A transformer time series data intelligent monitoring system and method based on prediction model

Abstract

The present invention relates to the technical field of time series data prediction, and specifically to a transformer time series data intelligent supervision system and method based on a prediction model. The present invention collects the operating parameters of the transformer during operation to obtain the change cycle of the operating parameters, extracts and detects the transformer oil according to the change cycle as the time interval, establishes a time series prediction model for the gas concentration of dissolved gas in the transformer oil, tests the time series prediction model, obtains the average absolute percentage error, obtains the optimal extraction and detection interval time according to the average absolute percentage error analysis, extracts and detects the transformer oil according to the optimal extraction and detection interval time, establishes various dissolved gas concentration time series prediction models, and performs prediction and supervision on the transformer. Finding the optimal extraction and detection interval time first ensures accurate prediction, and secondly avoids excessive data collection and waste of resources.

Description

Technical Field

[0001] The present invention relates to the technical field of time series data prediction, and in particular to a transformer time series data intelligent supervision system and method based on a prediction model. Background Art

[0002] A time series forecasting model is a model used to predict future trends or values ​​of time series data. In time series data, data points are arranged in chronological order. The main purpose of a time series forecasting model is to predict values ​​or trends at future time points based on past data patterns and regularities.

[0003] During the use of the transformer, the parameters of the transformer will change with time. In order to protect the transformer, transformer oil needs to be added to the transformer. The function of the transformer oil is insulation and heat dissipation. However, as the parameters of the transformer change, the transformer oil will react and produce dissolved gas. When the concentration of the dissolved gas reaches a certain level, the transformer will be dangerous. Therefore, a time series prediction model of dissolved gas is needed to monitor the safety of the transformer. However, when establishing the time series prediction model, it is necessary to find the optimal extraction detection interval time to collect the dissolved gas concentration. First, to ensure accurate prediction, and secondly to avoid excessive data collection and waste of resources. Summary of the invention

[0004] The purpose of the present invention is to provide a transformer time series data intelligent supervision system and method based on a prediction model to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a transformer time series data intelligent supervision method based on a prediction model, the method comprising the following steps:

[0006] Step S100: collecting operating parameters of the transformer during operation by means of a detection device, and obtaining a variation cycle of the operating parameters by analyzing a value set of the operating parameters;

[0007] Step S200: extracting and testing the transformer oil at intervals according to the change cycle of the operating parameters of the transformer during operation, and obtaining a gas concentration set of dissolved gas in the transformer oil;

[0008] Step S300: extract and detect the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, perform EMD empirical mode decomposition, input the decomposition result into the RNN recurrent network model, establish a time series prediction model of the gas concentration of dissolved gas obtained by extracting and detecting the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, test the time series prediction model, and obtain the mean absolute percentage error of the time series prediction model;

[0009] Step S400: sort the mean absolute percentage errors, analyze and obtain the optimal extraction and detection interval, extract and detect the transformer oil according to the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm when the predicted value exceeds the threshold.

[0010] Further, step S100 includes the following steps:

[0011] Step S101: by placing a detection device, the operating parameters of the transformer are collected at random time intervals to obtain various operating parameter value sets;

[0012] Step S102: Draw the i-th operating parameter value set into a curve graph, analyze the curve graph of the i-th operating parameter of the transformer, record the time period t when the slope value changes twice from the origin of the curve graph, and calculate the slope value according to the formula where a i It is expressed as the change period of the i-th operating parameter when the transformer is working, t1, t2, ...t n Represented as the 1st, 2nd, ..., nth time period.

[0013] The above steps are equivalent to the transformer operating parameters changing when the transformer is working. The operating parameters of the transformer are collected and a curve graph is drawn. According to the changing law of the curve graph, the changing cycle of the operating parameters of the transformer is obtained, which supports the subsequent analysis.

