A transformer state quantity change trend prediction method and system

Abstract

The application discloses a transformer state quantity change trend prediction method and system, and the method steps are as follows: historical data of each perception quantity of the transformer is extracted and pretreated; a suitable prediction model and a prediction step are matched according to the change rule of the historical data of each state quantity; and each state quantity prediction data is analyzed and diagnosed. The system comprises a data extraction module, an analysis and judgment module and a result display module. The application can perform short-term trend prediction according to the historical data of the transformer state quantity, early warning of the fault defects of the transformer, and has high practical value for guaranteeing the safe operation of the power system.

Description

TECHNICAL FIELD

[0001] The present application relates to power equipment state prediction, in particular to a transformer state quantity change trend prediction method and system. BACKGROUND

[0002] In recent years, with the rapid development of national economy, the scale of power transformation equipment is expanding, and the traditional periodic operation and maintenance mode of equipment cannot meet the requirements of intelligent and lean management under the new situation.

[0003] The power transformer is the core equipment in the power system, mainly undertakes the tasks of voltage transformation and power transmission, and its safe operation is an important factor to ensure the stability of the power grid and the quality of power supply. At present, the service life of the transformers put into operation in China is increasing, and the related equipment materials are aging, the insulation performance is reduced, and the short-circuit impact resistance is weak, etc. Therefore, it is urgent to strengthen the monitoring of the health degree of such transformers and the early warning of potential fault defects, and to carry out targeted operation and maintenance to reduce the probability of failure and the loss caused by transformer failure. Therefore, the trend prediction of the state quantity of the power transformer is of great significance. SUMMARY

[0004] The purpose of the present application is to provide a transformer state quantity change trend prediction method and system, so as to predict the short-term trend of each state quantity of the power transformer, to warn potential fault defects, to improve the efficiency of transformer operation and maintenance, and to ensure the safe operation of the power grid.

[0005] The technical scheme of the present application is a transformer state quantity change trend prediction method, characterized in that it comprises the following steps:

[0006] (1) extracting the historical data of each state quantity of the transformer;

[0007] The historical data of each state quantity of the transformer includes oil temperature data, winding temperature data, current amplitude data, hydrogen data, carbon monoxide daily increment data, carbon dioxide daily increment data, methane data, ethylene data, acetylene data, ethane data, total hydrocarbon data, moisture data, local data, sleeve dielectric loss factor data, sleeve capacitance deviation data, iron core grounding current data and clamp grounding current data.

[0008] (2) judging the change rule of the historical data of each state quantity, and matching the corresponding prediction model and prediction step;

[0009] The step (2) is specifically:

[0010] judging the data change trend of each state quantity;

[0011] A, when the data trend is basically unchanged, linear fitting model is adopted, and the prediction step length per iteration is 1 / n of the historical data length, n can be adjusted according to the data length. It is suitable for slowly growing state variables in the observation period, such as the content of some gases in oil chromatogram, micro water in oil, casing capacity deviation, etc.

[0012] B, when the data trend is monotonically increasing or decreasing, the grey prediction model is adopted, and the prediction step length per iteration is 1 / n of the historical data length. It is suitable for characteristic variables with obvious growth in the observation period, such as the content of some gases in oil chromatogram under fault or defect state.

[0013] Specifically:

[0014] (1) from the historical input data constructing cumulative generation sequence

[0015]

[0016] (2) constructing the nearest mean generation sequence from the cumulative generation sequence

[0017]

[0018] (3) establishing grey differential equation set

[0019] that is

[0020]

[0021] wherein, known, unknown a and b

[0022] (4) after a and b are obtained, the prediction model is obtained

[0023]

[0024] Using the model, 10% x n new data are obtained per prediction.

[0025] C, when the data trend is non-monotonic and there is an inflection point in the observation interval, an adaptive grey prediction model is adopted; in the interval of the relative change rate of adjacent two data , all historical data in the interval are selected to participate in prediction, and the prediction step length per iteration is approximately 3 / n of the data length in the interval; in the interval of the relative change rate of adjacent two data , only the last 1 / 2 data in the interval are selected to participate in prediction, and the prediction step length per iteration is approximately 2 / n of the data length in the interval. It is suitable for characteristic variables with approximate periodic variation in the observation period, such as oil temperature, winding temperature, core clamp ground current, etc.

