A method, device, equipment and medium for monitoring temperature of a converter transformer winding

Abstract

The application discloses a kind of converter transformer winding temperature monitoring method, device, equipment and medium, it is related to power transformer technical field, comprising the following steps: establishing the two-dimensional equivalent electromagnetic model of converter transformer, and electromagnetic simulation is carried out, obtain winding loss;Winding loss is as the heat source of two-dimensional flow-temperature field of two-dimensional equivalent flow-temperature model to two-dimensional equivalent flow-temperature model carries out fluid-temperature field simulation;With inlet oil temperature, inlet flow and ambient temperature as input variable, net side hot spot temperature, net side average temperature, valve side hot spot temperature and valve side average temperature as output value, construct linear regression model;Fine tuning model is constructed, linear regression model is combined with fine tuning model, obtain hybrid model.The application is with multiple independent variables as input, with winding hot spot and average temperature as output to establish the basic model based on linear regression and the fine tuning model with the error of winding hot spot and average temperature as output, and the two are combined to obtain hybrid model, the output result of the hybrid model can be with higher precision approximate simulation result, greatly improve the precision of winding temperature calculation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power transformers, in particular to a converter transformer winding temperature monitoring method, device, equipment and medium. BACKGROUND

[0002] As the core equipment of the DC power transmission system, the operation reliability of the converter transformer is related to its service life, and is more related to the safety and stability of the power system. The winding temperature is an important factor restricting the safe operation of the transformer and affecting the service life of the transformer, so it is of great significance to accurately and quickly obtain the winding temperature of the converter transformer for the design and safe operation of the converter transformer.

[0003] The existing winding temperature detection methods include indirect calculation method and thermal simulation measurement method, which calculate the temperature through a regression model. Although the winding hot spot temperature of the transformer can be approximately calculated, and the real heat conduction process can be basically reflected, since these methods are only for the winding hot spot temperature and the regression model involves fewer independent variables, all important factors affecting the winding temperature distribution of the transformer are not covered, which easily causes insufficient calculation result precision. SUMMARY

[0004] The present application provides a converter transformer winding temperature monitoring method, device, equipment and medium, which solves the existing problems.

[0005] The present application provides a converter transformer winding temperature monitoring method, which comprises the following steps:

[0006] A two-dimensional equivalent electromagnetic model of the converter transformer is established, electromagnetic simulation is performed on the two-dimensional equivalent electromagnetic model, and the winding loss under the rated operating condition is obtained;

[0007] A two-dimensional equivalent flow-thermal model of the converter transformer is established, the winding loss is taken as a heat source of the fluid-temperature field of the two-dimensional equivalent flow-thermal model, fluid-temperature field simulation is performed on the two-dimensional equivalent flow-thermal model, three parameters of the inlet oil temperature, the inlet flow rate and the ambient temperature are changed respectively, and the grid side hot spot temperature, the grid side average temperature, the valve side hot spot temperature and the valve side average temperature under different conditions are obtained;

[0008] A linear regression model is constructed with the inlet oil temperature, the inlet flow rate and the ambient temperature as input variables and the grid side hot spot temperature, the grid side average temperature, the valve side hot spot temperature and the valve side average temperature as output values;

[0009] The grid side hot spot temperature error, the grid side average temperature error, the valve side hot spot temperature error and the valve side average temperature error are calculated, a fine tuning model is constructed with the inlet oil temperature, the inlet flow rate and the ambient temperature as input and the grid side hot spot temperature error, the grid side average temperature error, the valve side hot spot temperature error and the valve side average temperature error as output values, the linear regression model and the fine tuning model are combined, and a hybrid model is obtained.

[0010] The inlet oil temperature, the inlet flow rate and the ambient temperature are input to the mixing model to obtain the fine-tuned mesh side hot spot temperature, the mesh side average temperature, the valve side hot spot temperature and the valve side average temperature.

[0011] Preferably, a two-dimensional equivalent electromagnetic model of the converter transformer is established, and electromagnetic simulation is performed on the two-dimensional equivalent electromagnetic model to obtain the winding loss under the rated operating condition, including the following steps:

[0012] A simplified two-dimensional equivalent electromagnetic model is constructed by using the electromagnetic module of the COMSOL software.

