Transformer hotspot inversion method based on time convolutional neural network, and electronic equipment

Abstract

The invention discloses a transformer hotspot inversion method based on a time convolutional neural network, and electronic equipment. The method comprises the steps that a first training data set is acquired, and the first training data set is constructed based on an orthogonal design mode and a finite element simulation mode; performing data dimension reduction processing on the first training data set to obtain a target training data set; training the initial hotspot inversion model by using the target training data set, and determining a target hotspot inversion model; and using the target hot spot inversion model to predict hot spot data of the target transformer to obtain a prediction result, the prediction result including a target hot spot temperature value and a target hot spot position. According to the invention, the technical problems of low accuracy and high cost of the determination method of the hot spot temperature value and the hot spot position when the three phases of the transformer are unbalanced in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer hotspot inversion, and in particular to a transformer hotspot inversion method and electronic equipment based on a time convolutional neural network. Background Art

[0002] As a key equipment in the power system, the reliability and stability of the transformer are crucial to the safe operation of the entire power grid. Hot spot temperature is an important parameter to measure the operating status of the transformer. It refers to the temperature of the hottest point in the transformer winding. When the transformer is in a three-phase unbalanced operating state, the hot spot position is no longer fixed on the low-voltage or high-voltage winding of a certain phase, but moves dynamically between the three phases with the imbalance of the three-phase load. Accurately measuring and monitoring the hot spot temperature and hot spot position of the transformer is of great significance for preventing transformer overheating failures, extending equipment life, and improving operating efficiency. At present, the measurement methods of transformer hot spot temperature and hot spot position mainly have the following problems:

[0003] (1) Direct measurement method: By embedding the optical fiber sensor directly into the transformer winding conductor to directly obtain the hot spot temperature, it can provide real-time and accurate hot spot temperature and hot spot location. However, since it needs to be repeatedly operated during transformer operation, it will affect the oil flow distribution, and repeated measurements are required under different working conditions, resulting in high costs.

[0004] (2) Thermal simulation method: Based on the simplified model of load guideline IEC 354, the hot spot temperature and location of the winding are estimated by measuring the oil temperature at the top of the transformer and the temperature rise of the winding relative to the oil temperature at the top. This method is simple to operate, but due to the simplification of the model, the estimation result has a large error.

[0005] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0006] The embodiment of the present invention provides a transformer hotspot inversion method and electronic device based on a time convolutional neural network, so as to at least solve the technical problems of low accuracy and high cost in the method for determining the hotspot temperature value and the hotspot position when the three-phase of the transformer is unbalanced in the related art.

[0007] According to one embodiment of the present invention, a transformer hotspot inversion method based on a time convolutional neural network is provided, comprising: obtaining a first training data set, wherein the first training data set is constructed based on an orthogonal design method and a finite element simulation method; performing data dimensionality reduction processing on the first training data set to obtain a target training data set; using the target training data set to train an initial hotspot inversion model to determine a target hotspot inversion model; using the target hotspot inversion model to predict the hotspot data of a target transformer to obtain a prediction result, wherein the prediction result includes a target hotspot temperature value and a target hotspot position.

[0008] Optionally, obtaining the first training data set includes: obtaining preset three-phase load factor parameters, preset ambient temperature parameters and preset ambient wind speed parameters; combining the preset three-phase load factor parameters, preset ambient temperature parameters and preset ambient wind speed parameters using an orthogonal design method to determine multiple groups of training parameters; simulating the multiple groups of training parameters using a finite element simulation method to obtain temperature field distribution data corresponding to each group of training parameters; and determining the first training data set based on the multiple groups of training parameters and the temperature field distribution data.

[0009] Optionally, determining the first training data set based on multiple sets of training parameters and temperature field distribution data includes: determining three corresponding shell temperature measurement points of the target transformer based on the temperature field distribution data; controlling the target transformer to operate according to the multiple sets of training parameters, and obtaining temperature data corresponding to each set of training parameters at the shell temperature measurement points; determining the first training data set based on the temperature data and the multiple sets of training parameters.

[0010] Optionally, performing data dimensionality reduction processing on the first training data set to obtain a target training data set includes: normalizing the first training data set to obtain a second training data set; and performing data dimensionality reduction processing on the second training data set using a correlation analysis method to obtain a target training data set.

[0011] Optionally, the method also includes: obtaining test parameters, wherein the test parameters include: test three-phase load factor parameters, test environment temperature parameters and test environment wind speed parameters; using the test parameters to control the target transformer to obtain the actual hotspot temperature value of the target transformer; inputting the temperature data corresponding to the test parameters into the target hotspot inversion model to obtain the inverted hotspot temperature value of the target transformer; and determining the temperature error of the target hotspot inversion model based on the actual hotspot temperature value and the inverted hotspot temperature value.

[0012] Optionally, the method also includes: obtaining test parameters, wherein the test parameters include: test three-phase load factor parameters, test environment temperature parameters and test environment wind speed parameters; using the test parameters to control the target transformer to obtain the actual hotspot position of the target transformer; inputting the temperature data corresponding to the test parameters into the target hotspot inversion model to obtain the inverted hotspot position of the target transformer; comparing the actual hotspot position with the inverted hotspot position to determine the positioning accuracy of the target hotspot inversion model.