[0026] (3) analyzing and diagnosing the prediction data of each state quantity, if the state quantity prediction value is abnormal, diagnosing the components and fault causes that are likely to fail, and giving early warning of fault defects, wherein the state quantity prediction value being abnormal refers to exceeding the limit value or reaching a state that needs to be focused on.

[0027] Specifically,

[0028] (3.1) the prediction data of each state quantity obtained from step (2) is Y i ={y1,y2,…,y k}, wherein i represents the i-th state quantity, and k is the prediction data quantity of the i-th state quantity;

[0029] (3.2) for the prediction data of each state quantity, it is judged whether y j ,j∈[1,k] exceeds the limit value specified in the regulation specification, or is close to the attention value although it does not exceed the limit value, and then early warning information is given, and the time required for y j subscript to exceed the limit value or reach the attention value is obtained;

[0030] (3.3) for the state quantity showing abnormality, the fault defect component to which it belongs is located, the fault cause is inferred from the combination mode of the abnormal state quantity and the performance of the abnormal data, the operation and maintenance personnel are reminded to track and pay attention, and the maintenance plan is arranged.

[0031] A transformer state quantity change trend prediction system, the system adopts the transformer state quantity change trend prediction method described above, comprising the following modules:

[0032] A data extraction module for extracting historical data of each state quantity of the transformer;

[0033] An analysis and judgment module for judging the change rule of the historical data of each state quantity and matching the corresponding prediction model and prediction step length, and for analyzing and diagnosing the prediction data of each state quantity, if the state quantity prediction value is abnormal, diagnosing the components and fault causes that are likely to fail, and giving early warning of fault defects;

[0034] A result display module for displaying the curve of the change rule of the historical data of each state quantity, for displaying the alarm information of the state quantity prediction value being abnormal, and for displaying the components and fault causes that are likely to fail.

[0035] The historical data of each state quantity of the transformer includes oil temperature data, winding temperature data, current amplitude data, hydrogen data, carbon monoxide daily increment data, carbon dioxide daily increment data, methane data, ethylene data, acetylene data, ethane data, total hydrocarbon data, moisture data, local data, sleeve medium loss factor data, sleeve capacitance deviation data, core grounding current data, and clamp grounding current data.

[0036] The step of the analysis and judgment module judging the historical data change rule of each state quantity and matching the corresponding prediction model and prediction step length is:

[0037] When the data trend is basically unchanged, a linear fitting model is used, and the prediction step length is 1 / n of the historical data length each time, and n can be adjusted according to the data length;

[0038] When the data trend is monotonously increasing or decreasing, a grey prediction model is used, and the prediction step length is 1 / n of the historical data length each time;

[0039] When the data trend is non-monotonic and there is an inflection point in the observation interval, an adaptive grey prediction model is used; in the interval of the relative change rate of adjacent two data , all historical data in the interval are selected to participate in prediction, and the prediction step length is approximately 3 / n of the data length of the interval each time; in the interval of the relative change rate of adjacent two data , only the last 1 / 2 data in the interval are selected to participate in prediction, and the prediction step length is approximately 2 / n of the data length of the interval each time.

[0040] The state quantity prediction value anomaly refers to exceeding a limit value or reaching a state that needs to be focused on.

[0041] A computer storage medium has a computer program stored thereon, and the computer program is executed by a processor to implement the transformer state quantity change trend prediction method.

[0042] A computer device includes a storage, a processor, and a computer program stored on the storage and executable on the processor, and the processor implements the transformer state quantity change trend prediction method when executing the computer program.

[0043] Advantages: Compared with the prior art, the present application has the following advantages: the present application can perform short-term trend prediction according to historical data of transformer state quantities, and early warning of transformer faults and defects, which has high practical value for ensuring safe operation of a power system. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 The step flowchart of the method of the present application;

[0045] Figure 2 The transformer state quantity change trend prediction flowchart in the embodiment of the present application;

[0046] Figure 3 The transformer state quantity prediction data analysis and diagnosis flowchart in the embodiment of the present application. DETAILED DESCRIPTION

[0047] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

[0048] As Figure 1 shown, a method for predicting the change trend of transformer status quantities is as follows:

[0049] Step S1: Collect historical data of each transformer status quantity and preprocess the data.