[0013] The electromagnetic property parameters of each component in the two-dimensional equivalent electromagnetic model and the operating parameters of the converter transformer are set according to the type of the converter transformer under study.

[0014] The grid division of each component in the two-dimensional equivalent electromagnetic model is refined.

[0015] The two-dimensional equivalent electromagnetic model after the grid division is subjected to steady-state electromagnetic simulation to calculate the DC resistance loss and the eddy current loss of the winding of the converter transformer.

[0016] The DC resistance loss and the eddy current loss of the winding are added to obtain the winding loss under the rated operating condition.

[0017] Preferably, the winding includes a mesh side winding and a valve side winding.

[0018] The DC resistance loss of the winding is calculated by the following formula:

[0019] P dc =I 2 R dc

[0020] In the formula, P dc is the DC resistance loss of the winding, I is the effective value of the winding current, R dc is the DC resistance of the coil.

[0021] The eddy current loss of the winding is calculated by the following formula:

[0022]

[0023] In the formula, P eddy is the axial or radial eddy current loss on the winding, d is the average diameter of the winding, a is the width of the winding, b is the height of the winding, ω is the current frequency, B av is the average magnetic flux density on the winding, and σ is the electrical conductivity.

[0024] Preferably, a two-dimensional equivalent flow-thermal model of the converter transformer is established, and the winding loss is taken as a heat source of a fluid-temperature field of the two-dimensional equivalent flow-thermal model to perform fluid-temperature field simulation on the two-dimensional equivalent flow-thermal model, including the following steps:

[0025] A simplified two-dimensional equivalent flow-thermal model is constructed by using a flow-thermal module of COMSOL software.

[0026] According to the type of the converter transformer to be studied, material thermal property parameters of each component of the two-dimensional equivalent flow-thermal model are set.

[0027] Each component of the two-dimensional equivalent flow-thermal model is subjected to grid division in a partition refinement manner.

[0028] The winding loss is loaded as a heat source into the fluid-temperature field, and a boundary condition is set to approximately simulate an oil circulation process in which the transformer oil floats upward due to a decrease in the heating density in actual operation.

[0029] The flow-thermal equation is solved in a steady state to obtain the temperature distribution inside the converter transformer and the hot spot temperature, the average temperature on the grid side, the hot spot temperature on the valve side, and the average temperature on the valve side.

[0030] Preferably, the linear regression model is as follows:

[0031]

[0032] In the formula, Y i is an output, Y1, Y2, Y3, and Y4 are the hot spot temperature on the grid side, the average temperature on the grid side, the hot spot temperature on the valve side, and the average temperature on the valve side, X j is an input, X1, X2, and X3 are the inlet oil temperature, the inlet flow rate, and the ambient temperature, k ij and b j are model parameters.

[0033] Preferably, the hot spot temperature error on the grid side, the average temperature error on the grid side, the hot spot temperature error on the valve side, and the average temperature error on the valve side are calculated, including the following steps:

[0034] The simulation values of the hot spot temperature on the grid side, the average temperature on the grid side, the hot spot temperature on the valve side, and the average temperature on the valve side of the two-dimensional equivalent flow-thermal model are obtained.

[0035] The errors obtained by subtracting the simulation values from the predicted values of the hot spot temperature on the grid side, the average temperature on the grid side, the hot spot temperature on the valve side, and the average temperature on the valve side output by the linear regression model are calculated respectively.

[0036] Preferably, the linear regression model is combined with the fine-tuning model, and the combination method includes the following steps:

[0037] The same group of input variables is input into the linear regression model and the fine-tuning model respectively, the outputs of the two models are added correspondingly, and the final output of the hybrid model is obtained.

[0038] A converter transformer winding temperature monitoring device comprises:

[0039] A loss module is configured to establish a two-dimensional equivalent electromagnetic model of the converter transformer, perform electromagnetic simulation on the two-dimensional equivalent electromagnetic model, and obtain winding loss under a rated operating condition.

