A temperature monitoring method and related device for power transformer

Abstract

The present application discloses a temperature monitoring method and related devices for a power transformer, which relates to the field of non-electrical protection of transformers. The temperature data of various positions on the power transformer are collected by intelligent temperature measuring bolts installed in the power transformer at various heat source points where the temperature difference with the external environment reaches a preset temperature difference threshold, and then the collected temperature data is input into a temperature characteristic fault diagnosis model for processing to obtain a temperature characteristic fault diagnosis result. The intelligent temperature measuring bolt in the present application can be installed in any position in the power transformer that meets the above-mentioned installation conditions, that is, it can comprehensively collect the temperature data of various heat source points of the power transformer, and naturally it can also collect the internal temperature of the power transformer. After that, the temperature data is processed by a temperature characteristic fault diagnosis model including a multi-scale convolution module, a densely stacked network module, a residual module, and a compression and incentive attention mechanism module to obtain an accurate temperature characteristic fault diagnosis result.

Description

Technical Field

[0001] The present application relates to the technical field of transformer non-electrical quantity protection, and in particular to a temperature monitoring method and related device for a power transformer. Background Art

[0002] A power transformer is a static electrical device that is mainly used to convert an AC voltage (current) of a certain value into another or several different voltages (currents) with the same frequency.

[0003] Power transformers are vital equipment in power systems, and their stable operation is crucial to ensuring the reliable transmission, flexible distribution, and safe use of electric energy. Among the many parameters that need to be detected in power transformers, temperature is one of the core indicators for monitoring the status of power transformers.

[0004] In the known technology, traditional temperature sensors are used to collect temperature data of power transformers, and the collected temperature data is transmitted to the monitoring station in real time. Further, with the help of wireless sensor networks and gateways, real-time monitoring of the operating status of power transformers is achieved. Real-time monitoring of the operating status of power transformers provides a key reference for the safety and stability of the power system.

[0005] The traditional temperature sensor is an infrared camera device, which specifically uses an infrared camera to detect the temperature of the power transformer. However, this monitoring method cannot monitor the internal temperature of the power transformer. Summary of the invention

[0006] In view of the above problems, the present application provides a temperature monitoring method and related device for a power transformer to achieve the purpose of comprehensively monitoring the temperature of the power transformer. The specific scheme is as follows:

[0007] A first aspect of the present application provides a temperature monitoring method for a power transformer, comprising:

[0008] At least the temperature data of the power transformer is obtained through each temperature measurement node on the wireless link; each temperature measurement node is each intelligent temperature measurement bolt installed in the power transformer at a heat source point where the temperature difference with the external environment reaches a preset temperature difference threshold;

[0009] The temperature data is input into the temperature feature fault diagnosis model for processing to obtain the temperature feature fault diagnosis result; the temperature feature fault diagnosis model is configured to process the temperature data in the order of a multi-scale convolution module, a dense stacked network module, a residual module, and a compression and excitation attention mechanism module, and then output the temperature feature fault diagnosis result.

[0010] In one possible implementation, the design process of the wireless link includes:

[0011] A two-way information exchange model is used to determine the temperature measurement node pairs that need to be time synchronized;

[0012] The time synchronization between temperature measurement node pairs is achieved through the receiver-receiver synchronization model.

[0013] In a possible implementation, the operation process of the temperature characteristic fault diagnosis model includes:

[0014] The temperature data is input into the multi-scale convolution module; the multi-scale convolution module uses filters of different scales to perform convolution fusion on the temperature data for a preset number of times to obtain feature vectors at different scales;

[0015] The feature vectors at different scales are input into the dense stacking network module; the dense stacking network module performs feature fusion on the feature vectors at different scales to obtain fused features;

[0016] The fused features are input into the residual module; the residual module processes the fused features by skip connection to obtain a deep feature representation suitable for propagation in a deep network;

[0017] The deep feature representation is input into the compression and excitation attention mechanism module; after the compression and excitation attention mechanism module inputs the deep feature representation into the global average pooling layer and the fully connected layer in sequence, the feature channel is generated by modeling and the weight of the feature channel is adjusted to obtain the temperature fault feature;

[0018] The temperature fault feature is compared with each sub-fault feature in the fault feature set in the temperature feature fault diagnosis model to obtain the fault diagnosis result of the temperature fault feature.

[0019] In a possible implementation, the compression and excitation attention mechanism module inputs the deep feature representation into the global average pooling layer and the fully connected layer in sequence, models the generated feature channels, and then adjusts the weights of the feature channels to obtain the temperature fault features, including:

[0020] The compression and excitation attention mechanism module inputs the deep feature representation into the global average pooling layer, processes the deep feature representation using a compression operation, and obtains a first processing result;

[0021] Inputting the first processing result into the fully connected layer, processing the first processing result by using an excitation operation to obtain a second processing result;

[0022] The second processing result is processed by a scaling operation to obtain a temperature fault feature.

[0023] In a possible implementation, the method is applied to a data processing end, and a wireless channel with the same frequency band is used between each temperature measurement node on the wireless link and the data processing end.

[0024] In a possible implementation, the intelligent temperature measuring bolt includes a temperature sensor, a thermoelectric generator and a heat sink; the temperature sensor is installed on the temperature sensing surface inside the intelligent temperature measuring bolt; the thermoelectric generator is placed inside the intelligent temperature measuring bolt; and the heat sink is installed on the top of the intelligent temperature measuring bolt.

