary of the Invention
[0004] This invention provides a method, apparatus, equipment, medium, and product for predicting transformer faults, which can improve the prediction accuracy and efficiency of transformer fault prediction and enhance its generalizability.
[0005] In a first aspect, embodiments of the present invention provide a transformer fault prediction method, comprising:
[0006] A temperature field prediction sample set is generated through finite element analysis. The temperature field prediction sample set includes multiple state influence parameter groups for simulating different fault states of transformers, and temperature field sample data corresponding to each state influence parameter group.
[0007] A deep neural network model is trained based on the temperature field prediction sample set to obtain the target deep neural network model.
[0008] The target deep neural network model is used to determine the temperature field prediction data corresponding to the group of parameters affecting the state to be predicted.
[0009] The predicted temperature field data is compared with the reference temperature field data of the transformer when it is not in a fault state to determine the fault prediction result of the transformer.
[0010] Secondly, embodiments of the present invention provide a transformer fault prediction device, comprising:
[0011] The sample generation module is used to generate a temperature field prediction sample set through finite element analysis. The temperature field prediction sample set includes multiple state influence parameter groups for simulating different fault states of transformers, and temperature field sample data corresponding to each state influence parameter group.
[0012] The model training module is used to train a deep neural network model based on the temperature field prediction sample set to obtain the target deep neural network model.
[0013] The temperature field prediction module is used to determine the temperature field prediction data corresponding to the group of parameters affecting the state to be predicted through the target deep neural network model.
[0014] The fault prediction module is used to compare the predicted temperature field data with the reference temperature field data of the transformer when it is not in a fault state, and to determine the fault prediction result of the transformer.
[0015] Thirdly, embodiments of the present invention provide an electronic device, including:
[0016] At least one processor; and
[0017] A memory communicatively connected to the at least one processor; wherein,
[0018] The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the method as described in the first aspect.
[0019] Fourthly, embodiments of the present invention provide a computer-readable storage medium storing computer instructions that cause a processor to execute the method described in the first aspect.
[0020] Fifthly, embodiments of the present invention provide a computer program product, the computer program product including a computer program, which, when executed by a processor, implements the method described in the first aspect.
[0021] The technical solution of this invention generates a temperature field prediction sample set through finite element analysis. This sample set includes multiple state influence parameter groups for simulating different fault states of a transformer, and temperature field sample data corresponding to each state influence parameter group. A deep neural network model is trained based on this sample set to obtain a target deep neural network model. The target deep neural network model is used to determine the temperature field prediction data corresponding to the state influence parameter groups to be predicted. The predicted temperature field data is compared with reference temperature field data of the transformer when it is not in a fault state to determine the transformer's fault prediction result. This solution integrates finite element simulation and deep learning technologies to achieve transformer temperature field prediction and transformer fault inversion. It overcomes the limitations of traditional methods in modeling high-dimensional temperature fields, improves the prediction accuracy of transformer temperature fields, and further improves the prediction accuracy of transformer fault prediction. The coupling of multiple state influence parameters is fully considered during temperature field prediction, which enhances generalization. Using deep learning to achieve transformer temperature field prediction and further achieve fault prediction improves the prediction efficiency of transformer fault prediction.
[0022] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description
[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0024] Figure 1 This is a flowchart of a transformer fault prediction method provided in Embodiment 1 of the present invention;
[0025] Figure 2 This is a flowchart of a transformer fault prediction method provided in Embodiment 2 of the present invention;
[0026] Figure 3 This is a schematic diagram of the structure of a deep neural network model provided in Embodiment 2 of the present invention;
[0027] Figure 4 This is a schematic diagram of the structure of a transformer fault prediction device according to Embodiment 3 of the present invention;
[0028] Figure 5 This is a schematic diagram of the structure of an electronic device that implements an embodiment of the present invention. Detailed Implementation
[0029] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.
[0030] It should be noted that the terms "first," "second," etc., used in this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0031] Example 1
[0032] Figure 1 This is a flowchart of a transformer fault prediction method according to Embodiment 1 of the present invention. This embodiment is applicable to situations involving transformer fault prediction. The method can be executed by a transformer fault prediction device, which can be implemented in software and / or hardware and integrated into an electronic device. Furthermore, the electronic device includes, but is not limited to, computers, laptops, etc.