[0014] Further, step S200 includes the following steps:

[0015] Step S201: During the operation of the transformer, the transformer oil is treated according to a i The gas concentration of the rth dissolved gas in the transformer oil is recorded and obtained as a i The gas concentration set of the rth dissolved gas in the time interval is W ir ;

[0016] Step S202: According to the formula Where a represents the reference variation period, a1, a2, ..., a i They are respectively represented as the change cycles of the 1st, 2nd, ..., ith operating parameters when the transformer is working. The transformer oil is extracted and tested at the time interval of the reference change cycle, and the gas concentration of the rth dissolved gas in the transformer oil is recorded. The gas concentration set of the rth dissolved gas at the time interval of the reference change cycle is W r .

[0017] The above steps are equivalent to adding transformer oil inside the transformer to avoid danger to the transformer. However, during the use of the transformer, the transformer oil will produce dissolved gas. By extracting and testing the transformer oil according to the change cycle of the operating parameters, a gas concentration set of the dissolved gas is obtained. According to the change cycle of various operating parameters, a reference change cycle is established, and the transformer oil is extracted and tested according to the reference change cycle to obtain a reference set of dissolved gas concentration, which supports subsequent analysis.

[0018] Further, step S300 includes the following steps:

[0019] Step S301: W ir , W r Perform EMD empirical mode decomposition respectively, by inputting W ir , W r Into the EMD algorithm, W ir , W r Decompose into IMF components, obtain the time series of each IMF component, and determine the time series of the IMF component with the largest Pearson coefficient as the key sequence based on the Pearson coefficient of each IMF component;

[0020] Step S302: The key sequence data is integrated into the original time series data as features, and the integrated sequence data set is input into the RNN recurrent network model to establish W ir , W r The time series prediction model is M ir 、M r ;

[0021] Step S303: By ir 、M r Test and get M ir 、M r The mean absolute percentage error b ir 、b r , where b ir 、b r Respectively, the time series prediction model is M ir 、M r The mean absolute percentage error.

[0022] The above steps are equivalent to establishing different time series prediction models based on the dissolved gas concentration sets collected at different change periods, testing the different time series prediction models, and obtaining different mean absolute percentage errors, which supports subsequent analysis.

[0023] Further, step S400 includes the following steps:

[0024] Step S401: The average absolute percentage errors obtained by testing the time series prediction model of the gas concentration of the rth dissolved gas with a change period of various operating parameters as the time interval are aggregated to obtain {b 1r , b 2r , ..., b ir}, where b 1r , b 2r , ..., b ir are respectively expressed as the average absolute percentage error obtained by testing the time series prediction model of the gas concentration of the rth dissolved gas with the change period of the 1st, 2nd, ...ith operating parameters as the time interval. When the value in the set is less than b r When the value in the set is greater than or equal to b, the set is sorted from small to large, and the change period of the operating parameter of the minimum value of the set is the optimal extraction detection interval of the transformer oil. r When the transformer is working, the change cycles of various operating parameters are arranged from small to large, and the minimum value of the change cycle of the operating parameters is a min , the maximum value of the change cycle of the operating parameters is a max , according to the formula Where c1 and c2 represent the minimum value of the change cycle of the operating parameter and the reference change cycle, and the maximum value of the change cycle of the operating parameter and the average change cycle of the reference change cycle, respectively. By extracting and testing the transformer oil at time intervals of c1 and c2, the gas concentration of the rth dissolved gas in the transformer oil is recorded, and the gas concentration set W of the rth dissolved gas at time intervals of c1 and c2 is obtained. c1r , W c2r , establish W c1r , W c2r The time series prediction model is M c1r 、M c2r , the average absolute percentage error obtained through testing is b c1r 、b c2r ;

[0025] Step S402: b r 、b c1r 、b c2r Sorting from small to large, the minimum value change cycle is the optimal extraction and detection interval of transformer oil;

[0026] Step S403: extract and test the transformer oil according to the optimal extraction and detection interval, record the gas concentrations of various dissolved gases in the transformer oil, obtain a gas concentration set of various dissolved gases, establish a gas concentration time series prediction model for various dissolved gases, and predict the gas concentrations of various dissolved gases by inputting the predicted future time step. When the predicted gas concentration value of a dissolved gas exceeds a threshold, issue an early warning to maintain the transformer.

