An intelligent identification method for fault events of temperature transmitter

Abstract

The present invention discloses a method for intelligently identifying fault events of a temperature transmitter, which relates to the technical field of temperature transmitters and comprises a data acquisition module, which is used to collect temperature data of i monitoring points on a winding component at preset time intervals within a unit time t; a data processing module; a feature extraction module, which respectively extracts the first feature data of the i-th monitoring point, generates an abnormal monitoring point based on the first feature data of the i-th monitoring point, and collects the second feature data of the abnormal monitoring point; and a model prediction module. The method for intelligently identifying fault events of a temperature transmitter provided by the present invention can directly predict the fault type of the winding based on the second feature data input into a trained machine learning model, thereby overcoming the defects of the temperature transmitter in the prior art, which relies solely on collecting temperature changes to make fault judgments, has insufficient judgment accuracy, and secondly cannot identify the fault type.

Description

Technical Field

[0001] The present invention relates to the technical field of temperature transmitters, and in particular to a method for intelligently identifying fault events of a temperature transmitter. Background Art

[0002] A temperature transmitter is a device that converts physical measurement signals or ordinary electrical signals into standard electrical signal outputs or outputs in the form of communication protocols. It is an instrument that converts temperature variables into transmittable standardized output signals. It is mainly used for the measurement and control of temperature parameters in industrial processes.

[0003] A Chinese patent with authorization announcement number CN106197729B discloses a high-precision temperature transmitter, including a temperature sensor, a signal processing circuit and a control module connected in sequence, the control module is connected to a wireless signal receiving module, a storage circuit, a power supply circuit, a display device and an external interface, the external interface includes multiple USB interfaces, the signal processing circuit includes a temperature compensation circuit, the signal processing circuit processes the signal measured by the temperature sensor and sends it to the control module, the control module controls the display device to display, the storage circuit stores the temperature signal, and the external interface provides an interface for data access of external devices; the signal processing circuit includes a temperature compensation circuit, and the temperature transmitter has high measurement accuracy.

[0004] As mentioned in the above application, the existing temperature transmitters are generally installed in production equipment, pipelines, containers and other locations to monitor temperature changes and transmit real-time data to the control system to determine whether a fault occurs in the production process, that is, by monitoring the temperature changes of the equipment, it is determined whether the equipment fails. Its single reliance on temperature changes for fault prediction has low accuracy and cannot accurately analyze the type of fault, making it inconvenient to formulate reasonable maintenance methods in a timely and effective manner. Summary of the invention

[0005] In order to solve the above problems, the present invention provides a method for intelligently identifying fault events of a temperature transmitter.

[0006] The present invention adopts the following technical solution, a method for intelligently identifying fault events of a temperature transmitter, wherein the temperature transmitter is installed on a winding component of a transformer, comprising:

[0007] The data acquisition module is a temperature sensor, which is respectively installed at i monitoring points evenly distributed on the winding component inside the transformer. A three-dimensional coordinate system is established with the length and width of the bottom area of ​​the winding component and the height of the winding component to obtain the coordinate data of the i monitoring points. The a temperature sensor is used to collect the temperature data of the i monitoring points on the winding component at preset time intervals within a unit time t; wherein i is a positive integer greater than 1; a=i;

[0008] The data processing module is used to perform data preprocessing on the data collected by the temperature sensor, and then respectively establish a temperature data set for the temperature data collected at i monitoring points, that is, the temperature data set of the i-th monitoring point is: in, It represents the temperature data collected for the nth time at the ith monitoring point, where n represents the number of times the temperature data is collected, wherein the data preprocessing includes data noise reduction, amplification and filtering processing.

[0009] A feature extraction module extracts first feature data of the ith monitoring point, generates an abnormal monitoring point based on the first feature data of the ith monitoring point, and collects second feature data of the abnormal monitoring point;

[0010] The model prediction module inputs the collected second feature data into the trained machine learning model for predicting the winding fault type, predicts the fault type of the winding component, and wirelessly transmits the fault type to the terminal PC.

