Transformer monitoring method and device
By using preset models of convolution modules, attention mechanisms and long and short-term memory modules, combined with transformer equipment data and historical temperature data for temperature prediction, the problem of low accuracy of transformer abnormal temperature rise monitoring and professional dependence is solved, and high Accuracy and low-cost monitoring effects.
Patent Information
- Application Number
- CN202411272422.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-11
- Publication Date
- 2024-12-13
AI Technical Summary
In the prior art, the method of identifying abnormal temperature rise of transformers depends on temperature data and is susceptible to interference from external factors, resulting in misjudgment or misjudgment, and requires professionals to operate and analyze, and it is difficult for non-professionals to interpret thermal imaging images.
By obtaining the transformer equipment data, including load coefficient, ambient temperature and humidity, the temperature data prediction is achieved using the preset models of the convolution module, attention mechanism and long-term memory module, and abnormal temperature rise monitoring is performed in combination with historical temperature data to improve monitoring of monitoring Accuracy and reduce misjudgment.
It improves the accuracy and effectiveness of abnormal temperature rise monitoring of transformer, without professional operation and analysis, saves labor and costs, and reduces misjudgment.
Smart Images
Figure CN119128439A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of transformers, and in particular to a transformer monitoring method and device. Background Art
[0002] As one of the core equipment of the power system, the normal operation of the transformer is crucial to ensure the stability and safety of the entire power system. Abnormal temperature rise of the transformer is one of the common fault phenomena. If it is not discovered and handled in time, it may cause serious consequences such as equipment damage, power outages and even fire.
[0003] In the related art, there are two methods for identifying abnormal temperature rise of transformers: collecting the changes in transformer temperature to identify whether the transformer is in an abnormal temperature rise state; and using thermal imagers and other equipment to monitor the temperature distribution on the transformer surface in real time and intuitively display the temperature conditions through thermal images.
[0004] However, abnormality identification based solely on temperature may be interfered by external factors, resulting in misjudgment or missed judgment. Although thermal imaging technology can quickly capture the temperature distribution of the target, its measurement accuracy may be affected by the surface characteristics of the target, resulting in insufficient accuracy. In addition, thermal imaging technology requires professionals to operate and analyze. Non-professionals may not be able to correctly interpret thermal imaging images or misjudge the location and severity of the fault. Summary of the invention
[0005] The embodiments of the present application provide a transformer monitoring method and device to solve the problem that the transformer abnormal temperature rise has low accuracy and requires professional personnel to operate and analyze.
[0006] In a first aspect, an embodiment of the present application provides a transformer monitoring method, comprising:
[0007] Acquire equipment data of the transformer within a first preset time period; wherein the equipment data includes data associated with the temperature of the transformer;
[0008] Determining temperature data of the transformer according to the device data;
[0009] Based on the temperature data and the historical temperature data of the transformer, the predicted temperature data of the transformer in the next time period after the first preset time period is obtained; wherein the historical temperature data is used to characterize the historical temperature change data of the transformer;
[0010] The transformer is monitored for abnormal temperature rise according to the predicted temperature data.
[0011] In a possible implementation, the device data includes a load factor of the transformer, an ambient temperature of the transformer, and an ambient humidity of the transformer;
[0012] Determining the temperature data of the transformer according to the device data includes:
[0013] Input the device data into a trained preset model; wherein the preset model is trained based on the device data and real temperature data of multiple transformers, and the preset model includes a convolution module, an attention mechanism module, and a long short-term memory module;
[0014] Extracting local features of the device data according to the convolution module to obtain a first feature vector;
[0015] According to the attention mechanism module, extract strongly correlated features of the first feature vector to obtain a second feature vector;
[0016] Based on the long short-term memory module, feature selection processing is performed on the second feature vector to obtain temperature data of the transformer.
[0017] In a possible implementation manner, before obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer, the method further includes:
[0018] Using the first preset time period as a time keyword, filtering in the first preset database to obtain the historical equipment data of the transformer within the second preset time period; wherein the first preset database stores the correspondence between the preset time period and the equipment data of the transformer;
[0019] The historical device data is input into the preset model to obtain the historical temperature data of the transformer output by the preset model; wherein the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0020] In a possible implementation manner, before obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer, the method further includes:
[0021] Using the first preset time period as a time keyword, historical temperature data of the transformer within a second preset time period is obtained by screening in a second preset database; wherein the second preset database stores a correspondence between preset time periods and temperature data of the transformer, and the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0022] In a possible implementation manner, it is characterized in that obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer includes:
[0023] Acquire temperature change data of the transformer in a time period most similar to a next time period after the first preset time period from the historical temperature data;
[0024] The predicted temperature data of the transformer in a next time period after the first preset time period is obtained according to the temperature change data and the temperature data.
[0025] In a possible implementation, it is characterized in that the abnormal temperature rise monitoring of the transformer according to the predicted temperature data includes:
[0026] Detecting whether the predicted temperature data contains data exceeding a preset temperature threshold;
[0027] If there is data exceeding the preset temperature threshold in the predicted temperature data, it is determined that the transformer has a risk of abnormal temperature rise, and an early warning message is issued; wherein the early warning message is used to prompt that the transformer has a risk of abnormal temperature rise.
