Method for analyzing temperature distribution of stator core of hydro-generator

Abstract

The application provides a hydroelectric generator stator core temperature distribution analysis method, aiming at other common analysis methods of large generator core temperature, including monthly maximum value, same period and period comparison method, when the difference between the same period and period comparison is large, further in-depth analysis is conducted. At the same time, the maximum temperature deviation of all tiles of the bearing at the same time is analyzed, when the temperature deviation value is large, further in-depth analysis is conducted. The existing large generator bearing tile temperature analysis method has less content, cannot analyze the temperature state of multiple tiles, and cannot analyze the state and trend of tile temperature in the spatial dimension. The hydroelectric generator stator core temperature distribution analysis method has remarkable beneficial effects in improving real-time and accuracy, comprehensive data analysis and expanding bearing tile temperature analysis, and is helpful to guarantee the safe and stable operation of the hydroelectric generator.

Description

Technical Field

[0001] This invention relates to a method for analyzing the temperature distribution of the stator core of a hydro-generator, and more particularly to a method for analyzing the temperature distribution of the stator core of a hydro-generator. Background Technology

[0002] A hydroelectric generator is a device that converts the energy of flowing water into electrical energy, and its stator core is one of the core components of the generator. The temperature distribution of the stator core has a significant impact on the operational stability and lifespan of the generator. Therefore, research on analytical methods for the temperature distribution of the hydroelectric generator stator core is of great importance.

[0003] Traditional methods for analyzing the temperature distribution of the stator core in hydro-generators are primarily based on empirical formulas. These formulas are empirical rules derived from experimental data or theoretical derivations and can be used to preliminarily estimate the temperature distribution of the stator core. However, due to the complex and variable operating conditions of hydro-generators, and the significant differences between different models, the accuracy of empirical formulas is somewhat limited.

[0004] Another commonly used method is finite element simulation. Finite element simulation involves establishing a mathematical model, dividing the stator core into multiple small elements, and performing numerical calculations based on boundary conditions and material properties to obtain the temperature distribution of the stator core. This method can simulate the temperature distribution of the stator core relatively accurately, but it requires a significant amount of computational resources and time.

[0005] In recent years, with the development of sensor and data processing technologies, methods for analyzing the temperature distribution of hydro-generator stator cores based on sensor data have gradually attracted attention. This method involves arranging multiple temperature sensors on the stator core to collect temperature data in real time. Through an industrial internet platform, time-series data from all hydropower station equipment is aggregated into a database, and then processed and analyzed using data analysis algorithms provided by the technology platform to obtain the temperature distribution of the stator core. This method has advantages such as real-time performance and high accuracy, enabling better monitoring and control of the hydro-generator's operating status. Summary of the Invention

[0006] The main objective of this invention is to provide a method for analyzing the temperature distribution of the stator core of a hydro-generator, thereby solving the problems of poor real-time performance and insufficient accuracy of existing methods.

[0007] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for analyzing the temperature distribution of the stator core of a hydro-generator, comprising the following steps:

[0008] S1. Arrange multiple temperature sensors on the stator core;

[0009] S2. Real-time temperature data of the stator core is collected using a data acquisition device;

[0010] S3. Process and store the collected temperature data;

[0011] S4. Through the industrial internet platform, the core temperature time series data is aggregated into the database, and then the operating parameters and running status parameters of the unit are preprocessed regularly.

[0012] S5. Analyze the processed temperature data using data analysis algorithms;

[0013] S6. Generate a temperature distribution diagram and a temperature change trend diagram of the stator core based on the analysis results;

[0014] S7. Based on the core temperature distribution and changing trend, diagnose and issue early warning events and abnormal events;

[0015] S8. Based on the manual determination of early warning events and the special analysis of abnormal events, conclusions and handling suggestions are given. Finally, based on the previous statistics and analysis, suggestions for equipment operation and maintenance are put forward, including suggestions for equipment operation and equipment maintenance.

[0016] In the preferred embodiment, the temperature sensor is connected to the data acquisition device via wired or wireless means.

[0017] In the preferred embodiment, the data processing steps include data filtering, noise reduction, and calibration.

[0018] In the preferred embodiment, the data analysis algorithm includes steps of extracting temperature distribution features and predicting and analyzing temperature change trends;

[0019] Median analysis was used to analyze the core temperatures of all units.

