A contact temperature prediction method and terminal of GIS disconnector

Abstract

The application discloses a GIS disconnector contact temperature prediction method and a terminal. First, electromagnetic field simulation is performed on the GIS disconnector to obtain losses generated by each component, the obtained losses are introduced into a temperature-fluid field as heat sources to perform temperature field calculation, in this way, multi-physical field simulation is established for the GIS disconnector in an indirect coupling mode, the simulation accuracy is improved, and a training set and a test set required for prediction can be obtained based on simulation results. A sparse chaotic polynomial prediction model is constructed, and an input variable subset of the sparse chaotic polynomial prediction model is determined, and the contact temperature of the GIS disconnector is predicted through the sparse chaotic polynomial prediction model and the input variable subset. In the GIS disconnector contact temperature prediction, the data samples are less, and the prediction through the polynomial chaotic theory can maintain high prediction accuracy, and because there is no pending setting parameter in the algorithm, the model does not need to be adjusted according to the data, and the applicability is high.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of GIS device temperature prediction, and particularly relates to a GIS disconnector contact temperature prediction method and a terminal. BACKGROUND

[0002] The gas insulated switchgear (GIS) disconnector has the advantages of simple operation and maintenance, long service life, high reliability, small area occupation, etc., and is widely used and has strong versatility. However, in actual production and field operation, there are still problems such as individual part manufacturing process defects, immature worker assembly technology, and harsh field environment. These problems are particularly prominent with increasing voltage level and operating time, and may cause many faults such as mechanical failure and thermal failure.

[0003] Among them, thermal failure is an important part of GIS failure and one of the main forms of failure. Internal defects, contact wear and poor contact of GIS are difficult to be found by existing monitoring technology, and when the fault continues to develop, it will cause serious harm to the GIS device. Long-time operation of the device can greatly increase the probability of failure. Under the limited temperature measurement means, the internal temperature of the GIS disconnector can be obtained through multi-physical field simulation. The internal contact temperature is a direct temperature rise indicator for the thermal failure of the GIS disconnector, and the contact temperature inversion through the shell temperature of the GIS disconnector is an effective means of monitoring the internal contact temperature rise. On this basis, temperature inversion also emerged.

[0004] Chinese patent CN201911141385.2 discloses a GIS disconnector contact temperature prediction method, device and readable storage medium. The method is based on shell temperature and environmental temperature inversion, and a GIS disconnector contact temperature regression prediction model is constructed simply and conveniently, and then the internal temperature of the GIS contact is predicted.

[0005] Chinese patent CN201910368964.4 discloses a GIS disconnector, a temperature calculation method and device. The method establishes an equivalent thermal circuit model for the disconnector by using an equivalent thermal circuit method. The equivalent thermal circuit method is based on the similarity between thermal circuits and electrical circuits. Similar to resistance, the equivalent thermal circuit method uses the resistance of heat in the transmission path to equivalent the attenuation of heat in the transmission path, and is a kind of thermal analysis method that uses heat flow to replace current. The method uses related methods for solving the circuit to solve the thermal circuit. The method has the characteristics of simple theoretical basis, fast calculation speed, easy modeling, obvious parameter relationship and high accuracy.

[0006] The Chinese patent CN202310335390.7 discloses a GIS disconnector contact temperature long-period prediction model method, comprising the following steps: finite element temperature field simulation constructs a GIS disconnector structure model; according to the SVR algorithm, a mapped SVR model is formed; a dynamic load prediction model is selected; the dynamic load of the GIS disconnector structure model is predicted; parameter adjustment is performed to obtain a model with the lowest prediction error; the contact temperature long-period prediction model of the GIS disconnector is obtained; the trend and periodic change trend can be fitted, this method is suitable for the daily periodic and annual periodic load change law of the power system, and is also suitable for the daily periodic and annual periodic change law of environmental factors, and the prediction effect is improved.

