CN116644663A - A Transformer Hotspot Temperature Inversion Prediction Method Based on Oil Temperature at Characteristic Measurement Points - Google Patents

Abstract

The application discloses a transformer hot spot temperature inversion prediction method based on characteristic measurement point oil temperature, which comprises the following steps: acquiring the characteristic measuring point oil temperature of the oil immersed transformer and the monitoring information of the relevant state quantity of the characteristic measuring point for several days, and constructing multi-element time sequence sample data of a characteristic measuring point oil temperature prediction model; calculating a correlation coefficient matrix of the sample data, and extracting a characteristic vector of the oil temperature of the characteristic measuring point; then taking the characteristic vector as multidimensional input, taking the oil temperature of the characteristic measuring point as multidimensional output, establishing a multivariate time sequence prediction model, verifying the accuracy of the multivariate time sequence prediction model on sample data, and realizing single-step prediction (real-time rolling prediction) or multi-step prediction (short-term prediction) of the oil temperature of the characteristic measuring point; then, obtaining hot spot temperatures under different working conditions through simulation calculation of transformer temperature field distribution, and constructing inversion samples of the hot spot temperatures by combining monitoring data of the oil temperatures of the characteristic measuring points; and finally, establishing a hot spot temperature and characteristic measuring point oil temperature relation model, and verifying the effectiveness and accuracy of the model on an inversion sample, so that inversion prediction of a hot spot temperature future value is realized. The application can rapidly realize real-time or short-term prediction of the hot spot temperature by utilizing the relevant state quantity historical information which is easy to acquire in the transformer monitoring system, and has a certain reference value for the operation and maintenance management of the large-scale oil-immersed transformer.

Description

A Transformer Hotspot Temperature Inversion Prediction Method Based on Oil Temperature at Characteristic Measurement Points

technical field

The invention relates to the technical field of hot spot temperature monitoring of oil-immersed transformers, in particular to a transformer hot spot temperature inversion prediction method based on oil temperature at characteristic measuring points.

Background technique

The transformer is an important connection between the power grid and the generator set, and its safe and stable operation is crucial to ensure the long-term safety of the entire system. According to relevant statistical analysis at home and abroad, the aging or damage of transformer insulation is an important cause of transformer accidents, and winding is one of the main parts of transformer failure. Winding insulation is closely related to transformer temperature rise, and with the commissioning of large-scale high-voltage oil-immersed transformers, the problem of temperature rise becomes more prominent. If the hot spot temperature of the transformer exceeds the threshold, it will easily cause transformer accidents and directly affect the safety of the substation; if the hot spot temperature is too low, it means that the transformer capacity is not fully utilized, and the economy of the substation needs to be improved. Therefore, monitoring the hot spot temperature of large oil-immersed transformers is conducive to reducing the failure rate of transformers, prolonging the service life of transformers, and ensuring the economic benefits of transformers.

At present, for large-scale oil-immersed transformers, it is difficult to arrange distributed optical fiber temperature sensors to directly monitor the hot-spot temperature in batches, and the operation and maintenance costs are high. Therefore, for large-scale oil-immersed transformers, the engineering site usually uses indirect methods to estimate the hot-spot temperature Forecasting, the most commonly used method at this stage is to indirectly estimate the hot spot temperature from the monitoring value of the top oil temperature according to the empirical model in the national standard GB/T 1094.7 (corresponding to IEC60076-7). However, this empirical model ignores some nonlinear characteristics of transformer oil, and cannot reflect the influence of transformer operating state changes on hot spot temperature. With the wide application of artificial intelligence technology in the power system, using the historical operation data that is easy to obtain in the transformer monitoring system except for the hot spot temperature, the intelligent algorithm is used to realize the continuous prediction of the hot spot temperature, which is conducive to the real-time monitoring of the transformer operating status And the timely adjustment of the operating mode. However, most of the hotspot temperature prediction models established in the existing literature are the hotspot temperature test measurement data, and the acquisition of samples is limited by the test conditions and the actual operation status of the transformer; in addition, the hotspot temperature prediction model in some literatures can only The measured data of other states calculates the hotspot temperature at the current moment, but does not reflect the change of the hotspot temperature in the time domain.