[0028] In a possible implementation manner, it is characterized in that, before inputting the device data into the trained preset model, the method further includes:
[0029] Inputting equipment data of multiple transformers into a preset model to obtain temperature prediction data corresponding to each transformer output by the preset model;
[0030] According to the actual temperature data of each transformer, the temperature prediction data corresponding to each transformer and the preset loss function, the model parameters of the preset model are adjusted until a trained preset model is obtained.
[0031] In a possible implementation, before inputting the equipment data of the plurality of transformers into the preset model, the method further includes:
[0032] Data cleaning is performed on the equipment data of multiple transformers; wherein the data cleaning is used to remove invalid or abnormal data in the equipment data.
[0033] In a second aspect, an embodiment of the present application provides a transformer monitoring device, the device comprising:
[0034] An acquisition unit, configured to acquire equipment data of the transformer within a first preset time period; wherein the equipment data includes data associated with the temperature of the transformer;
[0035] A first processing unit, configured to determine temperature data of the transformer according to the device data;
[0036] a second processing unit, configured to obtain predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and historical temperature data of the transformer; wherein the historical temperature data is used to characterize historical temperature change data of the transformer;
[0037] A monitoring unit is used to monitor abnormal temperature rise of the transformer according to the predicted temperature data.
[0038] In a possible implementation, the device data includes a load factor of the transformer, an ambient temperature of the transformer, and an ambient humidity of the transformer;
[0039] The first processing unit is specifically configured to:
[0040] Input the device data into a trained preset model; wherein the preset model is trained based on the device data and real temperature data of multiple transformers, and the preset model includes a convolution module, an attention mechanism module, and a long short-term memory module;
[0041] Extracting local features of the device data according to the convolution module to obtain a first feature vector;
[0042] According to the attention mechanism module, extract strongly correlated features of the first feature vector to obtain a second feature vector;
[0043] Based on the long short-term memory module, feature selection processing is performed on the second feature vector to obtain temperature data of the transformer.
[0044] The embodiment of the present application provides a transformer monitoring method and device, which determines the temperature data within the first preset time period according to the equipment data within the first preset time period that is associated with the transformer temperature, and on this basis, combines the historical temperature data of the transformer to predict the transformer temperature in the next time period after the first preset time period, so that the transformer is monitored for abnormal temperature rise according to the predicted transformer temperature. Among them, the temperature data of the transformer within the preset time period and the historical temperature data are combined to perform temperature prediction, which improves the accuracy of temperature prediction, and further improves the accuracy and effectiveness of abnormal temperature rise monitoring of the transformer based on the predicted temperature data, and does not require professional personnel to operate and analyze, saving manpower and costs, and reducing misjudgments caused by operational and analytical errors of relevant personnel. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0046] Figure 1 is a flowchart of the transformer monitoring method provided in the embodiment of the present application;
[0047] Figure 2 is a flowchart of another transformer monitoring method provided in an embodiment of the present application;
[0048] Figure 3 It is a structural schematic diagram of a transformer monitoring device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0049] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.
[0050] The applicant has found that in the related art, transformer abnormality monitoring based solely on temperature may be interfered by external factors, resulting in misjudgment or missed judgment; the monitoring method based on thermal imaging technology, although thermal imaging technology can quickly capture the temperature distribution of the target, its measurement accuracy may be affected by the surface characteristics of the target, resulting in insufficient accuracy, and thermal imaging technology requires professionals to operate and analyze, and non-professionals may not be able to correctly interpret thermal imaging images, resulting in misjudgment of the location and severity of the transformer abnormality.
[0051] With the idea of improving the accuracy of transformer abnormality monitoring, in the implementation mode of the present application, the transformer temperature data is predicted in combination with the temperature data and historical temperature data of the transformer within a preset time period, thereby improving the accuracy of temperature prediction, and further improving the accuracy and effectiveness of transformer abnormal temperature rise monitoring based on the predicted temperature data. Moreover, no professional personnel are required to operate and analyze, thus saving manpower and costs, and reducing misjudgments caused by operational and analytical errors of relevant personnel.
[0052] In order to make the purpose, technical solutions and advantages of the present application clearer, specific embodiments will be described below in conjunction with the accompanying drawings.
[0053] Figure 1 The implementation flow chart of the transformer monitoring method provided in the embodiment of the present application is described in detail as follows:
[0054] Step 101, obtaining equipment data of a transformer within a first preset time period; wherein the equipment data includes data associated with a temperature of the transformer.
[0055] Exemplarily, this embodiment acquires transformer equipment data associated with the transformer temperature within a first preset time period.
[0056] In a possible implementation, the present embodiment automatically acquires transformer device data in response to a trigger of a preset time condition. The preset time condition may instruct the electronic device to automatically acquire device data within a preset time period at a certain time frequency. For example, the preset time condition instructs the electronic device to acquire transformer device data within an hour before each hour.
