Converter transformer internal winding temperature monitoring method based on digital twinning

Abstract

The invention belongs to the field of converter transformer temperature monitoring, and relates to a converter transformer winding temperature monitoring method based on digital twinning, and the method comprises the steps: building a converter transformer electromagnetic thermal coupling two-dimensional simulation model, and inputting actual operation current data as an excitation source into the two-dimensional simulation model for simulation calculation; performing electromagnetic simulation according to ampere-turn balance to obtain a winding temperature rise distribution cloud picture under different working conditions; inputting the data under the operation condition into a machine learning model for training to obtain a rapid calculation model based on a radial basis function network; a real-time current value of each winding during operation is collected by a sensor, the real-time current values are imported into a rapid calculation model, second-level real-time display of winding temperature distribution is realized, and the load is fed back and adjusted according to a set alarm threshold value. According to the method, a multi-working-condition electromagnetic thermal simulation model is established, temperature field simulation data are extracted and input into a machine learning model for data analysis, and the converter transformer winding temperature field digital twinborn body is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of converter transformer temperature monitoring, and specifically relates to a converter transformer winding temperature monitoring method based on digital twins, which is particularly suitable for ultra-high voltage direct current transmission projects and is also applicable to conventional transformers. Background Art

[0002] In high-voltage direct current (HVDC) transmission grid projects, converter transformers are crucial for the grid's normal operation, playing a vital role in the AC-DC conversion of power systems. Ultra-high voltage (UHV) converter transformers, with their complex structures, high grid-side voltage levels, large capacities, and heavy operating loads, have exposed issues such as high summer operating temperatures and a lack of current margin. This has led to a high risk of equipment explosions, posing significant challenges to the operation and maintenance of UHV converter transformers.

[0003] Due to the complex insulation structure of converter transformers, fiber-optic temperature measurement devices cannot be installed within the windings. Furthermore, online multi-physics field simulation capabilities are limited, resulting in inadequate means for verifying the insulation, temperature rise, and mechanical performance of in-service equipment. Furthermore, the use of general-purpose fluid computation and analysis software for winding temperature field simulation analysis results in a large mesh size due to the small size of the winding oil passages (3mm-5mm). The solution time ranges from several hours to several days, and each change in boundary conditions requires a solution. Therefore, breakthroughs in converter transformer hotspot prediction technology based on temperature field simulation analysis are urgently needed to effectively improve converter transformer winding temperature monitoring capabilities and ensure the export of large quantities of clean energy. Summary of the Invention

[0004] In order to solve the above technical problems, the present invention provides a method for monitoring the winding temperature of a converter transformer based on digital twins. By inputting the ambient temperature, the number of coolers opened, and the current, the digital twin technology is used to synchronize the virtual model of the converter transformer and the real-time sensor data, thereby realizing real-time monitoring of the winding temperature changes and displaying them with a cloud map effect. When the hot spot temperature is too high, the cooler fails, or the load is too large, the operation and maintenance personnel are promptly reminded to adjust the number of coolers in operation and the operating load to a safe operating state to prevent the hot spot temperature from being too high and the insulation life from being lost too quickly, thereby maximizing the safe operation of the converter transformer and avoiding false alarms and malfunctions, thereby effectively ensuring the power supply. The technical solution adopted by the present invention to achieve the above purpose is: A digital twin-based converter transformer winding temperature monitoring method implements real-time converter transformer winding temperature monitoring based on on-site multi-operating load conditions and a multi-physics field fast calculation model based on machine learning. The method specifically includes the following steps: S1. Analyze the historical operating data of the converter transformer, extract the current data of the historical operation of the converter transformer under multiple operating conditions, establish a two-dimensional simulation model of electromagnetic thermal coupling of the converter transformer, and input the actual operating current data as an excitation source into the two-dimensional simulation model for simulation calculation, which can be performed using TranCalc professional software; S2. Perform electromagnetic simulation based on ampere-turn balance to calculate the current distribution of each winding of the converter transformer under different operating conditions, calculate the loss of each winding and each coil, and the loss distribution of each winding and each coil, and then perform temperature field analysis to obtain the temperature rise distribution cloud map of the converter transformer winding under different operating conditions; S3. Input the data under the operating conditions of the converter transformer into the machine learning model for training to obtain a fast calculation model based on the fast calculation method of the radial basis function network.

[0005] S4. Use sensors to collect the real-time current values ​​of each winding during the operation of the converter transformer, import the real-time current values ​​into the above-mentioned fast calculation model, realize the real-time display of the temperature distribution of the converter transformer winding in seconds, and provide timely feedback and adjust the load size according to the set alarm threshold.