[0020] By using an industrial internet platform, the median core temperature of all units is calculated within the current analysis period. The median core temperatures of multiple units are compared and analyzed, and month-on-month and year-on-year analyses are added. This allows us to analyze the differences between the overall core temperature and other units of the same / different / adjacent / non-adjacent type in the power plant.

[0021] This analysis is not only helpful in discovering the overall temperature changes of components, but also in discovering the overall temperature differences between components of different units. For components with large differences, subsequent special analyses such as manual analysis can be carried out.

[0022] In the preferred scheme, the analytical algorithm for extracting temperature distribution features is a combination of singular value decomposition and wavelet transform. This step includes:

[0023] A1. First, collect and organize the core temperature data of all units to form a temperature dataset. ,in Indicates the first A collection of temperature values ​​for the unit at different time points. The total number of units. For time points;

[0024] Temperature dataset The matrix formed Perform singular value decomposition. ,in and It is an orthogonal matrix. It is a diagonal matrix, and the elements on its diagonal are singular values. Before selection... The left and right singular vectors corresponding to the larger singular values ​​form a new matrix. and Then the temperature distribution characteristic matrix ;

[0025] A2. Then, for the temperature time series collected by each sensor, wavelet transform is used to further extract local features; let the temperature change over time be... The formula for calculating wavelet coefficients is:

[0026] ;

[0027] in It is the average temperature. These are wavelet basis functions. These are wavelet coefficients. and These are the scale and translation parameters, respectively.

[0028] A3. Combine the wavelet coefficients with the feature matrix obtained from singular value decomposition to obtain the final temperature distribution feature matrix. ,in It is a weighting function used to adjust the contribution of wavelet coefficients in the feature matrix.

[0029] In the preferred scheme, the core temperature is analyzed using median analysis. The method for calculating the median temperature is as follows:

[0030] B1. Calculate the median temperature of all units during the current analysis period. First, the temperature data of all units are merged into one set. ,in This represents the total number of all temperature data.

[0031] Then to Sort, if If it is an odd number, then the median ;like If it is even, then the median ;

[0032] B2. For each unit, calculate its median temperature during the current analysis period. Similarly, the temperature data for each unit are sorted, and the median is calculated separately for odd and even numbers.

[0033] formula: , ;in, It is the median temperature of all units. It is the first The median temperature of the unit. This indicates the position after sorting in ascending order. Temperature value at location, It is the total number of all temperature data. It is the first Total temperature data for each unit;

[0034] The purpose of the formula is to more accurately reflect the central trend of temperature data by calculating the median, thus resisting the influence of outliers.

[0035] In the preferred scheme, the core temperature is analyzed using median analysis. The median core temperatures of multiple units are compared and analyzed. The median comparison analysis method is as follows:

[0036] C1. Conduct comparative analysis and calculate the difference value. Simultaneously, both month-on-month and year-on-year analyses are performed. The month-on-month analysis compares the temperature with the median temperature of the previous period to calculate the month-on-month growth rate. ;

[0037] Year-on-year analysis compares the temperature with the median temperature of the same period last year to calculate the year-on-year growth rate. ;

[0038] C2. Standardized difference is used to measure the degree of difference between two sets of data. The calculation formula is:

[0039] ;

[0040] in It is the first The median of the group of data. It is the average of all medians. It is the standard deviation;

[0041] The purpose of the above formula is to understand the degree of deviation of a unit from the average level through comparative analysis and standardized difference, and to discover potential problematic units.

[0042] In the preferred embodiment, the overall temperature trend among different unit components is predicted based on the overall temperature change of the components. The prediction method includes:

[0043] D1. Combine kernel-based support vector regression and ARIMA models to predict temperature change trends, using a Gaussian kernel function. ,in These are kernel parameters that map temperature time-series data to a high-dimensional feature space, where linear regression is then performed.

[0044] D2. Algorithm Construction and Optimization: ,subject to , , ,in and It is the regression coefficient. and It is a slack variable. It is a penalty parameter. It is an insensitive loss parameter. It is to input data Functions that map to a high-dimensional feature space;

[0045] D3. By solving the optimization problem in step D2, the regression function is obtained. It is used to make preliminary predictions of temperature values ​​at future points in time;

[0046] D4. Simultaneously, the ARIMA model is used to further predict the trend of the temperature series, which is: The expression for the ARIMA model is: ,in It is the predicted temperature value. It is the mean term. It is the autoregressive part. It is the difference order. It is a white noise process;

[0047] D5. Finally, the prediction results of the SVR and ARIMA models are weighted and fused to obtain the final predicted value of temperature change trend.