[0025] A second aspect of the present application provides a temperature monitoring device for a power transformer, comprising:

[0026] an acquisition unit and a processing unit; wherein:

[0027] An acquisition unit is used to acquire at least the temperature data of the power transformer through each temperature measurement node on the wireless link; each temperature measurement node is each intelligent temperature measurement bolt installed in the power transformer at a heat source point position where the temperature difference with the external environment reaches a preset temperature difference threshold;

[0028] The processing unit is used to input the temperature data into the temperature characteristic fault diagnosis model for processing to obtain the temperature characteristic fault diagnosis result; the temperature characteristic fault diagnosis model is configured to process the temperature data in the order of a multi-scale convolution module, a dense stacked network module, a residual module and a compression and excitation attention mechanism module, and then output the temperature characteristic fault diagnosis result.

[0029] A third aspect of the present application provides a computer program product, comprising computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the above methods for monitoring the temperature of a power transformer.

[0030] A fourth aspect of the present application provides a temperature monitoring device for a power transformer, comprising at least one processor and a memory connected to the processor, wherein:

[0031] The memory is used to store computer programs;

[0032] The processor is used to execute the computer program so that the temperature monitoring device of the power transformer can implement any one of the temperature monitoring methods for the power transformer described above.

[0033] A fifth aspect of the present application provides a computer storage medium, which carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the temperature monitoring methods for power transformers described above.

[0034] By means of the above technical scheme, the temperature monitoring method and related device of the power transformer provided by the present application, the present application collects the temperature data of each position on the power transformer through the intelligent temperature measuring bolts installed in the power transformer, at each heat source point position whose temperature difference with the external environment reaches a preset temperature difference threshold, and then inputs the collected temperature data into the temperature characteristic fault diagnosis model for processing to obtain the temperature characteristic fault diagnosis result. The intelligent temperature measuring bolt in the present application can be installed in any position in the power transformer that meets the above installation conditions, that is, it can comprehensively collect the temperature data of each heat source point of the power transformer, and naturally it can also collect the internal temperature of the power transformer. After that, the temperature characteristic fault diagnosis model including a multi-scale convolution module, a densely stacked network module, a residual module and a compression and incentive attention mechanism module is used to process these temperature data to obtain accurate temperature characteristic fault diagnosis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and other features, advantages and aspects of the embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the accompanying drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and the originals and elements are not necessarily drawn to scale.

[0036] Figure 1 A schematic diagram of a flow chart of a temperature monitoring method for a power transformer provided in the present application;

[0037] Figure 2 An example diagram of the ITMB-1000 intelligent temperature measuring bolt provided for this application;

[0038] Figure 3 Diagnostic accuracy curve example diagram and loss example diagram of the temperature characteristic fault diagnosis model provided in this application;

[0039] Figure 4 A connection example diagram of the temperature monitoring method for a power transformer provided in the present application;

[0040] Figure 5 A schematic diagram of a structure of a temperature monitoring device for a power transformer provided in the present application;

[0041] Figure 6 A schematic diagram of the structure of the temperature monitoring device for the power transformer provided in this application. DETAILED DESCRIPTION

[0042] The following describes the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. The terms used in the implementation method section of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0043] The embodiments of the present application are described below in conjunction with the accompanying drawings. Those skilled in the art will appreciate that, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of the present application are also applicable to similar technical problems.

[0044] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and need not be used to describe a specific order or sequential order. It should be understood that the terms used in this way can be interchangeable under appropriate circumstances, which is only to describe the distinction mode adopted by the objects of the same attributes when describing in the embodiments of the present application. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, so that the process, method, system, product or equipment comprising a series of units need not be limited to those units, but may include other units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0045] Temperature is one of the core indicators for monitoring power transformers. Existing temperature monitoring technology for power transformers mainly relies on infrared cameras, which use infrared thermal imaging technology to monitor the operating status of power transformers in real time, detect potential overheating problems in a timely manner, and prevent failures.

[0046] However, the temperature monitoring technology of the power transformer in the known technology has the following problems:

[0047] It can only capture temperature changes on the surface of the transformer, but cannot monitor the temperature inside the power transformer; infrared temperature measurement technology is easily affected by the external environment, such as sunlight, radiation from surrounding objects, etc., resulting in inaccurate temperature measurement data; infrared camera equipment requires regular calibration and maintenance to ensure its measurement accuracy and stability, but due to the complexity and precision of the equipment, maintenance work often requires professional personnel, increasing operation and maintenance costs; infrared camera equipment relies on external power supply and has a limited scope of application.

[0048] In order to solve the above problems, the present application provides a temperature monitoring method and related devices for a power transformer.

[0049] Optional, see Figure 1 , a flow chart of a temperature monitoring method for a power transformer provided in the present application.

[0050] like Figure 1 As shown, the temperature monitoring method of the power transformer includes the following steps:

[0051] Step 101, obtaining at least the temperature data of the power transformer through each temperature measurement node on the wireless link; each temperature measurement node is each intelligent temperature measurement bolt installed in the power transformer at a heat source point position whose temperature difference with the external environment reaches a preset temperature difference threshold.