[0013] Optionally, the target hotspot inversion model is used to predict the hotspot data of the target transformer, and the prediction results include: obtaining the current temperature data at the shell temperature measurement point; inputting the current temperature data into the target hotspot inversion model to obtain the target hotspot temperature value and the target hotspot position.

[0014] According to another aspect of one of the embodiments of the present invention, there is also provided a transformer hotspot inversion device based on a time convolutional neural network, comprising: an acquisition module, used to acquire a first training data set, wherein the first training data set is constructed based on an orthogonal design method and a finite element simulation method; a processing module, used to perform data dimensionality reduction processing on the first training data set to obtain a target training data set; a determination module, used to train an initial hotspot inversion model using the target training data set to determine a target hotspot inversion model; a prediction module, used to predict the hotspot data of a target transformer using the target hotspot inversion model to obtain a prediction result, wherein the prediction result includes a target hotspot temperature value and a target hotspot position.

[0015] Optionally, the acquisition module is also used to obtain preset three-phase load coefficient parameters, preset ambient temperature parameters and preset ambient wind speed parameters; the determination module is also used to combine the preset three-phase load coefficient parameters, preset ambient temperature parameters and preset ambient wind speed parameters using an orthogonal design method to determine multiple sets of training parameters; the processing module is also used to simulate multiple sets of training parameters using a finite element simulation method to obtain temperature field distribution data corresponding to each set of training parameters; the determination module is also used to determine a first training data set based on multiple sets of training parameters and temperature field distribution data.

[0016] Optionally, the determination module is also used to: determine three corresponding shell temperature measurement points of the target transformer based on the temperature field distribution data; control the target transformer to operate according to multiple sets of training parameters, and obtain temperature data corresponding to each set of training parameters at the shell temperature measurement points; and determine the first training data set based on the temperature data and multiple sets of training parameters.

[0017] Optionally, the processing module is further used to: perform normalization processing on the first training data set to obtain a second training data set; and perform data dimension reduction processing on the second training data set using a correlation analysis method to obtain a target training data set.

[0018] Optionally, the acquisition module is also used to acquire test parameters, wherein the test parameters include: test three-phase load factor parameters, test environment temperature parameters and test environment wind speed parameters; the determination module is also used to: use the test parameters to control the target transformer to obtain the actual hotspot temperature value of the target transformer; input the temperature data corresponding to the test parameters into the target hotspot inversion model to obtain the inverted hotspot temperature value of the target transformer; based on the actual hotspot temperature value and the inverted hotspot temperature value, determine the temperature error of the target hotspot inversion model.

[0019] Optionally, the acquisition module is also used to obtain test parameters, wherein the test parameters include: test three-phase load factor parameters, test environment temperature parameters and test environment wind speed parameters; the determination module is also used to: use the test parameters to control the target transformer to obtain the actual hotspot position of the target transformer; input the temperature data corresponding to the test parameters into the target hotspot inversion model to obtain the inverted hotspot position of the target transformer; compare the actual hotspot position with the inverted hotspot position to determine the positioning accuracy of the target hotspot inversion model.

[0020] Optionally, the acquisition module is also used to obtain current temperature data at the temperature measurement point of the shell; the prediction module is also used to input the current temperature data into the target hotspot inversion model to obtain the target hotspot temperature value and the target hotspot position.

[0021] According to one embodiment of the present invention, there is also provided an electronic device, comprising: a memory storing an executable program; and a processor for running the program, wherein the above-mentioned transformer hotspot inversion method based on time convolutional neural network is executed when the program is running.

[0022] According to one embodiment of the present invention, a computer-readable storage medium is also provided, including: the computer-readable storage medium includes a stored executable program, wherein when the executable program is running, the device where the storage medium is located is controlled to execute the above-mentioned transformer hotspot inversion method based on time convolutional neural network.

[0023] According to one embodiment of the present invention, a computer program product is also provided, including: a computer program, which implements the above-mentioned transformer hotspot inversion method based on time convolutional neural network when executed by a processor.

[0024] In an embodiment of the present invention, a first training data set is obtained, a data dimension reduction process is performed on the first training data set to obtain a target training data set, and an initial hotspot inversion model is trained using the target training data set to determine a target hotspot inversion model. The target hotspot inversion model is used to predict the hotspot data of a target transformer to obtain a prediction result. By using the target training data set to construct a target hotspot inversion model and using the target hotspot inversion model to predict the hotspot data of the target transformer, the hotspot data acquisition process of the transformer is simplified and the accuracy of the hotspot data is improved, thereby achieving the technical effect of improving the accuracy of the method for determining the hotspot temperature value and the hotspot position and reducing the cost of the method for determining the hotspot temperature value and the hotspot position, thereby solving the technical problem of low accuracy and high cost of the method for determining the hotspot temperature value and the hotspot position when the three-phase imbalance of the transformer in the related art. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0026] Figure 1 It is a hardware structure block diagram of a transformer hot spot inversion method based on a time convolutional neural network according to one embodiment of the present invention;

[0027] Figure 2 is a flow chart of a transformer hot spot inversion method based on a temporal convolutional neural network according to one embodiment of the present invention;

[0028] Figure 3 is a schematic diagram of another transformer hot spot inversion method based on a time convolutional neural network according to one embodiment of the present invention;

[0029] Figure 4 is a flow chart of another transformer hot spot inversion method based on a time convolutional neural network according to one embodiment of the present invention;