[0027] The above steps are equivalent to sorting and comparing the mean absolute percentage errors, and obtaining the optimal extraction and detection interval based on the comparison results. The transformer oil is extracted and tested according to the optimal extraction and detection interval to obtain the gas concentration set of various dissolved gases, and a time series prediction model for various dissolved gases is established. When the prediction result is abnormal, the transformer is maintained.

[0028] In order to better implement the above method, a transformer time series data intelligent supervision system based on a prediction model is also proposed, which is characterized in that the system includes a transformer parameter collection module, a gas concentration collection module, a time series prediction model module, and an analysis and early warning module:

[0029] Transformer parameter module: collects the operating parameters of the transformer during operation through a detection device, and obtains the change cycle of the operating parameters by analyzing the value set of the operating parameters;

[0030] Gas concentration collection module: extracts and detects transformer oil at time intervals according to the change cycle of the operating parameters of the transformer, and obtains the gas concentration set of dissolved gas in the transformer oil;

[0031] Time series prediction model module: extract and detect the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, perform EMD empirical mode decomposition, input the decomposition result into the RNN recurrent network model, establish a time series prediction model of the gas concentration of dissolved gas obtained by extracting and detecting the gas concentration set by taking the change period of various operating parameters as the time interval, test the time series prediction model, and obtain the mean absolute percentage error of the time series prediction model;

[0032] Analysis and early warning module: Sort the mean absolute percentage errors, analyze and obtain the optimal extraction and detection interval, extract and test the transformer oil according to the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm when the predicted value exceeds the threshold.

[0033] Furthermore, the transformer parameter module includes a transformer parameter acquisition unit and a transformer parameter change cycle analysis unit;

[0034] Transformer parameter collection unit: collects the operating parameters of the transformer during operation through the detection device;

[0035] Transformer parameter change cycle analysis unit: Create a curve chart for the collected data and analyze the change cycle of the operating parameters of the transformer when it is working.

[0036] Furthermore, the gas concentration collection module includes a standard gas concentration collection unit and a reference gas concentration collection unit;

[0037] Collecting standard gas concentration unit: extract and detect the dissolved gas in transformer oil according to the change cycle of different transformer operating parameters to obtain the gas concentration of dissolved gas in transformer oil;

[0038] Collection reference gas concentration unit: Analyze the change cycle of various operating parameters of the transformer to obtain the reference cycle, extract and detect the dissolved gas in the transformer oil according to the reference cycle, and obtain the gas concentration of the dissolved gas in the transformer oil.

[0039] Further, the time series prediction model module includes a time series prediction model establishment unit and a time series prediction model test unit;

[0040] Establishing a time series prediction model unit: extracting the detected gas concentration set of dissolved gas according to the change period of various operating parameters to establish a time series prediction model;

[0041] Test time series prediction model unit: extract the gas concentration set of dissolved gas detected during the change period of various operating parameters to establish a time series prediction model for testing, and obtain the average absolute percentage error of the time series prediction model.

[0042] Further, the analysis and warning module includes an analysis and collection time interval unit and an establishment warning unit;

[0043] Analyzing and collecting time interval unit: analyzing the mean absolute percentage error of the time series prediction model to obtain the optimal collection time interval;

[0044] Establish an early warning unit: extract and test the transformer oil at the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm if the prediction result is abnormal.

[0045] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention obtains the change cycle of the operating parameters by collecting the operating parameters during operation of the transformer, extracts and detects the transformer oil according to the change cycle as the time interval, establishes a time series prediction model for the gas concentration of the dissolved gas in the transformer oil, tests the time series prediction model, obtains the average absolute percentage error, obtains the optimal extraction and detection interval time according to the average absolute percentage error analysis, extracts and detects the transformer oil according to the optimal extraction and detection interval time, establishes various dissolved gas concentration time series prediction models, and performs prediction and supervision on the transformer. Finding the optimal extraction and detection interval time firstly ensures accurate prediction, and secondly avoids excessive data collection and waste of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0047] Figure 1 It is a flow chart of a transformer time series data intelligent supervision method based on a prediction model of the present invention;