[0033] like Figure 1 As shown, the method includes:
[0034] S110. A temperature field prediction sample set is generated through finite element analysis. The temperature field prediction sample set includes multiple state influence parameter groups for simulating different fault states of the transformer, and temperature field sample data corresponding to each state influence parameter group.
[0035] In this embodiment of the invention, the transformer is not limited. It can be an oil-immersed transformer or any transformer that requires fault prediction. The specific transformer can be selected according to the actual application requirements.
[0036] A state-influence parameter set can be a combination of multiple state-influence parameters, which are parameters that can affect the operating state of a transformer. In one embodiment, the state-influence parameter set includes at least ambient temperature, winding load, and inlet flow velocity. Ambient temperature can be the temperature of the environment surrounding the transformer, such as 20℃ to 50℃; winding load can be the load rate of the transformer windings, usually expressed as a percentage, such as 70% to 120% winding load, which can be divided into low-voltage load and high-voltage load; inlet flow velocity can be the flow rate of the cooling medium (such as insulating oil or air) entering the transformer cooling system.
[0037] In practical applications, transformer fault conditions can be simulated by changing one or more of the ambient temperature, winding load, and inlet flow velocity. That is, a combination of ambient temperature, winding load, and inlet flow velocity capable of simulating any fault condition of the transformer is defined as a set of state-influence parameters, and multiple sets of state-influence parameters are defined in the same way.
[0038] There is no limit to the number of state-affecting parameter groups. For example, corresponding state-affecting parameter groups can be set for both the fault states that the transformer may encounter in actual operation and the fault states that the transformer may theoretically encounter.
[0039] The temperature field sample data corresponding to the state influence parameter group can be understood as the temperature field data corresponding to the operation of the transformer with the state influence parameter group. The temperature field data describes the temperature distribution of various parts of the transformer.
[0040] Finite element analysis can be understood as a technique that uses mathematical approximation methods to numerically simulate complex engineering structures or physical fields. It decomposes a continuous physical system (such as a temperature field) into a finite number of discrete "elements", establishes a mathematical model for each element, and then assembles the elements into a whole system through the connection relationships between the elements, finally solving for an approximate distribution of physical quantities (such as temperature distribution).
[0041] In this step, finite element analysis software can be used to operate the transformer in a virtual environment under multiple state influence parameter groups as determined above. Finite element analysis is then performed to obtain temperature field sample data under each state influence parameter group, i.e., the temperature field data of the transformer under different fault states. Each state influence parameter group and its corresponding temperature field sample data are used as a temperature field prediction sample, generating multiple temperature field prediction samples to obtain a temperature field prediction sample set. The finite element analysis software can be any software capable of performing finite element analysis; no limitation is made here.
[0042] S120. Train a deep neural network model based on the temperature field prediction sample set to obtain the target deep neural network model.
[0043] In this step, the state influence parameter set (i.e., the combination of ambient temperature, winding load and inlet flow velocity) in the temperature field prediction sample set can be used as the model input, and the temperature field sample data corresponding to the state influence parameter set can be used as the model output to train the deep neural network model.
[0044] During training, the input model's state influences the parameter set through multi-level abstraction and feature extraction in the forward propagation phase. Mean squared error constrains the entire training process, and the model parameters are continuously updated in the backpropagation phase. For the trained deep neural network model, model validation can be performed to verify the accuracy of the model's predictions. The training and validation processes are combined until the model's prediction accuracy reaches the expected level, resulting in the target deep neural network model.
[0045] S130. Using the target deep neural network model, determine the temperature field prediction data corresponding to the group of parameters affecting the state to be predicted.
[0046] As shown in S120, during the training phase, the input of the deep neural network model is the state influence parameter set, and the output is the temperature field sample data corresponding to the state influence parameter set.
[0047] In this step, during the application phase of the target deep neural network model, the set of parameters affecting the state to be predicted can be input into the target deep neural network model, and the target deep neural network model can output the temperature field prediction data corresponding to the set of parameters affecting the state to be predicted.
[0048] The set of parameters to be predicted for the state influence can be the set of parameters collected from the transformer for which fault prediction is to be performed, namely, the ambient temperature, winding load, and inlet flow velocity of the transformer for which fault prediction is to be performed. The temperature field prediction data can be the temperature field data predicted for the transformer for which fault prediction is to be performed.
[0049] S140. The predicted temperature field data is compared with the reference temperature field data of the transformer when it is not in a fault state to determine the fault prediction result of the transformer.