[0040] A prediction module is configured to establish a two-dimensional equivalent flow-thermal model of the converter transformer, perform flow-temperature field simulation on the two-dimensional equivalent flow-thermal model by taking the winding loss as a heat source of a flow-temperature field of the two-dimensional equivalent flow-thermal model, and change three parameters of an inlet oil temperature, an inlet flow rate and an ambient temperature to obtain grid-side hotspot temperature, grid-side average temperature, valve-side hotspot temperature and valve-side average temperature under different conditions.

[0041] A construction module is configured to take the inlet oil temperature, the inlet flow rate and the ambient temperature as input variables, and take the grid-side hotspot temperature, the grid-side average temperature, the valve-side hotspot temperature and the valve-side average temperature as output values to construct a linear regression model.

[0042] A hybrid module is configured to calculate grid-side hotspot temperature error, grid-side average temperature error, valve-side hotspot temperature error and valve-side average temperature error, and construct a fine-tuning model by taking the inlet oil temperature, the inlet flow rate and the ambient temperature as input and taking the grid-side hotspot temperature error, the grid-side average temperature error, the valve-side hotspot temperature error and the valve-side average temperature error as output values, combine the linear regression model and the fine-tuning model, and obtain a hybrid model.

[0043] A fine-tuning module is configured to input the inlet oil temperature, the inlet flow rate and the ambient temperature into the hybrid model to obtain fine-tuned grid-side hotspot temperature, grid-side average temperature, valve-side hotspot temperature and valve-side average temperature.

[0044] A computer device comprises a memory, a processor and a computer program stored on the memory and executable on the processor, and the processor implements the above-mentioned converter transformer winding temperature monitoring method when executing the program.

[0045] A computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the above-mentioned converter transformer winding temperature monitoring method.

[0046] Compared with the prior art, the present application has the following beneficial effects:

[0047] The application establishes a magnetic field model of the converter transformer and simulates, obtains winding loss under a rated operating condition, then constructs a two-dimensional equivalent flow-heat model, then takes inlet oil temperature, inlet flow and ambient temperature as independent variables, inputs to the two-dimensional equivalent flow-heat model with the winding loss as a heat source, obtains winding hot spot temperature, average temperature under different conditions and analyzes. With multiple independent variables as input, winding hot spot temperature and average temperature as output, a basic model based on linear regression and a fine-tuning model with the error of winding hot spot temperature and average temperature as output are established, and the two are combined, and finally a hybrid model with inlet oil temperature, inlet flow and ambient temperature as input and corresponding grid side hot spot temperature, grid side average temperature, valve side hot spot temperature and valve side average temperature as output is obtained. The output result of the hybrid model can approximate the simulation result with high accuracy, greatly improving the accuracy of temperature calculation. BRIEF DESCRIPTION OF DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0049] Figure 1 The flow chart of the winding temperature monitoring method of the converter transformer of the present application;

[0050] Figure 2 The schematic diagram of the two-dimensional equivalent electromagnetic model of the converter transformer of the present application;

[0051] Figure 3 The grid division effect diagram of the two-dimensional equivalent electromagnetic model of the converter transformer of the present application in the electromagnetic field;

[0052] Figure 4 The magnetic flux density distribution diagram of the converter transformer of the present application;

[0053] Figure 5 The schematic diagram of the two-dimensional equivalent flow-heat model of the converter transformer of the present application;

[0054] Figure 6 The grid division effect diagram of the two-dimensional equivalent flow-heat model of the converter transformer of the present application in the fluid-temperature field;

[0055] Figure 7 The temperature distribution cloud chart of the converter transformer of the present application;

[0056] Figure 8 The upper flow velocity distribution cloud chart of the converter transformer of the present application;

[0057] Figure 9A lower flow velocity distribution cloud chart of the converter transformer of the application;

[0058] Figure 10 A fine-tuning model structure diagram of the application;

[0059] Figure 11 A network-side winding hot spot temperature prediction value of the hybrid model when the ambient temperature is 20 DEG C of the application;

[0060] Figure 12 A network-side winding average temperature prediction value of the hybrid model when the ambient temperature is 20 DEG C of the application;

[0061] Figure 13 A valve-side winding hot spot temperature prediction value of the hybrid model when the inlet oil temperature is 40 DEG C of the application;

[0062] Figure 14 A valve-side winding hot spot temperature prediction value of the hybrid model when the inlet oil temperature is 40 DEG C of the application. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the application will be apparently and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative work fall within the protection scope of the application.