[0048] In the preferred embodiment, the position of the temperature sensor is rationally arranged according to the structural characteristics and temperature distribution law of the stator core.

[0049] This invention provides a method for analyzing the temperature distribution of the stator core of a hydro-generator. By arranging multiple temperature sensors on the stator core and using a data acquisition device to collect temperature data in real time, and then using an industrial internet platform to aggregate the time-series temperature data into a database for processing and analysis, this method effectively solves the problems of poor real-time performance and insufficient accuracy in existing methods. It can reflect the temperature changes of the stator core in a timely and accurate manner, providing a reliable guarantee for the safe operation of the equipment.

[0050] It not only employs unique data analysis algorithms to analyze processed temperature data, generating stator core temperature distribution maps and temperature change trend maps, but also diagnoses and issues early warning events and abnormal events based on the core temperature distribution and change trends. Furthermore, it performs manual verification and thematic analysis, providing conclusions and handling suggestions, and finally proposes equipment operation and maintenance recommendations. This achieves end-to-end management from data acquisition to analysis, early warning, and processing.

[0051] For the analysis of core temperature in large generators, in addition to the core analytical steps of this method, the method also incorporates year-on-year and month-on-month comparisons of monthly maximum values. When the year-on-year and month-on-month differences are significant, further in-depth analysis can be conducted to more comprehensively grasp the changing trends and anomalies of core temperature.

[0052] Existing methods for analyzing bearing temperature in large generators are limited in scope. This method, however, can not only analyze the maximum temperature deviation of all bearings at any given moment, but also conduct in-depth analysis when the temperature deviation is large. Furthermore, it can analyze the temperature status and trend of the bearings in a spatial dimension, thus overcoming the shortcomings of existing methods and providing richer information for the overall operational status assessment of the generator.

[0053] In summary, this method for analyzing the temperature distribution of the stator core of a hydro-generator has significant benefits in improving real-time performance and accuracy, providing comprehensive data analysis, and expanding the analysis of bearing bearing temperature, thus helping to ensure the safe and stable operation of the hydro-generator. Attached Figure Description

[0054] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0055] Figure 1 This is a table of median core temperature analysis data from the present invention; Detailed Implementation

[0056] Example 1

[0057] A method for analyzing the temperature distribution of the stator core of a hydro-generator includes the following steps:

[0058] S1. Arrange multiple temperature sensors on the stator core;

[0059] S2. Real-time temperature data of the stator core is collected using a data acquisition device;

[0060] S3. Process and store the collected temperature data;

[0061] S4. Through the industrial internet platform, the core temperature time series data is aggregated into the database, and then the operating parameters and running status parameters of the unit are preprocessed periodically.

[0062] S5. Analyze the processed temperature data using data analysis algorithms;

[0063] S6. Generate a temperature distribution diagram and a temperature change trend diagram of the stator core based on the analysis results;

[0064] S7. Based on the core temperature distribution and changing trend, diagnose and issue early warning events and abnormal events;

[0065] S8. Based on the manual determination of early warning events and the special analysis of abnormal events, conclusions and handling suggestions are given. Finally, based on the previous statistics and analysis, suggestions for equipment operation and maintenance are put forward, including suggestions for equipment operation and equipment maintenance.

[0066] In the preferred embodiment, the temperature sensor is connected to the data acquisition device via wired or wireless means.

[0067] In the preferred embodiment, the data processing steps include data filtering, noise reduction, and calibration.

[0068] In the preferred embodiment, the data analysis algorithm includes steps of extracting temperature distribution features and predicting and analyzing temperature change trends;

[0069] Median analysis was used to analyze the core temperatures of all units.

[0070] By using an industrial internet platform, the median core temperature of all units is calculated within the current analysis period. The median core temperatures of multiple units are compared and analyzed, and month-on-month and year-on-year analyses are added. This allows us to analyze the differences between the overall core temperature and other units of the same / different / adjacent / non-adjacent type in the power plant.

[0071] This analysis is not only helpful in discovering the overall temperature changes of components, but also in discovering the overall temperature differences between components of different units. For components with large differences, subsequent special analyses such as manual analysis can be carried out.