[0052] The temperature acquisition component of the power transformer provided in the present application is an intelligent temperature measuring bolt, which is specifically installed at the heating point (also called heat source) of the power transformer where wiring cannot be carried out and bolts can be used for fastening. The intelligent temperature measuring bolt includes but is not limited to an intelligent temperature measuring bolt body, a temperature sensor, a thermoelectric generator and a heat sink.

[0053] Temperature sensors are mainly used to collect temperature data of electrical connection points and are generally installed on the temperature sensing surface inside the intelligent temperature measuring bolt.

[0054] The radiator is mainly used for direct heat conduction with the thermoelectric generator to help dissipate heat. The radiator is generally installed on the top of the intelligent temperature measuring bolt.

[0055] Thermoelectric generators are mainly used to convert thermal energy into electrical energy to power temperature sensors, radiators and other components. Specifically, the thermoelectric generator converts the temperature difference between the heating point of the power transformer and the intelligent temperature measuring bolt radiator into electrical energy. It is understandable that the thermoelectric generator can only perform thermoelectric effect when the temperature difference between the heating point of the power transformer and the radiator reaches a preset threshold, wherein the temperature of the radiator is equivalent to the ambient temperature. Exemplarily, the preset threshold of the temperature difference can be 5K, that is, when the temperature difference between the heating point of the power transformer and the environment is greater than 5K, the thermoelectric generator can convert thermal energy into electrical energy to power other components. It should be noted that the thermal energy consumed by the thermoelectric generator for the thermoelectric effect is very small, and the thermal energy of the power transformer is in a state of constant replenishment, so the thermal energy consumed by the thermoelectric generator will not affect the acquisition accuracy of the temperature sensor.

[0056] It should be noted that the intelligent temperature measuring spiral in this application can be selected as ITMB-1000. The performance of this model of intelligent temperature measuring spiral is relatively excellent. If intelligent temperature measuring bolts or other intelligent temperature measuring components with better performance appear in the future, they can be replaced.

[0057] The intelligent temperature measuring bolt model ITMB-1000 mainly includes a bolt body and a temperature measuring device. The bolt body mainly includes a screw and a nut, which are used to tighten the electrical contacts; the temperature measuring device mainly includes a PCB circuit board, a heat sink and a thermoelectric generator; the PCB (Printed Circuit Board) is provided with a temperature sensor and a transmitting antenna for real-time temperature monitoring and data transmission; the thermoelectric generator has a cold end and a hot end, and uses the temperature difference between the electrical contacts at the heat sink end to convert into electrical energy to power the PCB board; the heat sink is located on one side of the cold end of the thermoelectric generator, and directly conducts heat with the cold end to help dissipate heat. The temperature sensor of this model of intelligent temperature measuring bolt adopts an IP66 protection grade design, can work in a temperature environment of -20℃ to 85℃, and has a service life of 20 years.

[0058] For example, see Figure 2 , an example diagram of the ITMB-1000 model intelligent temperature measuring bolt provided in this application.

[0059] like Figure 2 As shown, the heat input surface, heat sink, antenna and indicator light are visible on the surface of the intelligent temperature measuring bolt. The heat input surface is the temperature sensing surface of the temperature sensor in the intelligent temperature measuring bolt; the heat sink is used to help dissipate heat; the antenna refers to the transmitting antenna placed in the intelligent temperature measuring bolt; the indicator light is used to display the working status or warning information of the intelligent temperature measuring bolt, for example, to show whether the intelligent temperature measuring bolt is in normal working state (whether data is collected normally). In addition, when the collected temperature of the power transformer exceeds the preset safety threshold, the indicator light may flash or change color to remind the operation and maintenance personnel to check and deal with potential problems in time.

[0060] The intelligent temperature measuring bolts installed at the heating point of the power transformer and the multifunctional data acquisition and monitoring device located at the data processing end form a wireless link. The temperature measurement nodes in the wireless link are the intelligent temperature measuring bolts installed in the power transformer. The frequency bands of the wireless channels between the intelligent temperature measuring bolts and the multifunctional data acquisition and monitoring device at the data processing end are consistent. For example, the frequency band of the wireless channel can be 433MHz. It should be noted that the frequency band of the wireless channel can be replaced, but it is necessary to ensure that the frequency bands of the wireless channels between the intelligent temperature measuring bolts on the wireless link and between the intelligent temperature measuring bolts and the multifunctional data acquisition and monitoring device are consistent, and the selected channel is available in the actual application environment.

[0061] It should be noted that the selection range of the installation position of the intelligent temperature measuring bolt is: the heating part of the power transformer that can be fastened with bolts, that is, the intelligent temperature measuring bolt can be installed at any position that meets the above conditions. Furthermore, a more precise installation position is preferably selected at the heating part of the power transformer that can be fastened with bolts and the temperature difference with the external environment reaches a preset threshold, and the preset threshold here can be set to 10K.

[0062] The installation orientation of the intelligent temperature measuring bolt is also very important, which is of great significance for the collection of radio frequency and data. Specifically, the intelligent temperature measuring bolt should be installed in a position where the bolt radiator is not subjected to intensive shear or vibration mechanical stress; the radiator should be kept at a certain distance from the heat source, and the fins of the radiator should be facing the heat source and perpendicular to the ground to facilitate the use of natural convection heat dissipation; avoid placing the intelligent temperature measuring bolt in the horizontal direction. If placed in the horizontal direction, the collected heat energy will be greatly reduced. The best installation position is the side or bottom of the heat source point; if there is an electric fan or ventilation fan to generate forced convection, it can greatly improve the data collection performance of the intelligent temperature measuring bolt.