[0050] In this embodiment of the invention, when a transformer fails and the type, location, and extent of the failure vary, the transformer's state-affecting parameters and temperature distribution will change accordingly. Therefore, the transformer's temperature distribution can be used to inversely determine whether a transformer has failed, as well as the type, location, and extent of the failure.
[0051] Temperature field reference data can be the temperature field data of the transformer when it is not in a fault state, that is, when the transformer is in normal operating condition, and can be determined according to the actual application needs.
[0052] In this step, a comparison algorithm is used to compare the predicted temperature field data with the reference temperature field data to determine whether there are any abnormal deviations between the predicted and reference data, and to identify the location and extent of such deviations. Then, combined with the thermal characteristics of transformer faults, the algorithm infers whether a fault has occurred inside the transformer, and if so, the type, location, and extent of the fault, thus determining the transformer fault prediction result. The comparison algorithm can be any algorithm capable of comparing temperature field data; no specific limitation is imposed here.
[0053] The technical solution of this invention generates a temperature field prediction sample set through finite element analysis. This sample set includes multiple state influence parameter groups for simulating different fault states of a transformer, and temperature field sample data corresponding to each state influence parameter group. A deep neural network model is trained based on this sample set to obtain a target deep neural network model. The target deep neural network model is used to determine the temperature field prediction data corresponding to the state influence parameter groups to be predicted. The predicted temperature field data is compared with reference temperature field data of the transformer when it is not in a fault state to determine the transformer's fault prediction result. This solution integrates finite element simulation and deep learning technologies to achieve transformer temperature field prediction and transformer fault inversion. It overcomes the limitations of traditional methods in modeling high-dimensional temperature fields, improves the prediction accuracy of transformer temperature fields, and further improves the prediction accuracy of transformer fault prediction. The coupling of multiple state influence parameters is fully considered during temperature field prediction, which enhances generalization. Using deep learning to achieve transformer temperature field prediction and further achieve fault prediction improves the prediction efficiency of transformer fault prediction.
[0054] Example 2
[0055] Figure 2 This is a flowchart of a transformer fault prediction method according to Embodiment 2 of the present invention. This embodiment is based on Embodiment 1 above, further refining the generation of a temperature field prediction sample set through finite element analysis; further refining the training of a deep neural network model based on the temperature field prediction sample set to obtain a target deep neural network model; and further refining the comparison of the temperature field prediction data with the temperature field reference data of the transformer when it is not in a fault state to determine the fault prediction result of the transformer.
[0056] like Figure 2 As shown, the method includes:
[0057] S111. Using the three-dimensional transformer model in the finite element analysis software, combined with multiple state influence parameter groups used to simulate different fault states of the transformer, temperature field sample data corresponding to each state influence parameter group are generated.
[0058] In this step, finite element analysis software can be used to select the three-dimensional model of the transformer for which fault prediction is required. The selected three-dimensional model of the transformer is then run in a virtual environment under different states affecting the parameter group. Finite element analysis is then performed to obtain temperature field sample data under each state affecting the parameter group.
[0059] S112. Take each state-affected parameter group and the corresponding temperature field sample data as a temperature field prediction sample, and generate a temperature field prediction sample set.
[0060] S121. Based on the temperature field prediction sample set, divide the sample set into a training sample set and a test sample set.
[0061] In this step, the temperature field prediction sample set is divided into a training sample set and a test sample set according to a certain sample ratio. For example, the training sample set consists of 90% of the temperature field prediction samples in the temperature field prediction sample set, and the test sample set consists of 10% of the temperature field prediction samples in the temperature field prediction sample set excluding the training sample set.
[0062] S122. The state influence parameter group in the training sample set is used as the model input, and the temperature field sample data corresponding to the input state influence parameter group is used as the model output to train the deep neural network model.
[0063] Figure 3 This is a schematic diagram of the structure of a deep neural network model provided according to Embodiment 2 of the present invention. Figure 3 As shown, the model includes an input layer, an output layer, and multiple hidden layers. The input layer takes in the state influence parameter set as input. After processing through multiple hidden layers, the output should be the temperature field sample data corresponding to the state influence parameter set. Figure 3 The model structure shown is used to train a deep neural network model, allowing the model to learn the relationship between the parameter set and temperature field sample data based on the state of the model. Optionally, the number of hidden layers can be 4, and the number of neurons in each layer can be 128.
