A method for predicting main transformer oil temperature out-of-limit

Abstract

The application discloses a kind of methods for predicting the oil temperature of main transformer to cross the border, comprising the following steps: step one, data analysis;Step two, feature screening;Step three, matrix generation;Step four, model optimization;Step five, oil temperature prediction;Step six, cross-border prediction;The model constructed by adopting Lasso regression algorithm is used to predict the oil temperature of main transformer, so that the safety of main transformer is controlled, the accuracy of main transformer oil temperature prediction is improved, the accuracy of main transformer oil temperature cross-border judgment is improved, the risk of main transformer oil temperature cross-border is reduced, the problems of brittle rupture of insulation paperboard inside main transformer, serious aging of winding insulation and serious deterioration of insulating oil are avoided, the safety and stability of main transformer are improved, the service life of main transformer is prolonged, and the prediction model constructed by algorithm can test and evaluate the prediction performance and training optimization by loading new data, continuously improve the accuracy of prediction.

Description

Technical Field

[0001] This invention relates to the field of transformer oil temperature anomaly identification technology, specifically a method for predicting main transformer oil temperature exceeding the limit. Background Technology

[0002] The safety of substation operation is crucial to the safety of production in various industries and the safety and property of the general public, thus attracting widespread social attention. Substation operation is the source of power grid safety and the forefront of power grid operation and management. The temperature of the main transformer oil directly affects the safety and stability of the main transformer. If the main transformer oil temperature rises abnormally, it may cause damage to transformer equipment and endanger personal safety. Unsafe transformer operation will seriously affect enterprise development and the improvement of social and economic benefits. The main transformer oil temperature is affected by factors such as load current, voltage, and temperature. Since many factors such as temperature and load current are random, and the transformer power is constantly changing, the transformer oil temperature is also constantly changing during operation. These factors interact with each other, making accurate prediction of oil temperature quite difficult.

[0003] However, there is currently little research on methods for predicting transformer oil temperature in the market, making it difficult to fully control the safety of the transformer. During the use of the transformer, the temperature may easily exceed the limit, causing the internal insulation paperboard to become brittle and crack, the winding insulation to age and the insulating oil to deteriorate, losing its proper insulation function, inevitably shortening the service life of the transformer, and significantly reducing its safety and stability.

[0004] Therefore, it is necessary to design a method to predict the main transformer oil temperature exceeding the limit. Summary of the Invention

[0005] The purpose of this invention is to provide a method for predicting the over-limit of main transformer oil temperature, so as to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for predicting the oil temperature exceeding the limit of a main transformer, comprising the following steps: Step 1, data analysis; Step 2, feature selection; Step 3, matrix generation; Step 4, model optimization; Step 5, oil temperature prediction; Step 6, limit exceeding prediction;

[0007] In step one above, data on the changes in oil temperature of a single device in summer, the distribution of oil temperature at various time points in summer, the changes in oil temperature of a single device in winter, and the distribution of oil temperature at various time points in winter are used to analyze the relationship between current and oil temperature, the relationship between current load rate and oil temperature, the relationship between climate temperature and oil temperature, and the relationship between various attributes and oil temperature, so as to obtain the relevant data characteristics of oil temperature and various features.

[0008] In step two above, the relevant data characteristics obtained in step one are analyzed using data such as the main transformer oil temperature telemetry history, main transformer ledger, current load, historical meteorological information, main transformer manufacturer information, main transformer defect process and weather forecast. Useful features with high correlation to main transformer oil temperature are selected and processed into a wide table of useful features.

[0009] In step three above, the Pearson correlation coefficient is used to calculate the correlation coefficient between oil temperature and n useful features. The calculation formula is as follows, where X represents the vector value of oil temperature and Y represents the vector value of a feature:

[0010]

[0011] The Pearson coefficient ranges from -1 to 1. The closer the absolute value of the correlation coefficient is to 1, the more correlated the two attributes are. The sign of the value indicates whether the two attributes are positively or negatively correlated. A positive value indicates a positive correlation and a negative value indicates a negative correlation. A 1*n correlation coefficient matrix is ​​generated to verify the correctness of the data in the useful feature wide table in step two and to generate a correlation coefficient report for production personnel to use.