[0072] In the preferred scheme, the analytical algorithm for extracting temperature distribution features is a combination of singular value decomposition and wavelet transform. This step includes:

[0073] A1. First, collect and organize the core temperature data of all units to form a temperature dataset. ,in Indicates the first A collection of temperature values ​​for the unit at different time points. The total number of units. For time points;

[0074] Temperature dataset The matrix formed Perform singular value decomposition. ,in and It is an orthogonal matrix. It is a diagonal matrix, and the elements on its diagonal are singular values. Before selection... The left and right singular vectors corresponding to the larger singular values ​​form a new matrix. and Then the temperature distribution characteristic matrix ;

[0075] A2. Then, for the temperature time series collected by each sensor, wavelet transform is used to further extract local features; let the temperature change over time be... The formula for calculating wavelet coefficients is:

[0076] ;

[0077] in It is the average temperature. These are wavelet basis functions. These are wavelet coefficients. and These are the scale and translation parameters, respectively.

[0078] A3. Combine the wavelet coefficients with the feature matrix obtained from singular value decomposition to obtain the final temperature distribution feature matrix. ,in It is a weighting function used to adjust the contribution of wavelet coefficients in the feature matrix.

[0079] formula: , , , .in, It is a matrix composed of temperature data. , and These are the matrices obtained from singular value decomposition and the diagonal matrix. The selected first A matrix consisting of the left singular vectors corresponding to the larger singular values. It is the temperature distribution characteristic matrix obtained from singular value decomposition. These are wavelet coefficients. These are wavelet basis functions. and These are the scale and translation parameters, It is the final temperature distribution feature matrix. It is a weighting function. The purpose of this formula is to extract the main and local features of temperature data by combining singular value decomposition and wavelet transform, thereby reducing the data dimensionality and capturing both the local and global characteristics of temperature distribution.

[0080] In the preferred scheme, the core temperature is analyzed using median analysis. The method for calculating the median temperature is as follows:

[0081] B1. Calculate the median temperature of all units during the current analysis period. First, the temperature data of all units are merged into one set. ,in This represents the total number of all temperature data.

[0082] Then to Sort, if If it is an odd number, then the median ;like If it is even, then the median ;

[0083] B2. For each unit, calculate its median temperature during the current analysis period. Similarly, the temperature data for each unit are sorted, and the median is calculated separately for odd and even numbers.

[0084] formula: , ;in, It is the median temperature of all units. It is the first The median temperature of the unit. This indicates the position after sorting in ascending order. Temperature value at location, It is the total number of all temperature data. It is the first Total temperature data for each unit;

[0085] The purpose of the formula is to more accurately reflect the central trend of temperature data by calculating the median, thus resisting the influence of outliers.

[0086] In the preferred scheme, the core temperature is analyzed using median analysis. The median core temperatures of multiple units are compared and analyzed. The median comparison analysis method is as follows:

[0087] C1. Conduct comparative analysis and calculate the difference value. Simultaneously, both month-on-month and year-on-year analyses are performed. The month-on-month analysis compares the temperature with the median temperature of the previous period to calculate the month-on-month growth rate. ;

[0088] Year-on-year analysis compares the temperature with the median temperature of the same period last year to calculate the year-on-year growth rate. ;

[0089] C2. Standardized difference is used to measure the degree of difference between two sets of data. The calculation formula is:

[0090] ;

[0091] in It is the first The median of the group of data. It is the average of all medians. It is the standard deviation;

[0092] formula: , , , .in, It is the difference value. and These are the month-on-month growth rate and the year-on-year growth rate, respectively. It is the standardized difference. It is the median temperature of all units. It is the first The median temperature of the unit. It is the average of all medians. It is the standard deviation. The purpose of this formula is to understand the degree of deviation of a unit from the average level through comparative analysis and standardized deviation, and to discover potential problem units.

[0093] In the preferred embodiment, the overall temperature trend among different unit components is predicted based on the overall temperature change of the components. The prediction method includes:

[0094] D1. Combine kernel-based support vector regression and ARIMA models to predict temperature change trends, using a Gaussian kernel function. ,in These are kernel parameters that map temperature time-series data to a high-dimensional feature space, where linear regression is then performed.