[0030] Figure 5 It is a structural block diagram of a transformer hot spot inversion device based on a time convolutional neural network according to one embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] The transformer hotspot inversion method embodiment based on time convolutional neural network provided in the embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 is a hardware structure diagram of a transformer hot spot inversion method based on a time convolutional neural network according to one embodiment of the present invention. Figure 1 As shown, the computer terminal 10 (or electronic device 10) may include one or more (102a, 102b, ..., 102n are used to illustrate) processors (the processor may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission module 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0034] It should be noted that the one or more processors and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or electronic device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0035] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the transformer hot spot inversion method based on time convolutional neural network in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, the above-mentioned transformer hot spot inversion method based on time convolutional neural network is realized. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0036] The transmission module 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission module 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission module 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0037] The display may be, for example, a touch screen type liquid crystal display (LCD), which may enable a user to interact with a user interface of the computer terminal 10 (or electronic device).

[0038] It should be noted that, in some optional embodiments, the above Figure 1 The computer device (or electronic device) shown may include hardware elements (including circuits), software elements (including computer code stored on a computer-readable medium), or a combination of hardware elements and software elements. It should be noted that Figure 1This is merely one example of a particular embodiment and is intended to illustrate the types of components that may be present in the above-described computer device (or electronic device).

[0039] According to an embodiment of the present invention, a method embodiment of a transformer hotspot inversion method based on a temporal convolutional neural network is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0040] Figure 2 is a flow chart of a transformer hot spot inversion method based on a temporal convolutional neural network according to one embodiment of the present invention, such as Figure 2 As shown, the method comprises the following steps:

[0041] Step S20, obtaining a first training data set, wherein the first training data set is constructed based on an orthogonal design method and a finite element simulation method;

[0042] Specifically, first, the orthogonal design method is used to systematically generate training samples covering a variety of operating conditions, including combinations of different load factors, ambient temperatures, and wind speeds. Then, the finite element simulation technology is used to simulate the internal temperature field of the transformer under these conditions, and the first training data set is constructed based on this. The data set contains the temperature data at the temperature measurement points of the transformer shell under various working conditions, the hot spot temperature value, and the hot spot location information.

[0043] Step S22, performing data dimensionality reduction processing on the first training data set to obtain a target training data set;

[0044] Specifically, since the first training data set contains a large number of features, in order to improve the efficiency of model training and prediction performance, data dimensionality reduction processing is required. Through correlation analysis, the high-dimensional features in the first training data set are reduced to fewer dimensions, thereby obtaining the target training data set. This process helps to remove redundant information and retain the most relevant features for hotspot temperature prediction.

[0045] Step S24, training the initial hotspot inversion model using the target training data set to determine the target hotspot inversion model;

[0046] In step S24, the initial hotspot inversion model is constructed based on a temporal convolutional neural network (TCN).

[0047] Specifically, the target training data set is used to train the initial TCN model. Through training, the model can learn the relationship between the temperature data of the temperature measurement point and the temperature value and position of the hot spot, thereby optimizing the model parameters and finally determining the target TCN model.

[0048] Step S26, using the target hotspot inversion model to predict the hotspot data of the target transformer to obtain a prediction result, wherein the prediction result includes the target hotspot temperature value and the target hotspot position.

[0049] Specifically, after obtaining the target hotspot inversion model, it is applied to real-time monitoring of the hotspot condition of the target transformer. The temperature data collected at the shell temperature measurement point in real time will be used as input and sent to the model for prediction. The model will output the prediction results, including the temperature value and location of the target hotspot.

[0050] Based on the above steps S20 to S26, a first training data set is obtained, a data dimension reduction process is performed on the first training data set to obtain a target training data set, and an initial hotspot inversion model is trained using the target training data set to determine a target hotspot inversion model, and the hotspot data of the target transformer are predicted using the target hotspot inversion model to obtain a prediction result. By using the target training data set to construct a target hotspot inversion model and using the target hotspot inversion model to predict the hotspot data of the target transformer, the hotspot data acquisition process of the transformer is simplified and the accuracy of the hotspot data is improved, thereby achieving the technical effect of improving the accuracy of the method for determining the hotspot temperature value and the hotspot position and reducing the cost of the method for determining the hotspot temperature value and the hotspot position, thereby solving the technical problems of low accuracy and high cost of the method for determining the hotspot temperature value and the hotspot position when the three-phase unbalanced transformer is in the related art.

[0051] Optionally, in step S20, obtaining a first training data set includes:

[0052] Step S201, obtaining a preset three-phase load factor parameter, a preset ambient temperature parameter, and a preset ambient wind speed parameter;

[0053] Step S202, combining preset three-phase load factor parameters, preset ambient temperature parameters and preset ambient wind speed parameters by using an orthogonal design method to determine multiple sets of training parameters;

[0054] Step S203, simulating multiple sets of training parameters using a finite element simulation method to obtain temperature field distribution data corresponding to each set of training parameters;

[0055] Step S204: determining a first training data set based on multiple sets of training parameters and temperature field distribution data.