[0057] Optionally, the device data may be screened data that has a strong correlation with the transformer temperature, such as the load factor of the transformer, the ambient temperature of the transformer, and the ambient humidity of the transformer.
[0058] Step 102: Determine temperature data of the transformer according to the equipment data.
[0059] Exemplarily, this embodiment determines corresponding temperature data of the transformer, that is, temperature data of the transformer within a first preset time period, based on the acquired device data.
[0060] In a feasible implementation, the device data is input into a preset model to obtain the temperature data of the transformer output by the model. The model may be a deep learning network structure model or other models with data processing, data analysis and data output capabilities.
[0061] In a feasible implementation manner, the temperature data corresponding to the device data is calculated based on a theoretical calculation model.
[0062] Step 103, based on the temperature data and the historical temperature data of the transformer, obtain predicted temperature data of the transformer in a next time period after the first preset time period; wherein the historical temperature data is used to characterize the historical temperature change data of the transformer.
[0063] Exemplarily, the historical temperature data of the transformer can reflect the temperature change trend of the transformer to a certain extent. Therefore, this embodiment can obtain the predicted temperature data of the transformer in the next time period after the first preset time period based on the temperature data of the transformer in the first preset time period and the historical temperature data of the transformer.
[0064] In one example, the historical temperature data of the transformer includes the transformer temperature data of the first preset time period and the next time period after the first preset time period. The first preset time period can be understood as the historical time period most similar to the first preset time period. Here, the similarity between time periods can be determined based on the information such as the month and season corresponding to different time periods, such as the afternoon of July 30 this summer and the afternoon of July 30 last summer have the highest similarity.
[0065] Step 104: monitor the transformer for abnormal temperature rise according to the predicted temperature data.
[0066] Exemplarily, this embodiment monitors the abnormal temperature rise of the transformer based on the obtained predicted temperature data, that is, the predicted temperature data of the next time period after the first preset time period. For example, if there is temperature data exceeding the preset temperature threshold in the predicted temperature data, it is determined that the transformer has an abnormal temperature rise, and an early warning is issued.
[0067] In one example, the present embodiment is pre-configured to obtain transformer equipment data at every hour of the day. For example, at eight o'clock in the morning, the transformer equipment data from seven to eight o'clock is obtained, and the transformer temperature data from seven to eight o'clock today is obtained. Then, based on the historical temperature data of the transformer in the same period, the predicted temperature data for the time periods such as eight to nine o'clock today, or eight to ten o'clock, is predicted. Finally, based on the predicted temperature data, the transformer is monitored for abnormal temperature rise in the corresponding time period.
[0068] In summary, the embodiment of the present application determines the temperature data within the first preset time period based on the equipment data within the first preset time period that is associated with the transformer temperature, thereby improving the accuracy of transformer temperature identification. On this basis, combined with the historical temperature data of the transformer, the transformer temperature in the next time period after the first preset time period is predicted, wherein the historical temperature data of the transformer is fully considered, and the accuracy of temperature prediction is further improved, thereby improving the accuracy and effectiveness of abnormal temperature rise monitoring of the transformer based on the predicted temperature data, and no professional personnel are required to operate and analyze, saving manpower and costs, and reducing misjudgments caused by operational and analytical errors of relevant personnel.
[0069] In order to improve the accuracy and effectiveness of abnormal temperature rise monitoring of the transformer based on the predicted temperature data, it is necessary to further improve the accuracy of the temperature data corresponding to the first preset time period. There are many factors that affect the temperature data of the transformer, among which the equipment data is particularly important. The following further introduces how to determine the temperature data of the transformer based on the equipment data.
[0070] In some embodiments, the device data includes a load factor of the transformer, an ambient temperature of the transformer, and an ambient humidity of the transformer; determining the temperature data of the transformer according to the device data includes:
[0071] The device data is input into a trained preset model; wherein the preset model is trained based on the device data and real temperature data of multiple transformers, and the preset model includes a convolution module, an attention mechanism module, and a long short-term memory module. Then, according to the convolution module, local features of the device data are extracted to obtain a first feature vector. According to the attention mechanism module, strongly correlated features of the first feature vector are extracted to obtain a second feature vector. Based on the long short-term memory module, feature selection processing is performed on the second feature vector to obtain the temperature data of the transformer.
[0072] Exemplarily, this embodiment determines the temperature data of the transformer corresponding to the device data based on a pre-trained preset model, wherein the preset model includes a convolution module, an attention mechanism module, and a long short-term memory module. For example, according to the convolution module, the local features of the device data are extracted to obtain a first feature vector, and then according to the attention mechanism module, the strongly correlated features of the first feature vector are extracted to obtain a second feature vector, and finally, based on the long short-term memory module, the second feature vector is subjected to feature selection processing to obtain the temperature data of the transformer.