[0012] In step four above, 70% of the data from the useful feature wide table obtained in step two, representing a specific time period, is used to construct an initial model using the Lasso regression algorithm. The calculation formula is as follows, where real represents the true value and predict represents the predicted value. The average value represents the true observed values:

[0013] Root mean square error:

[0014]

[0015] Average percentage error:

[0016]

[0017] Goodness of fit:

[0018]

[0019] The remaining 30% of the data was used to test the initial model, using RMSE, MAPE, and R... 2 Evaluation metrics assess the quality of the initial model. RMSE represents the root mean square error; a smaller value indicates a better model. R... 2 The value represents the goodness of fit; the larger the value, the better the model. MAPE represents the mean absolute percentage error; the smaller the value, the better the model. After multiple training sessions, an optimized oil temperature prediction model is obtained.

[0020] In step five above, the latest data of each useful feature in the last 6 hours are loaded into the oil temperature prediction model optimized in step three to obtain the oil temperature prediction value of each main transformer in the next 6 hours.

[0021] In step six above, the predicted oil temperature value obtained in step five is compared with the preset oil temperature threshold and the oil temperature exceeding threshold to obtain the predicted oil temperature exceeding threshold of the main transformer in the next 6 hours.

[0022] According to the above technical solution, in step three, the Pearson coefficient ranges from [-1, 1]. The closer the absolute value of the correlation coefficient is to 1, the more correlated the two attributes are. The sign of the value indicates whether the two attributes are positively or negatively correlated. A positive value indicates a positive correlation, and a negative value indicates a negative correlation.

[0023] According to the above technical solution, in step four, RMSE represents the root mean square error; the smaller the value, the better the model. 2 The value indicates the goodness of fit; a larger value indicates a better model. MAPE represents the mean absolute percentage error; a smaller value indicates a better model.

[0024] According to the above technical solution, in step six, when the maximum value of the predicted oil temperature is less than the preset oil temperature threshold, the oil temperature is considered to be normal for the next 6 hours.

[0025] According to the above technical solution, in step six, when the maximum value of the predicted oil temperature is greater than the preset oil temperature threshold but less than the oil temperature exceeding the limit threshold, it is considered that the oil temperature in the next 6 hours has reached the warning value but has not exceeded the limit.

[0026] According to the above technical solution, in step six, if the maximum value of the predicted oil temperature is greater than the oil temperature exceeding the threshold, then the oil temperature is considered to be exceeding the limit in the next 6 hours.

[0027] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: The method for predicting transformer oil temperature exceeding the limit uses the Lasso regression algorithm to construct a transformer oil temperature prediction model from the transformer's relevant feature data. This facilitates personnel's comprehensive control over the transformer's safety status, improves the accuracy of transformer oil temperature prediction, enhances the accuracy of transformer oil temperature exceeding the limit judgment, reduces the risk of transformer oil temperature exceeding the limit, avoids problems such as brittle and cracked insulation paperboard inside the transformer, severe aging of winding insulation, and severe deterioration of insulating oil, improves the safety and stability of the transformer, extends the transformer's service life, and the algorithm prediction model can continuously improve the accuracy of prediction by loading new data for prediction performance testing, evaluation, training, and optimization. Attached Figure Description

[0028] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0029] Figure 1 This is a flowchart of the method of the present invention;

[0030] Figure 2 This is a graph showing the relationship between current and oil temperature in an embodiment of the present invention;

[0031] Figure 3 This is a graph showing the relationship between current load rate and oil temperature in an embodiment of the present invention;

[0032] Figure 4 This is a graph showing the relationship between climate temperature and oil temperature in an embodiment of the present invention;

[0033] Figure 5 This is a correlation diagram between various attributes and oil temperature in the embodiments of the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] Please see Figure 1-5 This invention provides a technical solution: a method for predicting the oil temperature exceeding the limit in a main transformer, comprising the following steps: Step 1, data analysis; Step 2, feature selection; Step 3, matrix generation; Step 4, model optimization; Step 5, oil temperature prediction; Step 6, limit prediction.

[0036] In step one above, data on the changes in oil temperature of a single device in summer, the distribution of oil temperature at various time points in summer, the changes in oil temperature of a single device in winter, and the distribution of oil temperature at various time points in winter are used to analyze the relationship between current and oil temperature, the relationship between current load rate and oil temperature, the relationship between climate temperature and oil temperature, and the relationship between various attributes and oil temperature, so as to obtain the relevant data characteristics of oil temperature and various features.