[0064] A converter transformer winding temperature monitoring method, in particular a winding temperature monitoring method based on inlet oil temperature, inlet flow rate and ambient temperature under rated operating conditions of a converter transformer, a flow chart as shown in Figure 1 The method comprises the following steps:

[0065] Step 1: Establish a two-dimensional equivalent electromagnetic model of the converter transformer, perform electromagnetic simulation on the two-dimensional equivalent electromagnetic model, and obtain the winding loss under the rated operating conditions.

[0066] The core components of the converter transformer are the winding and the core, the core is made of a plurality of silicon steel sheets stacked, and the winding is made of copper, which is divided into a voltage regulating winding, a network-side winding and a valve-side winding from inside to outside, and is concentrically wound on the core column. The transformer is filled with insulating oil inside, which plays a role of insulation and heat dissipation.

[0067] To improve the computational efficiency, the physical model is simplified as follows: components such as clamps, pads, paperboard, etc. that have little effect on the magnetic field analysis in electromagnetic simulation are ignored, and a two-dimensional magnetic field model is established in a line cake unit to maximize the reflection of the distribution of the magnetic flux leakage. To simulate the flow of transformer oil inside, the insulation cylinder and baffle are modeled in detail, but since the thermal conductivity of the insulation cylinder and baffle inside and outside the oil duct is small, the heat exchange between these components and the outside and oil flow is not considered, that is, the temperature boundary condition of this part is set as a thermal insulation boundary condition.

[0068] The electromagnetic module of the multi-physical field simulation software COMSOL is used to build a simplified two-dimensional equivalent electromagnetic model. Taking a single-phase double-winding converter transformer with the model number ZZDFPZ-234900 / 220-20(D) as an example, the size parameters of each component are extracted to establish an equivalent simplified model as shown in Figure 2 . The electromagnetic property parameters of each component and the operating parameters of the transformer are set according to the model of the converter transformer being studied. The set electromagnetic property parameters are shown in Table 1. Since the applied magnetic field will cause the ferromagnetic material to exhibit nonlinear magnetization behavior, the B-H curve of the core needs to be set.

[0069] Table 1: Electromagnetic parameters of each component of the transformer body

[0070]

[0071] To improve the computational efficiency without affecting the accuracy, a partitioned manual meshing method is used for meshing, and the net side and valve side winding parts, which are the main heat sources, are divided in a relatively fine manner, while the transformer oil and oil tank parts are divided in a relatively coarse manner. Finally, the mesh quality is checked. The meshing results of the two-dimensional equivalent electromagnetic model of the converter transformer are shown in Figure 3 .

[0072] The steady-state solver is selected, and the rated current is applied on the net side and valve side winding, respectively. The current directions of the net side and valve side winding are opposite. The distribution of the magnetic flux density modulus of the transformer under rated load is shown in Figure 4 . The DC resistance loss P dc and eddy current loss P eddy of the net side and valve side winding are calculated according to the formula.

[0073] P dc = I 2 R dc

[0074] where P dc is the DC resistance loss of the winding, unit W; I is the effective value of the winding current, unit A; Rdc DC resistance of coil, unit Ω .

[0075]

[0076] where P eddy is the axial or radial eddy current loss on the winding, unit W; d is the average diameter of the winding, unit m; a is the winding width, unit m; b is the winding height, unit m; ω is the current frequency, unit rad / s; σ is the electrical conductivity, unit S / m; B av is the average magnetic flux density on the winding, unit T.

[0077] The total loss of the winding is the sum of the total eddy current loss and the resistance loss, which is 632.8 kW in this embodiment.

[0078] Second step: Establishing the two-dimensional equivalent flow-thermal model of the converter transformer, taking the winding loss as the heat source of the fluid-temperature field of the two-dimensional flow-thermal model, simulating the fluid-temperature field of the two-dimensional equivalent flow-thermal model, changing the three parameters of the inlet oil temperature, the inlet flow rate and the ambient temperature respectively, and obtaining the winding hot spot and average temperature data samples under different conditions.