[0056] Specifically, since the three-phase load loss of the transformer is random, different transformer load factors, ambient temperatures and ambient wind speeds are set at different time nodes. Then, the preset three-phase load factor parameters, preset ambient temperature parameters and preset ambient wind speed parameters are combined using the orthogonal design method to obtain multiple sets of training parameters. Finite element simulation is used to perform temperature field simulation on each set of training parameters, calculate the temperature field distribution of the transformer, and obtain the temperature field distribution data corresponding to each set of training parameters. Through finite element simulation, each set of training parameters will generate a temperature field distribution diagram, which shows the specific distribution of the temperature inside and around the transformer under specific conditions, including key data such as hot spot location, temperature gradient, and oil medium temperature change trend.

[0057] For example, since the loss at the load end is random, the hot spot position of the three-phase unbalanced transformer may appear in any phase. Considering the different load factors of each phase, four different load factors of 1.3, 1.0, 0.7, and 0.5 are set. In addition, four ambient temperatures and four ambient wind speeds are set. Using the orthogonal design method, 16 sets of training parameters are obtained from 5 factors and 4 levels. Each set of training parameters is simulated using the finite element simulation method to obtain the temperature field distribution diagram corresponding to each set of training parameters.

[0058] For example, there are five factors, namely, load factor of phase A, load factor of phase B, load factor of phase C, ambient temperature, and ambient wind speed. Each factor has four different levels, for example, the four levels of load factor of phase A are 1.3, 1.0, 0.7, and 0.5.

[0059] Based on the above steps S201 to S204, the preset three-phase load factor parameters, the preset ambient temperature parameters and the preset ambient wind speed parameters are obtained; the preset three-phase load factor parameters, the preset ambient temperature parameters and the preset ambient wind speed parameters are combined by using an orthogonal design method to determine multiple groups of training parameters; the multiple groups of training parameters are simulated by using a finite element simulation method to obtain temperature field distribution data corresponding to each group of training parameters; a first training data set is determined based on multiple groups of training parameters and temperature field distribution data. The orthogonal design can efficiently screen key parameter combinations, reduce measurement costs and time consumption for model training, and the finite element simulation can provide temperature field distribution data that is close to the actual value, which can accurately reflect the heat transfer and distribution inside the transformer even under complex working conditions, thereby improving the quality of the first training data set, helping the model to learn accurate hot spot temperature and position prediction rules, and being able to systematically and comprehensively construct a high-quality training data set, optimize the model training process, and improve the accuracy and robustness of the prediction.

[0060] Optionally, in step S204, determining a first training data set based on multiple sets of training parameters and temperature field distribution data includes:

[0061] Step S2041, determining three corresponding shell temperature measurement points of the target transformer based on the temperature field distribution data;

[0062] Specifically, observe the temperature field distribution diagram corresponding to each set of training parameters, and find the difference in temperature gradient in the transformer when the three-phase windings are respectively at the hot spot position. It can be obtained that: when the three phases of the transformer are balanced, the transformer oil flows evenly in all directions, and the temperature gradient shows that the upper temperature is high, the lower temperature is low, and the temperature difference of each phase is not large; when the three phases of the transformer are unbalanced, the transformer oil is mainly concentrated above the hot spot phase, and the temperature shows that the temperature above the phase with a large load coefficient is greater than that of the other two phases. It can be seen that the difference in temperature gradient becomes a key indicator for judging the hot spot phase. Select the representative temperature gradient change trend of each phase when phases A, B, and C are respectively selected as the hot spot position, select the streamline points according to the temperature gradient, and the temperature change of the streamline points when they are hot spots in different phases is the most significant. Select several points closest to the streamline points on the transformer shell as the shell temperature measurement points.

[0063] Step S2042, controlling the target transformer to operate according to the multiple sets of training parameters, and obtaining temperature data corresponding to each set of training parameters at the shell temperature measurement point;

[0064] Step S2043: determine the first training data set based on the temperature data and multiple sets of training parameters.

[0065] Specifically, the transformer is controlled to operate according to each set of training parameters, and the temperature data at the temperature measuring point of the shell is collected during the operation of the transformer. The temperature data and the corresponding training parameters form a first training data set. For example, there are 16 sets of training parameters, each set of training parameters runs for 5 hours, and the temperature at the temperature measuring point of the shell is measured every 15 minutes, and a total of 320 sets of training samples are obtained to form the first training data set.

[0066] Based on the above steps S2041 to S2043, three corresponding shell temperature measurement points of the target transformer are determined based on the temperature field distribution data; the target transformer is controlled to operate according to multiple groups of training parameters, and the temperature data corresponding to each group of training parameters at the shell temperature measurement points are obtained; the first training data set is determined based on the temperature data and the multiple groups of training parameters, and the temperature measurement points are determined based on the temperature field distribution data and the first training data set that fully covers the working conditions is constructed, which can not only ensure the adequacy and effectiveness of the model training stage, but also improve the accuracy, robustness and adaptability of the target hotspot inversion model in practical applications.

[0067] Optionally, in step S22, performing data dimension reduction processing on the first training data set to obtain a target training data set includes:

[0068] Step S221, normalizing the first training data set to obtain a second training data set;

[0069] Specifically, when the initial hotspot inversion model processes features, if the range of feature values ​​varies greatly, some features will dominate the model learning, while other important but smaller features will be ignored. Therefore, normalization is used to adjust the value ranges of different features in the first data set to a similar scale to obtain the second data set.

[0070] Furthermore, in the first training data set, the purpose of normalization is to ensure that each data has the same weight and influence when it is input into the model, so as to avoid the model being biased towards certain data due to differences in data scales, thereby helping the model to learn and use all features more evenly.