[0064] In one embodiment, the training process of the deep neural network model includes a forward propagation stage and a backpropagation stage. The forward propagation stage is implemented using an activation function for performing a binary classification task, and the backpropagation stage updates the model parameters using a mean squared error loss function.
[0065] During the forward propagation phase, the l-th hidden layer of the deep neural network model can be represented as follows:
[0066] a i,l =σ(y i,l )=σ(W l a i,l-1 +b l )
[0067] Where i represents the sample number in the training sample set; a i,l a represents the output of the i-th sample after activation in the l-th layer. i,l-1 This represents the output of the i-th sample after activation in layer (l-1); y i,l W represents the weighted sum of the i-th sample at layer l, i.e., the unactivated raw output; l b is the weight matrix of the l-th layer; l σ is the bias of the l-th layer, used to adjust the output; σ is the activation function used to perform binary classification tasks, such as the Sigmoid activation function.
[0068] During the forward propagation phase, starting from the input layer, 'a' is calculated layer by layer for each hidden layer. i,l and a i,l The input is passed to the next layer l+1, and so on, until it reaches the output layer, completing the forward propagation from input to prediction.
[0069] Throughout the training process, constraints are applied using the mean squared error loss function, which can be expressed as follows:
[0070]
[0071] Where MSE is the mean squared error; N is the number of samples in the training sample set; x i For the predicted value of the output layer for sample i (usually the output layer a) i,L (L is the output layer); y i Let be the actual value of the output layer for sample i, i.e., the temperature field sample data for sample i. The goal of training is to minimize the MSE, making the predicted value as close as possible to the actual value.
[0072] Unlike the forward propagation phase, the core of the backward propagation phase is to calculate the parameter W using MSE. l and b l The gradient is calculated, and then the parameters are updated to make the MSE smaller and smaller.
[0073] Specifically, the error term in the backpropagation phase can be expressed as follows:
[0074] δ i,l =(W l+1 ) T δ i,l+1 ⊙σ′(y i,l )
[0075] Where, δ i,l Let δ be the error term for the i-th sample in the l-th layer. i,l+1 W represents the error term for the i-th sample in the (l+1)-th layer. l+1 Here is the weight matrix for the (l+1)th layer; ⊙ represents element-wise multiplication (also known as the Hadamard product); σ′(yi,l ) is a σ function in y i,l The derivative at point .
[0076] The parameters W of the l-th layer can be updated according to the following formula. l and b l Where α is the learning rate:
[0077]
[0078]
[0079] Through the backpropagation phase, δ is calculated layer by layer, starting from the output layer L. i,l Update W l and b l This makes the MSE smaller and smaller.
[0080] S123. The trained deep neural network model is verified using the test sample set, and the model is evaluated by utilizing the uncertainty between the predicted and actual values of the temperature field sample data involved in the verification.
[0081] By inputting the parameter set of the state influence of the test samples into the trained deep neural network model, the trained deep neural network model outputs the predicted value of the corresponding temperature field sample data, and the model is evaluated by the uncertainty between the predicted value and the actual value.
[0082] Uncertainty can be expressed as uncertainty = |predicted value - actual value| / actual value × 100%, which means the reliability of the predicted value. Generally, the smaller the uncertainty, the higher the prediction accuracy of the model, and vice versa.
[0083] S124. If the model evaluation passes, the trained deep neural network model will be used as the target deep neural network model.
[0084] In this context, "model evaluation passed" can be understood as the uncertainty between the predicted value and the actual value being lower than the set uncertainty threshold (e.g., 1%). In this case, the trained deep neural network model is used as the target deep neural network model.
[0085] S130. Using the target deep neural network model, determine the temperature field prediction data corresponding to the group of parameters affecting the state to be predicted.
[0086] S141. Compare the predicted temperature field data with the reference temperature field data to determine the temperature offset for each temperature field position of the transformer.
[0087] In this step, for each temperature field location, the temperature value corresponding to that location in the predicted temperature field data and the temperature value corresponding to that location in the reference temperature field data are determined. The difference between the two determined temperature values is then defined as the temperature offset. Here, the temperature field location can be understood as the spatial distribution of any point inside the transformer within the temperature field.
[0088] S142. Determine the target offset that exceeds the set offset threshold from the temperature offset.
[0089] In other words, if the target offset is a temperature offset that exceeds the set offset threshold, it indicates that the temperature at the temperature field position corresponding to the target offset has an abnormal deviation. The magnitude of the target offset can also indicate the degree of abnormal deviation.