[0079] Referring to Figure 5 , a simplified two-dimensional equivalent flow-thermal model is constructed using the flow-thermal module of the multi-physics simulation software COMSOL . The transformer oil is the main internal cooling medium of the oil-immersed transformer, and the setting of its thermal characteristics is crucial to the internal temperature distribution of the transformer. According to the type of the converter transformer studied, the thermal property parameters of each component material are set, as shown in Table 2.

[0080] Table 2 Thermal properties of materials of each component of the transformer

[0081]

[0082] For the fluid-temperature field, the solution region of the transformer shell and the oil and other fluids with slow changes in temperature and flow rate is coarsely meshed. The regions with significant changes in temperature, such as the grid side and valve side windings and the solid-liquid interface, are finely meshed, and the mesh quality is checked. The meshing results of the two-dimensional equivalent flow-thermal model of the converter transformer are shown in Figure 6 .

[0083] The winding losses of the grid side and valve side are loaded as heat sources into the fluid-temperature field, and the ambient temperature and the initial temperature of the transformer are both set to 27℃. The transformer inlet adopts a velocity inlet boundary condition, the inlet oil temperature is 50℃, and the inlet flow rate is 120m 3h. The outlet adopts pressure outlet boundary condition, the shell wall surface is set as no-slip wall surface condition, and the gravitational acceleration is set as 9.81 m / s 2 , the velocity is negative along the z axis, and the transformer oil realizes oil circulation under the joint influence of its own physical properties and gravitational acceleration. Finally, the temperature distribution cloud map and the flow velocity cloud map of the converter transformer are shown in Figures 7-9 .

[0084] Under the rated operating condition of the transformer, i.e., the winding loss is 632.8 kW, different inlet oil temperatures, different inlet flow rates, different ambient temperatures, etc. are set, and the temperature distribution and temperature rise data samples under different inlet oil temperatures, inlet flow rates and ambient temperatures are obtained. The temperature rise data samples include the hot spot temperature on the grid side, the average temperature on the grid side, the hot spot temperature on the valve side and the average temperature on the valve side.

[0085] Third step: a linear regression model is constructed with the inlet oil temperature, the inlet flow rate and the ambient temperature as input variables, and the hot spot temperature on the grid side, the average temperature on the grid side, the hot spot temperature on the valve side and the average temperature on the valve side as output variables.

[0086] Based on the simulation data samples, the inlet oil temperature, the inlet flow rate and the ambient temperature set in the simulation are taken as three-dimensional input variables, denoted as X1, X2 and X3 respectively, and the hot spot temperature on the grid side, the average temperature on the grid side, the hot spot temperature on the valve side and the average temperature on the valve side obtained by simulation calculation are taken as four-dimensional output values, denoted as Y1, Y2, Y3 and Y4 respectively. A “three-input, four-output” linear regression model corresponding to the same input and output meanings is established, and then the least square method is used to fit the samples to determine all the model parameters of the established “three-input, four-output” linear regression model. This linear regression model will serve as a basic model, and its model parameters will remain unchanged in the following steps. The specific formula of the linear regression basic model is:

[0087]

[0088] The specific model parameters k ij of the linear regression basic model (retaining 4 decimal places) are shown in Table 3, and the specific model parameters b j of the linear regression basic model (retaining 4 decimal places) are shown in Table 4.

[0089] Table 3 Model parameters k of the linear regression basic model ij

[0090]

[0091] Table 4 Model parameters b of the linear regression basic model j

[0092]

[0093] Fourth step: build the fine-tuning model, combine the linear regression model with the fine-tuning model to get the hybrid model.

[0094] The errors of the four temperature prediction values output by the base model are calculated respectively, and then the quicksort algorithm is used to find the maximum values of the absolute values of the four temperature errors, denoted as M1, M2, M3, M4, and the specific values (retaining 4 decimal places) are shown in Table 5.