Contents of the invention

In order to solve the above technical problems, the present invention proposes a transformer hotspot temperature inversion prediction method based on the oil temperature of characteristic measuring points, combined with finite element simulation and artificial intelligence algorithm, and constructs an analysis sample with transformer multi-state monitoring data and hotspot temperature simulation data , can realize rapid and continuous prediction of hot spot temperature, and has certain reference value for the operation and maintenance management of large oil-immersed transformers.

In the first aspect, the present invention provides a transformer hotspot temperature inversion prediction method based on oil temperature at characteristic measuring points, including:

Step A, according to several days of monitoring information of the oil temperature at the characteristic measuring point of the oil-immersed transformer and other relevant state quantities, construct the multivariate time series sample data of the oil temperature prediction model at the characteristic measuring point;

Step B, performing normalization processing on the multivariate time series sample data described in step A, and calculating its correlation coefficient matrix, and extracting the feature vector of the oil temperature at the feature measuring point;

Step C, using the feature vector described in step B as the multidimensional input, and using the oil temperature of the characteristic measuring point as the multidimensional output, to establish a multivariate time series prediction model, and verify its validity and accuracy on the sample data described in step A, so that Single-step forecast (real-time rolling forecast) or multi-step forecast (short-term forecast) for oil temperature at characteristic measuring points;

Step D. Carry out simulation calculation on the temperature field distribution of the transformer, extract the hot spot temperature and the oil temperature of the characteristic measuring point under different working conditions to construct the inversion sample of the hot spot temperature;

Step E: Establish a relationship model between the hot spot temperature and the oil temperature at the characteristic measuring point, and verify its validity and accuracy on the inversion sample described in step D, so as to realize the inversion prediction of the future value of the hot spot temperature.

As a preferred technical solution of the present invention: in the step A, a plurality of transformer oil temperature measurement points that are easy to obtain monitoring data are selected as characteristic measurement points based on the hot spot temperature inversion principle of the oil-immersed transformer, such as the top layer oil temperature, outlet Oil port temperature, oil inlet temperature, etc., combined with monitoring information of other related state variables, such as voltage, current, active power, reactive power, cooler temperature, etc., construct multivariate time series sample data as shown in the following formula:

In the formula, is the i-th dimension state quantity of the j-th group of data, m is the total dimension of all state quantities including the oil temperature of the characteristic measuring point, and n is the total length of the time series, that is, the total number of groups of sample data.

As a kind of preferred technical scheme of the present invention: in described step B, comprehensively consider the linear correlation and the nonlinear correlation of sequence, calculate Pearson-Spearman mixed correlation coefficient matrix by following formula:

In the formula, Represent the average value of the p and q-dimensional state quantities of n groups of data; arrange the p and q two-dimensional state quantity data x _p , x _q in descending order to form new column vectors x′ _p , x′ _q , /> and /> The positions in x′ _p , x′ _q are respectively denoted as The value is between [-1, 1], /> Indicates a complete negative correlation, /> means completely irrelevant, /> means a perfect positive correlation.

As a preferred technical solution of the present invention: in the step C, the multivariate time series prediction model is a CNN-GRU model, its basic structure includes a CNN feature extraction module and a GRU prediction module, and the CNN feature extraction module contains a convolutional layer, a pool layer and flattening layer, the GRU prediction module includes the GRU recurrent unit and the fully connected layer, the number of convolutional layers in the specific CNN module, the number of convolution kernels, the size of the convolution kernel, and the number of recurrent unit layers in the GRU module, the cycle The number of units is further designed through the prediction effect of the sample data.

As a preferred technical solution of the present invention: in the step D, the finite element simulation calculation of multi-physics field coupling is performed on the temperature field distribution of the transformer under different working conditions, and the hot spot temperature under various working conditions is extracted therefrom, combined with the corresponding Based on the monitoring data of oil temperature at characteristic measuring points under working conditions, an inversion sample of hot spot temperature is constructed and normalized.