[0011] As a further description of the above technical solution: the first characteristic data of the i-th monitoring point includes the temperature change rate value of the i-th monitoring point, the temperature fluctuation value of the i-th monitoring point and the average temperature value of the i-th monitoring point.

[0012] As a further description of the above technical solution: the method for obtaining the temperature change rate value of the i-th monitoring point includes:

[0013]

[0014] Where W i is the temperature change rate value of the i-th monitoring point;

[0015] The method for obtaining the average temperature value of the i-th monitoring point includes:

[0016]

[0017] Where, Tpj i is the average temperature value of the i-th monitoring point;

[0018] The method for obtaining the temperature fluctuation value of the i-th monitoring point includes:

[0019]

[0020] Where, BDX i is the temperature fluctuation value of the i-th monitoring point.

[0021] As a further description of the above technical solution: the method for generating an abnormal monitoring point based on the first characteristic data of the i-th monitoring point includes:

[0022] Get the abnormal coefficient of the i-th monitoring point;

[0023]

[0024] In the formula, YCXS i is the abnormal coefficient of the ith monitoring point, and is the weight factor, and and Both are greater than 0.

[0025] Preset abnormal coefficient threshold YC1; Then the i-th monitoring point is marked as an abnormal monitoring point. Otherwise, when The i-th monitoring point is not marked as an abnormal monitoring point.

[0026] As a further description of the above technical solution: the second characteristic data of the abnormal monitoring points include the number of abnormal monitoring points, the average temperature value of m abnormal monitoring points, the average temperature change rate value of m abnormal monitoring points, the average temperature fluctuation value of m abnormal monitoring points, and the distribution relationship of the m abnormal monitoring points on the winding components.

[0027] As a further description of the above technical solution: the number of abnormal monitoring points can be obtained by directly obtaining the number of marked abnormal monitoring points among i monitoring points, and recorded as m abnormal monitoring points.

[0028] As a further description of the above technical solution: the method for obtaining the average temperature value PJm of the m abnormal monitoring points includes:

[0029]

[0030] Where, Tpj m represents the average temperature value of the mth abnormal monitoring point;

[0031] The method for obtaining the average temperature change rate value WBm of m abnormal monitoring points includes:

[0032]

[0033] Where W m Indicates the temperature change rate value of the mth abnormal monitoring point;

[0034] The method for obtaining the average temperature fluctuation value YBDX of m abnormal monitoring points includes:

[0035]

[0036] Where, BDX m Represents the temperature fluctuation value of the mth abnormal monitoring point.

[0037] As a further description of the above technical solution: the method for obtaining the distribution relationship of the m abnormal monitoring points on the winding component includes:

[0038] Obtain coordinate data of m abnormal monitoring points among i monitoring points, m∈{1, 2, ..., i}, extract the abnormal monitoring points, and map the positions of the abnormal monitoring points in the winding component to the blank background layer based on the coordinate data of the abnormal monitoring points to obtain an extraction pattern with the distribution relationship of the m abnormal monitoring points, that is, obtain the distribution relationship of the m abnormal monitoring points on the winding component;

[0039] The distribution relationship includes: local small area distribution, that is, m abnormal monitoring points are distributed on a local small area of ​​the winding, that is, the distribution range of the abnormal monitoring points is less than one quarter of the area of ​​the entire winding component;

[0040] Large-area uniform distribution, that is, the m abnormal monitoring points are evenly distributed over a large area of ​​the entire winding, that is, the distribution range of the abnormal monitoring points is greater than half of the area of ​​the entire winding component;

[0041] Local large-area distribution, that is, m abnormal monitoring points are evenly distributed over a large area of ​​the entire winding, that is, the distribution range of the abnormal monitoring points is between one quarter and one half of the surface area of ​​the entire winding component, including one quarter and one half;

[0042] Example: If m abnormal monitoring points are distributed on the back or side of the winding component, it is a local large-area distribution.