[0090] S143. Based on the number of target offsets and the temperature field position corresponding to the target offsets, determine the fault prediction result of the transformer.
[0091] In practical applications, if the number of target offsets exceeds the set number, it can be interpreted as a transformer failure. The location of the failure can be determined by the temperature field position corresponding to the target offset, and the degree of failure can be determined by the magnitude of the target offset, thereby determining the transformer failure prediction result.
[0092] The technical solution of this invention uses a three-dimensional transformer model in finite element analysis software, combined with multiple state influence parameter groups used to simulate different fault states of the transformer, to generate temperature field sample data corresponding to each state influence parameter group, thereby generating a temperature field prediction sample set, which can enrich the sample. By dividing the temperature field prediction sample set, one part is used for training and the other part is used for testing, a deep neural network model is trained to obtain a target deep neural network model, aiming to achieve accurate prediction of the transformer temperature field through the model. The transformer fault prediction result is obtained by inverting the transformer temperature field, which can improve the ability to monitor and manage transformer fault states.
[0093] Example 3
[0094] Figure 4 This is a schematic diagram of a transformer fault prediction device according to Embodiment 3 of the present invention. This embodiment is applicable to situations involving transformer fault prediction, such as... Figure 4 As shown, the specific structure of the device includes:
[0095] The sample generation module 41 is used to generate a temperature field prediction sample set through finite element analysis. The temperature field prediction sample set includes multiple state influence parameter groups for simulating different fault states of transformers, and temperature field sample data corresponding to each state influence parameter group.
[0096] Model training module 42 is used to train a deep neural network model based on the temperature field prediction sample set to obtain a target deep neural network model;
[0097] Temperature field prediction module 43 is used to determine the temperature field prediction data corresponding to the group of parameters affecting the state to be predicted through the target deep neural network model.
[0098] The fault prediction module 44 is used to compare the temperature field prediction data with the temperature field reference data of the transformer when it is not in a fault state, and to determine the fault prediction result of the transformer.
[0099] The transformer fault prediction device provided in this embodiment generates a temperature field prediction sample set through a sample generation module using finite element analysis. This sample set includes multiple state influence parameter groups for simulating different fault states of the transformer, and temperature field sample data corresponding to each state influence parameter group. A model training module trains a deep neural network model based on this sample set to obtain a target deep neural network model. The temperature field prediction module uses the target deep neural network model to determine the temperature field prediction data corresponding to the state influence parameter groups to be predicted. The fault prediction module compares the temperature field prediction data with reference temperature field data of the transformer when it is not in a fault state to determine the transformer fault prediction result. This scheme integrates finite element simulation and deep learning technologies to achieve transformer temperature field prediction and transformer fault inversion. It overcomes the limitations of traditional methods in modeling high-dimensional temperature fields, improves the prediction accuracy of transformer temperature fields, and further improves the prediction accuracy of transformer fault prediction. The coupling of multiple state influence parameters is fully considered during temperature field prediction, which enhances generalization. Using deep learning to achieve transformer temperature field prediction and further achieve fault prediction improves the prediction efficiency of transformer fault prediction.
[0100] Furthermore, the state-affecting parameter set includes at least ambient temperature, winding load, and inlet flow rate.
[0101] Furthermore, the sample generation module 41 is specifically used for:
[0102] Using the three-dimensional transformer model in the finite element analysis software, combined with multiple state influence parameter groups used to simulate different fault states of the transformer, temperature field sample data corresponding to each state influence parameter group are generated.
[0103] Each state-affected parameter group and its corresponding temperature field sample data are used as a temperature field prediction sample to generate a temperature field prediction sample set.
[0104] Furthermore, the model training module 42 is specifically used for:
[0105] Based on the temperature field prediction sample set, a training sample set and a test sample set are divided;
[0106] The state influence parameter set in the training sample set is used as the model input, and the temperature field sample data corresponding to the input state influence parameter set is used as the model output to train the deep neural network model.
[0107] The trained deep neural network model is validated using the test sample set, and the model is evaluated by utilizing the uncertainty between the predicted and actual values of the temperature field sample data involved in the validation.
[0108] If the model evaluation is successful, the trained deep neural network model will be used as the target deep neural network model.
[0109] Furthermore, the training process of the deep neural network model includes a forward propagation stage and a back propagation stage. The forward propagation stage is implemented through an activation function used to perform a binary classification task, and the back propagation stage updates the model parameters through a mean squared error loss function.