[0073] In an example, the first preset time period can be one day, one hour, etc. Taking one day as an example, the input of the preset model is a 3×144 dimensional tensor. Among them, 3 represents the three main factors that determine the hot spot temperature of the transformer, which are the load factor, the temperature of the environment where the transformer equipment is located, and the humidity of the environment where the transformer equipment is located, denoted as in and Represent the load factor, ambient temperature and ambient humidity at time t respectively. 144 represents a time series of one day, and the data recording interval is 10 minutes. Therefore, the input form of the data is x=[x1,x2,...,x 143 ,x 144 ].
[0074] First, since the time series model cannot mine the potential relationship between non-continuous data, the input is first fed into a 1D convolution to extract the local features of the data. After that, the data is converted into a 3×48 first eigenvector to achieve feature aggregation and dimensionality reduction.
[0075] Secondly, the first feature vector is then passed to the attention module to extract the strongly relevant features of the data. After calculating the self-attention mechanism, it is expanded into a 1×144 vector, which is the second feature vector.
[0076] Finally, the second eigenvector is fed into the long short-term memory module to obtain the transformer temperature data, which is expressed as y = [y1, y2, ..., y 143 ,y 144 ].
[0077] Among them, one of the purposes of the one-dimensional convolution module in this embodiment is to reduce the dimension of the time step to speed up the forward and backward propagation of the model and reduce the computational cost. Another purpose is to aggregate information while keeping the feature dimension of the transformer unchanged. Therefore, it is necessary to transpose the input data first so that it changes from 3×144 to 144×3 dimensions. Then, this vector will be multiplied by 48 convolution kernels with different parameters. The calculation formula for convolution is:
[0078] Y=σ(W·X+b)
[0079] Where: Y is the extracted feature; σ is the sigmoid activation function; W is the weight matrix; X is the time series data; b is the bias vector.
[0080] The dimension of each convolution kernel is 1×144, which changes the feature vector from 144×3 to 1×3. After the one-dimensional convolution, a total of 48 vectors are generated, which are then concatenated together to form a complete 48×3 vector. The vector is then transposed and batch normalization and maximum pooling are performed, ultimately generating a 3×48 vector. Compared to the original input data, the temporal information is aggregated and reduced to a dimension of 48 time steps, allowing the model to learn more refined and efficient features. In terms of the dimension of the transformer feature, it maintains the original dimension size, maintaining the interpretability of the data.
[0081] In the attention mechanism of this embodiment, in one example, for the vector zi, it is multiplied by three matrices W respectively. Q , W K , W V , we get three features, namely query: Qi = W Q ×zi, key: Ki=W K ×zi, value: Vi=W V×zi. Then the global attention weight is calculated, first with the query Q1 of zi and the key Ki of all other vectors:
[0082]
[0083]
[0084] Where: d k Represents the dimension of the vector; e li is the calculation weight; a li is the global attention weight, and then the corresponding a li Multiplying with Vi and adding them together, calculates the output vector h1 of z1
[0085]
[0086] Perform the above steps for each zi and get the corresponding output vector, and you will get a vector with a dimension of 3 × 48. Then flatten the vector and finally get a vector h with a dimension of 1 × 144.
[0087] The LSTM network module in this embodiment effectively solves the gradient vanishing and gradient exploding problems by introducing a gated structure. The cell memory element added to the LSTM model gives it good memory capabilities and has been widely used in time series prediction. The LSTM model has three stages: forget stage, memory stage, and output stage. Forget stage: Specifically, a forget gate f is used. t To control what needs to be retained and what needs to be forgotten in the previous state.
[0088] f t =σ(W f [y t-1 -1,h t ]+b f )
[0089] Where: W f is the weight matrix; b f is the deviation vector; f t is the retention degree value; y t-1 is the output at the previous moment; f t is the input vector.
[0090] Memory stage: This stage is used to selectively remember the input of this stage. It mainly selectively remembers the input h t , the current input content is calculated by the previously To indicate that the gate signal is i t control:
[0091] i t =σ(Wi [y t-1 ,ht]+b i )
[0092]
[0093]
[0094] Where: W i , W c is the weight matrix; b i , b c is the deviation vector; C t-1 is the memory state of the previous moment; i t is the memory level value at the current moment; is the intermediate state; Ct is the memory state at the current moment.
[0095] Output stage: This stage determines what the output of the current state is. The main way is through o t To control, the result Ct of the previous stage is also scaled and modified by a tanh activation function:
[0096] o t =σ(W0[y t-1 ,h t ]+b0)
[0097] y t =o t tanh(C t )
[0098] Where: W0 is the weight matrix; b0 is the bias vector; o t is the output degree value; y t It is the output at the current moment, representing the transformer temperature value at the corresponding moment.
[0099] In this embodiment, the transformer temperature is identified and calculated based on the transformer load factor, the temperature of the transformer environment, and the humidity of the transformer environment, and the transformer parameters and external factors affecting the transformer temperature data are comprehensively considered to improve the accuracy of transformer temperature identification. In addition, the transformer temperature at the corresponding moment is identified and calculated through a trained preset model, which further improves the accuracy of the output temperature data.
[0100] The above-mentioned embodiment introduces how to identify device data based on a model to obtain corresponding temperature data. The following further introduces how to train a preset model to obtain a trained preset model.