[0043] As a further description of the above technical solution: the training method of the machine learning model for predicting winding fault type includes:

[0044] Collecting a historical training data set of the temperature transmitter, wherein the historical training data set includes the collected second characteristic data and the winding component fault type corresponding to the second characteristic data;

[0045] Convert the collected historical training data into a corresponding set of feature vectors;

[0046] Each group of feature vectors is used as the input of the machine learning model, the machine learning model takes the winding component failure type corresponding to each group of second feature data as the output, the winding component failure type actually corresponding to each group of second feature data is used as the prediction target, and minimizing the loss function value of the machine learning model is used as the training target; training is stopped when the loss function value of the machine learning model is less than or equal to the preset target loss value.

[0047] As a further description of the above technical solution: the winding fault types include: overload fault, short circuit fault, poor ventilation fault and poor lubrication fault;

[0048] It should be noted that the above-mentioned different fault types have different areas (equivalent to the number of abnormal monitoring points) and regions (i.e. the distribution positions of abnormal monitoring points) that cause overheating of the winding components, as well as the temperature rise speed (i.e. the temperature change rate) and temperature fluctuation of the winding components;

[0049] Example: Overload fault is caused by overloading the transformer, which will increase the winding current and generate more heat, causing the temperature of the entire winding to rise normally;

[0050] Short circuit fault, that is, a short circuit inside the winding will cause the current to increase abnormally, generate a lot of heat, and cause the temperature of the local area of ​​the winding to rise rapidly;

[0051] Poor ventilation fault. The transformer is poorly ventilated and cannot effectively dissipate heat, which will cause the winding temperature to rise. Poor ventilation may be caused by fan failure, vent blockage, etc. Poor ventilation causes the winding temperature to rise slowly, and the local large area of ​​the windward side of the winding has the characteristic of slow temperature generation;

[0052] Poor lubrication failure, that is, poor lubrication inside the winding will increase mechanical friction, causing a slow increase in local temperature over a small area.

[0053] Beneficial effects:

[0054] The present invention provides a method for intelligently identifying fault events of a temperature transmitter. By setting i evenly distributed monitoring points on the winding components inside the transformer, the surface temperature of the winding components is monitored, so that the number of abnormal monitoring points on the entire winding component, the average temperature value of the abnormal monitoring points, the average temperature change rate value of the abnormal monitoring points, the average temperature fluctuation value of the abnormal monitoring points, and the distribution relationship of the abnormal monitoring points on the winding components can be obtained. Then, based on these feature data, they are input into a trained machine learning model, so that the fault type of the winding can be directly predicted, thereby overcoming the problem that the temperature transmitter in the prior art relies solely on collecting temperature changes to make fault judgments, which, on the one hand, has insufficient judgment accuracy, and secondly, cannot identify the fault type, and is not convenient for timely and reasonable formulation of maintenance methods. Therefore, the method for intelligently identifying fault events of a temperature transmitter can not only realize high-precision judgment of whether a winding fault occurs, but also identify the fault type, so that maintenance personnel can formulate maintenance plans in a timely manner according to the fault type. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The present invention will be further explained below in conjunction with the accompanying drawings and embodiments:

[0056] Figure 1 A module diagram of a method for intelligently identifying fault events of a temperature transmitter provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0057] In order to make the technical means, creative features, objectives and effects of the present invention easy to understand, the present invention is further described below with reference to specific diagrams. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other without conflict.

[0058] Example 1

[0059] See also Figure 1 The embodiment of the present invention provides a technical solution: a method for intelligently identifying fault events of a temperature transmitter, wherein the temperature transmitter is installed on a winding component of a transformer and is applied to temperature monitoring of the winding component of the transformer, comprising:

[0060] The data acquisition module comprises a temperature sensors, which are respectively installed at i monitoring points evenly distributed on the winding components inside the transformer. A three-dimensional coordinate system is established with the length and width of the bottom area of ​​the winding components and the height of the winding components to obtain the coordinate data of the i monitoring points. The a temperature sensors are used to collect the temperature data of the i monitoring points on the winding components at preset time intervals within a unit time t; wherein i is a positive integer greater than 1; preferably, the preset time interval is 1S, the unit time t is 1 minute, and a=i.