[0071] Step S222: perform data dimension reduction processing on the second training data set by using correlation analysis to obtain a target training data set.

[0072] Specifically, data dimensionality reduction is used to reduce the amount of input data while retaining as much useful information as possible. Using correlation analysis algorithms to perform parameter dimensionality reduction can identify the features most relevant to hotspot temperature prediction. For example, each set of training parameters runs for 5 hours, and the temperature at the shell temperature measurement point is measured every 15 minutes. Then each set of data in the second training data set contains the temperature data at each shell temperature measurement point and each set of training parameters. Not all data in each set of data are equally important for predicting hotspot temperature and hotspot location. Using correlation analysis algorithms to perform parameter dimensionality reduction can retain the features most relevant to hotspot temperature prediction and reduce the complexity and time required for model training.

[0073] Based on the above steps S221 to S222, the first training data set is normalized to obtain a second training data set; the second training data set is subjected to data dimensionality reduction processing using correlation analysis to obtain a target training data set. Normalization and dimensionality reduction processing can improve model training efficiency, enhance prediction accuracy, reduce overfitting risks, and simplify model structure, so that the established model is not only fast and accurate in calculation when dealing with transformer hot spot inversion problems, but also easy to deploy and apply in actual engineering environments.

[0074] Optionally, the method further comprises:

[0075] Step S301, obtaining test parameters, wherein the test parameters include: a test three-phase load factor parameter, a test environment temperature parameter, and a test environment wind speed parameter;

[0076] Step S302, using the test parameters to control the target transformer to obtain the actual hot spot temperature value of the target transformer;

[0077] Step S303, inputting the temperature data corresponding to the test parameters into the target hot spot inversion model to obtain the inverted hot spot temperature value of the target transformer;

[0078] Step S304, determining the temperature error of the target hotspot inversion model based on the actual hotspot temperature value and the inverted hotspot temperature value.

[0079] Specifically, the three-phase load factor, ambient temperature, and ambient wind speed of the transformer are changed to obtain the test parameters. The target transformer is controlled to operate under the test parameters. On the one hand, the actual hot spot temperature value of the target transformer is directly measured using the measurement tool. On the other hand, the temperature data at the temperature measuring point of the shell is collected to obtain the test data set. For example, under each set of test parameters, the transformer is operated for 5 hours, and the temperature at the temperature measuring point of the transformer shell is collected every 15 minutes to obtain the test data set.

[0080] Furthermore, the test data set is input into the target hotspot inversion model to obtain the inverted hotspot temperature value of the target transformer. The actual hotspot temperature value is compared with the inverted hotspot temperature value to determine the temperature error of the target hotspot inversion model.

[0081] Based on the above steps S301 to S304, the test parameters are obtained; the target transformer is controlled by using the test parameters to obtain the actual hot spot temperature value of the target transformer; the temperature data corresponding to the test parameters is input into the target hot spot inversion model to obtain the inverted hot spot temperature value of the target transformer; based on the actual hot spot temperature value and the inverted hot spot temperature value, the temperature error of the target hot spot inversion model is determined. The comparison process can provide a specific direction for the improvement of the model, improve its reliability and generalization ability in practical applications, and ensure that the model is not only effective in theory, but also can stably perform hot spot inversion in engineering practice.

[0082] Optionally, the method further comprises:

[0083] Step S401, obtaining test parameters, wherein the test parameters include: a test three-phase load factor parameter, a test environment temperature parameter, and a test environment wind speed parameter;

[0084] Step S402, controlling the target transformer using the test parameters to obtain the actual hot spot position of the target transformer;

[0085] Step S403, inputting the temperature data corresponding to the test parameters into the target hot spot inversion model to obtain the inversion hot spot position of the target transformer;

[0086] Step S404, comparing the actual hotspot position with the inverted hotspot position to determine the positioning accuracy of the target hotspot inversion model.

[0087] Specifically, the three-phase load factor, ambient temperature, and ambient wind speed of the transformer are changed to obtain the test parameters. The target transformer is controlled to operate under the test parameters. On the one hand, the actual hot spot position of the target transformer is directly measured using the measurement tool. On the other hand, the temperature data at the temperature measurement point of the shell is collected to obtain the test data set. For example, under each set of test parameters, the transformer is operated for 5 hours, and the temperature at the temperature measurement point of the transformer shell is collected every 15 minutes to obtain the test data set.

[0088] Furthermore, the test data set is input into the target hotspot inversion model to obtain the inversion hotspot position of the target transformer, where the location where the hotspot temperature appears is the hotspot position. The actual hotspot position and the inversion hotspot position are compared to determine the positioning accuracy of the target hotspot inversion model.

[0089] Exemplarily, 20 test samples in the test data set are randomly selected, and the temperature error and positioning accuracy of the target hotspot inversion model are calculated and determined by comparing the measured values ​​and the inversion values, as shown in Table 1.

[0090] According to the measured values ​​and inversion values ​​in Table 1, the temperature error between the hotspot inversion result of the transformer and the actual measured value is controlled within 2°C, and the hotspot phase positioning accuracy is 100%.

[0091] Furthermore, according to the temperature error and positioning accuracy of the target hotspot inversion model, the target hotspot inversion model can be further adjusted to improve its prediction accuracy.