[0063] In addition, before installing the intelligent temperature measuring bolt, the installation surface of the power transformer can be cleaned. Specifically, acetone or anhydrous ethanol can be used for cleaning to remove rust, oil and other debris. For better thermal performance, thermal conductive silicone grease can be filled between the heat source surface of the power transformer and the temperature sensor of the intelligent temperature measuring bolt.

[0064] It should also be noted that the temperature sensor of the intelligent temperature measuring bolt can also collect the transmission energy of the antenna and the wireless link signal strength of the intelligent temperature measuring bolt to which the temperature sensor belongs. The transmission energy and the wireless link signal strength are sent to the multifunctional data acquisition and monitoring device together with the collected temperature data. The multifunctional data acquisition and monitoring device uses the transmission energy to evaluate whether the transmission energy of the current antenna of the intelligent temperature measuring bolt is stable, and uses the wireless link signal strength to evaluate the signal quality of the intelligent temperature measuring bolt.

[0065] In summary, the intelligent temperature measuring bolts used in the temperature monitoring method of the power transformer provided in the present application can not only realize independent power supply through the self-generation function, but also transmit temperature data through wireless transmission technology, which is suitable for temperature monitoring inside the power transformer.

[0066] Optionally, the data processing end obtains the temperature data of the power transformer from each intelligent temperature measuring bolt installed in the power transformer through a wireless channel of a preselected frequency band.

[0067] For the wireless link mentioned above, the design needs to be completed before the power transformer temperature monitoring. The design process of the wireless link includes the following steps:

[0068] Step 1: Use a two-way information exchange model to determine the temperature measurement node pairs that need to be time synchronized.

[0069] The design of the wireless link mainly needs to complete the time synchronization of each smart temperature measuring bolt installed in the power transformer. In this application, a two-way information exchange model and a receiver-receiver synchronization model are mainly used to complete the time synchronization process of each smart bolt. It can be understood that after each smart temperature measuring bolt completes the time synchronization, the design of the wireless link is completed.

[0070] In the present application, a two-way information exchange model is used to determine the temperature measurement node pairs that need to perform two-way information exchange, and specifically, the two-way information exchange model is used to determine the temperature measurement node pairs that perform time synchronization.

[0071] Optionally, a two-way information exchange model is used to perform time synchronization on the temperature measurement node pair.

[0072] Step 2: Realize time synchronization between temperature measurement node pairs through the receiver-receiver synchronization model.

[0073] Then, the receiver-receiver synchronization model is used to complete the time synchronization between the temperature measurement node pairs.

[0074] It should be noted that the time synchronization between each smart temperature measuring bolt on the wireless link is completed through the above two steps, the purpose is to ensure the time consistency of data collection between each smart temperature measuring bolt and the accuracy of subsequent data transmission. In addition, during the time synchronization process, the smart temperature measuring bolts also use wireless channels for communication. The frequency band of the wireless channel is consistent with the frequency band of the wireless channel used by the subsequent smart temperature measuring bolts and the data processing end for data transmission. It can be understood that the smart temperature measuring bolts have wireless communication functions.

[0075] Taking a group of adjacent temperature measurement nodes A and B as an example, with temperature measurement node B as a reference node, temperature measurement node A needs to be synchronized with temperature measurement node B.

[0076] Assume that when the Kth round of information exchange is completed, the temperature measurement node A will be at time T 1,k Send a synchronization message to the temperature measurement node B, and the temperature measurement node B will record the time T when the synchronization message is received 2,k , and reply to the temperature measurement node A with a timestamp T 3,k Finally, the temperature measurement node A will record the time T when it receives the reply message from the temperature measurement node B. 4,k After N rounds of time message exchange, message synchronization is finally achieved.

[0077] In this process, the time synchronization calculation formula in the receiver-receiver synchronization model can be expressed as:

[0078] ;

[0079] ;

[0080] Where f is the time skew of temperature measurement node A relative to temperature measurement node B, is the time offset of temperature measurement node A relative to temperature measurement node B, is a fixed time delay, X k is the variable time delay from temperature measurement node A to temperature measurement node B, Y k is the variable time delay from temperature measurement node B to temperature measurement node A.

[0081] Correspondingly, substituting all the collected timestamp information into the time synchronization formula in the receiver-receiver synchronization model in matrix form yields:

[0082] ;

[0083] Among them, 1 2Nis the number of synchronization rounds, X1 is the initial position of the temperature measurement node A in the synchronization process, and X N is the final position of the temperature measurement node A during the synchronization process, Y1 is the initial coordinate of the temperature measurement node A during the synchronization process, and Y N is the final coordinate of the temperature measurement node A during the synchronization process; T 1,1 , T 1,N , T 4,1 , T 4,N is a timestamp, which indicates the time record of the temperature measurement node A sending and receiving synchronization messages during the time message exchange process from the 1st round to the Nth round; T 2,1 , T 2,N , T 3,1 , T 3,N It is also a timestamp, indicating the time record of the temperature measurement node B receiving and sending time messages in multiple rounds.