[0061] The data processing module is used to perform data preprocessing on the data collected by the temperature sensor, and then respectively establish a temperature data set for the temperature data collected at i monitoring points, that is, the temperature data set of the i-th monitoring point is: in, It represents the temperature data collected for the nth time at the ith monitoring point, where n represents the number of times the temperature data is collected, wherein the data preprocessing includes data noise reduction, amplification and filtering processing.

[0062] It should be noted that the noise reduction processing, that is, the sliding average processing of the data, smoothes the signal by calculating the average value of the data in the window to reduce the impact of noise; the amplification processing, that is, the use of an amplifier to linearly amplify the signal, increase the amplitude of the signal, and improve the resolution and sensitivity of the signal; the filtering processing, that is, the use of a bandpass filter to select the signal within a specific frequency range and remove noise of other frequencies, is often used for the analysis and extraction of specific frequency components. The noise reduction, amplification and filtering processing of the data are all existing technologies.

[0063] The feature extraction module extracts the first feature data of the ith monitoring point respectively, generates an abnormal monitoring point based on the first feature data of the ith monitoring point, and collects the second feature data of the abnormal monitoring point.

[0064] The first characteristic data of the i-th monitoring point includes a temperature change rate value of the i-th monitoring point, a temperature fluctuation value of the i-th monitoring point, and an average temperature value of the i-th monitoring point.

[0065] The method for extracting the temperature change rate value of the i-th monitoring point includes:

[0066]

[0067] Where W i is the temperature change rate value of the i-th monitoring point.

[0068] The method for obtaining the average temperature value of the i-th monitoring point includes:

[0069]

[0070] Where, Tpj i is the average temperature value of the ith monitoring point.

[0071] The method for obtaining the temperature fluctuation value of the i-th monitoring point includes:

[0072]

[0073] Where, BDX i is the temperature fluctuation value of the i-th monitoring point.

[0074] The method for generating an abnormal monitoring point based on the first characteristic data of the i-th monitoring point includes:

[0075]

[0076] In the formula, YCXS i is the abnormal coefficient of the ith monitoring point, and is the weight factor, and and Both are greater than 0.

[0077] It should be noted that the size of the weight factor is a specific value obtained by quantifying each data to facilitate subsequent comparison. The size of the weight factor depends on the number of comprehensive parameters and the preliminary setting of the corresponding weight factor for each set of comprehensive parameters by technical personnel in this field.

[0078] Preset abnormal coefficient threshold YC1; Then the i-th monitoring point is marked as an abnormal monitoring point. Otherwise, when The i-th monitoring point is not marked as an abnormal monitoring point.

[0079] The second characteristic data of the abnormal monitoring points include the number of abnormal monitoring points, the average temperature value of the m abnormal monitoring points, the average temperature change rate value of the m abnormal monitoring points, the average temperature fluctuation value of the m abnormal monitoring points, and the distribution relationship of the m abnormal monitoring points on the winding components.

[0080] The number of abnormal monitoring points can be obtained by directly obtaining the number of marked abnormal monitoring points among i monitoring points, and recorded as m abnormal monitoring points.

[0081] The method for obtaining the average temperature value PJm of m abnormal monitoring points includes:

[0082]

[0083] Where, Tpj m Represents the average temperature value of the mth abnormal monitoring point.

[0084] The method for obtaining the average temperature change rate value WBm of m abnormal monitoring points includes:

[0085]

[0086] Where W m Indicates the temperature change rate value of the mth abnormal monitoring point.

[0087] The method for obtaining the average temperature fluctuation value YBDX of m abnormal monitoring points includes:

[0088]

[0089] Where, BDX m Represents the temperature fluctuation value of the mth abnormal monitoring point.