[0007] However, in the research on the GIS disconnector contact temperature prediction method, the SVR algorithm or the equivalent thermal circuit method used in the above method is too simple, and the contact temperature prediction accuracy cannot be guaranteed, and there is a large error between the predicted contact temperature and the actual contact temperature. Moreover, the contact temperature prediction method established by the above method is only applicable to the corresponding type of GIS disconnector, and when the type is changed, a new contact temperature prediction model needs to be established, and the algorithm parameters need to be recalculated, so the applicability is poor. SUMMARY

[0008] The technical problem to be solved by the present application is to provide a GIS disconnector contact temperature prediction method and terminal, which can improve the prediction accuracy and applicability of the GIS disconnector contact temperature prediction.

[0009] In order to solve the above technical problems, the technical scheme adopted by the present application is:

[0010] A GIS disconnector contact temperature prediction method, comprising the steps of:

[0011] An electromagnetic field simulation model of the GIS disconnector is established, and the losses of each component in the GIS disconnector are obtained by solving the electromagnetic field simulation model, the losses are introduced as heat source information into the temperature-fluid field of the GIS disconnector, and the temperature field of the GIS disconnector is obtained;

[0012] According to the temperature field and operating conditions of the GIS disconnector, a training set and a test set containing the load current, contact resistance and environmental temperature of the GIS disconnector are established;

[0013] A sparse chaotic polynomial prediction model is constructed, the sparse chaotic polynomial prediction model is trained through the training set, the sparse chaotic polynomial prediction model is evaluated through the test set, and an input variable subset of the sparse chaotic polynomial prediction model is determined, and the contact temperature of the GIS disconnector is predicted through the sparse chaotic polynomial prediction model and the input variable subset.

[0014] To solve the above technical problems, another technical solution adopted by the present application is:

[0015] A contact temperature prediction terminal of a GIS disconnector comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements each step of the contact temperature prediction method of the GIS disconnector when executing the computer program.

[0016] The present application has the beneficial effects that: the electromagnetic field simulation is first performed on the GIS disconnector to obtain the loss generated by each component, the obtained loss is introduced as a heat source into a temperature-fluid field for temperature field calculation, in this way, the multi-physical field simulation is established in an indirect coupling manner for the GIS disconnector to improve the simulation accuracy, and the training set and the test set required for prediction can be obtained based on the simulation result. A sparse chaotic polynomial prediction model is constructed, and an input variable subset of the sparse chaotic polynomial prediction model is determined, and the contact temperature of the GIS disconnector is predicted through the sparse chaotic polynomial prediction model and the input variable subset. In the GIS disconnector contact temperature prediction, the data samples are often less, in this way, the high prediction accuracy can be maintained through the polynomial chaotic theory, the algorithm required for prediction is matched, and because there is no pending setting parameter in the algorithm, the algorithm model does not need to be adjusted according to the data, and can be applied to various GIS disconnector contact temperature predictions, and has good applicability. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 A flow chart of a contact temperature prediction method of a GIS disconnector according to an embodiment of the present application;

[0018] Figure 2 A schematic diagram of a contact temperature prediction terminal of a GIS disconnector according to an embodiment of the present application;

[0019] Figure 3 A technical roadmap of a contact temperature prediction method of a GIS disconnector according to an embodiment of the present application;

[0020] Figure 4 A prediction temperature error diagram according to an embodiment of the present application.

[0021] LABEL EXPLANATION:

[0022] 1. A GIS disconnector contact temperature prediction terminal; 2. a memory; 3. a processor. DETAILED DESCRIPTION

[0023] To illustrate the technical content of the present application, the purposes and effects achieved are described in detail below in conjunction with the embodiments and the accompanying drawings.

[0024] Please refer to Figure 1 The embodiment of the present application provides a GIS disconnector contact temperature prediction method, comprising the steps of:

[0025] An electromagnetic field simulation model of the GIS disconnector is established, and the losses of each component in the GIS disconnector are obtained by solving the electromagnetic field simulation model, and the losses are introduced as heat source information into the temperature-fluid field of the GIS disconnector to obtain the temperature field of the GIS disconnector.