[0006] Preferably, in step S3, the fast calculation method based on radial basis function comprises the following specific steps: S3.1. Determine the variables and operating range in the simulation calculation based on the calculation file. Here, the per-unit value of the current is taken, that is, the ratio of the operating current to the rated current. Based on the on-site operating conditions of the converter transformer, the operating range is determined to be 0.1 to 1.2 pu (pu represents the per-unit value of the rated current, and 1.2 pu represents 1.2 times the rated current). S3.2. Input the number of samples, use the Latin hypercube sampling method to generate a list of sample points; S3.3. Replace the sample points generated by the simulation software with the sample points of the actual operating conditions of the converter transformer; S3.4, read in the n source files obtained by simulation calculation in step S1; S3.5. Select the radial basis function algorithm, input the test sample ratio, and perform rapid calculation model training; S3.6. Check the error between the test sample and the calculation result of the fast calculation model; if the error meets the preset threshold, output the fast calculation model; if the error does not meet the preset threshold, adjust the data and recalculate.

[0007] Because the multi-physics coupling calculation process for converter transformers involves nonlinear physical equations and material parameters, the input and output parameters of the machine learning model exhibit a nonlinear mapping relationship. Furthermore, the output of the machine learning model is field data for all model nodes. Therefore, a fast calculation method based on a radial basis function network is employed for the model. The data in the sample library is divided into training and test sets in a ratio of 7:3 or 8:2. A fast calculation method based on a radial basis function network is used to calculate the error between the test data and the measured data. If the error meets a preset threshold, the output is packaged as a fast calculation model. The fast calculation model is essentially a reduced-order model. The meaning of a reduced-order model is that in normal simulations, the calculation time for a single operating condition is generally long. For example, in this case, even on a well-equipped workstation, the calculation time for a single operating condition is approximately 30 minutes. However, after processing previously calculated simulation data, the fast calculation model can display the corresponding cloud map for different current values ​​within 30 seconds.

[0008] Preferably, in step S3.1, the calculation file includes a simulation calculation file obtained by calculation using ANSYS Fluent software, and a file exported by Fluent software.

[0009] Preferably, in step S3.6, the adjustment data is to adjust the number of simulation calculation conditions and input boundary conditions.

[0010] Preferably, in step S3, the data under the operating condition of the converter transformer includes temperature distribution, oil flow velocity distribution, pressure distribution and density distribution.

[0011] Preferably, in step S1, the current data of the historical operation of the converter transformer under multiple operating conditions is extracted by monitoring and collecting the daily load operation curve of the actual operation of the converter transformer, and analyzing the daily load operation curve to obtain the current value of each winding.

[0012] Beneficial effects of the present invention: This method establishes a multi-operating condition electromagnetic thermal simulation model, extracts temperature field simulation data, inputs the data into a machine learning model for data analysis, and then generates a fast calculation model. The fast calculation model is used to quickly display the temperature field cloud map, thereby realizing the digital twin of the converter transformer temperature field. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1is a flow chart of a winding temperature monitoring method according to an embodiment of the present invention; Figure 2 2 is a schematic diagram of a two-dimensional simulation model of an electromagnetic thermocouple according to an embodiment of the present invention; Figure 3 is a schematic diagram of simulation calculation results of an embodiment of the present invention; Figure 4 is a schematic diagram of calculation results of a fast calculation model according to an embodiment of the present invention; Figure 5 It is a schematic diagram of the error between the simulation calculation results of the embodiment of the present invention and the calculation results of the fast calculation model. DETAILED DESCRIPTION

[0014] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0015] like Figure 1 As shown, the embodiment of the present invention is a method for monitoring the temperature of a converter transformer winding based on digital twins. The specific implementation is as follows: The historical operating data of the converter transformer is compiled to extract the operating current values ​​of each converter transformer winding under multiple on-site operating conditions. The daily load curves of the converter transformer during actual operation are monitored and collected. The current values ​​of each winding corresponding to the daily load curves are analyzed and extracted. The winding current values ​​are recorded and used as the excitation source input for the electromagnetic-thermal coupling simulation. The data set of daily load curves and current values ​​should be as extensive as possible, preferably more than 100, and the data should vary widely, covering as many load conditions as possible during actual operation of the converter transformer to provide a sufficient data analysis foundation for subsequent machine learning steps.

[0016] Establish a two-dimensional simulation model of electromagnetic thermal coupling of converter transformer. Figure 2 As shown in the figure, taking the ±800kV converter transformer as an example, the CAD drawing software is used to draw the two-dimensional simulation model of the electromagnetic thermal coupling of the converter transformer, and the axisymmetric model is used for modeling.

[0017] Ansys Fluent simulation software was used to perform electromagnetic and thermal field simulations to calculate the winding loss and loss distribution under different operating conditions. Here, the current values ​​were divided into operating conditions of 0.1, 0.2, 0.4, 0.5, 0.6, 0.8, 1.0, 1.05, 1.1, and 1.2 pu, and the winding loss and loss distribution of the converter transformer were calculated respectively.

[0018] The WHC module in TranCalc software is used to establish a calculation model for the electromagnetic-thermal coupling two-dimensional simulation model of the converter transformer. The actual operating current data is input as the excitation source to calculate the current distribution, magnetic circuit distribution, winding loss, and winding temperature rise.