[0050] Collect historical data of transformer status quantities in the past 24h according to the actual situation, such as oil chromatogram data, partial discharge data, bushing capacitance data, and core grounding data. Specifically include: oil temperature, winding temperature, current amplitude, hydrogen, carbon monoxide, carbon dioxide, methane, ethylene, acetylene, ethane, total hydrocarbons, moisture, local maximum discharge amount, discharge times, bushing dielectric loss factor, bushing capacitance, core grounding current, and clamp grounding current.

[0051] Preprocess the historical data of each status quantity, including filling in missing values and replacing outliers.

[0052] Step S2: According to the change rules of the historical data of each status quantity, match the prediction model and prediction step length, as Figure 2 shown, specifically including the following steps:

[0053] Step S2-1: Determine the change rule of the status quantity data.

[0054] After preprocessing the historical data of the transformer status quantity, it basically shows three trends: basically unchanged, monotonically changing, and non-monotonic (with inflection points). Different change trends require different prediction models. Therefore, it is necessary to first determine the change rule of the data.

[0055] For example, in the past 24 hours, the sampling frequency is one point every ⑤ minutes, and the historical data sample points n = 288 are obtained.

[0056] The data sequence X = {x1, x2,..., x n} and the mean value

[0057] (1) In this data sequence, when the number N of x that satisfies i ≥ n×90% = 259.2 (take 260), it is considered that the data trend is basically unchanged;

[0058] (2) In this data sequence, when the number N of x that satisfies i < n×90%, it is considered that the data trend has changed. Then, count the number of x i+1 -x i <0 as N1, and the number of x i+1 -x iThe number of <0 is N2, when N1 (or N2) ≥ n×90% (take 260), it is considered that the data trend changes monotonously; when n×10% (take 30) < N1 (or N2) < n×90% (take 260), it is considered that the data trend is non-monotonic, and there is an inflection point in the interval.

[0059] Step S2-2, according to the change rule, match the prediction model and the prediction step.

[0060] (1) For the data sequence which basically remains unchanged, a linear function y=ax+b is used to fit the historical data, and the prediction data is about n×10% = 28.8 (take 30);

[0061] (2) For the sequence with a large change in data trend and monotonicity, a gray prediction model is used, and the prediction data is about n×10% = 28.8 (take 30), which is specifically:

[0062] From the historical input data Construct the cumulative generating sequence

[0063]

[0064] Construct the close-to-mean generating sequence from the cumulative generating sequence

[0065]

[0066] Establish the gray differential equation set

[0067] That is

[0068]

[0069] Wherein, Given, the unknown quantities a and b can be obtained

[0070] After a and b are obtained, the prediction model is obtained

[0071]

[0072] Using the model, 10%×n = 28.8 (take 30) prediction data is obtained.

[0073] (3) For the sequence with a large change in data trend and an inflection point in the interval, an adaptive gray prediction model is used, that is, according to the relative change rate of adjacent data in the sequence, the sample data participating in prediction and the step of prediction data are determined.

[0074] (a) In the interval of the relative change rate of adjacent data , all historical data in the interval are selected to participate in prediction, and the prediction step is about 30% of the length of data in the interval each time;

[0075] (b) The relative rate of change between adjacent data For each interval, only the last 50% of the data within that interval is selected for prediction, and the prediction step size for each iteration is approximately 20% of the data length of that interval.

[0076] Taking transformer condition parameters such as oil temperature and winding temperature as an example, the temperature trough generally occurs between 6:00 and 9:00 AM, and the peak generally occurs between 6:00 and 9:00 PM. For data prediction within these two intervals, scheme (b) is used; for other intervals, scheme (a) is used.

[0077] Step S2-3: Based on the prediction model and prediction step size, obtain the prediction data for each state variable.

[0078] Step S3: Analyze and diagnose the predicted data to provide early warnings of faults and defects, as detailed below:

[0079] Step S3-1: Obtain the predicted data Y for each state variable from step S2. i ={y1,y2,…,y k}, where i represents the i-th state variable, and k is the predicted data amount of the i-th state variable. For the predicted data of each state variable, determine y j Whether j∈[1,k] exceeds the limit specified in the regulations, or although not exceeding the limit, it is close to the attention value, and from y j The index can indicate the time required to exceed the limit or reach the attention value, thus providing early warning information.