[0084] Furthermore, the time synchronization formula in the receiver-receiver synchronization model between any two temperature measurement nodes A and B is:

[0085] ;

[0086] ;

[0087] in, and Indicates the receiving timestamp of temperature measurement node A and temperature measurement node B in the kth round of message exchange, and Indicates the clock frequency multiplier of temperature measurement node A and temperature measurement node B, T 1,k It is the time when the time message is sent from the temperature measuring node A in the kth round of message exchange. and is the fixed transmission delay between temperature measurement node A and temperature measurement node B, and represents the variable time delay from temperature measurement node A to temperature measurement node B and from temperature measurement node B to temperature measurement node A in the kth round, and Indicates the time offset between temperature measurement node A and temperature measurement node B.

[0088] Substituting all the timestamp information collected above into the time synchronization formula in the above receiver-receiver synchronization model in the form of a matrix, we can get:

[0089] ;

[0090] in, represents the difference between the receiving timestamps of temperature measuring node A and temperature measuring node B in multiple rounds; T 1,1 , T1,2 ,…,T 1,N It is the timestamp when the temperature measuring node A sends the synchronization message during the Nth round of message exchange, which is used to pair with the receiving timestamp of the temperature measuring node B to analyze the time difference between the two timestamps; It is the relative clock frequency factor between temperature measurement node A and temperature measurement node B, which is used to adjust the clock rate difference between the two temperature measurement nodes. It is the clock offset of temperature measurement node A relative to temperature measurement node B, which is used to indicate the initial difference in relative time between the two temperature measurement nodes; Indicates the fixed transmission delay in each of the 2N synchronizations; is a variable time delay, which indicates the dynamic delay change from temperature measurement node A to temperature measurement node B or from temperature measurement node B to temperature measurement node A in the i-th round of message exchange.

[0091] It should also be noted that if there is no relative clock skew between the two temperature measurement nodes, the clock offset The least squares estimate of can be written as:

[0092] ;

[0093] In summary, the wireless links to which each smart bolt belongs are designed based on a two-way information exchange model and a receiver-receiver synchronization model.

[0094] Step 102: input the temperature data into the temperature characteristic fault diagnosis model for processing to obtain the temperature characteristic fault diagnosis result; the temperature characteristic fault diagnosis model is configured to process the temperature data in the order of a multi-scale convolution module, a dense stacked network module, a residual module, and a compression and excitation attention mechanism module, and then output the temperature characteristic fault diagnosis result.

[0095] The temperature feature fault diagnosis model is trained and used based on convolutional neural networks. It is mainly used to diagnose whether there is a fault in the power transformer through the temperature data obtained from each intelligent temperature measurement bolt. The model mainly includes four modules: multi-scale convolution module, dense stacking network module, residual module and compression and excitation attention mechanism module, which can also be called SE (Squeeze-and-Excitation) attention mechanism module.

[0096] Among them, the multi-scale convolution module mainly uses filters of different scales to convolve the temperature data for a preset number of times to obtain feature vectors at different scales. This module can improve the convolutional network's ability to recognize features of different sizes and complexities, and improve the robustness of fault diagnosis.

[0097] The densely stacked network module receives feature vectors at different scales output from the multi-scale convolution module, performs feature fusion processing on the feature vectors at different scales to obtain fused features. This module can connect the output of each layer with the input of the subsequent layer to achieve efficient feature fusion, which helps to extract more detailed features and improve the ability to detect faults at an early stage.

[0098] The residual module receives the fused features output by the densely stacked network module, and processes the fused features through skip connections to obtain a deep feature representation suitable for propagation in a deep network. The skip connection can solve the gradient message problem in the deep network, so that the feature information can be effectively propagated in the deep network, which can improve the training effect and accuracy of the temperature feature fault diagnosis model.

[0099] The SE attention mechanism module receives the deep feature representation output by the residual module. After the deep feature representation is input into the global average pooling layer and the fully connected layer in sequence, the module generates feature channels through modeling and adjusts the weights of the feature channels to obtain the temperature fault features. The weights of the generated feature channels are used to weight the feature channels, enhance the weights of important features, suppress irrelevant features, improve the feature expression ability of the temperature feature fault diagnosis model, and further improve the fault identification accuracy of the temperature feature fault diagnosis model.

[0100] Finally, the temperature fault features output by the SE attention mechanism module are compared with the various sub-fault features in the fault feature set generated during the training process in the temperature feature fault diagnosis model. Specifically, the fault feature set is the model training process. By pre-setting the fault data, these pre-set different fault data are processed in the convolutional neural network to obtain waveforms corresponding to different fault data. After multiple trainings, various different waveforms corresponding to different fault features are obtained to form a fault feature set. After the temperature feature fault diagnosis model processes the temperature data to obtain the fault feature, it will judge the fault feature and compare it with the various fault features in the fault feature set. When there are sub-fault features in the fault feature set that are similar to or exactly the same as the fault feature, the power transformer is diagnosed as the fault type indicated by the sub-fault feature.

[0101] In summary, the temperature feature fault diagnosis model integrates multi-scale feature fusion, residual learning and attention mechanism. The model can accurately and efficiently extract and diagnose fault features when processing complex temperature signal data, and has higher robustness and accuracy.

[0102] For example, see Figure 3 , the diagnostic accuracy curve example graph and loss example graph of the temperature characteristic fault diagnosis model provided in this application.

[0103] Depend on Figure 3It can be seen that at the 20th iteration, the training accuracy of the temperature feature fault diagnosis model has reached 96% and tends to be stable.