[0090] The method for obtaining the distribution relationship of m abnormal monitoring points on the winding component includes:

[0091] Obtain the coordinate data of m abnormal monitoring points among i monitoring points, m∈{1, 2, ..., i}, extract the abnormal monitoring points, and map the positions of the abnormal monitoring points in the winding to the blank background layer based on the coordinate data of the abnormal monitoring points to obtain an extracted pattern with the distribution relationship of the m abnormal monitoring points, that is, obtain the distribution relationship of the m abnormal monitoring points on the winding components.

[0092] The distribution relationship includes: local small area distribution, that is, m abnormal monitoring points are distributed on a local small area of ​​the winding, that is, the distribution range of the abnormal monitoring points is less than one quarter of the area of ​​the entire winding component.

[0093] Large-area uniform distribution, that is, the m abnormal monitoring points are evenly distributed over a large area of ​​the entire winding, that is, the distribution range of the abnormal monitoring points is greater than one-half of the area of ​​the entire winding component.

[0094] Local large-area distribution, that is, m abnormal monitoring points are evenly distributed over a large area of ​​the entire winding, that is, the distribution range of the abnormal monitoring points is between one quarter and one half of the surface area of ​​the entire winding component, including one quarter and one half.

[0095] Example: If m abnormal monitoring points are distributed on the back or side of the winding component, it is a local large-area distribution.

[0096] Specifically, the intelligent identification method of fault events of a temperature transmitter is provided with i uniformly distributed monitoring points on the winding component, and then a temperature sensors are arranged to collect temperature data of the i monitoring points on the winding component at preset time intervals within a unit time, and a feature extraction module is provided to extract the first feature data of each monitoring point, and then a number of i monitoring points are analyzed based on the first feature data to obtain m abnormal monitoring points, and then the number of abnormal monitoring points, the average temperature value of the abnormal monitoring points, the average temperature change rate value of the abnormal monitoring points, the average temperature fluctuation value of the abnormal monitoring points, and the distribution relationship of the abnormal monitoring points on the winding component are obtained based on the m abnormal monitoring points, so that a variety of special data can be obtained to determine whether a fault occurs and the type of fault, thereby overcoming the prior art that relies solely on temperature changes to perform fault judgment, resulting in low accuracy of its judgment.

[0097] Example 2

[0098] Reference Figure 1 Based on the above embodiment, this embodiment adds a model prediction module. The model prediction module inputs the collected second feature data into the trained machine learning model for predicting the winding fault type, predicts the winding fault type, and wirelessly transmits the fault type to the terminal PC.

[0099] The training method of the machine learning model for predicting winding fault type includes:

[0100] A historical training data set of the temperature transmitter is collected, wherein the historical training data set includes the collected second characteristic data and a winding component fault type corresponding to the second characteristic data.

[0101] The collected historical training data is converted into a corresponding set of feature vectors.

[0102] Each group of feature vectors is used as the input of the machine learning model, the machine learning model takes the winding component failure type corresponding to each group of second feature data as the output, the winding component failure type actually corresponding to each group of second feature data is used as the prediction target, and minimizing the loss function value of the machine learning model is used as the training target; training is stopped when the loss function value of the machine learning model is less than or equal to the preset target loss value.

[0103] The machine learning model is any one of a deep neural network model or a deep belief network model.

[0104] The machine learning model loss function value is the mean square error;

[0105] Mean square error is one of the commonly used loss functions. By transforming the loss function formula Minimization is used as the goal to train the model so that the machine learning model better fits the data, thereby improving the performance and accuracy of the model;

[0106] In the loss function, MSE is the loss function value of the machine learning model, x is the feature vector group number; K is the number of feature vector groups; yx is the fault type predicted by the xth feature vector group, is the fault type actually corresponding to the xth group of feature vectors;

[0107] Other model parameters of the machine learning model, target loss value, optimization algorithm, training set test set validation set ratio and loss function optimization are all achieved through actual engineering and continuously experimentally tuned.