[0037] In step two above, the relevant data characteristics obtained in step one are analyzed using data such as the main transformer oil temperature telemetry history, main transformer ledger, current load, historical meteorological information, main transformer manufacturer information, main transformer defect process and weather forecast. Useful features with high correlation to main transformer oil temperature are selected and processed into a wide table of useful features.

[0038] In step three above, the Pearson correlation coefficient is used to calculate the correlation coefficient between oil temperature and n useful features. The calculation formula is as follows, where X represents the vector value of oil temperature and Y represents the vector value of a feature:

[0039]

[0040] The Pearson coefficient ranges from -1 to 1. The closer the absolute value of the correlation coefficient is to 1, the more correlated the two attributes are. The sign of the value indicates whether the two attributes are positively or negatively correlated. A positive value indicates a positive correlation and a negative value indicates a negative correlation. A 1*n correlation coefficient matrix is ​​generated to verify the correctness of the data in the useful feature wide table in step two and to generate a correlation coefficient report for production personnel to use.

[0041] In step four above, 70% of the data from the useful feature wide table obtained in step two, representing a specific time period, is used to construct an initial model using the Lasso regression algorithm. The calculation formula is as follows, where real represents the true value and predict represents the predicted value. The average value represents the true observed values:

[0042] Root mean square error:

[0043]

[0044] Average percentage error:

[0045]

[0046] Goodness of fit:

[0047]

[0048] The remaining 30% of the data was used to test the initial model, using RMSE, MAPE, and R... 2 Evaluation metrics assess the quality of the initial model. RMSE represents the root mean square error; a smaller value indicates a better model. R... 2 The value represents the goodness of fit; the larger the value, the better the model. MAPE represents the mean absolute percentage error; the smaller the value, the better the model. After multiple training sessions, an optimized oil temperature prediction model is obtained.

[0049] In step five above, the latest data of each useful feature in the last 6 hours are loaded into the oil temperature prediction model optimized in step three to obtain the oil temperature prediction value of each main transformer in the next 6 hours.

[0050] In step six above, the oil temperature prediction value obtained in step five is compared with the preset oil temperature threshold and the oil temperature over-limit threshold. When the maximum value of the oil temperature prediction value is less than the preset oil temperature threshold, the oil temperature is considered to be normal for the next 6 hours. When the maximum value of the oil temperature prediction value is greater than the preset oil temperature threshold but less than the oil temperature over-limit threshold, the oil temperature is considered to have reached the warning value but not exceeded the limit for the next 6 hours. When the maximum value of the oil temperature prediction value is greater than the oil temperature over-limit threshold, the oil temperature is considered to have exceeded the limit for the next 6 hours. The oil temperature over-limit prediction result of the main transformer for the next 6 hours is obtained.

[0051] The raw data required from 2016 to 2018 were used to predict the main transformer oil temperature using this method. The raw data and predicted data were then compiled and the resulting table of common attribute characteristics of abnormal oil temperature conditions is as follows:

[0052]

[0053]

[0054] The training test results are shown in the table below:

[0055]

[0056]

[0057]

[0058]

[0059] The results of the oil temperature prediction are shown in the table below:

[0060]

[0061]

[0062] The following table shows examples of the predicted results:

[0063]

[0064]

[0065] Based on the above, the advantages of this invention are as follows: This invention uses the Lasso regression algorithm to construct a main transformer oil temperature prediction model from the main transformer's relevant feature data, thereby facilitating personnel to have a comprehensive grasp of the main transformer's safety status, improving the accuracy of main transformer oil temperature prediction, enhancing the accuracy of main transformer oil temperature exceeding limits judgment, reducing the risk of main transformer oil temperature exceeding limits, avoiding problems such as the main transformer's internal insulation paperboard becoming brittle and cracking, severe aging of winding insulation, and severe deterioration of insulating oil, improving the safety and stability of the main transformer, extending the main transformer's service life, and the algorithm prediction model can continuously improve the prediction accuracy by loading new data for prediction performance testing, evaluation, training, and optimization.