As a preferred technical solution of the present invention: in the step E, the relationship model between the hot spot temperature and the oil temperature at the characteristic measuring point is a PSO-BPNN model; the step E includes the following steps E1 to E7:

Step E1, determine the topological structure of BPNN, that is, the number of units in its input layer, hidden layer and output layer;

Step E2, determine the PSO structure, use the BPNN training error as the particle fitness, and initialize the particle swarm;

Step E3, calculate the fitness value of the current position of each particle, so as to measure the pros and cons of the position, and use the position with the maximum fitness value found by the i-th particle in the d-th dimension under the current number of iterations Denoted as P _id , that is, the individual historical optimal solution; the position with the maximum fitness value found by all particles in the d-th dimension under the current iteration number is denoted as P _gd , that is, the group historical optimal solution.

Step E4, according to the individual historical optimal solution and the group historical optimal solution of the particle swarm, use the following formula to update the velocity and position of each particle to ensure that the particle moves towards a better solution:

v _id (k+1)＝ωv _id +c ₁ r ₁ (P _id (k)-x _id (k))+c ₂ r ₂ (P _gd (k)-x _id (k))

x _id (k+1)＝x _id (k)+v _id (k+1)

In the formula, k is the number of iterations; x _id and v _id are respectively the position and velocity of the i-th particle in the d-dimensional target solution space; ω is the inertia weight of the velocity; c ₁ is the individual acceleration coefficient of a single particle, and c ₂ is The population acceleration coefficient of the particle swarm; r ₁ and r ₂ are random constants in the range [0, 1];

Step E5, repeat step E3 and step E4 until the set termination condition is reached, such as reaching the maximum number of iterations, the fitness meets the preset threshold or the running time exceeds the limit, etc., and output the group history optimal calculated in the last iteration The solution is taken as the global optimal solution, that is, the PSO-optimized BPNN weights and biases;

Step E6, initialize the BPNN weight and bias with the output of the particle swarm, and train and verify the relationship model between the hot spot temperature and the oil temperature at the characteristic measuring point on the hot spot temperature inversion sample described in step D;

Step E7, based on the future value of the oil temperature of the characteristic measuring point predicted in the aforementioned step C, and the relationship model between the hot spot temperature and the oil temperature of the characteristic measuring point described in step E6, realize real-time or short-term prediction of the future value of the hot spot temperature.

In a second aspect, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and operable on the processor, and the processor implements the steps of the above method when executing the computer program.

In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.

The present invention proposes a transformer hotspot temperature inversion prediction method based on the oil temperature of characteristic measuring points. The sample data during modeling does not depend on the hot spot temperature measurement test, so it is not limited by the test conditions and the actual operating state of the transformer; The quick prediction of the future value of temperature has certain reference value for the operation and maintenance management of large oil-immersed transformers.

Description of drawings

Fig. 1 is a schematic flow chart of a transformer hotspot temperature inversion prediction method based on characteristic measuring point oil temperature in an embodiment of the present invention;

2 is a schematic diagram of the CNN-GRU network structure used to establish the feature measuring point oil temperature prediction model in the embodiment of the present invention;

Fig. 3 is a schematic diagram of the BPNN structure used to establish the relationship model between the hot spot temperature and the oil temperature of the characteristic measuring point in the embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

As shown in Figure 1, a transformer hotspot temperature inversion prediction method based on oil temperature at characteristic measuring points includes the following steps:

Step A. For the large-scale oil-immersed transformer, based on the hot spot temperature inversion principle, select the top oil temperature T _{top_oil} , the oil outlet temperature T _{oil_out} , and the oil inlet temperature T _{oil_in} which are easy to obtain monitoring data as the characteristic measuring point oil of the transformer temperature; other relevant state quantities include high voltage side phase voltage U _A , U _B , U _C , high voltage side line voltage U _AB , U _BC , U _CA , high voltage side current I _A , I _B , I _C , active power P, reactive power Power Q, frequency f, power factor cosφ, cooler water outlet temperature T _{water_out} and cooler water inlet temperature T _{water_in} ; oil-immersed transformer characteristic measuring point oil temperature and other related state quantity monitoring data sampling time is 3 minutes, extract The 5-day monitoring data constructs multivariate time series sample data as shown in the following formula:

In the formula, is the i-th dimension state quantity of the j-th group of data, /> respectively represent T _{oil_out} , T _{oil_in} , T _{top_oil} , T _{water_in} , T _{water_out} , U _A , U _B , U _C , U _AB , U _BC , U _CA , I _A , I _B , I _C , P, Q, f, cos φ.