[0108] Winding fault types include: overload fault, short circuit fault, poor ventilation fault and poor lubrication fault.

[0109] It should be noted that the above-mentioned different fault types cause different areas (equivalent to the number of abnormal monitoring points) and regions (i.e. the distribution positions of abnormal monitoring points) of overheating of winding components, as well as the temperature rise rate (i.e. the temperature change rate) and temperature volatility of winding components.

[0110] Example: Overload fault is caused by overloading the transformer. Overload will increase the winding current, which will generate more heat, causing the temperature of the entire winding to rise normally.

[0111] A short circuit fault, that is, a short circuit inside the winding, will cause the current to increase abnormally, generate a large amount of heat, and cause the temperature of the local area of ​​the winding to rise rapidly.

[0112] Poor ventilation fault, poor ventilation of the transformer, inability to effectively dissipate heat, will cause the winding temperature to rise. Poor ventilation may be caused by fan failure, vent blockage, etc. Poor ventilation causes the winding temperature to rise slowly, and the local large area of ​​the windward side of the winding shows the characteristic of slow temperature generation.

[0113] Poor lubrication failure, that is, poor lubrication inside the winding will increase mechanical friction, causing a slow increase in local temperature over a small area.

[0114] Specifically, the intelligent identification method of fault events of this temperature transmitter can directly output the predicted fault type by inputting the collected second feature data into the trained machine learning model for predicting the winding fault type. That is, the fault type can be directly determined by the collected data, and the cause of winding heating can be determined. Compared with traditional temperature transmitters, they can only determine whether a fault has occurred by collecting temperature changes, but cannot determine the type of fault, which causes maintenance personnel to judge the fault type by themselves and then formulate maintenance strategies, which affects maintenance efficiency and has a low degree of intelligence.

[0115] The basic principles, main features and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and the description in the specification are only to illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for intelligently identifying fault events of a temperature transmitter, wherein the temperature transmitter is installed on a winding component of a transformer, characterized in that: include: The data acquisition module is a temperature sensor, which is respectively installed at i monitoring points evenly distributed on the winding component inside the transformer. A three-dimensional coordinate system is established with the length and width of the bottom area of ​​the winding component and the height of the winding component to obtain the coordinate data of the i monitoring points. The a temperature sensor is used to collect the temperature data of the i monitoring points on the winding component at preset time intervals within a unit time t; wherein i is a positive integer greater than 1; a=i; The data processing module is used to perform data preprocessing on the data collected by the temperature sensor, and then respectively establish a temperature data set for the temperature data collected by i monitoring points, that is, the temperature data set of the i-th monitoring point is: in, represents the temperature data collected at the i-th monitoring point for the nth time, where n represents the number of times the temperature data is collected, and the data preprocessing includes data noise reduction, amplification and filtering; A feature extraction module extracts first feature data of the ith monitoring point, generates an abnormal monitoring point based on the first feature data of the ith monitoring point, and collects second feature data of the abnormal monitoring point; the second feature data of the abnormal monitoring point includes the number of abnormal monitoring points, the average temperature value of the m abnormal monitoring points, the average temperature change rate value of the m abnormal monitoring points, the average temperature fluctuation value of the m abnormal monitoring points, and the distribution relationship of the m abnormal monitoring points on the winding component; The model prediction module inputs the collected second feature data into the trained machine learning model for predicting the winding fault type, predicts the fault type of the winding component, and wirelessly transmits the fault type to the terminal PC.

2. The method for intelligently identifying fault events of a temperature transmitter according to claim 1, characterized in that: The first characteristic data of the i-th monitoring point includes a temperature change rate value of the i-th monitoring point, a temperature fluctuation value of the i-th monitoring point, and an average temperature value of the i-th monitoring point.