[0066] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0067] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting excessive oil temperature in a main transformer, comprising the following steps: Step 1, Data Analysis; Step 2, Feature Selection; Step 3, Matrix Generation; Step 4, Model Optimization; Step 5, Oil Temperature Prediction; Step 6, Boundary Exceeding Prediction; Its key feature is: In step one above, data on the changes in oil temperature of a single device in summer, the distribution of oil temperature at various time points in summer, the changes in oil temperature of a single device in winter, and the distribution of oil temperature at various time points in winter are used to analyze the relationship between current and oil temperature, the relationship between current load rate and oil temperature, the relationship between climate temperature and oil temperature, and the relationship between various attributes and oil temperature, so as to obtain the relevant data characteristics of oil temperature and various features. In step two above, the relevant data characteristics obtained in step one are analyzed using data such as the main transformer oil temperature telemetry history, main transformer ledger, current load, historical meteorological information, main transformer manufacturer information, main transformer defect process and weather forecast. Useful features with high correlation to main transformer oil temperature are selected and processed into a wide table of useful features. In step three above, the Pearson correlation coefficient is used to calculate the correlation coefficient between oil temperature and n useful features. The calculation formula is as follows, where X represents the vector value of oil temperature and Y represents the vector value of a feature: The Pearson coefficient ranges from -1 to 1. The closer the absolute value of the correlation coefficient is to 1, the more correlated the two attributes are. The sign of the value indicates whether the two attributes are positively or negatively correlated. A positive value indicates a positive correlation and a negative value indicates a negative correlation. A 1*n correlation coefficient matrix is ​​generated to verify the correctness of the data in the useful feature wide table in step two and to generate a correlation coefficient report for production personnel to use. In step four above, 70% of the data from the useful feature wide table obtained in step two, representing a specific time period, is used to construct an initial model using the Lasso regression algorithm. The calculation formula is as follows, where real represents the true value and predict represents the predicted value. The average value represents the true observed values: Root mean square error: Average percentage error: Goodness of fit: The remaining 30% of the data was used to test the initial model, using RMSE, MAPE, and R... 2 Evaluation metrics assess the quality of the initial model. RMSE represents the root mean square error; a smaller value indicates a better model. R... 2 The value represents the goodness of fit; the larger the value, the better the model. MAPE represents the mean absolute percentage error; the smaller the value, the better the model. After multiple training sessions, an optimized oil temperature prediction model is obtained. In step five above, the latest data of each useful feature in the last 6 hours are loaded into the oil temperature prediction model optimized in step three to obtain the oil temperature prediction value of each main transformer in the next 6 hours. In step six above, the predicted oil temperature value obtained in step five is compared with the preset oil temperature threshold and the oil temperature exceeding threshold to obtain the predicted oil temperature exceeding threshold of the main transformer in the next 6 hours.

2. The method for predicting transformer oil temperature exceeding the limit according to claim 1, characterized in that: In step three, the Pearson coefficient ranges from -1 to 1. The closer the absolute value of the correlation coefficient is to 1, the more correlated the two attributes are. The sign of the value indicates whether the two attributes are positively or negatively correlated. A positive value indicates a positive correlation, and a negative value indicates a negative correlation.

3. The method for predicting transformer oil temperature exceeding the limit according to claim 1, characterized in that: In step four, RMSE represents the root mean square error; the smaller the value, the better the model. 2 The value indicates the goodness of fit; a larger value indicates a better model. MAPE represents the mean absolute percentage error; a smaller value indicates a better model.

4. The method for predicting transformer oil temperature exceeding the limit according to claim 1, characterized in that: In step six, if the maximum value of the predicted oil temperature is less than the preset oil temperature threshold, the oil temperature is considered to be normal for the next 6 hours.

5. The method for predicting transformer oil temperature exceeding the limit according to claim 1, characterized in that: In step six, when the maximum value of the predicted oil temperature is greater than the preset oil temperature threshold but less than the oil temperature exceeding the limit threshold, it is considered that the oil temperature in the next 6 hours has reached the warning value but has not exceeded the limit.

6. The method for predicting transformer oil temperature exceeding the limit according to claim 1, characterized in that: In step six, if the maximum value of the predicted oil temperature is greater than the oil temperature exceeding the threshold, then the oil temperature is considered to be exceeding the limit for the next 6 hours.

Images