[0026] According to the temperature field and operating conditions of the GIS disconnector, a training set and a test set containing the load current, contact resistance and ambient temperature of the GIS disconnector are established.

[0027] A sparse chaotic polynomial prediction model is constructed, the sparse chaotic polynomial prediction model is trained by the training set, the sparse chaotic polynomial prediction model is evaluated by the test set, and the input variable subset of the sparse chaotic polynomial prediction model is determined, and the contact temperature of the GIS disconnector is predicted by the sparse chaotic polynomial prediction model and the input variable subset.

[0028] From the above description, the beneficial effects of the present application are that the GIS disconnector is first simulated by electromagnetic field, the losses of each component are obtained, the obtained losses are introduced as heat source into the temperature-fluid field for temperature field calculation, in this way, the multi-physical field simulation is established for the GIS disconnector in an indirect coupling manner to improve the simulation accuracy, and the training set and the test set required for prediction can be obtained based on the simulation results. A sparse chaotic polynomial prediction model is constructed, and the input variable subset of the sparse chaotic polynomial prediction model is determined, and the contact temperature of the GIS disconnector is predicted by the sparse chaotic polynomial prediction model and the input variable subset. In the GIS disconnector contact temperature prediction, the data samples are often less, in this way, the high prediction accuracy can be maintained by the polynomial chaos theory, which is consistent with the algorithm required for prediction, and because there are no pending parameters in the algorithm, the algorithm model does not need to be adjusted according to the data, and can be applied to various GIS disconnector contact temperature predictions, and has good applicability.

[0029] Further, the losses of each component in the GIS disconnector obtained by solving the electromagnetic field simulation model comprise:

[0030] Solving the electromagnetic field simulation model in combination with the contact resistance equivalence, proximity effect and skin effect of the GIS disconnector, the loss of each component in the GIS disconnector is obtained.

[0031] As can be seen from the above description, the contact resistance equivalence of the GIS disconnector is considered in the solving process, which can avoid overheating and damage caused by current concentration; the proximity effect is considered, which can improve the prediction ability of the model for actual current distribution and loss; and the skin effect is considered, which can more accurately simulate the distribution and loss of high-frequency current in the conductor.

[0032] Further, the loss is introduced as heat source information into the temperature-fluid field of the GIS disconnector to obtain the temperature field of the GIS disconnector, including:

[0033] The loss obtained by solving the electromagnetic field simulation model is introduced as heat source information into the temperature-fluid field of the GIS disconnector, and the temperature-fluid field is solved in combination with heat conduction, heat radiation and gas-solid boundary layer network division to obtain the overall temperature distribution information of the GIS disconnector.

[0034] As can be seen from the above description, the heat conduction is considered in the solving process, which can more truly reflect the temperature distribution and change inside the device; the heat radiation is considered, which can more accurately simulate the heat transfer process in the GIS disconnector; and the fine boundary layer network division is considered, which can more accurately simulate the interaction between gas and solid, including heat exchange and flow characteristics, thereby improving the accuracy of the temperature-fluid field simulation.

[0035] Further, the sparse chaotic polynomial prediction model is constructed, including:

[0036] Obtaining a random factor x i , the random factor x i is converted into a random variable ξ i subject to a standard distribution i , the contact temperature Y of the GIS disconnector is expressed as a chaotic polynomial function of the random variable ξ i by orthogonal polynomials:

[0037]

[0038] In the formula, d represents the number of variables, b0, b i , b ij , b ijk all represent the coefficients of the chaotic polynomial, and Hm represents the Hermite orthogonal polynomial of order m.

[0039] As can be known from the above description, the polynomial chaos theory represents the random output by using the orthogonal polynomial of the random input, and can improve the accuracy of the model without requiring a large number of samples.