[0104] It should be noted that the formulas involved in the training process of each module in the temperature characteristic fault diagnosis model are as follows:

[0105] The multi-scale convolution module uses filters of different scales to perform multiple convolutions on the temperature data to obtain feature vectors at different scales. Then, the feature vectors at different scales are fused and further extracted to obtain feature vectors of different fine-grained sizes. The setting of the multi-scale convolution module can improve the robustness of the convolutional neural network.

[0106] The calculation formula is as follows:

[0107] ;

[0108] Among them, Y represents the output data after one-dimensional multi-scale convolution fusion, X i =[X1, X2, X3, …, X n ] represents the feature vectors extracted at different scales, and Concat represents the process of feature fusion.

[0109] After that, the feature vectors of different fine-grained levels are input into the dense stacking network module. The Concat1 layer fuses the input layer with the output layer after the first convolution layer. The Concat2 layer fuses the output layer after the first convolution layer with the output layer after the second convolution layer. The Concat3 layer fuses the output layer after the first, second, and third convolutions of the input layer with the temperature data. It should be noted that the dense stacking network module is conducive to mining the subtle features of temperature data and discovering early faults of power transformers. The calculation formula of this module can be:

[0110] ;

[0111] Among them, C i is the output after the fusion of the i-th feature, X is the temperature data, is the weight of the convolution, b i is the bias of the convolution.

[0112] Furthermore, the fused features output from the densely stacked network module are input into the residual module. The residual module is usually composed of multiple residual blocks. When the input dimension and output dimension of a residual block are the same, the jump connection structure is a simple identity mapping; when the input dimension and output dimension of the residual block are different, a 1×1 convolution channel is added at the jump connection to keep the two dimensions consistent.

[0113] The calculation formula of the residual module can be as follows:

[0114] ;

[0115] Among them, L is the total number of residual blocks, l represents the number of layers of residual blocks, represents the input of the l-th layer residual block, represents the output of the residual block, F(·) represents the identity mapping function, Represents the bias or weight matrix of layer l.

[0116] The residual module obtains a deep feature representation suitable for propagation in a deep network. Then, the deep feature representation output from the residual module is input into the SE attention mechanism module, which mainly includes three working steps:

[0117] First, the deep feature representation is input into the global average pooling layer, and the deep feature representation is processed by a compression operation to obtain a first processing result. The compression operation may also be called a Squeeze operation.

[0118] The calculation formula for the Squeeze operation can be as follows:

[0119] ;

[0120] in, is the predicted value for a specific category, , for a given one-dimensional input, C is the number of channels, L is the sequence length, , is the global average pooling output, For Channel and location The characteristic value of .

[0121] Then, the first processing result is input into the fully connected layer, and the first processing result is processed by an excitation operation to obtain a second processing result. The excitation operation may also be called an Excitation operation.

[0122] The calculation formula of the Excitation operation can be as follows:

[0123] ;

[0124] Among them, s is the output of the fully connected layer, represents the activation function of Sigmoid, represents the activation function of ReLU, , is the weight of the fully connected layer, , is the weight of the fully connected layer, and z is the reduction ratio.

[0125] Finally, the second processing result is processed by a scaling operation to obtain the temperature fault feature. Specifically, the calculated feature channel is multiplied by the original channel feature, thereby performing feature calibration. The scaling operation can also be called a Scale operation.

[0126] The calculation formula for this operation can be as follows:

[0127] ;

[0128] in, is the scaled eigenvalue, is the weight coefficient of channel c.

[0129] Finally, record the results.

[0130] In summary, the temperature monitoring method of the power transformer provided in the present application collects temperature data at various positions on the power transformer through the intelligent temperature measuring bolts installed in the power transformer at various heat source points whose temperature difference with the external environment reaches a preset temperature difference threshold, and then inputs the collected temperature data into the temperature characteristic fault diagnosis model for processing to obtain the temperature characteristic fault diagnosis result. The intelligent temperature measuring bolt in the present application can be installed in any position in the power transformer that meets the above-mentioned installation conditions, that is, it can comprehensively collect the temperature data of various heat source points of the power transformer, and naturally it can also collect the internal temperature of the power transformer. After that, the temperature characteristic fault diagnosis model including a multi-scale convolution module, a densely stacked network module, a residual module and an SE attention mechanism module is used to process these temperature data to obtain accurate temperature characteristic fault diagnosis results.

[0131] It should be particularly noted that the temperature monitoring method for power transformers provided in the present application can monitor the temperature inside the power transformer through intelligent temperature measuring bolts, breaking the limitations of traditional infrared camera monitoring equipment, and has good market applicability. It can also be expanded to the closed busbar connections of circuit breakers, the closed busbar connections of bushings, the inside of bushing equalizers, and other power equipment such as switch cabinets and rectifiers, and has broad application prospects.

[0132] For example, see Figure 4 , a connection example diagram of a temperature monitoring method for a power transformer provided in the present application.

[0133] The temperature data is collected by the intelligent temperature measuring bolts installed on the power transformer, and the collected temperature data is transmitted to the multifunctional data acquisition monitoring device through the wireless channel. The monitoring device analyzes and processes the temperature data and performs fault diagnosis. The monitoring and diagnosis results are sent to the mobile phone APP or Web for display.

[0134] A method for monitoring the temperature of a power transformer provided in an embodiment of the present application is introduced above. A device for executing the method for monitoring the temperature of a power transformer will be introduced below.