[0101] In some embodiments, before inputting the device data into the trained preset model, the method further includes:
[0102] The equipment data of multiple transformers are input into the preset model to obtain the temperature prediction data corresponding to each transformer output by the preset model. According to the actual temperature data of each transformer, the temperature prediction data corresponding to each transformer and the preset loss function, the model parameters of the preset model are adjusted until a trained preset model is obtained.
[0103] In some embodiments, before inputting the equipment data of the plurality of transformers into the preset model, the method further comprises:
[0104] Data cleaning is performed on the equipment data of multiple transformers; wherein the data cleaning is used to remove invalid or abnormal data in the equipment data. In this way, the accuracy of the subsequent processing results can be improved by using the data after data cleaning for subsequent processing.
[0105] Exemplarily, before the device data is identified and calculated to obtain the corresponding temperature data, the preset model needs to be trained in advance in this embodiment. This includes: obtaining the device data and real temperature data of multiple transformers, cleaning the data, removing invalid or abnormal data in the data, and then inputting the device data of multiple transformers into the preset model to obtain the temperature prediction data corresponding to each transformer output by the preset model; adjusting the model parameters of the preset model according to the real temperature data of each transformer, the temperature prediction data corresponding to each transformer, and the preset loss function until a trained preset model is obtained.
[0106] In one example, in this embodiment, the training of the preset model includes the following steps: (1) data acquisition and collection: a large amount of transformer equipment data and its corresponding transformer temperature data are acquired from the database of multiple transformer sites; (2) the acquired data is then cleaned, abnormal temperature rise data is screened out according to the screening conditions, and invalid or abnormal data is removed to construct a data set; (3) the data set is randomly divided into a training set and a test set according to a certain ratio; (4) again, the prepared training set is sequentially passed through the deep learning network structure model provided and built by this application, and in the process of feature extraction layer by layer, the data features are fused at different scales, and finally the feature data sequence is clustered to take the loss function, and the error is back-propagated to correct the network parameters; (5) the test set is passed into the network to test the accuracy of the network model to determine whether the convergence conditions are met; (6) after steps 4 and 5 are completed, the network weights obtained from the training are brought into the deep network again, and the training and verification are repeated for 400 rounds. After the loss function and the average accuracy converge, the training is stopped. (7) The trained weight file is extracted and deployed on the edge. At this time, the model can identify the temperature of the transformer.
[0107] In this embodiment, the model is trained using the equipment data of multiple transformers and their corresponding real temperature data, and the loss function and average accuracy are used as convergence targets to obtain a trained preset model. A large amount of data training and verification improves the preset model's ability to recognize transformer equipment data and capture the correlation between transformer equipment data and transformer temperature data.
[0108] The above-mentioned embodiment introduces the training of the preset model. The following further introduces how to obtain the historical temperature data of the transformer.
[0109] In some embodiments, before obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer, the method further includes:
[0110] The first preset time period is used as a time keyword, and historical equipment data of the transformer within the second preset time period is obtained by screening in the first preset database; wherein the first preset database stores a correspondence between the preset time period and the equipment data of the transformer.
[0111] The historical device data is input into the preset model to obtain the historical temperature data of the transformer output by the preset model; wherein the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0112] Exemplarily, this embodiment can use the first preset time period as the time keyword, and filter out the historical equipment data of the transformer within the second preset time period in the first preset database; wherein the first preset database stores the correspondence between the preset time period and the equipment data of the transformer; and then input the historical equipment data into the preset model to obtain the historical temperature data of the transformer output by the preset model; the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0113] In one example, the first preset database stores equipment data of the transformer in different historical time periods; the first preset time period is from 8:00 to 9:00 on June 1st of this year, then this embodiment uses 8:00 to 9:00 on June 1st as the time keyword, and filters out the historical equipment data of previous years in the first preset database, for example, the historical equipment data of the transformer from 8:00 to 10:00 on June 1st last year, and then inputs the historical equipment data into the preset model to obtain the historical temperature data of the transformer from 8:00 to 10:00 on June 1st last year output by the preset model.
[0114] In some embodiments, before obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer, the method further includes:
[0115] Using the first preset time period as the time keyword, historical temperature data of the transformer within the second preset time period is screened in the second preset database; wherein the second preset database stores a correspondence between the preset time period and the temperature data of the transformer, and the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0116] Exemplarily, the present embodiment may also use the first preset time period as a time keyword, and filter out the historical temperature data of the transformer within the second preset time period in the second preset database; wherein the second preset database stores the correspondence between the preset time period and the temperature data of the transformer, and the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0117] In one example, the second preset database stores historical temperature data of the transformer in different historical time periods; the first preset time period is 8:00 to 9:00 on June 1st of this year, then this embodiment can use 8:00 to 9:00 on June 1st as the time keyword, and filter out the historical temperature data of the transformer in previous years in the second preset database, for example, last year from 8:00 to 10:00, or from 7:00 to 11:00 on June 1st.
[0118] Optionally, in this embodiment, data filtering items can be customized, such as data filtering range, filtering time period, etc.