[0040] Further, the input variable subset of the sparse chaos polynomial prediction model is determined, and the method comprises the following steps of:

[0041] The input variable subset having the greatest influence on the output result of the sparse chaos polynomial prediction model is selected by using the least angle regression method.

[0042] As can be known from the above description, by only retaining the polynomial terms having significant contribution to the output variable, the number of required polynomial terms is greatly reduced, the complexity of the model in processing high-dimensional data is effectively reduced, and the calculation efficiency is significantly improved on the premise of maintaining the accuracy of the output response.

[0043] Please refer to Figure 2 The other embodiment of the present application provides a GIS disconnector contact temperature prediction terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and each step of the GIS disconnector contact temperature prediction method is realized when the processor executes the computer program.

[0044] The GIS disconnector contact temperature prediction method and terminal can improve the prediction accuracy and applicability of the GIS disconnector contact temperature prediction, and the following specific embodiments are described.

[0045] Embodiment one

[0046] Please refer to Figure 1 and Figure 3 A GIS disconnector contact temperature prediction method comprises the following steps:

[0047] S1, an electromagnetic field simulation model of a GIS disconnector is established, and the loss of each component in the GIS disconnector is obtained by solving the electromagnetic field simulation model, the loss is introduced as heat source information into a temperature-fluid field of the GIS disconnector, and a temperature field of the GIS disconnector is obtained.

[0048] In the method, the electromagnetic field simulation model is solved in combination with the contact resistance equivalence, the proximity effect and the skin effect of the GIS disconnector, and the loss of each component in the GIS disconnector is obtained.

[0049] Specifically, the embodiment considers the equivalent contact resistance of the GIS disconnector in the solving process. The actual contact resistance is included in the model, which can more truly reflect the distribution of current at the contact point and avoid overheating and damage caused by current concentration. When the current flows through the conductor, an induced current is generated between the adjacent conductors, which affects the current distribution and electric field loss. Therefore, considering the proximity effect can improve the prediction ability of the model for actual current distribution and loss. When the current is distributed in the conductor, it tends to concentrate near the surface of the conductor. Therefore, considering the skin effect can more accurately simulate the distribution and loss of high-frequency current in the conductor.

[0050] The loss obtained by solving the electromagnetic field simulation model is introduced as heat source information into the temperature-fluid field of the GIS disconnector, and the temperature-fluid field is solved by combining heat conduction, heat radiation, and gas-solid boundary layer network division to obtain the overall temperature distribution information of the GIS disconnector.

[0051] Specifically, since the GIS disconnector contains various metal components and insulating gas, the heat conduction between these components and gas is an important factor affecting temperature distribution. Accurate simulation of heat conduction process can more truly reflect the temperature distribution and change inside the device. In a high-temperature environment, heat radiation is an important heat transfer method that cannot be ignored. Considering heat radiation can more accurately simulate the heat transfer process in the GIS disconnector. Fine boundary layer network division can more accurately simulate the interaction between gas and solid, including heat exchange and flow characteristics, thereby improving the accuracy of temperature-fluid field simulation.

[0052] Therefore, by means of indirect coupling of multiple physical fields of the GIS disconnector, multiple physical field simulation of the GIS disconnector can be realized, which can ensure the accuracy of simulation and facilitate subsequent data set required for prediction.

[0053] S2, according to the temperature field and operating conditions of the GIS disconnector, a training set and a test set containing the load current, contact resistance and ambient temperature of the GIS disconnector are established.

[0054] Specifically, due to the limited simulation and test sample data, in order to meet the needs of engineering application, it is very important to establish a reasonable temperature inversion training set and test set. Therefore, the training set and test set fully consider various different operating conditions of the GIS disconnector, and the selected training set and test set are as follows:

[0055] (1) Considering the different states of contact contact of GIS disconnectors, three contact states are fully considered in the training set and the test set, because different contact states have an impact on temperature inversion and point selection optimization. The contact resistance of the training set is set to 3 μΩ, 20 μΩ, 40 μΩ, and 60 μΩ, and the contact resistance of the test set is set to 5 μΩ, 15 μΩ, 25 μΩ, and 35 μΩ.