[0095] D2. Algorithm Construction and Optimization: ,subject to , , ,in and It is the regression coefficient. and It is a slack variable. It is a penalty parameter. It is an insensitive loss parameter. It is to input data Functions that map to a high-dimensional feature space;

[0096] D3. By solving the optimization problem in step D2, the regression function is obtained. It is used to make preliminary predictions of temperature values ​​at future points in time;

[0097] D4. Simultaneously, the ARIMA model is used to further predict the trend of the temperature series, which is: The expression for the ARIMA model is: ,in It is the predicted temperature value. It is the mean term. It is the autoregressive part. It is the difference order. It is a white noise process;

[0098] D5. Finally, the prediction results of the SVR and ARIMA models are weighted and fused to obtain the final predicted value of temperature change trend.

[0099] formula: , ,subject to , , , , .in, It is a kernel function used to map data to a high-dimensional feature space; the optimization problem is used to solve for regression coefficients. It is the regression function of SVR, used to predict temperature values; This is the temperature value predicted by the ARIMA model. It is the mean term. It is the autoregressive part. It is the difference order. It is a white noise process. The advantage of this formula is that by combining SVR and ARIMA models, it can fully utilize the advantages of both methods to improve the accuracy of temperature change trend prediction.

[0100] In the preferred embodiment, the position of the temperature sensor is rationally arranged according to the structural characteristics and temperature distribution law of the stator core.

[0101] Example 2

[0102] Further explanation in conjunction with Example 1, such as Figure 1The structure shown illustrates the analysis of maximum stator core temperatures exceeding limits for all units. Using an industrial internet platform, the maximum stator core temperature for the current month for all units in all power plants is calculated. Bearing components exceeding the stator core temperature threshold are listed as equipment experiencing abnormal events, allowing for subsequent manual analysis and other specialized analyses.

[0103] Median analysis of stator core temperatures for all units. Using an industrial internet platform, the median stator core temperature for the current month was calculated for all units across all power plants. The median stator core temperature of the same unit from multiple units within a single power plant was compared and analyzed, incorporating month-on-month and year-on-year comparisons. Figure 1 As shown, Figure 1 In the diagram, 1F~32F represent the unit number, and the X-axis represents the temperature value. This allows for a direct analysis of the differences in the overall stator core temperature between the stator core and other units of the same or different models in the power plant, as well as the differences between the overall stator core temperature and adjacent or non-adjacent units. When the differences are significant, such as a temperature difference exceeding 5°C, further in-depth analysis, such as manual analysis, is conducted.

[0104] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for analyzing the temperature distribution of the stator core of a hydro-generator, characterized by: Includes the following steps: S1. Arrange multiple temperature sensors on the stator core; S2. Real-time temperature data of the stator core is collected using a data acquisition device; S3. Process and store the collected temperature data; S4. Through the industrial internet platform, the core temperature time series data is aggregated into the database, and then the operating parameters and running status parameters of the unit are preprocessed regularly. S5. Analyze the processed temperature data using data analysis algorithms; S6. Generate a temperature distribution diagram and a temperature change trend diagram of the stator core based on the analysis results; S7. Based on the core temperature distribution and changing trend, diagnose and issue early warning events and abnormal events; S8. Based on the manual determination of early warning events and the special analysis of abnormal events, conclusions and handling suggestions are given. Finally, based on the previous statistics and analysis, suggestions for equipment operation and maintenance are put forward, including suggestions for equipment operation and equipment maintenance. The data analysis algorithm includes steps such as temperature distribution feature extraction and temperature change trend prediction and analysis. Median analysis was used to analyze the core temperatures of all units. By using an industrial internet platform, the median core temperature of all units is calculated within the current analysis period. The median core temperatures of multiple units are compared and analyzed, and month-on-month and year-on-year analyses are added. This allows us to analyze the differences between the overall core temperature and other units of the same / different / adjacent / non-adjacent type in the power plant. This analysis is not only helpful in discovering the overall temperature changes of components, but also in discovering the overall temperature differences between components of different units. For components with large differences, subsequent manual analysis or special analysis can be carried out. The analytical algorithm for extracting temperature distribution features employs a combination of singular value decomposition and wavelet transform. This process includes the following steps: A1. First, collect and organize the core temperature data of all units to form a temperature dataset. ,in Indicates the first A collection of temperature values ​​for the unit at different time points. The total number of units. For time points; Temperature dataset The matrix formed Perform singular value decomposition. ,in and It is an orthogonal matrix. It is a diagonal matrix whose diagonal elements are singular values. Selecting the first... The left and right singular vectors corresponding to the larger singular values ​​form a new matrix. and Then the temperature distribution characteristic matrix ; A2. Then, for the temperature time series collected by each sensor, wavelet transform is used to further extract local features; let the temperature change over time be... The formula for calculating wavelet coefficients is: ； in It is the average temperature. These are wavelet basis functions. These are wavelet coefficients. and These are the scale and translation parameters, respectively. A3. Combine the wavelet coefficients with the feature matrix obtained from singular value decomposition to obtain the final temperature distribution feature matrix. ,in It is a weighting function used to adjust the contribution of wavelet coefficients in the feature matrix.