[0092] Specifically, Figure 3 is a schematic diagram of another transformer hot spot inversion method based on a time convolutional neural network according to one embodiment of the present invention, such as Figure 3 As shown, the method includes the following steps: obtaining test parameters, wherein the test parameters include: test three-phase load factor parameters, test environment temperature parameters and test environment wind speed parameters; using the test parameters to control the target transformer to obtain the actual value of the hot spot of the target transformer; inputting the temperature data corresponding to the test parameters into the target hot spot inversion model to obtain the inversion value of the hot spot of the target transformer; comparing the actual value with the inversion value to determine the temperature error and positioning accuracy of the target hot spot inversion model.

[0093] Table 1 The measured values ​​and inverted values ​​corresponding to the 20 test samples and the temperature errors of the target hotspot inversion model

[0094]

[0095] Based on the above steps S401 to S404, the test parameters are obtained; the target transformer is controlled by using the test parameters to obtain the actual hot spot position of the target transformer; the temperature data corresponding to the test parameters is input into the target hot spot inversion model to obtain the inverted hot spot position of the target transformer; the actual hot spot position and the inverted hot spot position are compared to determine the positioning accuracy of the target hot spot inversion model. The comparison process can provide a specific direction for the improvement of the model, improve its reliability and generalization ability in practical applications, and ensure that the model is not only effective in theory, but also can stably perform hot spot inversion in engineering practice.

[0096] Optionally, in step S26, the target hotspot inversion model is used to predict the hotspot data of the target transformer, and the prediction results include:

[0097] Step S261, obtaining current temperature data at the temperature measurement point of the housing;

[0098] Specifically, during the operation of the transformer, the temperature data at the shell temperature measurement points are collected in real time. These temperature data reflect the external thermal state of the transformer at a certain moment. The shell temperature measurement points are determined based on the previous analysis and are located at a position that can sensitively reflect internal thermal changes. Therefore, the temperature data at the shell temperature measurement points can be used to understand the thermal distribution inside the transformer.

[0099] Step S262, input the current temperature data into the target hotspot inversion model to obtain the target hotspot temperature value and the target hotspot position.

[0100] Specifically, the current temperature data of the shell temperature measurement point is input into the previously trained target hotspot inversion model. The target hotspot inversion model is a temporal convolutional neural network (TCN) model. After being fully trained using the target training data set Jinxiu Valley, it has been able to predict the hotspot temperature and hotspot position inside the transformer based on the input temperature data. When the target hotspot inversion model receives the current temperature data, it will use its internal weights and parameters to process these data through time series analysis and convolution operations to output the target hotspot temperature value and target hotspot position. If the hotspot temperature value predicted by the model exceeds the safe range, or the hotspot position persists on one phase, it indicates a potential three-phase imbalance problem, and immediate measures need to be taken, such as adjusting the load or conducting further inspections, to prevent the transformer from overheating or damage.

[0101] Based on the above steps S261 to S262, the current temperature data at the shell temperature measuring point is obtained; the current temperature data is input into the target hotspot inversion model to obtain the target hotspot temperature value and the target hotspot position, and the target hotspot inversion model is used to predict the hotspot data of the target transformer, thereby achieving the purpose of simplifying the hotspot data acquisition process of the transformer and improving the accuracy of the hotspot data, thereby achieving the technical effect of improving the accuracy of the method for determining the hotspot temperature value and the hotspot position and reducing the cost of the method for determining the hotspot temperature value and the hotspot position, thereby solving the technical problems of low accuracy and high cost of the method for determining the hotspot temperature value and the hotspot position when the three-phase unbalanced transformer is in the related art.

[0102] Figure 4 is a flowchart of another transformer hot spot inversion method based on a temporal convolutional neural network according to one embodiment of the present invention, such as Figure 4 As shown, the method comprises the following steps:

[0103] Step S401, obtaining a preset three-phase load factor parameter, a preset ambient temperature parameter, and a preset ambient wind speed parameter;

[0104] Step S402, combining preset three-phase load factor parameters, preset ambient temperature parameters and preset ambient wind speed parameters by using an orthogonal design method to determine multiple sets of training parameters;

[0105] Step S403, simulating multiple sets of training parameters using a finite element simulation method to obtain temperature field distribution data corresponding to each set of training parameters;

[0106] Step S404, determining three corresponding shell temperature measurement points of the target transformer based on the temperature field distribution data;

[0107] Step S405, controlling the target transformer to operate according to the multiple sets of training parameters, and obtaining temperature data corresponding to each set of training parameters at the shell temperature measurement point;

[0108] Step S406, determining the first training data set based on the temperature data and multiple sets of training parameters;

[0109] Step S407, normalizing the first training data set to obtain a second training data set;

[0110] Step S408, performing data dimension reduction processing on the second training data set by using correlation analysis to obtain a target training data set;

[0111] Step S409, training the initial hotspot inversion model using the target training data set to determine the target hotspot inversion model;

[0112] Step S410, obtaining current temperature data at the temperature measurement point of the housing;

[0113] Step S411, input the current temperature data into the target hotspot inversion model for prediction, and obtain the target hotspot temperature value and the target hotspot position.