[0048] Figure 2 It is a structural schematic diagram of a transformer time series data intelligent supervision system based on a prediction model of the present invention;

[0049] Figure 3 It is a schematic diagram of an embodiment of a method for intelligent monitoring of transformer time series data based on a prediction model of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] See also Figure 1 The present invention provides a technical solution: a dynamic sensing adjustment method for a multi-axis robot arm, the method comprising the following steps:

[0052] Step S100: collecting operating parameters of the transformer during operation by means of a detection device, and obtaining a variation cycle of the operating parameters by analyzing a value set of the operating parameters;

[0053] Wherein, step S100 includes the following steps:

[0054] Step S101: by placing a detection device, the operating parameters of the transformer are collected at random time intervals to obtain various operating parameter value sets;

[0055] Step S102: Draw the i-th operating parameter value set into a curve graph, analyze the curve graph of the i-th operating parameter of the transformer, record the time period t when the slope value changes twice from the origin of the curve graph, and calculate the slope value according to the formula where a i It is expressed as the change period of the i-th operating parameter when the transformer is working, t1, t2, ...t n Represented as the 1st, 2nd, ..., nth time period.

[0056] like Figure 3 As shown, f1, f2, f3, and f4 are respectively the positive and negative change points of the slope value of the oil temperature curve of the transformer oil. For example, the first time period when the slope value changes twice is the time period from the origin to f2, and the second time period is the time period from f2 to f4.

[0057] For example, when the transformer is working, the operating parameters include the oil temperature of the transformer oil and the load current of the transformer. Through data collection and calculation, it is found that the oil temperature change cycle of the transformer oil is 1.5 hours, and the load current change cycle of the transformer is 2.5 hours.

[0058] Step S200: extracting and testing the transformer oil at intervals according to the change cycle of the operating parameters of the transformer during operation, and obtaining a gas concentration set of dissolved gas in the transformer oil;

[0059] Wherein, step S200 includes the following steps:

[0060] Step S201: During the operation of the transformer, the transformer oil is treated according to a i The gas concentration of the rth dissolved gas in the transformer oil is recorded and obtained as a i The gas concentration set of the rth dissolved gas in the time interval is W ir ;

[0061] Step S202: According to the formula Where a represents the reference variation period, a1, a2, ..., a i They are respectively represented as the change cycles of the 1st, 2nd, ..., ith operating parameters when the transformer is working. The transformer oil is extracted and tested at the time interval of the reference change cycle, and the gas concentration of the rth dissolved gas in the transformer oil is recorded. The gas concentration set of the rth dissolved gas at the time interval of the reference change cycle is W r .

[0062] For example, the transformer oil is extracted and tested according to the oil temperature change cycle of the transformer oil, that is, the extraction and testing is performed at a time interval of 1.5 hours, and the concentration of dissolved gas hydrogen generated by the transformer oil is recorded to obtain a hydrogen concentration set at a time interval of 1.5 hours;

[0063] Extract and test the transformer oil according to the load current change cycle of the transformer, that is, extract and test at a time interval of 2.5 hours, record the concentration of dissolved gas hydrogen generated in the transformer oil, and obtain a hydrogen concentration set at a time interval of 2.5 hours;

[0064] According to the formula, 2 hours is the reference change period, and the transformer oil is extracted and tested according to the reference change period, that is, the extraction and testing are performed at a time interval of 2 hours, and the concentration of dissolved gas hydrogen generated by the transformer oil is recorded to obtain a hydrogen concentration set at a time interval of 2 hours;

[0065] Step S300: extract and detect the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, perform EMD empirical mode decomposition, input the decomposition result into the RNN recurrent network model, establish a time series prediction model of the gas concentration of dissolved gas obtained by extracting and detecting the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, test the time series prediction model, and obtain the mean absolute percentage error of the time series prediction model;

[0066] Wherein, step S300 includes the following steps:

[0067] Step S301: W ir , W r Perform EMD empirical mode decomposition respectively, by inputting W ir , W r Into the EMD algorithm, W ir , W r Decompose into IMF components, obtain the time series of each IMF component, and determine the time series of the IMF component with the largest Pearson coefficient as the key sequence based on the Pearson coefficient of each IMF component;

[0068] Step S302: The key sequence data is integrated into the original time series data as features, and the integrated sequence data set is input into the RNN recurrent network model to establish W ir , W r The time series prediction model is M ir 、M r ;

[0069] Step S303: By ir 、Mr Test and get M ir 、M r The mean absolute percentage error b ir 、b r , where b ir 、b r Respectively, the time series prediction model is M ir 、M r The mean absolute percentage error.

[0070] For example, the hydrogen concentration sets collected at time intervals of 1.5 hours, 2.5 hours, and 2 hours are used to establish a time series prediction model for hydrogen concentration collected at time intervals of 1.5 hours, 2.5 hours, and 2 hours through EMD empirical mode decomposition and RNN recurrent network model. By testing different time series prediction models, the average absolute percentage error of the time series prediction model for hydrogen concentration collected at time intervals of 1.5 hours is 10.5%.

[0071] The average absolute percentage error of the time series prediction model for hydrogen concentration collected at 2.5-hour intervals is 12.3%

[0072] The average absolute percentage error of the time series prediction model of hydrogen concentration collected at 2-hour intervals is 9.85%

[0073] Step S400: sort the mean absolute percentage errors, analyze and obtain the optimal extraction and detection interval, extract and detect the transformer oil according to the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm when the predicted value exceeds the threshold.

[0074] Wherein, step S400 includes the following steps:

[0075] Step S401: The average absolute percentage errors obtained by testing the time series prediction model of the gas concentration of the rth dissolved gas with a change period of various operating parameters as the time interval are aggregated to obtain {b 1r , b 2r , ..., b ir}, where b 1r , b 2r , ..., b ir are respectively expressed as the average absolute percentage error obtained by testing the time series prediction model of the gas concentration of the rth dissolved gas with the change period of the 1st, 2nd, ...ith operating parameters as the time interval. When the value in the set is less than b rWhen the value in the set is greater than or equal to b, the set is sorted from small to large, and the change period of the operating parameter of the minimum value of the set is the optimal extraction detection interval of the transformer oil. r When the transformer is working, the change cycles of various operating parameters are arranged from small to large, and the minimum value of the change cycle of the operating parameters is a min , the maximum value of the change cycle of the operating parameters is a max , according to the formula Where c1 and c2 represent the minimum value of the change cycle of the operating parameter and the reference change cycle, and the maximum value of the change cycle of the operating parameter and the average change cycle of the reference change cycle, respectively. By extracting and testing the transformer oil at time intervals of c1 and c2, the gas concentration of the rth dissolved gas in the transformer oil is recorded, and the gas concentration set W of the rth dissolved gas at time intervals of c1 and c2 is obtained. c1r , W c2r , establish W c1r , W c2r The time series prediction model is M c1r 、M c2r , the average absolute percentage error obtained through testing is b c1r 、b c2r ;

[0076] Step S402: b r 、b c1r 、b c2r Sorting from small to large, the minimum value change cycle is the optimal extraction and detection interval of transformer oil;

[0077] Step S403: extract and test the transformer oil according to the optimal extraction and detection interval, record the gas concentrations of various dissolved gases in the transformer oil, obtain a gas concentration set of various dissolved gases, establish a gas concentration time series prediction model for various dissolved gases, and predict the gas concentrations of various dissolved gases by inputting the predicted future time step. When the predicted gas concentration value of a dissolved gas exceeds a threshold, issue an early warning to maintain the transformer.