[0119] In this embodiment, the first preset time period is used as the time keyword, the corresponding transformer historical equipment data is searched in the corresponding database, and it is input into the trained preset model to obtain the corresponding historical temperature data, or the historical temperature data is directly screened in the corresponding database, which can improve the speed and accuracy of data acquisition.
[0120] The above-mentioned embodiment introduces how to obtain the historical temperature data of the transformer. The following further introduces how to obtain the predicted temperature data of the transformer in the next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer.
[0121] In some embodiments, based on the temperature data and the historical temperature data of the transformer, the predicted temperature data of the transformer in the next time period after the first preset time period is obtained, including:
[0122] The temperature variation data of the transformer in a time period most similar to a next time period after the first preset time period is obtained from the historical temperature data.
[0123] According to the temperature change data and the temperature data, the predicted temperature data of the transformer in the next time period after the first preset time period is obtained.
[0124] Exemplarily, this embodiment determines, in the historical temperature data, the temperature change data of the next time period after the first preset time period relative to the temperature data of the first preset time period, and then obtains the predicted temperature data of the transformer in the next time period after the first preset time period based on the temperature change data and the temperature data. For example, the temperature change data and the temperature data corresponding to the first preset time period are numerically superimposed to obtain the predicted temperature data of the transformer in the next time period after the first preset time period.
[0125] This embodiment analyzes and obtains predicted temperature data based on the change data of the same period data in the historical temperature data and the temperature data of the previous time period, thereby improving the accuracy of temperature prediction.
[0126] The above-mentioned embodiment describes how to obtain predicted temperature data. The following further describes how to monitor abnormal temperature rise of the transformer based on the predicted temperature data.
[0127] In some embodiments, abnormal temperature rise monitoring of the transformer is performed according to the predicted temperature data, including:
[0128] Check whether the predicted temperature data contains data exceeding a preset temperature threshold.
[0129] If there is data in the predicted temperature data that exceeds the preset temperature threshold, it is determined that the transformer has a risk of abnormal temperature rise and an early warning message is issued; wherein, the early warning message is used to indicate that the transformer has a risk of abnormal temperature rise.
[0130] Exemplarily, this embodiment detects whether there is data in the predicted temperature data that exceeds a preset temperature threshold. If it is determined that there is data in the predicted temperature data that exceeds the preset temperature threshold, it is determined that the transformer has a risk of abnormal temperature rise and an early warning message is issued; wherein the early warning message is used to prompt that the transformer has a risk of abnormal temperature rise.
[0131] Optionally, the warning information may include one or more warnings such as light warning, sound warning, and information warning or telephone warning sent to technical personnel.
[0132] This embodiment monitors and warns the transformer temperature based on a set temperature threshold, and can promptly remind relevant personnel when the transformer has a risk of abnormal temperature rise, thereby reducing the risk of transformer failure.
[0133] Figure 2 Another implementation flow chart of a transformer monitoring method provided in an embodiment of the present application is described in detail as follows:
[0134] Step 201, obtaining equipment data of a transformer within a first preset time period; wherein the equipment data includes data associated with a temperature of the transformer.
[0135] In one example, the equipment data includes a load factor of a transformer, an ambient temperature of the transformer, and an ambient humidity of the transformer.
[0136] Step 202, input the equipment data of multiple transformers into the preset model to obtain the temperature prediction data corresponding to each transformer output by the preset model; according to the actual temperature data of each transformer, the temperature prediction data corresponding to each transformer and the preset loss function, adjust the model parameters of the preset model until a trained preset model is obtained.
[0137] In one example, before step 202, data cleaning is performed on the equipment data of multiple transformers; wherein the data cleaning is used to remove invalid or abnormal data in the equipment data.
[0138] Step 203: Determine the temperature data of the transformer according to the equipment data.
[0139] In one example, step 203 includes the following steps:
[0140] The device data is input into a trained preset model; wherein the preset model is trained based on the device data and real temperature data of multiple transformers, and the preset model includes a convolution module, an attention mechanism module, and a long short-term memory module.
[0141] According to the convolution module, local features of the device data are extracted to obtain a first feature vector.
[0142] According to the attention mechanism module, strongly correlated features of the first feature vector are extracted to obtain the second feature vector.
[0143] Based on the long short-term memory module, feature selection processing is performed on the second eigenvector to obtain the temperature data of the transformer.
[0144] Step 204, using the first preset time period as a time keyword, filter and obtain historical equipment data of the transformer within a second preset time period; wherein the second preset time period includes the first preset time period and a next time period after the first preset time period.
[0145] In a feasible implementation, the first preset time period is used as a time keyword, and the historical equipment data of the transformer within the second preset time period is screened in the first preset database; wherein the first preset database stores the correspondence between the preset time period and the equipment data of the transformer; the historical equipment data is input into the preset model to obtain the historical temperature data of the transformer output by the preset model; wherein the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0146] In another feasible implementation, the first preset time period is used as a time keyword, and the historical temperature data of the transformer within the second preset time period is screened in the second preset database; wherein the second preset database stores the correspondence between the preset time period and the temperature data of the transformer, and the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0147] Step 205 : obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer.