[0114] Based on the above steps S401 to S411, a first training data set is obtained, a data dimension reduction process is performed on the first training data set to obtain a target training data set, and an initial hotspot inversion model is trained using the target training data set to determine a target hotspot inversion model. The target hotspot inversion model is used to predict the hotspot data of the target transformer to obtain a prediction result. By using the target training data set to construct a target hotspot inversion model and using the target hotspot inversion model to predict the hotspot data of the target transformer, the hotspot data acquisition process of the transformer is simplified and the accuracy of the hotspot data is improved, thereby achieving the technical effect of improving the accuracy of the method for determining the hotspot temperature value and the hotspot position and reducing the cost of the method for determining the hotspot temperature value and the hotspot position, thereby solving the technical problems of low accuracy and high cost of the method for determining the hotspot temperature value and the hotspot position when the three-phase unbalanced transformer is in the related art.

[0115] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention 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 storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods of various embodiments of the present invention.

[0116] In an embodiment of the present invention, a transformer hot spot inversion device based on a time convolutional neural network is also provided, and the device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0117] Figure 5 is a structural block diagram of a transformer hot spot inversion device based on a time convolutional neural network according to one embodiment of the present invention, such as Figure 5 As shown, the device comprises:

[0118] An acquisition module 501 is used to acquire a first training data set, wherein the first training data set is constructed based on an orthogonal design method and a finite element simulation method;

[0119] A processing module 502 is used to perform data dimension reduction processing on the first training data set to obtain a target training data set;

[0120] A determination module 503 is used to train the initial hotspot inversion model using the target training data set to determine the target hotspot inversion model;

[0121] The prediction module 504 is used to predict the hotspot data of the target transformer using the target hotspot inversion model to obtain a prediction result, wherein the prediction result includes the target hotspot temperature value and the target hotspot position.

[0122] Optionally, the acquisition module 501 is also used to obtain preset three-phase load coefficient parameters, preset ambient temperature parameters and preset ambient wind speed parameters; the determination module 503 is also used to combine the preset three-phase load coefficient parameters, preset ambient temperature parameters and preset ambient wind speed parameters using an orthogonal design method to determine multiple groups of training parameters; the processing module 502 is also used to simulate multiple groups of training parameters using a finite element simulation method to obtain temperature field distribution data corresponding to each group of training parameters; the determination module 503 is also used to determine a first training data set based on multiple groups of training parameters and temperature field distribution data.

[0123] Optionally, the determination module 503 is also used to: determine three corresponding shell temperature measurement points of the target transformer based on the temperature field distribution data; control the target transformer to operate according to multiple sets of training parameters, and obtain temperature data corresponding to each set of training parameters at the shell temperature measurement points; determine the first training data set based on the temperature data and multiple sets of training parameters.

[0124] Optionally, the processing module 502 is further used to: perform normalization processing on the first training data set to obtain a second training data set; and perform data dimension reduction processing on the second training data set by using a correlation analysis method to obtain a target training data set.

[0125] Optionally, the acquisition module 501 is also used to obtain test parameters, wherein the test parameters include: test three-phase load factor parameters, test environment temperature parameters and test environment wind speed parameters; the determination module 503 is also used to: use the test parameters to control the target transformer to obtain the actual hotspot temperature value of the target transformer; input the temperature data corresponding to the test parameters into the target hotspot inversion model to obtain the inverted hotspot temperature value of the target transformer; based on the actual hotspot temperature value and the inverted hotspot temperature value, determine the temperature error of the target hotspot inversion model.

[0126] Optionally, the acquisition module 501 is also used to obtain test parameters, wherein the test parameters include: test three-phase load factor parameters, test environment temperature parameters and test environment wind speed parameters; the determination module 503 is also used to: use the test parameters to control the target transformer to obtain the actual hotspot position of the target transformer; input the temperature data corresponding to the test parameters into the target hotspot inversion model to obtain the inverted hotspot position of the target transformer; compare the actual hotspot position with the inverted hotspot position to determine the positioning accuracy of the target hotspot inversion model.

[0127] Optionally, the acquisition module 501 is also used to acquire current temperature data at the temperature measurement point of the shell; the prediction module 504 is also used to input the current temperature data into the target hotspot inversion model to obtain the target hotspot temperature value and the target hotspot position.

[0128] It should be noted that the above modules can be implemented by software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.

[0129] According to one embodiment of the present invention, there is also provided an electronic device, comprising: a memory storing an executable program; and a processor for running the program, wherein the above-mentioned transformer hotspot inversion method based on time convolutional neural network is executed when the program is running.

[0130] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0131] Step S1, obtaining a first training data set, wherein the first training data set is constructed based on an orthogonal design method and a finite element simulation method;

[0132] Step S2, performing data dimensionality reduction processing on the first training data set to obtain a target training data set;

[0133] Step S3, training the initial hotspot inversion model using the target training data set to determine the target hotspot inversion model;

[0134] Step S4, using the target hotspot inversion model to predict the hotspot data of the target transformer to obtain a prediction result, wherein the prediction result includes the target hotspot temperature value and the target hotspot position.

[0135] According to one embodiment of the present invention, a computer-readable storage medium is also provided, including: the computer-readable storage medium includes a stored executable program, wherein when the executable program is running, the device where the storage medium is located is controlled to execute the above-mentioned transformer hotspot inversion method based on time convolutional neural network.