[0056] (2) Considering different environmental temperatures, GIS disconnectors have different environmental temperatures in different regions, so environmental temperatures from 0°C to 30°C are considered. The environmental temperature of the training set is set to 0°C, 10°C, 20°C, and 30°C, and the environmental temperature of the test set is set to 15°C and 25°C.

[0057] (3) Considering different load currents, GIS disconnectors are rarely in full load, i.e., rated current 3150A, in actual operation, and more are less than rated current. Therefore, the training set is set to 1200A, 1800A, 2400A, and 3150A, and the test set is set to 2000A and 3000A.

[0058] Therefore, the algorithm test set of the embodiment is shown in Table 1:

[0059] Table 1 Test Set

[0060]

[0061]

[0062] S3, a sparse chaotic polynomial prediction model is constructed, the sparse chaotic polynomial prediction model is trained through the training set, the sparse chaotic polynomial prediction model is evaluated through the test set, and an input variable subset of the sparse chaotic polynomial prediction model is determined. The contact temperature of the GIS disconnector is predicted through the sparse chaotic polynomial prediction model and the input variable subset.

[0063] wherein a random factor x i is obtained i is converted into a random variable ξ i that obeys a standard distribution i The contact temperature Y of the GIS disconnector is expressed as a chaotic polynomial function of the random variable ξ

[0064]

[0065] In the formula, d represents the number of variables, b0, b i , b ij , and b ijk all represent the coefficients of the chaotic polynomial, Hermite orthogonal polynomials of order m.

[0066] In this embodiment, the number of undetermined coefficients in formula (1) is:

[0067]

[0068] where the m-order expansion of the Hermite orthogonal polynomial is:

[0069]

[0070] When the expansion order of the Hermite orthogonal polynomial increases, that is, m is larger, the accuracy of the prediction of the chaotic polynomial is also improved, that is, the predicted value Y is more accurate, but the number of coefficients to be determined is also larger. Therefore, the expansion order is not the higher the better, and the expansion order needs to be selected according to the accuracy required by the specific calculation.

[0071] In this embodiment, when the expansion order of the Hermite orthogonal polynomial is greater than or equal to 3, increasing the expansion order cannot significantly improve the accuracy of the overall algorithm, and therefore a three-order expansion is selected.

[0072] The expression of the chaotic polynomial expansion to the third order is:

[0073]

[0074] The minimum angle regression method is used to select the input variable subset that most affects the output result of the sparse chaotic polynomial prediction model.

[0075] Specifically, the pure PC (chaotic polynomial) method has problems when the input variable is high-dimensional. If the selected subset is not in the interval with high probability, it is very likely to cause a relatively large calculation error. This leads to the problem of optimal selection of the subset. Therefore, the least angle regression (LAR) method is used to select the input variable subset that most affects the output response Y.

[0076] The LAR optimization model is as follows:

[0077] min J(b) = ||y'-Ψ'b||, Ψ'∈R N×(P+1) y'∈R N×1

[0078]

[0079] In the formula, b represents the coefficient of the chaotic polynomial, y' = [y'1, y'2,..., y'N] represents the output response, and Ψ' represents the input variable subset. N ]T decentralized data representing output response, Ψ' represents an orthogonal polynomial matrix, standardized data, R represents a real number vector space, N represents a sample size, P represents a feature number, and λ is a given value.

[0080] wherein, given a λ value, the least square method can solve b. If the λ value is a small value, a sparse model can be obtained by dimension reduction.

[0081] Then, the error of the predicted value is evaluated:

[0082] The mean square error MSE is the average of the sum of squares of errors, and is the most commonly used index for measuring the error of a regression model. In the SVR grid search optimization parameter, MSE is also used. When the MSE value is closer to 0, it indicates that the fitting effect of the algorithm model is good, and the predicted value is closer to the true value y i .