[0078] For example, the average absolute percentage error is sorted, and the minimum value is the time series prediction model of hydrogen concentration collected at a time interval of 2 hours. By calculation, the average change period of the minimum value of the change period and the reference change period is 1.75 hours, and the average change period of the maximum value of the change period and the reference change period is 2.25 hours. The transformer oil is extracted and tested at time intervals of 1.75 and 2.25 hours, and the concentration of dissolved gas hydrogen generated by the transformer oil is recorded to obtain a hydrogen concentration set at time intervals of 1.75 and 2.25 hours. The average absolute percentage error of the time series prediction model of hydrogen concentration collected at a time interval of 1.75 hours is 9.3%;

[0079] The average absolute percentage error of the time series prediction model of hydrogen concentration collected at an interval of 2.25 hours is 10.2%. By comparison, the optimal collection time interval is 1.75 hours.

[0080] The transformer oil is extracted and tested at intervals of 1.75 hours, and the concentrations of various dissolved gases produced by the transformer oil are recorded to obtain a gas concentration set of various dissolved gases. A time series prediction model for various dissolved gases is established, and the gas concentrations of various dissolved gases are predicted by inputting the predicted future time step. If there is an abnormal dissolved gas, an alarm is issued to maintain the transformer.

[0081] See also Figure 2 , also provides a transformer time series data intelligent supervision system based on the prediction model, the system includes: a transformer parameter collection module, a gas concentration collection module, a time series prediction model module, and an analysis and early warning module:

[0082] Transformer parameter module: collects the operating parameters of the transformer during operation through a detection device, and obtains the change cycle of the operating parameters by analyzing the value set of the operating parameters;

[0083] Among them, the transformer parameter module includes a transformer parameter acquisition unit and a transformer parameter change cycle analysis unit;

[0084] Transformer parameter collection unit: collects the operating parameters of the transformer during operation through the detection device;

[0085] Transformer parameter change cycle analysis unit: Create a curve chart for the collected data and analyze the change cycle of the operating parameters of the transformer when it is working.

[0086] Gas concentration collection module: extracts and detects transformer oil at time intervals according to the change cycle of the operating parameters of the transformer, and obtains the gas concentration set of dissolved gas in the transformer oil;

[0087] Wherein, the gas concentration collection module includes a standard gas concentration collection unit and a reference gas concentration collection unit;

[0088] Collecting standard gas concentration unit: extract and detect the dissolved gas in transformer oil according to the change cycle of different transformer operating parameters to obtain the gas concentration of dissolved gas in transformer oil;

[0089] Collection reference gas concentration unit: Analyze the change cycle of various operating parameters of the transformer to obtain the reference cycle, extract and detect the dissolved gas in the transformer oil according to the reference cycle, and obtain the gas concentration of the dissolved gas in the transformer oil.

[0090] Time series prediction model module: extract and detect the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, perform EMD empirical mode decomposition, input the decomposition result into the RNN recurrent network model, establish a time series prediction model of the gas concentration of dissolved gas obtained by extracting and detecting the gas concentration set by taking the change period of various operating parameters as the time interval, test the time series prediction model, and obtain the mean absolute percentage error of the time series prediction model;

[0091] Among them, the time series prediction model module includes a time series prediction model establishment unit and a time series prediction model test unit;

[0092] Establishing a time series prediction model unit: extracting the detected gas concentration set of dissolved gas according to the change period of various operating parameters to establish a time series prediction model;

[0093] Test time series prediction model unit: extract the gas concentration set of dissolved gas detected during the change period of various operating parameters to establish a time series prediction model for testing, and obtain the average absolute percentage error of the time series prediction model.

[0094] Analysis and early warning module: Sort the mean absolute percentage errors, analyze and obtain the optimal extraction and detection interval, extract and test the transformer oil according to the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm when the predicted value exceeds the threshold.

[0095] Among them, the analysis and warning module includes an analysis and collection time interval unit and an establishment warning unit;

[0096] Analyzing and collecting time interval unit: analyzing the mean absolute percentage error of the time series prediction model to obtain the optimal collection time interval;

[0097] Establish an early warning unit: extract and test the transformer oil at the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm if the prediction result is abnormal.