[0019] The WHC module calculates winding losses and converts these data into per-coil volumetric heat flux. This processed data is then input into ANSYS Fluent software for temperature field analysis. This step simulates temperature rise based on steady-state conditions and involves solving the Fourier heat transfer equation for solid heat transfer and the convection heat dissipation equation for solid-to-fluid heat transfer.

[0020] The temperature value of each grid node in the temperature rise distribution results of the converter transformer winding is extracted, and combined with various operating conditions, the temperature values ​​and operating conditions and other data are input into the machine learning model. Data analysis and inversion calculation are performed based on the neural network algorithm, and a fast calculation model based on the radial basis function network fast calculation method suitable for any current excitation is generated.

[0021] like Figure 3-5 As shown, the error between the converter transformer temperature field simulation results and the fast calculation model calculation results is calculated. It can be seen that the error between the simulation calculation results and the fast calculation model calculation results meets the preset threshold requirement.

[0022] The substation monitors the real-time current of each winding of the converter transformer and imports a fast calculation model every 5 seconds to calculate the thermal distribution, achieving the second-level presentation of the thermal distribution of the converter transformer windings and realizing the digital twin of the converter transformer temperature field.

[0023] In the embodiments of the present invention, technical features not described in detail are all existing technologies or conventional technical means and will not be described in detail here.

[0024] Finally, it should be noted that the above embodiments are only specific implementation methods of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them, and the protection scope of the present invention is not limited thereto. Those skilled in the art should understand that any person skilled in the art can modify or easily think of changes to the technical solutions described in the above embodiments within the technical scope disclosed by the present invention, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A converter transformer winding temperature monitoring method based on digital twin, characterized in that: The following steps are involved: S1. Analyze the historical operating data of the converter transformer, extract the current data of the historical operation of the converter transformer under multiple operating conditions, establish a two-dimensional simulation model of electromagnetic thermal coupling of the converter transformer, and input the actual operating current data as an excitation source into the two-dimensional simulation model for simulation calculation; S2. Perform electromagnetic simulation based on ampere-turn balance to calculate the current distribution of each winding of the converter transformer under different operating conditions, calculate the loss of each winding and each coil, and the loss distribution of each winding and each coil, and then perform temperature field analysis to obtain the temperature rise distribution cloud map of the converter transformer winding under different operating conditions; S3. Inputting data under the operating conditions of the converter transformer into the machine learning model for training to obtain a fast calculation model based on a fast calculation method of a radial basis function network; S4. Use sensors to collect the real-time current values ​​of each winding during the operation of the converter transformer, import the real-time current values ​​into the above-mentioned fast calculation model, realize the real-time display of the temperature distribution of the converter transformer winding in seconds, and provide timely feedback and adjust the load size according to the set alarm threshold.

2. A converter transformer winding temperature monitoring method based on digital twin according to claim 1, characterized in that: In step S3, the fast calculation method based on radial basis function has the following specific steps: S3.

1. Determine the variables and operating range in the simulation calculation based on the calculation file. Here, the per-unit value of the current is taken, that is, the ratio of the operating current to the rated current. Based on the on-site operating conditions of the converter transformer, the operating range is determined to be 0.1~1.2pu, where pu represents the per-unit value of the rated current. S3.

2. Input the number of samples, use the Latin hypercube sampling method to generate a list of sample points; S3.

3. Replace the sample points generated by the simulation software with the sample points of the actual operating conditions of the converter transformer; S3.4, read in the n source files obtained by simulation calculation in step S1; S3.

5. Select the radial basis function algorithm, input the test sample ratio, and perform rapid calculation model training; S3.

6. Check the error between the test sample and the calculation result of the fast calculation model; if the error meets the preset threshold, output the fast calculation model; if the error does not meet the preset threshold, adjust the data and recalculate.

3. The method for monitoring converter transformer winding temperature based on digital twin according to claim 2, characterized in that: In step S3.1, the calculation file includes a simulation calculation file obtained by calculation using ANSYS Fluent software, and a file exported by Fluent software.

4. The method for monitoring converter transformer winding temperature based on digital twin according to claim 2, characterized in that: In step S3.6, the adjustment data is to adjust the number of simulation calculation conditions and input boundary conditions.

5. The method for monitoring converter transformer winding temperature based on digital twin according to claim 1, characterized in that: In step S3, the data under the operating condition of the converter transformer includes temperature distribution, oil flow velocity distribution, pressure distribution and density distribution.

6. The method for monitoring converter transformer winding temperature based on digital twin according to claim 1, characterized in that: In step S1, the current data of the historical operation of the converter transformer under multiple operating conditions is extracted by monitoring and collecting the daily load operation curve of the actual operation of the converter transformer, and analyzing the daily load operation curve to obtain the current value of each winding.

Images