[0095] Table 5 Maximum values of the absolute values of the four temperature errors

[0096]

[0097]

[0098] A "three-input, four-output" fine-tuning model based on neural network is established, with inlet oil temperature, inlet flow and ambient temperature as inputs, and net-side hot spot temperature error, net-side average temperature error, valve-side hot spot temperature error, valve-side average temperature error as outputs. 2M1, 2M2, 2M3, 2M4 are used as the amplitude limiting standards of the four outputs of the fine-tuning model respectively. Then, based on the open source pytorch code framework, the simulation set inlet oil temperature, inlet flow and ambient temperature in the sample are used as sample input, and the errors of the four temperatures obtained are used as sample labels to train the fine-tuning model. Finally, the parameters of the fine-tuning model after training are fixed, and combined with the base model to get the hybrid model. The combination method is to input the same group of input values into the base model and the fine-tuning model respectively, and add the outputs of the two models correspondingly to get the final output of the hybrid model. The model structure of the fine-tuning model is shown in Figure 10

[0099] Table 6 Training parameters of the fine-tuning model

[0100]

[0101] Fifth step: input the inlet oil temperature, inlet flow and ambient temperature into the hybrid model to get the winding temperature.

[0102] When the ambient temperature is 20℃, the net-side hot spot temperature and the net-side average temperature predicted by the hybrid model under different inlet oil temperatures and different inlet flows are shown in Figure 11 and Figure 12 respectively, where the circular discrete points are simulation samples and the grid surface is the hybrid model prediction value.

[0103] When the inlet oil temperature is 40℃, the valve-side hot spot temperature and the valve-side average temperature predicted by the hybrid model under different inlet flows and ambient temperatures are shown in Figure 13 and Figure 14 respectively, where the circular discrete points are simulation samples and the grid surface is the hybrid model prediction value. ​

[0104] Based on the same concept, the application also provides a converter transformer winding temperature monitoring device, comprising a loss module, a prediction module, a construction module, a hybrid module and a fine-tuning module.

[0105] The loss module is used to establish a two-dimensional equivalent electromagnetic model of the converter transformer, perform electromagnetic simulation on the two-dimensional equivalent electromagnetic model, and obtain winding loss under a rated operating condition.

[0106] The prediction module is used to establish a two-dimensional equivalent flow-thermal model of the converter transformer, perform flow-temperature field simulation on the two-dimensional equivalent flow-thermal model by taking the winding loss as a heat source of the flow-temperature field of the two-dimensional equivalent flow-thermal model, change three parameters of inlet oil temperature, inlet flow rate and ambient temperature respectively, and obtain grid-side hot spot temperature, grid-side average temperature, valve-side hot spot temperature and valve-side average temperature under different conditions.

[0107] The construction module is used to construct a linear regression model by taking the inlet oil temperature, the inlet flow rate and the ambient temperature as input variables and taking the grid-side hot spot temperature, the grid-side average temperature, the valve-side hot spot temperature and the valve-side average temperature as output values.

[0108] The hybrid module is used to calculate grid-side hot spot temperature error, grid-side average temperature error, valve-side hot spot temperature error and valve-side average temperature error, and construct a fine-tuning model by taking the inlet oil temperature, the inlet flow rate and the ambient temperature as input and taking the grid-side hot spot temperature error, the grid-side average temperature error, the valve-side hot spot temperature error and the valve-side average temperature error as output values, combine the linear regression model and the fine-tuning model, and obtain a hybrid model.

[0109] The fine-tuning module is used to input the inlet oil temperature, the inlet flow rate and the ambient temperature into the hybrid model, and obtain fine-tuned grid-side hot spot temperature, grid-side average temperature, valve-side hot spot temperature and valve-side average temperature.

[0110] The application also provides a computer device, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, and the processor implements the converter transformer winding temperature monitoring method when executing the program.

[0111] The application also provides a computer readable storage medium, and the storage medium stores a computer program, and the computer program is executed by the processor to implement the converter transformer winding temperature monitoring method.

[0112] Although the preferred embodiments of the application have been described, those skilled in the art can make further changes and modifications to the embodiments once they know the basic inventive concept. Therefore, the appended claims are intended to be interpreted as including all changes and modifications falling within the scope of the application.