[0135] See also Figure 5 , a schematic diagram of a structure of a temperature monitoring device for a power transformer provided in the present application. Figure 5 As shown, the device comprises:

[0136] Acquisition unit 10 and processing unit 20; wherein:

[0137] In one embodiment, the apparatus further comprises a wireless link design unit, configured to:

[0138] A two-way information exchange model is used to determine the temperature measurement node pairs that need to be time synchronized;

[0139] The time synchronization between temperature measurement node pairs is achieved through the receiver-receiver synchronization model.

[0140] In one embodiment, the processing unit 20 is specifically configured to:

[0141] The temperature data is input into the multi-scale convolution module; the multi-scale convolution module uses filters of different scales to perform convolution fusion on the temperature data for a preset number of times to obtain feature vectors at different scales;

[0142] The feature vectors at different scales are input into the dense stacking network module; the dense stacking network module performs feature fusion on the feature vectors at different scales to obtain fused features;

[0143] The fused features are input into the residual module; the residual module processes the fused features by skip connection to obtain a deep feature representation suitable for propagation in a deep network;

[0144] The deep feature representation is input into the compression and excitation attention mechanism module; after the compression and excitation attention mechanism module inputs the deep feature representation into the global average pooling layer and the fully connected layer in sequence, the feature channel is generated by modeling and the weight of the feature channel is adjusted to obtain the temperature fault feature;

[0145] The temperature fault feature is compared with each sub-fault feature in the fault feature set in the temperature feature fault diagnosis model to obtain the fault diagnosis result of the temperature fault feature.

[0146] In one embodiment, the processing unit 20 is specifically configured to:

[0147] The compression and excitation attention mechanism module inputs the deep feature representation into the global average pooling layer, processes the deep feature representation using a compression operation, and obtains a first processing result;

[0148] Inputting the first processing result into the fully connected layer, processing the first processing result by using an excitation operation to obtain a second processing result;

[0149] The second processing result is processed by a scaling operation to obtain a temperature fault feature.

[0150] In one embodiment, a wireless channel with the same frequency band is used between each temperature measurement node on the wireless link in the acquisition unit 10 and the data processing end.

[0151] In one embodiment, the smart temperature measuring bolt in the acquisition unit 10 includes a temperature sensor, a thermoelectric generator and a heat sink; the temperature sensor is installed on the temperature sensing surface inside the smart temperature measuring bolt; the thermoelectric generator is placed inside the smart temperature measuring bolt; and the heat sink is installed on the top of the smart temperature measuring bolt.

[0152] The present application also provides a temperature monitoring device for a power transformer. Figure 6 As shown, it shows a schematic diagram of the structure of the temperature monitoring device for the power transformer provided by the present application. The temperature monitoring device for the power transformer in the embodiment of the present application may include but is not limited to fixed terminals such as mobile phones, laptop computers, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 6 The temperature monitoring device for the power transformer shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0153] like Figure 6 As shown, the temperature monitoring device of the power transformer may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage device 608 to a random access memory (RAM) 603. When the temperature monitoring device of the power transformer is powered on, various programs and data required for the operation of the temperature monitoring device of the power transformer are also stored in the RAM 603. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0154] Typically, the following devices may be connected to the I / O interface 605: an input device 606 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 607 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 608 including, for example, a memory card, a hard disk, etc.; and a communication device 609. The communication device 609 may allow the temperature monitoring device of the power transformer to communicate with other devices wirelessly or by wire to exchange data. Although Figure 6 The temperature monitoring device of the power transformer having various devices is shown, but it should be understood that it is not required to implement or have all the devices shown. More or fewer devices may be implemented or have instead.

[0155] An embodiment of the present application also provides a computer program product including computer-readable instructions. When the computer-readable instructions are executed on an electronic device, the electronic device implements any one of the temperature monitoring methods for power transformers provided in the embodiments of the present application.

[0156] A computer storage medium is also provided in an embodiment of the present application. The storage medium carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the temperature monitoring methods for power transformers provided in the embodiments of the present application.

[0157] It should also be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the device embodiments provided by the present application, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines.

[0158] Through the description of the above implementation mode, the technicians in the field can clearly understand that the present application can be implemented by means of software plus necessary general hardware, and of course, it can also be implemented by special hardware including special integrated circuits, special CPUs, special memories, special components, etc. In general, all functions completed by computer programs can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be various, such as analog circuits, digital circuits or special circuits. However, for the present application, software program implementation is a better implementation mode in more cases. Based on such an understanding, the technical solution of the present application is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which is stored in a readable storage medium, such as a computer floppy disk, a U disk, a mobile hard disk, a ROM, a RAM, a disk or an optical disk, etc., including a number of instructions to enable a computer device (which can be a personal computer, a training device, or a network device, etc.) to execute the method of each embodiment of the present application.

[0159] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0160] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on the computer, the process or function according to the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website site, a computer, a training device or a data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, training device or data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a training device, a data center, etc. that includes one or more available media integration. Available media can be magnetic media, (e.g., floppy disk, hard disk, tape), optical media (e.g., DVD), or semiconductor media (e.g., solid-state hard disk (SSD)), etc.