[0083]

[0084] The mean absolute error MAE (also known as L1 norm loss) is used, that is, the average of the absolute values of errors. MAE can accurately reflect the size of the actual prediction error. MAE is used to evaluate the deviation between the true value yi and the fitted value . The smaller the MAE value, the better the model fitting, and the higher the model prediction accuracy.

[0085]

[0086] The mean absolute percentage error MAPE is used to express the model effect in percentage:

[0087]

[0088] In this embodiment, 80 samples are divided into a training set, a validation set and a test set, with a ratio of 7:1:2. The load current of the GIS disconnector, the environmental temperature and the 78 shell temperature measurement points are used as inputs, and the average value of the three-phase contact hot spot temperature is used as output. The calculated model indicators are e MSE , e MAE , e MAPE are 0.39, 0.44 and 0.47%, respectively.

[0089] The temperature predicted by the sparse chaotic polynomial is compared with the simulation temperature, as shown in Table 2 and Figure 4 . It can be obtained that the maximum absolute error of the contact predicted temperature obtained by the sparse chaotic polynomial prediction and the simulation temperature value is 1.28℃, and the minimum absolute error value is 0.00℃, the error is small, and the fitting effect is good.

[0090] Table 2 Sparse chaotic polynomial temperature prediction error

[0091]

[0092]

[0093] Embodiment two

[0094] Please refer to Figure 2 A contact temperature prediction terminal 1 of a GIS disconnector, comprising a memory 2, a processor 3, and a computer program stored on the memory 2 and executable on the processor 3, wherein the processor 3 implements each step of the contact temperature prediction method of the GIS disconnector according to the computer program.

[0095] In summary, the contact temperature prediction method and terminal of the GIS disconnector provided by the present application, when the shell temperature precision is limited by the precision of the temperature sensor and cannot be effectively improved due to environmental factors, the precision of the inversion algorithm is more important. Compared with traditional polynomial chaos, sparse polynomial chaos retains the ability to handle uncertainty, greatly reduces the number of polynomial terms required for calculation by retaining only the polynomial terms that have a significant contribution to the output variable, effectively reduces the complexity of the model when handling high-dimensional data, and significantly improves the calculation efficiency while maintaining the accuracy of the output response. The following advantages are obtained:

[0096] (1) Reduce the calculation cost: by retaining the polynomial terms that have a significant contribution to the output, the calculation amount required by the model is reduced.

[0097] (2) Improve the calculation efficiency: optimize the algorithm to reduce the number of necessary samples, suitable for small sample model, can quickly get accurate results.

[0098] (3) Solve high-dimensional problems: use the sparsity to handle multiple random input variables, effectively manage and reduce the complexity of high-dimensional problems.

[0099] The above is only an embodiment of the present application, and does not limit the patent range of the present application, any equivalent transformation or direct or indirect application in related technical fields using the content of the present application specification and drawings is also included in the patent protection range of the present application.