Step B, normalize the multivariate time series sample data described in step A, comprehensively consider the linear correlation and nonlinear correlation of the sequence, and calculate the Pearson-Spearman mixed correlation coefficient matrix according to the following formula:

In the formula, Represent the average value of the p and q-dimensional state quantities of n groups of data respectively; arrange the p and q two-dimensional state quantity data x _p and x _q in descending order to form a new column vector /> and /> at /> The positions in are denoted as /> The value is between [-1, 1], /> Indicates a complete negative correlation, /> means completely irrelevant, /> Indicates a complete positive correlation;

According to the calculation results of the correlation coefficient matrix, extract is the eigenvector of the oil temperature at the characteristic measuring point.

Step C, using the feature vector described in step B as the multidimensional input, and using the oil temperature of the characteristic measuring point as the multidimensional output, to establish a multivariate time series prediction model based on CNN-GRU. The basic structure of the CNN-GRU network includes a CNN feature extraction module and a GRU prediction module. The CNN feature extraction module includes a convolutional layer, a pooling layer, and a flattening layer. The GRU prediction module includes a GRU recurrent unit and a fully connected layer. According to the prediction effect of actual samples, the specific structure of the CNN-GRU network is determined as follows: CNN convolution layer is 2 layers, the number and size of convolution kernels are 64, 2 and 32, 2, respectively, and the convolution step size is 1; GRU The number of recurrent unit layers is 1, and the number of units is 200; the activation functions are all ReLU functions.

For the sample data described in step A, divide the training set and the test set by 8:2. The setting of the single-step prediction model is: predict the oil temperature of the characteristic measuring point at the next sampling time according to the historical monitoring information of the transformer for 1 hour, that is, the input of the single learning of the model is 20 sets of feature vectors, and the output is 1 set of target vectors in the future . The setting of the multi-step prediction model is: a CNN-GRU model that predicts the oil temperature of the characteristic measuring point in the next hour from the historical monitoring information of 5 hours, that is, the input of the single learning of the model is 100 sets of feature vectors, and the output is 20 sets in the future target vector. The verification results of the model on the test set are shown in the following table:

The verification results shown in the table can prove that the CNN-GRU model is effective and accurate in predicting the oil temperature of the characteristic measuring points, so that the CNN-GRU model can be used for single-step prediction of the oil temperature of the characteristic measuring points (real-time rolling prediction) or multi-step forecasting (short-term forecasting)

Step D: Carry out multi-physics coupling finite element simulation calculations on the temperature field distribution of 400 groups of transformers under different working conditions, extract hot spot temperatures under various working conditions, and combine the monitoring data of oil temperature at characteristic measuring points under corresponding working conditions , construct the inversion sample of hotspot temperature, normalize the inversion, and divide the training set and test set according to the ratio of 8:2.

Step E. Establish a PSO-BPNN-based relationship model between the hot spot temperature and the oil temperature at the characteristic measuring point, and verify its validity and accuracy on the inversion sample described in step D, so as to realize the inversion prediction of the future value of the hot spot temperature . Specifically, the following steps E1 to E7 are included:

Step E1, determine the topological structure of BPNN: take [T _{oil_out} , T _{oil_in} , T _{top_oil} ] as the input, take the hot spot temperature T _{hot_spot} as the output, and the hidden layer is 1 layer and 10 units;

Step E2, determine the PSO structure, set the number of particles to 80, the maximum number of iterations to 500, and use the BPNN training error as the particle fitness to initialize the particle swarm.

Step E3, calculate the fitness value of the current position of each particle, so as to measure the pros and cons of the position, and use the position with the maximum fitness value found by the i-th particle in the d-th dimension under the current number of iterations Denoted as P _id , that is, the individual historical optimal solution; the position with the maximum fitness value found by all particles in the d-th dimension under the current iteration number is denoted as P _gd , that is, the group historical optimal solution.