[0080] Step S3-2: For abnormal state quantities, the faulty or defective component to which they belong can be located.

[0081] Step S3-3: The cause of the fault can be inferred from the combination of abnormal state variables and the behavior of abnormal data.

[0082] Step S3-4: Based on the faulty component and possible causes, generate a fault analysis report, remind maintenance personnel to follow up and pay attention, and arrange a maintenance plan.

[0083] Taking the bushing medium loss factor as an example, the predicted data Y = {y1, y2, ..., y} is obtained from the historical 24-hour data in step S2. 30 According to the regulations, the normal range for the dielectric loss factor of 550kV bushings is [0.4, 0.7]. Each yi in the prediction sequence is checked to determine whether it is within the normal range or close to the limit (for example, [0.4, 0.45] and [0.65, 0.7] can be set as intervals requiring special attention); and the time i×Δt to reach the abnormal point can be calculated based on the index i of yi and the sampling time Δt, thus providing early warning information.

[0084] If the state quantity bushing dielectric loss factor data is abnormal, the fault component is determined as the bushing, and the possible causes of the fault are inferred as: general cap loosening, conductive rod loosening, poor contact of the bushing end screen, internal moisture or breakdown discharge of the bushing, etc. Then, whether the bushing is internally moist or discharged can be determined according to the bushing capacitance deviation data. Whether the end screen is in poor contact can be determined through the end screen dielectric loss test data, so as to gradually narrow down the fault range and give a diagnosis conclusion.

[0085] Finally, the prediction system pushes the prediction data and abnormal analysis report to the operation and maintenance personnel, reminds the components and state quantities that need to be focused on, and recommends the next test and maintenance plan, which can assist the operation and maintenance personnel in decision-making and greatly improve the intelligent operation and inspection level of the power transformation equipment.

[0086] In this embodiment, the state quantity prediction is displayed in the form of a curve graph, and the prediction data analysis and fault early warning are displayed in the form of text information.

[0087] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer usable storage media containing computer usable program code (including but not limited to disk storage, CD-ROM, optical storage, etc.). The solutions in the embodiments of the present application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted script language JavaScript.

[0088] The present application is described with reference to flowcharts and / or block diagrams according to the methods, devices (systems), and computer program products of the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one flow or multiple flows and / or blocks. Figure 1 The devices that implement the functions specified in one block or multiple blocks.

[0089] These computer program instructions can also be stored in a computer readable storage medium that can guide the computer or other programmable data processing devices to work in a specific manner, so that the instructions stored in the computer readable storage medium produce a manufactured product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one flow or multiple flows and / or blocks. Figure 1the function specified in the one or more blocks.

[0090] These computer program instructions can also be loaded into computer or other programmable data processing devices, so that a series of operations steps are performed on the computer or other programmable data processing devices to generate computer-implemented processes, thus the instructions executed on the computer or other programmable data processing devices provide processes for implementing the flows Figure 1 the flow or flows and / or blocks Figure 1 the steps of the function specified in the one or more blocks.

[0091] Although preferred embodiments of the application have been described, those skilled in the art will recognize that additional modifications and variations can be made thereto without departing from the spirit and scope of the application. It is therefore intended that the appended claims cover all such modifications and variations as fall within the scope of the application.

[0092] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore intended that the application be covered within the scope of the claims, and that the scope of the claims be interpreted not to be limited by the embodiments described above but by the claims themselves as permitted by statute.

Claims

1. A method for predicting the trend of transformer state variables, characterized in that, Includes the following steps: (1) Extract historical data of various state quantities of the transformer; (2) Determine the data change trend of each state variable: When the data trend is basically unchanged, a linear fitting model is used, and the prediction step size for each iteration is 1 / n of the historical data length; when the data trend is increasing or decreasing, a grey prediction model is used, and the prediction step size for each iteration is 1 / n of the historical data length; when the data trend is not monotonic and there is an inflection point in the observation interval, an adaptive grey prediction model is used; the relative change rate between two adjacent data points... The prediction process involves selecting all historical data within a given interval for prediction, with each iteration's prediction step size ≈ 3 / n of the data length within that interval; and considering the relative rate of change between adjacent data points. For each interval, only the last half of the data within that interval is selected for prediction. The prediction step size for each iteration is approximately equal to the data length of the interval (2 / n). Here, n is the number of historical data sample points, which is determined based on the data collection time span and sampling frequency. (3) Analyze and diagnose the predicted data of each state quantity. If the predicted value of the state quantity is abnormal, diagnose the components that may fail and the cause of the failure, and give early warning of the failure defects.