[0098] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0099] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A transformer time series data intelligent supervision method based on a prediction model, characterized in that: Methods include: Step S100: collecting operating parameters of the transformer during operation by means of a detection device, and obtaining a variation cycle of the operating parameters by analyzing a value set of the operating parameters; In S100, the specific steps are as follows: Step S101: by placing a detection device, the operating parameters of the transformer are collected at random time intervals to obtain various operating parameter value sets; Step S102: Draw the i-th operating parameter value set into a curve graph, analyze the curve graph of the i-th operating parameter of the transformer, record the time period t when the slope value changes twice from the origin of the curve graph, and calculate the slope value according to the formula where a i It is expressed as the change period of the i-th operating parameter when the transformer is working, t1, t2, ...t n Represented as the 1st, 2nd, ..., nth time period; Step S200: extracting and testing the transformer oil at intervals according to the change cycle of the operating parameters of the transformer during operation, and obtaining a gas concentration set of dissolved gas in the transformer oil; In S200, the specific steps are as follows: Step S201: During the operation of the transformer, the transformer oil is treated according to a i The gas concentration of the rth dissolved gas in the transformer oil is recorded and obtained as a i The gas concentration set of the rth dissolved gas in the time interval is W ir ; Step S202: According to the formula Where a represents the reference variation period, a1, a2, ..., a i They are respectively represented as the change cycles of the 1st, 2nd, ..., ith operating parameters when the transformer is working. The transformer oil is extracted and tested at the time interval of the reference change cycle, and the gas concentration of the rth dissolved gas in the transformer oil is recorded. The gas concentration set of the rth dissolved gas at the time interval of the reference change cycle is W r ; Step S300: extract and detect the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, perform EMD empirical mode decomposition, input the decomposition result into the RNN recurrent network model, establish a time series prediction model of the gas concentration of dissolved gas obtained by extracting and detecting the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, test the time series prediction model, and obtain the mean absolute percentage error of the time series prediction model; In S300, the specific steps are as follows: Step S301: W ir , W r Perform EMD empirical mode decomposition respectively, by inputting W ir , W r Into the EMD algorithm, W ir , W r Decompose into IMF components, obtain the time series of each IMF component, and determine the time series of the IMF component with the largest Pearson coefficient as the key sequence based on the Pearson coefficient of each IMF component; Step S302: The key sequence data is integrated into the original time series data as features, and the integrated sequence data set is input into the RNN recurrent network model to establish W ir , W r The time series prediction model is M ir 、M r ; Step S303: By ir 、M r Test and get M ir 、M r The mean absolute percentage error b ir 、b r , where b ir 、b r Respectively, the time series prediction model is M ir 、M r The mean absolute percentage error of Step S400: sort the mean absolute percentage errors, analyze and obtain the optimal extraction and detection interval, extract and detect the transformer oil according to the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm when the predicted value exceeds the threshold.

2. According to the method of intelligent monitoring of transformer time series data based on prediction model in claim 1, it is characterized in that: The step S400 includes the following steps: Step S401: The average absolute percentage errors obtained by testing the time series prediction model of the gas concentration of the rth dissolved gas with a change period of various operating parameters as the time interval are aggregated to obtain {b 1r , b 2r , ..., b ir }, where b 1r , b 2r , ..., b ir are respectively expressed as the average absolute percentage error obtained by testing the time series prediction model of the gas concentration of the rth dissolved gas with the change period of the 1st, 2nd, ...ith operating parameters as the time interval. When the value in the set is less than b r When the value in the set is greater than or equal to b, the set is sorted from small to large, and the change period of the operating parameter of the minimum value of the set is the optimal extraction detection interval of the transformer oil. r When the transformer is working, the change cycles of various operating parameters are arranged from small to large, and the minimum value of the change cycle of the operating parameters is a min , the maximum value of the change cycle of the operating parameters is a max , according to the formula Where c1 and c2 represent the minimum value of the change cycle of the operating parameter and the reference change cycle, and the maximum value of the change cycle of the operating parameter and the average change cycle of the reference change cycle, respectively. By extracting and testing the transformer oil at time intervals of c1 and c2, the gas concentration of the rth dissolved gas in the transformer oil is recorded, and the gas concentration set W of the rth dissolved gas at time intervals of c1 and c2 is obtained. c1r , W c2r , establish W c1r , W c2r The time series prediction model is M c1r 、M c2r , the average absolute percentage error obtained through testing is b c1r 、b c2r ; Step S402: b r 、b c1r 、b c2r Sorting from small to large, the minimum value change cycle is the optimal extraction and detection interval of transformer oil; Step S403: extract and test the transformer oil according to the optimal extraction and detection interval, record the gas concentrations of various dissolved gases in the transformer oil, obtain a gas concentration set of various dissolved gases, establish a gas concentration time series prediction model for various dissolved gases, and predict the gas concentrations of various dissolved gases by inputting the predicted future time step. When the predicted gas concentration value of a dissolved gas exceeds a threshold, issue an early warning to maintain the transformer.