3. The method for intelligently identifying fault events of a temperature transmitter according to claim 2, characterized in that: The method for obtaining the temperature change rate value of the i-th monitoring point includes: Where W i is the temperature change rate value of the i-th monitoring point; The method for obtaining the average temperature value of the i-th monitoring point includes: Where, Tpj i is the average temperature value of the i-th monitoring point; The method for obtaining the temperature fluctuation value of the i-th monitoring point includes: Where, BDX i is the temperature fluctuation value of the i-th monitoring point.

4. The method for intelligently identifying fault events of a temperature transmitter according to claim 3, characterized in that: The method for generating an abnormal monitoring point based on the first characteristic data of the i-th monitoring point comprises: Get the abnormal coefficient of the i-th monitoring point; In the formula, YCXS i is the abnormal coefficient of the ith monitoring point, and is the weight factor, and and All are greater than 0; Preset abnormal coefficient threshold YC1; when YCXS i >YC1, the i-th monitoring point is marked as an abnormal monitoring point. Otherwise, when YCXS i ≤YC1, the i-th monitoring point will not be marked as an abnormal monitoring point.

5. The method for intelligently identifying fault events of a temperature transmitter according to claim 4, characterized in that: The number of abnormal monitoring points can be obtained by directly acquiring the number of marked abnormal monitoring points among i monitoring points, and recorded as m abnormal monitoring points.

6. The method for intelligently identifying fault events of a temperature transmitter according to claim 4, characterized in that: The method for obtaining the average temperature value PJm of the m abnormal monitoring points includes: Where, Tpj m represents the average temperature value of the mth abnormal monitoring point; The method for obtaining the average temperature change rate value WBm of m abnormal monitoring points includes: Where W m Indicates the temperature change rate value of the mth abnormal monitoring point; The method for obtaining the average temperature fluctuation value YBDX of m abnormal monitoring points includes: Where, BDX m Represents the temperature fluctuation value of the mth abnormal monitoring point.

7. The method for intelligently identifying fault events of a temperature transmitter according to claim 4, characterized in that: The method for obtaining the distribution relationship of the m abnormal monitoring points on the winding component includes: Obtain coordinate data of m abnormal monitoring points among i monitoring points, m∈{1, 2, ..., i}, extract the abnormal monitoring points, and map the positions of the abnormal monitoring points in the winding component to the blank background layer based on the coordinate data of the abnormal monitoring points to obtain an extraction pattern with the distribution relationship of the m abnormal monitoring points, that is, obtain the distribution relationship of the m abnormal monitoring points on the winding component; The distribution relationship includes: local small area distribution, that is, m abnormal monitoring points are distributed on a local small area of ​​the winding component, that is, the distribution range of the abnormal monitoring points is less than one quarter of the area of ​​the entire winding component; Large-area uniform distribution, that is, the m abnormal monitoring points are evenly distributed over a large area of ​​the entire winding component, that is, the distribution range of the abnormal monitoring points is greater than one-half of the area of ​​the entire winding component; Local large-area distribution, that is, m abnormal monitoring points are evenly distributed over a large area of ​​the entire winding component, that is, the distribution range of the abnormal monitoring points is between one quarter and one half of the surface area of ​​the entire winding component, including one quarter and one half.

8. The method for intelligently identifying fault events of a temperature transmitter according to claim 1, characterized in that: The training method of the machine learning model for predicting winding fault types includes: Collecting a historical training data set of the temperature transmitter, wherein the historical training data set includes the collected second characteristic data and the winding component fault type corresponding to the second characteristic data; Convert the collected historical training data into a corresponding set of feature vectors; Each group of feature vectors is used as the input of the machine learning model, the machine learning model takes the winding component failure type corresponding to each group of second feature data as the output, the winding component failure type actually corresponding to each group of second feature data is used as the prediction target, and minimizing the loss function value of the machine learning model is used as the training target; training is stopped when the loss function value of the machine learning model is less than or equal to the preset target loss value.

9. The method for intelligently identifying fault events of a temperature transmitter according to claim 8, characterized in that: The winding fault types include: overload fault, short circuit fault, poor ventilation fault and poor lubrication fault.

Images