[0148] In one example, step 205 includes:
[0149] The temperature variation data of the transformer in a time period most similar to a next time period after the first preset time period is obtained from the historical temperature data.
[0150] According to the temperature change data and the temperature data, the predicted temperature data of the transformer in the next time period after the first preset time period is obtained.
[0151] Step 206: Monitor the transformer for abnormal temperature rise based on the predicted temperature data.
[0152] In one example, step 206 includes the following steps:
[0153] Check whether the predicted temperature data contains data exceeding a preset temperature threshold.
[0154] If there is data in the predicted temperature data that exceeds the preset temperature threshold, it is determined that the transformer has a risk of abnormal temperature rise and an early warning message is issued; wherein, the early warning message is used to indicate that the transformer has a risk of abnormal temperature rise.
[0155] The detailed implementation process of the above steps 201 to 206 can refer to the relevant descriptions of the above embodiments, which will not be repeated here. The execution order of the above steps 201 to 206 can be set according to actual needs and is not specifically limited here.
[0156] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0157] The following is an embodiment of the device of the present application. For details not described in detail, please refer to the corresponding method embodiment described above.
[0158] Figure 3 The structure diagram of the transformer monitoring device provided in the embodiment of the present application is shown. For the convenience of explanation, only the part related to the embodiment of the present application is shown, which is described in detail as follows:
[0159] like Figure 3 As shown, the transformer monitoring device 3 includes:
[0160] The acquisition unit 31 is used to acquire equipment data of the transformer within a first preset time period; wherein the equipment data includes data associated with the temperature of the transformer.
[0161] The first processing unit 32 is used to determine the temperature data of the transformer according to the equipment data.
[0162] The second processing unit 33 is used to obtain predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and historical temperature data of the transformer; wherein the historical temperature data is used to characterize historical temperature change data of the transformer.
[0163] The monitoring unit 34 is used to monitor the abnormal temperature rise of the transformer according to the predicted temperature data.
[0164] In a possible implementation, the device data includes a load factor of the transformer, an ambient temperature of the transformer, and an ambient humidity of the transformer.
[0165] The first processing unit 32 is specifically used for:
[0166] The device data is input into a trained preset model; wherein the preset model is trained based on the device data and real temperature data of multiple transformers, and the preset model includes a convolution module, an attention mechanism module, and a long short-term memory module.
[0167] According to the convolution module, local features of the device data are extracted to obtain a first feature vector.
[0168] According to the attention mechanism module, strongly correlated features of the first feature vector are extracted to obtain the second feature vector.
[0169] Based on the long short-term memory module, feature selection processing is performed on the second eigenvector to obtain the temperature data of the transformer.
[0170] In a possible implementation, before the second processing unit 33, the device 3 further includes a third processing unit, and the third processing unit is specifically configured to:
[0171] The first preset time period is used as a time keyword, and historical equipment data of the transformer within the second preset time period is obtained by screening in the first preset database; wherein the first preset database stores a correspondence between the preset time period and the equipment data of the transformer.
[0172] The historical device data is input into the preset model to obtain the historical temperature data of the transformer output by the preset model; wherein the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0173] In a possible implementation manner, the third processing unit is further configured to:
[0174] Using the first preset time period as the time keyword, historical temperature data of the transformer within the second preset time period is screened in the second preset database; wherein the second preset database stores a correspondence between the preset time period and the temperature data of the transformer, and the second preset time period includes the first preset time period and the next time period after the first preset time period.
[0175] In a possible implementation manner, the second processing unit 33 is specifically configured to:
[0176] The temperature variation data of the transformer in a time period most similar to a next time period after the first preset time period is obtained from the historical temperature data.
[0177] According to the temperature change data and the temperature data, the predicted temperature data of the transformer in the next time period after the first preset time period is obtained.
[0178] In a possible implementation, the monitoring unit 34 is specifically configured to:
[0179] Check whether the predicted temperature data contains data exceeding a preset temperature threshold.
[0180] If there is data in the predicted temperature data that exceeds the preset temperature threshold, it is determined that the transformer has a risk of abnormal temperature rise and an early warning message is issued; wherein, the early warning message is used to indicate that the transformer has a risk of abnormal temperature rise.
[0181] In a possible implementation, before the first processing unit 32, the device 3 further includes a training unit, and the training unit is specifically configured to:
[0182] The equipment data of multiple transformers are input into the preset model to obtain the temperature prediction data corresponding to each transformer output by the preset model.
[0183] According to the actual temperature data of each transformer, the temperature prediction data corresponding to each transformer and the preset loss function, the model parameters of the preset model are adjusted until a trained preset model is obtained.
[0184] In a possible implementation, before the training unit, the device 3 further includes:
[0185] The cleaning unit is used to perform data cleaning processing on the equipment data of multiple transformers; wherein the data cleaning processing is used to remove invalid or abnormal data in the equipment data.