[0136] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0137] Step S1, obtaining a first training data set, wherein the first training data set is constructed based on an orthogonal design method and a finite element simulation method;

[0138] Step S2, performing data dimensionality reduction processing on the first training data set to obtain a target training data set;

[0139] Step S3, training the initial hotspot inversion model using the target training data set to determine the target hotspot inversion model;

[0140] Step S4, using the target hotspot inversion model to predict the hotspot data of the target transformer to obtain a prediction result, wherein the prediction result includes the target hotspot temperature value and the target hotspot position.

[0141] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store computer programs.

[0142] According to one embodiment of the present invention, a computer program product is also provided, including: a computer program, which implements the above-mentioned transformer hotspot inversion method based on time convolutional neural network when executed by a processor.

[0143] Optionally, in this embodiment, the computer program product may be configured to store computer program instructions for performing the following steps:

[0144] Step S1, obtaining a first training data set, wherein the first training data set is constructed based on an orthogonal design method and a finite element simulation method;

[0145] Step S2, performing data dimensionality reduction processing on the first training data set to obtain a target training data set;

[0146] Step S3, training the initial hotspot inversion model using the target training data set to determine the target hotspot inversion model;

[0147] Step S4, using the target hotspot inversion model to predict the hotspot data of the target transformer to obtain a prediction result, wherein the prediction result includes the target hotspot temperature value and the target hotspot position.

[0148] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0149] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0150] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0151] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0152] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0153] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0154] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0155] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A transformer hotspot inversion method based on temporal convolutional neural network, characterized in that: include: Acquire a first training data set, wherein the first training data set is constructed based on an orthogonal design method and a finite element simulation method; Performing data dimensionality reduction processing on the first training data set to obtain a target training data set; Using the target training data set to train an initial hotspot inversion model, and determining the target hotspot inversion model; The target hotspot inversion model is used to predict the hotspot data of the target transformer to obtain a prediction result, wherein the prediction result includes a target hotspot temperature value and a target hotspot position.

2. The transformer hotspot inversion method based on time convolutional neural network according to claim 1 is characterized in that: Acquiring the first training data set includes: Obtaining preset three-phase load factor parameters, preset ambient temperature parameters, and preset ambient wind speed parameters; Combining the preset three-phase load factor parameter, the preset ambient temperature parameter and the preset ambient wind speed parameter by using the orthogonal design method to determine multiple sets of training parameters; The finite element simulation method is used to simulate the multiple sets of training parameters to obtain temperature field distribution data corresponding to each set of training parameters; The first training data set is determined based on the multiple sets of training parameters and the temperature field distribution data.

3. The transformer hotspot inversion method based on time convolutional neural network according to claim 2 is characterized in that: Determining the first training data set based on the multiple sets of training parameters and the temperature field distribution data includes: Determine three corresponding shell temperature measurement points of the target transformer based on the temperature field distribution data; Controlling the target transformer to operate according to the multiple sets of training parameters, and obtaining temperature data corresponding to each set of training parameters at the shell temperature measurement point; The first training data set is determined based on the temperature data and the plurality of sets of training parameters.

4. The transformer hotspot inversion method based on time convolutional neural network according to claim 1 is characterized in that: Performing data dimension reduction processing on the first training data set to obtain a target training data set includes: Normalizing the first training data set to obtain a second training data set; The second training data set is processed by data dimension reduction using a correlation analysis method to obtain a target training data set.

5. The transformer hot spot inversion method based on time convolutional neural network according to claim 1 is characterized in that: The method further comprises: Acquire test parameters, wherein the test parameters include: a test three-phase load factor parameter, a test environment temperature parameter, and a test environment wind speed parameter; Controlling the target transformer using the test parameters to obtain an actual hot spot temperature value of the target transformer; Inputting the temperature data corresponding to the test parameters into the target hotspot inversion model to obtain the inverted hotspot temperature value of the target transformer; Based on the actual hotspot temperature value and the inverted hotspot temperature value, a temperature error of the target hotspot inversion model is determined.

6. The transformer hot spot inversion method based on time convolutional neural network according to claim 1 is characterized in that: The method further comprises: Acquire test parameters, wherein the test parameters include: a test three-phase load factor parameter, a test environment temperature parameter, and a test environment wind speed parameter; Controlling the target transformer using the test parameters to obtain an actual hot spot position of the target transformer; Inputting the temperature data corresponding to the test parameters into the target hotspot inversion model to obtain the inversion hotspot position of the target transformer; The actual hotspot position is compared with the inverted hotspot position to determine the positioning accuracy of the target hotspot inversion model.

7. The transformer hot spot inversion method based on time convolutional neural network according to claim 3 is characterized in that: The target hotspot inversion model is used to predict the hotspot data of the target transformer, and the prediction results include: Acquire current temperature data at the temperature measuring point of the housing; The current temperature data is input into the target hotspot inversion model for prediction to obtain the target hotspot temperature value and the target hotspot position.

8. An electronic device, characterized in that: include: A memory storing an executable program; A processor is used to run the program, wherein when the program is run, the transformer hotspot inversion method based on a time convolutional neural network described in any one of claims 1 to 7 is executed.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored executable program, wherein when the executable program is running, the device where the storage medium is located is controlled to execute the transformer hotspot inversion method based on time convolutional neural network as described in any one of claims 1 to 7.

10. A computer program product, characterized in that It comprises a computer program, which, when executed by a processor, implements the transformer hot spot inversion method based on a time convolutional neural network according to any one of claims 1 to 7.

Images