2. The method for predicting the trend of transformer state variables according to claim 1, characterized in that, The historical data of various transformer state quantities include oil temperature data, winding temperature data, current amplitude data, hydrogen data, daily increase data of carbon monoxide, daily increase data of carbon dioxide, methane data, ethylene data, acetylene data, ethane data, total hydrocarbon data, moisture data, local data, bushing dielectric loss factor data, bushing capacitance deviation data, core grounding current data, and clamp grounding current data.

3. The method for predicting the trend of transformer state variables according to claim 1, characterized in that, The abnormal state value prediction mentioned in step (3) refers to exceeding the limit or reaching a state that requires special attention.

4. The method for predicting the trend of transformer state variables according to claim 1, characterized in that, Step (3) specifically involves: (3.1) The predicted data of each state variable obtained from step (2) is Y. i ={y1,y2,…,y k }, where i represents the i-th state variable, and k is the predicted data amount of the i-th state variable; (3.2) For the predicted data of each state variable, determine y j If j∈[1,k] exceeds the limit specified in the regulations, or if it is close to the warning value even though it has not exceeded the limit, an early warning message will be given in advance, starting from y j The time required for the index to exceed the limit or reach the attention value; (3.3) For abnormal state quantities, locate the faulty or defective component to which it belongs, infer the cause of the fault from the combination of abnormal state quantities and the performance of abnormal data, remind maintenance personnel to follow up and pay attention, and arrange maintenance plans.

5. A transformer state quantity change trend prediction system, wherein the system employs a transformer state quantity change trend prediction method as described in any one of claims 1-4, characterized in that, Includes the following modules: Data extraction module: used to extract historical data of various state quantities of the transformer; Analysis and Judgment Module: Used to judge the historical data change patterns of each state variable and match the corresponding prediction model and prediction step size. Specific steps include: (1) Determine the data change trend of each state variable; (2) When the data trend remains basically unchanged, a linear fitting model is used, and the prediction step size for each iteration is 1 / n of the historical data length. (3) When the data trend is increasing or decreasing, a gray prediction model is used, and the prediction step size is 1 / n of the historical data length. (4) When the data trend is not monotonic and an inflection point appears within the observation interval, an adaptive grey prediction model is adopted; the relative change rate between two adjacent data points... The prediction process involves selecting all historical data within a given interval for prediction, with each iteration's prediction step size ≈ 3 / n of the data length within that interval; and considering the relative rate of change between adjacent data points. For each interval, only the last half of the data within that interval is selected for prediction, and the prediction step size for each iteration is approximately 2 / n of the data length of that interval. Where n is the number of historical data sample points, which is determined according to the data collection time span and sampling frequency; the analysis and judgment module is also used to analyze and diagnose the predicted data of each state quantity. If the predicted value of the state quantity is abnormal, it diagnoses the components that may fail and the causes of the failure, and provides early warning of the failure defects. Results display module: used to display curves showing the historical data change patterns of each state variable, to display alarm information for abnormal predicted values ​​of state variables, and to display components that may fail and the causes of the failure.

6. The transformer state variable change trend prediction system according to claim 5, characterized in that, The historical data of various transformer state quantities include oil temperature data, winding temperature data, current amplitude data, hydrogen data, daily increase data of carbon monoxide, daily increase data of carbon dioxide, methane data, ethylene data, acetylene data, ethane data, total hydrocarbon data, moisture data, local data, bushing dielectric loss factor data, bushing capacitance deviation data, core grounding current data, and clamp grounding current data.

7. The transformer state variable change trend prediction system according to claim 5, characterized in that, The aforementioned abnormal state quantity prediction value refers to exceeding the limit or reaching a state that requires special attention.

8. A computer storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements a method for predicting the trend of transformer state variables as described in any one of claims 1-4.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a method for predicting the trend of transformer state variables as described in any one of claims 1-4.

Images