[0113] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A method of monitoring the temperature of a converter transformer winding, characterized by, The method comprises the following steps: establishing a two-dimensional equivalent electromagnetic model of the converter transformer, performing electromagnetic simulation on the two-dimensional equivalent electromagnetic model, and obtaining winding loss under a rated operating condition; establishing a two-dimensional equivalent flow-thermal model of the converter transformer, taking the winding loss as a heat source of a fluid-temperature field of the two-dimensional equivalent flow-thermal model, performing fluid-temperature field simulation on the two-dimensional equivalent flow-thermal model, changing three parameters of an inlet oil temperature, an inlet flow rate and an ambient temperature respectively, and obtaining a net-side hot spot temperature, a net-side average temperature, a valve-side hot spot temperature and a valve-side average temperature under different conditions; constructing a linear regression model with the inlet oil temperature, the inlet flow rate and the ambient temperature as input variables and the net-side hot spot temperature, the net-side average temperature, the valve-side hot spot temperature and the valve-side average temperature as output values; calculating a net-side hot spot temperature error, a net-side average temperature error, a valve-side hot spot temperature error and a valve-side average temperature error, and constructing a fine tuning model with the inlet oil temperature, the inlet flow rate and the ambient temperature as input and the net-side hot spot temperature error, the net-side average temperature error, the valve-side hot spot temperature error and the valve-side average temperature error as output, combining the linear regression model and the fine tuning model, and obtaining a hybrid model; inputting the inlet oil temperature, the inlet flow rate and the ambient temperature into the hybrid model to obtain fine-tuned net-side hot spot temperature, net-side average temperature, valve-side hot spot temperature and valve-side average temperature; calculating the net-side hot spot temperature error, the net-side average temperature error, the valve-side hot spot temperature error and the valve-side average temperature error, comprising the following steps: obtaining simulation values of the net-side hot spot temperature, the net-side average temperature, the valve-side hot spot temperature and the valve-side average temperature of the two-dimensional equivalent flow-thermal model; respectively calculating errors obtained by subtracting the simulation values from the predicted values of the net-side hot spot temperature, the net-side average temperature, the valve-side hot spot temperature and the valve-side average temperature output by the linear regression model.

2. A method of monitoring the temperature of a converter transformer winding as claimed in claim 1, characterized in that, establishing a two-dimensional equivalent electromagnetic model of the converter transformer, performing electromagnetic simulation on the two-dimensional equivalent electromagnetic model, and obtaining winding loss under a rated operating condition, comprising the following steps: building a simplified two-dimensional equivalent electromagnetic model through an electromagnetic module of COMSOL software; setting electromagnetic property parameters of materials of each component in the two-dimensional equivalent electromagnetic model and operating parameters of the converter transformer according to a type of the converter transformer studied; performing grid division on each component of the two-dimensional equivalent electromagnetic model in a partitioned and refined manner; performing steady-state electromagnetic simulation on the two-dimensional equivalent electromagnetic model after the grid division, and calculating direct current resistance loss and eddy current loss of the winding of the converter transformer; adding the direct current resistance loss and the eddy current loss of the winding to obtain the winding loss under the rated operating condition.

3. A method of monitoring the temperature of a converter transformer winding as claimed in claim 2, characterised in that, The winding comprises a net-side winding and a valve-side winding; The direct current resistance loss of the winding is calculated by the following formula: wherein is the winding DC resistance loss, is the winding current effective value, is the coil DC resistance; The eddy current loss of the winding is calculated by the following formula: wherein is the axial or radial eddy current loss in the winding, is the average diameter of the winding, is the width of the winding, is the height of the winding, is the current frequency, is the average magnetic flux density in the winding, is the electrical conductivity.