[0186] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0187] Those of ordinary skill in the art will appreciate that the templates, units, and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0188] If the module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned transformer monitoring method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. Computer-readable media may include: any entity or device that can carry computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory, random access memory, electrical carrier signal, telecommunication signal and software distribution medium, etc.
[0189] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.
Claims
1. A transformer monitoring method, characterized in that: The method comprises: Acquire equipment data of the transformer within a first preset time period; wherein the equipment data includes data associated with the temperature of the transformer; Determining temperature data of the transformer according to the device data; Based on the temperature data and the historical temperature data of the transformer, the predicted temperature data of the transformer in the next time period after the first preset time period is obtained; wherein the historical temperature data is used to characterize the historical temperature change data of the transformer; The transformer is monitored for abnormal temperature rise according to the predicted temperature data.
2. The transformer monitoring method according to claim 1, characterized in that: The equipment data includes the load factor of the transformer, the ambient temperature of the transformer, and the ambient humidity of the transformer; Determining the temperature data of the transformer according to the device data includes: Input the device data into a trained preset model; wherein the preset model is trained based on the device data and real temperature data of multiple transformers, and the preset model includes a convolution module, an attention mechanism module, and a long short-term memory module; Extracting local features of the device data according to the convolution module to obtain a first feature vector; According to the attention mechanism module, extract strongly correlated features of the first feature vector to obtain a second feature vector; Based on the long short-term memory module, feature selection processing is performed on the second feature vector to obtain temperature data of the transformer.
3. The transformer monitoring method according to claim 2, characterized in that: Before obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer, the method further includes: Using the first preset time period as a time keyword, filtering in the first preset database to obtain the historical equipment data of the transformer within the second preset time period; wherein the first preset database stores the correspondence between the preset time period and the equipment data of the transformer; The historical device data is input into the preset model to obtain the historical temperature data of the transformer output by the preset model; wherein the second preset time period includes the first preset time period and the next time period after the first preset time period.
4. The transformer monitoring method according to claim 1, characterized in that: Before obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer, the method further includes: Using the first preset time period as a time keyword, historical temperature data of the transformer within a second preset time period is obtained by screening in a second preset database; wherein the second preset database stores a correspondence between preset time periods and temperature data of the transformer, and the second preset time period includes the first preset time period and the next time period after the first preset time period.
5. The transformer monitoring method according to any one of claims 1 to 4, characterized in that: The step of obtaining predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and the historical temperature data of the transformer includes: Acquire temperature change data of the transformer in a time period most similar to a next time period after the first preset time period from the historical temperature data; The predicted temperature data of the transformer in a next time period after the first preset time period is obtained according to the temperature change data and the temperature data.
6. The transformer monitoring method according to any one of claims 1 to 4, characterized in that: The abnormal temperature rise monitoring of the transformer according to the predicted temperature data includes: Detecting whether the predicted temperature data contains data exceeding a preset temperature threshold; If there is data exceeding the preset temperature threshold in the predicted temperature data, it is determined that the transformer has a risk of abnormal temperature rise, and an early warning message is issued; wherein the early warning message is used to prompt that the transformer has a risk of abnormal temperature rise.
7. The transformer monitoring method according to any one of claims 2 to 4, characterized in that: Before inputting the device data into the trained preset model, the method further includes: Inputting equipment data of multiple transformers into a preset model to obtain temperature prediction data corresponding to each transformer output by the preset model; According to the actual temperature data of each transformer, the temperature prediction data corresponding to each transformer and the preset loss function, the model parameters of the preset model are adjusted until a trained preset model is obtained.
8. The transformer monitoring method according to claim 7, characterized in that: Before inputting the equipment data of the plurality of transformers into the preset model, the method further comprises: Data cleaning is performed on the equipment data of multiple transformers; wherein the data cleaning is used to remove invalid or abnormal data in the equipment data.
9. A transformer monitoring device, characterized in that: The device comprises: An acquisition unit, configured to acquire equipment data of the transformer within a first preset time period; wherein the equipment data includes data associated with the temperature of the transformer; A first processing unit, configured to determine temperature data of the transformer according to the device data; a second processing unit, configured to obtain predicted temperature data of the transformer in a next time period after the first preset time period based on the temperature data and historical temperature data of the transformer; wherein the historical temperature data is used to characterize historical temperature change data of the transformer; A monitoring unit is used to monitor abnormal temperature rise of the transformer according to the predicted temperature data.
10. The transformer monitoring device according to claim 9, characterized in that: The equipment data includes the load factor of the transformer, the ambient temperature of the transformer, and the ambient humidity of the transformer; The first processing unit is specifically configured to: Input the device data into a trained preset model; wherein the preset model is trained based on the device data and real temperature data of multiple transformers, and the preset model includes a convolution module, an attention mechanism module, and a long short-term memory module; Extracting local features of the device data according to the convolution module to obtain a first feature vector; According to the attention mechanism module, extract strongly correlated features of the first feature vector to obtain a second feature vector; Based on the long short-term memory module, feature selection processing is performed on the second feature vector to obtain temperature data of the transformer.