Claims

1. A method for monitoring the temperature of a power transformer, characterized in that: include: At least the temperature data of the power transformer is obtained through each temperature measurement node on the wireless link; each temperature measurement node is a smart temperature measurement bolt installed in the power transformer at a heat source point where the temperature difference with the external environment reaches a preset temperature difference threshold; The temperature data is input into a temperature characteristic fault diagnosis model for processing to obtain a temperature characteristic fault diagnosis result; the temperature characteristic fault diagnosis model is configured to process the temperature data in the order of a multi-scale convolution module, a dense stacked network module, a residual module, and a compression and excitation attention mechanism module, and then output the temperature characteristic fault diagnosis result; The operation process of the temperature characteristic fault diagnosis model includes: The temperature data is input into the multi-scale convolution module; the multi-scale convolution module uses filters of different scales to perform convolution fusion on the temperature data for a preset number of times to obtain feature vectors at different scales; Inputting the feature vectors at different scales into the dense stacking network module; the dense stacking network module performs feature fusion on the feature vectors at different scales to obtain fused features; The fused features are input into a residual module; the residual module processes the fused features by skip connection to obtain a deep feature representation suitable for propagation in a deep network; The deep feature representation is input into the compression and excitation attention mechanism module; after the compression and excitation attention mechanism module sequentially inputs the deep feature representation into the global average pooling layer and the fully connected layer, the feature channel is generated by modeling, and then the weight of the feature channel is adjusted to obtain the temperature fault feature; The temperature fault feature is compared with each sub-fault feature in the fault feature set in the temperature feature fault diagnosis model to obtain a fault diagnosis result of the temperature fault feature.

2. The temperature monitoring method of the power transformer according to claim 1, characterized in that: The design process of the wireless link includes: A two-way information exchange model is used to determine the temperature measurement node pairs that need to be time synchronized; The time synchronization between the temperature measurement node pairs is achieved through a receiver-receiver synchronization model.

3. The temperature monitoring method of the power transformer according to claim 1, characterized in that: The compression and excitation attention mechanism module inputs the deep feature representation into the global average pooling layer and the fully connected layer in sequence, generates feature channels by modeling, and then adjusts the weights of the feature channels to obtain temperature fault features, including: The compression and excitation attention mechanism module inputs the deep feature representation into the global average pooling layer, processes the deep feature representation using a compression operation, and obtains a first processing result; Inputting the first processing result into a fully connected layer, and processing the first processing result using an excitation operation to obtain a second processing result; The second processing result is processed by a scaling operation to obtain the temperature fault feature.

4. The temperature monitoring method of a power transformer according to claim 1, characterized in that: The method is applied to a data processing end, and a wireless channel with the same frequency band is used between each temperature measurement node on the wireless link and the data processing end.

5. The temperature monitoring method of a power transformer according to claim 1, characterized in that: The intelligent temperature measuring bolt includes a temperature sensor, a thermoelectric generator and a heat sink; the temperature sensor is installed on the temperature sensing surface inside the intelligent temperature measuring bolt; the thermoelectric generator is placed inside the intelligent temperature measuring bolt; and the heat sink is installed on the top of the intelligent temperature measuring bolt.

6. A temperature monitoring device for a power transformer, characterized in that: include: an acquisition unit and a processing unit; wherein: The acquisition unit is used to acquire at least the temperature data of the power transformer through each temperature measurement node on the wireless link; each temperature measurement node is a smart temperature measurement bolt installed in the power transformer at a heat source point position where the temperature difference with the external environment reaches a preset temperature difference threshold; The processing unit is used to input the temperature data into the temperature characteristic fault diagnosis model for processing to obtain a temperature characteristic fault diagnosis result; the temperature characteristic fault diagnosis model is configured to process the temperature data in the order of a multi-scale convolution module, a dense stacked network module, a residual module, and a compression and excitation attention mechanism module, and then output the temperature characteristic fault diagnosis result; The operation process of the temperature characteristic fault diagnosis model in the processing unit includes: The temperature data is input into the multi-scale convolution module; the multi-scale convolution module uses filters of different scales to perform convolution fusion on the temperature data for a preset number of times to obtain feature vectors at different scales; the feature vectors at different scales are input into the dense stacking network module; the dense stacking network module performs feature fusion on the feature vectors at different scales to obtain fusion features; the fusion features are input into the residual module; the residual module processes the fusion features by skip connection to obtain a deep feature representation suitable for propagation in a deep network; the deep feature representation is input into the compression and incentive attention mechanism module; the compression and incentive attention mechanism module inputs the deep feature representation into the global average pooling layer and the fully connected layer in sequence, models and generates feature channels, and then adjusts the weights of the feature channels to obtain temperature fault features; the temperature fault features are compared with each sub-fault feature in the fault feature set in the temperature feature fault diagnosis model to obtain the fault diagnosis result of the temperature fault features.

7. A computer program product, characterized in that It comprises computer-readable instructions, and when the computer-readable instructions are executed on an electronic device, the electronic device implements the temperature monitoring method for a power transformer as claimed in any one of claims 1 to 5.

8. A temperature monitoring device for a power transformer, characterized in that: The method comprises at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program so that the temperature monitoring device of the power transformer can implement the temperature monitoring method of the power transformer according to any one of claims 1 to 5.

9. A computer storage medium, characterized in that The computer storage medium carries one or more computer programs, and when the one or more computer programs are executed by an electronic device, the electronic device can implement the temperature monitoring method for a power transformer as described in any one of claims 1 to 5.

Images