Step E4, according to the individual historical optimal solution and the group historical optimal solution of the particle swarm, use the following formula to update the velocity and position of each particle to ensure that the particle moves towards a better solution:

v _id (k+1)＝ωv _id +c ₁ r ₁ (P _id (k)-x _id (k))+c ₂ r ₂ (P _gd (k)-x _id (k))

x _id (k+1)＝x _id (k)+v _id (k+1)

In the formula, k is the number of iterations; x _id and v _id are _the position and velocity of the i-th particle in the d-dimensional target solution space; is the individual acceleration coefficient of a single particle, c ₂ is the group acceleration coefficient of the particle swarm, and c ₁ =c ₂ =2; r ₁ and r ₂ are random constants in the range of [0, 1];

Step E5, repeat step E3 and step E4 until the set termination condition is reached, such as reaching the maximum number of iterations, the fitness meets the preset threshold or the running time exceeds the limit, etc., and output the group history optimal calculated in the last iteration The solution is taken as the global optimal solution, that is, the PSO-optimized BPNN weights and biases;

Step E6, initialize the BPNN weight and bias with the output of the particle swarm, train the relationship model between the hot spot temperature and the oil temperature of the characteristic measuring point on the hot spot temperature inversion sample training set described in step D, and verify it on the test set , the results show that the error between the observed value of the hot spot temperature sample and the fitted value of the PSO-BPNN model does not exceed 0.8°C, and the absolute value of the relative error does not exceed 1.5%, which can fully meet the engineering requirements.

Step E7, based on the future value of the oil temperature of the characteristic measuring point predicted in the aforementioned step C, and the relationship model between the hot spot temperature and the oil temperature of the characteristic measuring point described in step E6, realize real-time or short-term prediction of the future value of the hot spot temperature.

In addition, an embodiment of the present invention also proposes a computer device, including a memory, a processor, and a computer program stored in the memory and operable on the processor. When the processor executes the computer program, the steps of the above method are implemented.

In addition, an embodiment of the present invention also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.

Those skilled in the art should understand that the embodiments of the present application may be provided as methods, systems, or computer program products. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The above-mentioned embodiments only express several preferred implementations of the present application, but the present invention is not limited to the above-mentioned specific implementations, and the above-mentioned specific implementations are only illustrative, not restrictive. Under the enlightenment of the present invention, those of ordinary skill in the art can also make many forms of improvement and replacement without departing from the purpose of the present invention and the scope protected by the claims, and these all belong to the protection scope of the present invention .

Claims (8)