[0110] Furthermore, the fault prediction module 44 is specifically used for:
[0111] By comparing the predicted temperature field data with the reference temperature field data, the temperature offset corresponding to each temperature field position of the transformer is determined.
[0112] Determine the target offset that exceeds the set offset threshold from the temperature offset;
[0113] Based on the number of target offsets and the temperature field position corresponding to the target offsets, the fault prediction result of the transformer is determined.
[0114] The transformer fault prediction device provided in this embodiment of the invention can execute the transformer fault prediction method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.
[0115] Example 4
[0116] Figure 5 This is a schematic diagram of the structure of an electronic device implementing embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.
[0117] like Figure 5 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.
[0118] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0119] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as transformer fault prediction methods.
[0120] In some embodiments, the transformer fault prediction method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the transformer fault prediction method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the transformer fault prediction method by any other suitable means (e.g., by means of firmware).
[0121] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0122] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0123] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.
[0124] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0125] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.
[0126] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.
[0127] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.
[0128] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.
Claims
1. A method for predicting transformer faults, characterized in that, include: A temperature field prediction sample set is generated through finite element analysis. The temperature field prediction sample set includes multiple state influence parameter groups for simulating different fault states of transformers, and temperature field sample data corresponding to each state influence parameter group. A deep neural network model is trained based on the temperature field prediction sample set to obtain the target deep neural network model. The target deep neural network model is used to determine the temperature field prediction data corresponding to the group of parameters affecting the state to be predicted. The predicted temperature field data is compared with the reference temperature field data of the transformer when it is not in a fault state to determine the fault prediction result of the transformer.
2. The method according to claim 1, characterized in that, The state-affected parameter set includes at least ambient temperature, winding load, and inlet flow rate.
3. The method according to claim 1, characterized in that, A temperature field prediction sample set was generated using finite element analysis, including: Using the three-dimensional transformer model in the finite element analysis software, combined with multiple state influence parameter groups used to simulate different fault states of the transformer, temperature field sample data corresponding to each state influence parameter group are generated. Each state-affected parameter group and its corresponding temperature field sample data are used as a temperature field prediction sample to generate a temperature field prediction sample set.
4. The method according to claim 1, characterized in that, A deep neural network model is trained based on the temperature field prediction sample set to obtain a target deep neural network model, including: Based on the temperature field prediction sample set, a training sample set and a test sample set are divided; The state influence parameter set in the training sample set is used as the model input, and the temperature field sample data corresponding to the input state influence parameter set is used as the model output to train the deep neural network model. The trained deep neural network model is validated using the test sample set, and the model is evaluated by utilizing the uncertainty between the predicted and actual values of the temperature field sample data involved in the validation. If the model evaluation is successful, the trained deep neural network model will be used as the target deep neural network model.
5. The method according to claim 4, characterized in that, The training process of the deep neural network model includes a forward propagation stage and a back propagation stage. The forward propagation stage is implemented through an activation function used to perform a binary classification task, and the back propagation stage updates the model parameters through a mean squared error loss function.
6. The method according to claim 1, characterized in that, The predicted temperature field data is compared with the reference temperature field data of the transformer when it is not in a fault state to determine the fault prediction result of the transformer, including: By comparing the predicted temperature field data with the reference temperature field data, the temperature offset corresponding to each temperature field position of the transformer is determined. Determine the target offset that exceeds the set offset threshold from the temperature offset; Based on the number of target offsets and the temperature field position corresponding to the target offsets, the fault prediction result of the transformer is determined.
7. A transformer fault prediction device, characterized in that, include: The sample generation module is used to generate a temperature field prediction sample set through finite element analysis. The temperature field prediction sample set includes multiple state influence parameter groups for simulating different fault states of transformers, and temperature field sample data corresponding to each state influence parameter group. The model training module is used to train a deep neural network model based on the temperature field prediction sample set to obtain the target deep neural network model. The temperature field prediction module is used to determine the temperature field prediction data corresponding to the group of parameters affecting the state to be predicted through the target deep neural network model. The fault prediction module is used to compare the predicted temperature field data with the reference temperature field data of the transformer when it is not in a fault state, and to determine the fault prediction result of the transformer.
8. An electronic device, characterized in that, include: At least one processor; as well as A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1-6.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the method as described in any one of claims 1-6.
10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method as described in any one of claims 1-6.