4. A method of monitoring the temperature of a converter transformer winding as claimed in claim 1, characterized in that, establishing a two-dimensional equivalent flow-thermal model of the converter transformer, taking the winding loss as a heat source of a fluid-temperature field of the two-dimensional flow-thermal model, and performing fluid-temperature field simulation on the two-dimensional equivalent flow-thermal model, comprising the following steps: building a simplified two-dimensional equivalent flow-thermal model through a flow-thermal module of COMSOL software; setting thermal property parameters of materials of each component in the two-dimensional equivalent flow-thermal model; performing grid division on each component of the two-dimensional equivalent flow-thermal model in a partitioned and refined manner; The winding loss is loaded as a heat source into the fluid-temperature field, boundary conditions are set, and the oil circulation process in which the transformer oil floats upward due to the decrease in the heated density in actual operation is simulated approximately; The flow-heat equation is solved in the steady state to obtain the temperature distribution inside the converter transformer and the hot-spot temperature, the average temperature at the grid side, the hot-spot temperature at the valve side, and the average temperature at the valve side.

5. A method of monitoring the temperature of a converter transformer winding as claimed in claim 1, characterized in that, The linear regression model is as follows: wherein, is an output, , , , is a net-side hot-spot temperature, a net-side average temperature, a valve-side hot-spot temperature, and a valve-side average temperature, is an input, , , is an inlet oil temperature, an inlet flow rate, and an ambient temperature, and are model parameters.

6. A method of monitoring the temperature of a converter transformer winding as claimed in claim 1, characterized in that, The linear regression model is combined with the fine-tuning model, and the combination method includes the following steps: The same set of input variables is input into the linear regression model and the fine-tuning model, and the outputs of the two models are added correspondingly to obtain the final output of the hybrid model.

7. A converter transformer winding temperature monitoring device, characterized by Comprise: The loss module is configured to establish a two-dimensional equivalent electromagnetic model of the converter transformer, perform electromagnetic simulation on the two-dimensional equivalent electromagnetic model, and obtain the winding loss under the rated operating condition; The prediction module is configured to establish a two-dimensional equivalent flow-heat model of the converter transformer, perform fluid-temperature field simulation on the two-dimensional equivalent flow-heat model by taking the winding loss as a heat source of the fluid-temperature field of the two-dimensional equivalent flow-heat model, and change the inlet oil temperature, the inlet flow rate, and the ambient temperature to obtain the hot-spot temperature at the grid side, the average temperature at the grid side, the hot-spot temperature at the valve side, and the average temperature at the valve side under different conditions; The construction module is configured to take the inlet oil temperature, the inlet flow rate, and the ambient temperature as input variables, and take the hot-spot temperature at the grid side, the average temperature at the grid side, the hot-spot temperature at the valve side, and the average temperature at the valve side as output values to construct a linear regression model; The hybrid module is configured to calculate the hot-spot temperature error at the grid side, the average temperature error at the grid side, the hot-spot temperature error at the valve side, and the average temperature error at the valve side, and take the inlet oil temperature, the inlet flow rate, and the ambient temperature as input, and take the hot-spot temperature error at the grid side, the average temperature error at the grid side, the hot-spot temperature error at the valve side, and the average temperature error at the valve side as output values to construct a fine-tuning model, combine the linear regression model with the fine-tuning model, and obtain a hybrid model; The fine-tuning module is configured to input the inlet oil temperature, the inlet flow rate, and the ambient temperature into the hybrid model to obtain the fine-tuned hot-spot temperature at the grid side, the average temperature at the grid side, the hot-spot temperature at the valve side, and the average temperature at the valve side, and calculate the hot-spot temperature error at the grid side, the average temperature error at the grid side, the hot-spot temperature error at the valve side, and the average temperature error at the valve side, including the following steps: Obtain the simulation values of the hot-spot temperature at the grid side, the average temperature at the grid side, the hot-spot temperature at the valve side, and the average temperature at the valve side of the two-dimensional equivalent flow-heat model; Calculate the errors obtained by subtracting the simulation values from the predicted values of the hot-spot temperature at the grid side, the average temperature at the grid side, the hot-spot temperature at the valve side, and the average temperature at the valve side output by the linear regression model, respectively.

8. A computer device, comprising: The computer program stored in the storage medium and executable on the processor, when the processor executes the program, implements the converter transformer winding temperature monitoring method of any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and the computer program is executed by the processor to implement the converter transformer winding temperature monitoring method of any one of claims 1-6.

Images