1. A transformer hotspot temperature inversion prediction method based on characteristic measuring point oil temperature, is characterized in that, comprises: Step A, according to several days of monitoring information of the oil temperature at the characteristic measuring point of the oil-immersed transformer and other relevant state quantities, construct the multivariate time series sample data of the oil temperature prediction model at the characteristic measuring point; Step B, performing normalization processing on the multivariate time series sample data described in step A, and calculating its correlation coefficient matrix, and extracting the feature vector of the oil temperature at the feature measuring point; Step C, using the feature vector described in step B as the multidimensional input, and using the oil temperature of the characteristic measuring point as the multidimensional output, to establish a multivariate time series prediction model, and verify its validity and accuracy on the sample data described in step A, so that Single-step forecast (real-time rolling forecast) or multi-step forecast (short-term forecast) for oil temperature at characteristic measuring points; Step D. Carry out simulation calculation on the temperature field distribution of the transformer, extract the hot spot temperature and the oil temperature of the characteristic measuring point under different working conditions to construct the inversion sample of the hot spot temperature; Step E: Establish a relationship model between the hot spot temperature and the oil temperature at the characteristic measuring point, and verify its validity and accuracy on the inversion sample described in step D, so as to realize the inversion prediction of the future value of the hot spot temperature. 2. The method according to claim 1, characterized in that, in the aforementioned step A, a plurality of transformer oil temperature measurement points that are easy to obtain monitoring data are selected as characteristic measurement points based on the principle of oil-immersed transformer hot spot temperature inversion, such as the top Oil temperature, oil outlet temperature, oil inlet temperature, etc., combined with monitoring information of other related state variables, such as voltage, current, active power, reactive power, cooler temperature, etc., construct a multivariate time series sample as shown in the following formula data: In the formula, is the i-th dimension state quantity of the j-th group of data, m is the total dimension of all state quantities including the oil temperature of the characteristic measuring point, and n is the total length of the time series, that is, the total number of groups of sample data. 3. method according to claim 1, it is characterized in that, in aforementioned step B, comprehensively consider the linear correlation and nonlinear correlation of sequence, calculate Pearson-Spearman mixed correlation coefficient matrix by following formula: In the formula, Represent the average value of the p and q-dimensional state quantities of n groups of data; arrange the p and q two-dimensional state quantity data x _p , x _q in descending order to form new column vectors x′ _p , x′ _q , /> and /> The positions in x′ _p and x′ _q are respectively denoted as /> The value is between [-1, 1], /> Indicates a complete negative correlation, /> means completely irrelevant, /> Indicates a perfect positive correlation. 4. The method according to claim 1, wherein the multivariate time series prediction model in the aforementioned step C is a CNN-GRU model, and its basic structure includes a CNN feature extraction module and a GRU prediction module, and the CNN feature extraction module contains volume Product layer, pooling layer and flattening layer, GRU prediction module includes GRU recurrent unit and fully connected layer, the number of convolutional layers in the specific CNN module, the number of convolution kernels, the size of convolution kernel and the recurrent unit in the GRU module The number of layers and the number of recurrent units are further designed through the prediction effect of the sample data. 5. The method according to claim 1, characterized in that, the finite element simulation calculation of multi-physics field coupling is carried out to the transformer temperature field distribution under different working conditions in the aforementioned step D, and the hot spot temperature under multiple working conditions is extracted therefrom , combined with the monitoring data of the oil temperature at the characteristic measuring points under the corresponding working conditions, the inversion sample of the hot spot temperature is constructed and normalized. 6. The method according to claim 1, wherein the relational model of the hot spot temperature and the characteristic measuring point oil temperature in the aforementioned step E is a PSO-BPNN model, and the step E includes the following steps E1 to E7: Step E1, determine the topological structure of BPNN, that is, the number of units in its input layer, hidden layer and output layer; Step E2, determine the PSO structure, use the BPNN training error as the particle fitness, and initialize the particle swarm; Step E3, calculate the fitness value of the current position of each particle, so as to measure the pros and cons of the position, and calculate the position with the maximum fitness value found by the i-th particle in the d-th dimension under the current number of iterations Denoted as P _id , that is, the individual historical optimal solution; and the position with the maximum fitness value found by all particles in the d-th dimension under the current iteration number is denoted as P _gd , that is, the group historical optimal solution. Step E4, according to the individual historical optimal solution and the group historical optimal solution of the particle swarm, use the following formula to update the velocity and position of each particle to ensure that the particle moves towards a better solution: v _id (k+1)＝ωv _id +c ₁ r ₁ (P _id (k)-x _id (k))+c ₂ r ₂ (P _gd (k)-x _id (k)) x _id (k+1)＝x _id (k)+v _id (k+1) In the formula, k is the number of iterations; x _id and v _id are respectively the position and velocity of the i-th particle in the d-dimensional target solution space; ω is the inertia weight of the velocity; c ₁ is the individual acceleration coefficient of a single particle, and c ₂ is The population acceleration coefficient of the particle swarm; r ₁ and r ₂ are random constants in the range [0, 1]; Step E5, repeat step E3 and step E4 until the set termination condition is reached, such as reaching the maximum number of iterations, the fitness meets the preset threshold or the running time exceeds the limit, etc., and output the group history optimal calculated in the last iteration The solution is taken as the global optimal solution, that is, the PSO-optimized BPNN weights and biases; Step E6, initialize the BPNN weight and bias with the output of the particle swarm, and train and verify the relationship model between the hot spot temperature and the oil temperature at the characteristic measuring point on the hot spot temperature inversion sample described in step D; Step E7, based on the future value of the oil temperature of the characteristic measuring point predicted in the aforementioned step C, and the relationship model between the hot spot temperature and the oil temperature of the characteristic measuring point described in step E6, realize real-time or short-term prediction of the future value of the hot spot temperature. 7. A computer device, comprising a memory, a processor, and a computer program stored on the memory and operable on the processor, characterized in that, when the processor executes the computer program, any one of claims 1 to 6 is realized. A step of said method. 8. A computer-readable storage medium, on which a computer program is stored, wherein when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.