Method, system and equipment for cleaning and checking transformer oil chromatography online monitoring data and medium

Abstract

The invention belongs to the technical field of transformer monitoring, discloses a method, a system and equipment for cleaning and checking transformer oil chromatography online monitoring data and a medium, and aims to solve the problem of recognition accuracy. The method comprises the following steps: acquiring the concentration of the dissolved gas in the transformer oil to construct gas concentration time sequence data, identifying and eliminating missing values and / or zero values in the gas concentration time sequence data, filling by using a linear interpolation method, detecting other values by using a Z-Score method to determine an obviously abnormal value E1, and calculating the concentration of the dissolved gas in the transformer oil. Preprocessed gas concentration data are obtained from E1 data in gas concentration time sequence data, specific value characteristic data are constructed according to an IEC three-specific-value method and added, then points which do not belong to data clusters obtained through clustering are determined as outliers by applying a DBSCAN clustering algorithm, and the outliers are determined as outliers. And identifying the clustered data clusters with fuzzy boundaries, finding out points near the boundaries, marking the points as boundary points, summarizing to form an abnormal data set E < total > to be checked, approving the abnormal data set E < total > by adopting a Rogers ratio method, and judging whether the abnormal data set E < total > is abnormal or not.

Description

Technical Field

[0001] The present invention belongs to the technical field of transformer monitoring, and particularly relates to a method, system, device and medium for cleaning and verifying on-line monitoring data of transformer oil chromatography. Background Art

[0002] As an important device in the power system, the operating state of a transformer is directly related to the stability and reliability of the power system. By monitoring the transformer oil chromatography, potential fault hazards inside the transformer, such as overheating faults, discharge faults, etc., can be detected in a timely manner, so as to take corresponding maintenance and repair measures to avoid the further expansion of faults and ensure the normal operation of the power system. Transformer oil chromatography refers to the analysis of various gas components dissolved in transformer oil, and these gas components mainly include hydrogen (H2), methane (CH4), ethane (C2H6), ethylene (C2H4), acetylene (C2H2), carbon monoxide (CO), carbon dioxide (CO2), etc. Different types and degrees of transformer faults will cause changes in the types and contents of dissolved gases in the oil. Therefore, by analyzing the components and contents of dissolved gases in the oil, the operating state and fault type of the transformer can be judged.

[0003] Currently, common methods for cleaning and verifying transformer oil chromatography data include model-driven methods, classical machine learning classification algorithms and neural network algorithms. Specifically, model-driven methods include the David triangle method, the IEC three-ratio method, the Rogers ratio method and the Doernenburg ratio method. The core idea of these methods is to establish the correlation between gas content ratios and fault categories, and they have been the mainstream methods for abnormal analysis of oil chromatography data for a quite long time in the past. However, model-driven methods can only establish relatively simple mapping relationships, and the models do not have the ability to learn and self-improve, cannot capture new emerging rules in fault data, and the recognition accuracy is relatively low. Classical machine learning classification algorithms need to build models based on existing historical data, adjust internal model parameters through self-learning, establish a non-linear prediction model to identify abnormal data, or further use supervised learning classification methods to identify abnormal types. Neural network algorithms (including deep learning algorithms) can construct complex non-linear mapping relationships and can identify the hidden rules in the data through self-learning ability. Neural network algorithms need to carefully design the network structure and use high computing resources, and usually can obtain a relatively high recognition accuracy. However, classical machine learning classification algorithms and neural network algorithms need a large amount of historical data for self-learning, and the learning process also consumes high computing resources. At the same time, although they can fit the data well, they can only obtain an "implicit" mapping relationship between feature quantities and fault types, and the physical meaning is fuzzy, especially for deep learning methods, which lack interpretability. To sum up, the existing technologies still have defects in terms of recognition accuracy and interpretability. Summary of the Invention

[0004] In view of the above-mentioned shortcomings and deficiencies in the prior art, one of the objects of the present invention is to at least solve one or more of the above-mentioned problems in the prior art. In other words, one of the objects of the present invention is to provide a cleaning and verification method, system, equipment and medium for transformer oil chromatogram online monitoring data that meets one or more of the above-mentioned needs, so as to achieve the purpose of improving recognition accuracy.

[0005] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:

[0006] In a first aspect, the present invention provides a method for cleaning and verifying transformer oil chromatogram online monitoring data, comprising the steps of:

[0007] S1. Obtaining the concentration of dissolved gases in transformer oil within a preset time period and constructing gas concentration time series data, wherein the gases include hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide, and carbon dioxide;

[0008] S2. For the gas concentration time series data, identify and remove missing values and / or zero values therein, use linear interpolation to fill the removed data, and then use the Z-Score method to detect the remaining values, determine the obviously abnormal values and mark them as abnormal data E1;

[0009] S3. Eliminate the data marked as E1 from the gas concentration time series data to obtain preprocessed gas concentration data;

[0010] S4. Constructing ratio feature data according to the IEC three-ratio method and adding it to the preprocessed gas concentration data. Then, using the DBSCAN clustering algorithm, identifying points that do not belong to the clustered data clusters as outliers. At the same time, identifying data clusters with fuzzy boundaries after clustering and finding points near their boundaries as boundary points, thereby obtaining outlier data E2 and boundary point data E3;

[0011] S5. Aggregate data E1, E2 and E3 to form the abnormal data set E to be checked. total , Rogers ratio method is used to analyze the abnormal data set E total Verify and determine whether it is abnormal.

[0012] As a preferred solution, the linear interpolation method is used to fill in the eliminated data.

[0013] Where y1 and y2 are the gas measurement values of the two known data points before and after the missing value and / or zero value, x1 and x2 are the time corresponding to the two known data points before and after the missing value and / or zero value, and x missing are times where there are missing and / or zero values.