2. The method for analyzing the temperature distribution of the stator core of a hydro-generator according to claim 1, characterized in that: The temperature sensor is connected to the data acquisition device via wired or wireless means.

3. The method for analyzing the temperature distribution of the stator core of a hydro-generator according to claim 1, characterized in that: The data processing steps include data filtering, noise reduction, and calibration.

4. The method for analyzing the temperature distribution of the stator core of a hydro-generator according to claim 1, characterized in that: Median analysis was used to analyze the core temperature. The median core temperatures of multiple units were compared and analyzed. The median comparison analysis method was as follows: C1. Conduct comparative analysis and calculate the difference value. Simultaneously, both month-on-month and year-on-year analyses are performed. The month-on-month analysis compares the temperature with the median temperature of the previous period to calculate the month-on-month growth rate. ； Year-on-year analysis compares the temperature with the median temperature of the same period last year to calculate the year-on-year growth rate: ； in, It is the first The median temperature of the unit during the current analysis period. It is the first The median temperature of the unit in the previous time period. It is the first The median temperature of the Taiwanese units during the same period last year; C2. Standardized difference is used to measure the degree of difference between two sets of data. The calculation formula is: ； in It is the first The median of the group of data. It is the average of all medians. It is the standard deviation; The purpose of the above formula is to understand the degree of deviation of a unit from the average level through comparative analysis and standardized difference, and to discover potential problematic units.

5. The method for analyzing the temperature distribution of the stator core of a hydro-generator according to claim 1, characterized in that: Median analysis was used to analyze the core temperature. The median temperature was calculated as follows: B1. Calculate the median temperature of all units during the current analysis period. First, the temperature data of all units are merged into one set. ,in This represents the total number of all temperature data. Then to Sort, if If it is an odd number, then the median ;like If it is even, then the median ; B2. For each unit, calculate its median temperature during the current analysis period. Similarly, the temperature data for each unit are sorted, and the median is calculated separately for odd and even numbers. formula: , ;in, It is the median temperature of all units. It is the first The median temperature of the unit. This indicates the position after sorting in ascending order. Temperature value at location, It is the total number of all temperature data. It is the first Total temperature data for each unit; The purpose of the formula is to more accurately reflect the central trend of temperature data by calculating the median, thus resisting the influence of outliers.

6. The method for analyzing the temperature distribution of the stator core of a hydro-generator according to claim 1, characterized in that: The overall temperature trend among different unit components is predicted based on the overall temperature change of the components. The prediction methods include: D1. Combine kernel-based support vector regression and ARIMA models to predict temperature change trends, using a Gaussian kernel function. ,in These are kernel parameters that map temperature time-series data to a high-dimensional feature space, where linear regression is then performed. D2. Algorithm Construction and Optimization: ,subject to , , ,in It is a weight vector. It is a bias term. and It is a slack variable. It is a penalty parameter. It is an insensitive loss parameter. It is to input data Functions that map to a high-dimensional feature space; D3. By solving the optimization problem in step D2, the regression function is obtained. It is used to make preliminary predictions of temperature values ​​at future points in time; D4. Simultaneously, the ARIMA model is used to further predict the trend of the temperature series, which is: The expression for the ARIMA model is: ,in It is the predicted temperature value. It is the mean term. It is the autoregressive part. It is the difference order. It is a white noise process; D5. Finally, the prediction results of the SVR and ARIMA models are weighted and fused to obtain the final predicted value of temperature change trend.

7. The method for analyzing the temperature distribution of the stator core of a hydro-generator according to claim 1, characterized in that: The location of the temperature sensor is rationally arranged based on the structural characteristics and temperature distribution law of the stator core.

Images