3. A transformer time series data intelligent supervision system based on a prediction model, used to implement a transformer time series data intelligent supervision method based on a prediction model according to any one of claims 1 to 2, characterized in that: The system includes transformer parameter acquisition module, gas concentration acquisition module, time series prediction model module, and analysis and early warning module: Transformer parameter module: collects the operating parameters of the transformer during operation through a detection device, and obtains the change cycle of the operating parameters by analyzing the value set of the operating parameters; Gas concentration collection module: extracts and detects transformer oil at time intervals according to the change cycle of the operating parameters of the transformer, and obtains the gas concentration set of dissolved gas in the transformer oil; Time series prediction model module: extract and detect the gas concentration set of dissolved gas by taking the change period of various operating parameters as the time interval, perform EMD empirical mode decomposition, input the decomposition result into the RNN recurrent network model, establish a time series prediction model of the gas concentration of dissolved gas obtained by extracting and detecting the gas concentration set by taking the change period of various operating parameters as the time interval, test the time series prediction model, and obtain the mean absolute percentage error of the time series prediction model; Analysis and early warning module: Sort the mean absolute percentage errors, analyze and obtain the optimal extraction and detection interval, extract and test the transformer oil according to the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm when the predicted value exceeds the threshold.

4. A transformer time series data intelligent monitoring system based on a prediction model according to claim 3, characterized in that: The transformer parameter module includes a transformer parameter acquisition unit and a transformer parameter change cycle analysis unit; Transformer parameter collection unit: collects the operating parameters of the transformer during operation through the detection device; Transformer parameter change cycle analysis unit: Create a curve chart for the collected data and analyze the change cycle of the operating parameters of the transformer when it is working.

5. A transformer time series data intelligent monitoring system based on a prediction model according to claim 3, characterized in that: The gas concentration collection module includes a standard gas concentration collection unit and a reference gas concentration collection unit; Collecting standard gas concentration unit: extract and detect the dissolved gas in transformer oil according to the change cycle of different transformer operating parameters to obtain the gas concentration of dissolved gas in transformer oil; Collection reference gas concentration unit: Analyze the change cycle of various operating parameters of the transformer to obtain the reference cycle, extract and detect the dissolved gas in the transformer oil according to the reference cycle, and obtain the gas concentration of the dissolved gas in the transformer oil.

6. A transformer time series data intelligent monitoring system based on a prediction model according to claim 3, characterized in that: The time series prediction model module includes a time series prediction model establishment unit and a time series prediction model test unit; Establishing a time series prediction model unit: extracting the detected gas concentration set of dissolved gas according to the change period of various operating parameters to establish a time series prediction model; Test time series prediction model unit: extract the gas concentration set of dissolved gas detected during the change period of various operating parameters to establish a time series prediction model for testing, and obtain the average absolute percentage error of the time series prediction model.

7. A transformer time series data intelligent monitoring system based on a prediction model according to claim 3, characterized in that: The analysis and warning module includes an analysis and collection time interval unit and an establishment warning unit; Analyzing and collecting time interval unit: analyzing the mean absolute percentage error of the time series prediction model to obtain the optimal collection time interval; Establish an early warning unit: extract and test the transformer oil at the optimal extraction and detection interval, establish a time series prediction model for the gas concentration of various dissolved gases, and issue an alarm if the prediction result is abnormal.

Images