Claims

1. A method for predicting the contact temperature of a GIS disconnector, characterized in that, Including the following steps: An electromagnetic field simulation model of a GIS disconnect switch is established. The losses of each component in the GIS disconnect switch are obtained by solving the electromagnetic field simulation model. The losses are then used as heat source information and imported into the temperature-fluid field of the GIS disconnect switch to obtain the temperature field of the GIS disconnect switch. Based on the temperature field and operating conditions of the GIS disconnect switch, a training set and a test set are established, which include the load current, contact resistance and ambient temperature of the GIS disconnect switch. A sparse chaotic polynomial prediction model is constructed, trained using the training set, evaluated using the test set, and a subset of input variables is determined. The contact temperature of the GIS disconnect switch is then predicted using the sparse chaotic polynomial prediction model and the subset of input variables. The loss of each component in the GIS disconnector is obtained by solving the electromagnetic field simulation model, including: By combining the contact resistance equivalent, proximity effect and skin effect of the GIS disconnect switch, the electromagnetic field simulation model is solved to obtain the losses of each component in the GIS disconnect switch; The loss is imported as heat source information into the temperature-fluid field of the GIS disconnect switch to obtain the temperature field of the GIS disconnect switch, including: The loss obtained from solving the electromagnetic field simulation model is used as heat source information and imported into the temperature-fluid field of the GIS disconnector. The temperature-fluid field is solved by combining heat conduction, heat radiation and gas-solid boundary layer network decomposition to obtain the overall temperature distribution information of the GIS disconnector. The construction of the sparse chaotic multinomial prediction model includes: Obtaining random factors x i The random factors x i Transform into a random variable that follows a standard distribution. ξ i The contact temperature of the GIS disconnect switch is determined by an orthogonal polynomial. Y Represented as the random variable ξ i Chaotic polynomial function: In the formula, d Indicates the number of variables. b 0、 b i , b ij , b ijk All represent the coefficients of the chaotic polynomial. φ m (·) indicates that the order is m The Hermite orthogonal polynomials, where m takes values ​​of 1, 2, and 3 respectively.

2. The method for predicting the contact temperature of a GIS disconnector according to claim 1, characterized in that, Determining the subset of input variables for the sparse chaotic polynomial prediction model includes: The minimum angle regression method is used to select the subset of input variables that have the greatest impact on the output of the sparse chaotic multinomial prediction model.

3. A contact temperature prediction terminal for a GIS disconnector, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it performs the following steps: An electromagnetic field simulation model of a GIS disconnect switch is established. The losses of each component in the GIS disconnect switch are obtained by solving the electromagnetic field simulation model. The losses are then used as heat source information and imported into the temperature-fluid field of the GIS disconnect switch to obtain the temperature field of the GIS disconnect switch. Based on the temperature field and operating conditions of the GIS disconnect switch, a training set and a test set are established, which include the load current, contact resistance and ambient temperature of the GIS disconnect switch. A sparse chaotic polynomial prediction model is constructed, trained using the training set, evaluated using the test set, and a subset of input variables is determined. The contact temperature of the GIS disconnect switch is then predicted using the sparse chaotic polynomial prediction model and the subset of input variables. The loss of each component in the GIS disconnector is obtained by solving the electromagnetic field simulation model, including: By combining the contact resistance equivalent, proximity effect and skin effect of the GIS disconnect switch, the electromagnetic field simulation model is solved to obtain the losses of each component in the GIS disconnect switch; The loss is imported as heat source information into the temperature-fluid field of the GIS disconnect switch to obtain the temperature field of the GIS disconnect switch, including: The loss obtained from solving the electromagnetic field simulation model is used as heat source information and imported into the temperature-fluid field of the GIS disconnector. The temperature-fluid field is solved by combining heat conduction, heat radiation and gas-solid boundary layer network decomposition to obtain the overall temperature distribution information of the GIS disconnector. The construction of the sparse chaotic multinomial prediction model includes: Obtaining random factors x i The random factors x i Transform into a random variable that follows a standard distribution. ξ i The contact temperature of the GIS disconnect switch is determined by an orthogonal polynomial. Y Represented as the random variable ξ i Chaotic polynomial function: In the formula, d Indicates the number of variables. b 0、 b i , b ij , b ijk All represent the coefficients of the chaotic polynomial. φ m (·) indicates that the order is m The Hermite orthogonal polynomials, where m takes values ​​of 1, 2, and 3 respectively.

4. The contact temperature prediction terminal for a GIS disconnector according to claim 3, characterized in that, Determining the subset of input variables for the sparse chaotic polynomial prediction model includes: The minimum angle regression method is used to select the subset of input variables that have the greatest impact on the output of the sparse chaotic multinomial prediction model.

Images