[0014] As a preferred solution, the threshold value detected by the Z-Score method in step S2 is set to 3, and its calculation formula is

[0015]

[0016] In the formula, x is a certain remaining value, μ is the mean value of all remaining values, and σ is the standard deviation of all remaining values.

[0017] As a preferred solution, the parameters of the DBSCAN clustering algorithm in step S4 include the neighborhood radius ∈ and the minimum number of samples min_samples.

[0018] As a preferred solution, the ratio characteristic data constructed according to the IEC three-ratio method in step S4 includes CH4 / H2, C2H4 / C2H6, and C2H2 / C2H4.

[0019] As a preferred solution, the step of identifying the data clusters with fuzzy boundaries after clustering in step S4 and finding the points near their boundaries and marking them as boundary points is as follows:

[0020] Calculate the distance between the data point and its neighboring points to obtain the neighborhood density, and combine the density threshold to determine whether the data point is located on the boundary of the data cluster;

[0021] Or calculate the distance between the data point and the center of the data cluster, and combine the distance threshold to determine whether the data point is located on the boundary of the data cluster.

[0022] As a preferred solution, the abnormal data set E total has the expression E total = E1 ∪ E2 ∪ E3.

[0023] In a second aspect, the present invention provides a cleaning and verification system for on-line monitoring data of transformer oil chromatography, which is used to implement the cleaning and verification method as described in the first aspect.

[0024] In a third aspect, the present invention provides an electronic device, where the computer device includes a memory, a processor, and a computer program, and when the computer program is executed by the processor, it implements the cleaning and verification method as described in the first aspect.

[0025] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by the processor, it implements the cleaning and verification method as described in the first aspect.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] 1. The present invention constructs ratio feature data according to the IEC three-ratio method and adds the preprocessed gas concentration data. By using the DBSCAN clustering algorithm, it can not only identify the outliers that do not belong to the data clusters obtained by clustering, but also identify the boundary points of the data clusters with unclear boundaries after clustering, greatly enhancing the comprehensiveness of abnormal data identification.

[0028] 2. The present invention aggregates the abnormal data identified by different methods (the abnormal data E1 obtained by processing missing values and zero values, the outlier data E2 and boundary point data E3 obtained by cluster analysis) to form an abnormal data set Etotal to be verified, and uses the Rogers ratio method for verification, which can more accurately judge whether the abnormal data is truly abnormal and improves the reliability of abnormal data verification.

[0029] Furthermore, more detailed beneficial effects will be described in combination with specific embodiments in the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0031] Figure 1 It is a schematic flowchart of the cleaning and verification method in Embodiment 1 of the present invention.

[0032] Figure 2 It is a schematic diagram of the principle of the DBSCAN clustering algorithm in Embodiment 1 of the present invention.

[0033] Figure 3 It is a structural diagram of the electronic device provided in the embodiment of the present invention.

[0034] Reference Numerals in the Drawings:

[0035] 300, Electronic device;

[0036] 301, Processor; 302, Communication bus; 303, User interface; 304, Network interface; 305, Memory. SPECIFIC IMPLEMENTATION MANNER

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention.

[0038] In the following description, multiple embodiments of the present invention are provided, and different embodiments can be replaced or combined. Therefore, the present invention can also be considered to include all possible combinations of the same and / or different embodiments described. Thus, if one embodiment includes features A, B, and C, and another embodiment includes features B and D, then the present invention should also be considered to include embodiments that contain one or more of all other possible combinations of A, B, C, and D, even though such embodiments may not be explicitly described in the following text.

[0039] The following description provides examples and does not limit the scope, applicability, or examples set forth in the claims. Changes can be made to the functions and arrangements of the elements described without departing from the scope of the present invention. Various processes or components can be appropriately omitted, substituted, or added to each example. For example, the methods described can be performed in a different order than the order described, and various steps can be added, omitted, or combined. Additionally, the features described for some examples can be combined into other examples.

[0040] To facilitate a better understanding of the embodiments of the present invention, before explaining the specific embodiments of the present invention in detail, its application scenarios will be described first.

[0041] The cleaning and verification method described in the embodiments of this specification is applied to the process of transformer operation status monitoring and fault diagnosis. In these scenarios, the application of the cleaning and verification method can accurately detect abnormalities from the transformer oil chromatogram data, providing a reference for transformer faults and having good application prospects.

[0042] Embodiment 1:

[0043] As Figure 1 shown, this embodiment provides a method for cleaning and verifying on-line monitoring data of transformer oil chromatogram, including the following steps 1 to 6.

[0044] Step 1: Obtain the time series data of the dissolved gas concentration in transformer oil, including 7 characteristic gases: hydrogen (H2), methane (CH4), ethane (C2H6), ethylene (C2H4), acetylene (C2H2), carbon monoxide (CO), and carbon dioxide (CO2), and arrange them in chronological order to obtain the dissolved gas concentration data set in transformer oil.

[0045] Step 2: Preprocess the oil chromatogram data. Specifically, identify the missing values and zero values and remove them, and use linear interpolation to fill the data. For the remaining values, detect the significantly abnormal values based on Z-Score, record the corresponding data as E1, and wait for processing in step 6.

[0046] More specifically, assume that the missing gas measurement value y missingLocated between the known values y1 and y2, where x1 and x2 are the corresponding times. The linear interpolation formula is:

[0047]

[0048] In Equation (1), y1 and y2 are the gas measurement values of two known data points before and after the missing value and / or zero value, x1 and x2 are the corresponding times of the two known data points before and after the missing value and / or zero value, and x missing is the time where the missing value and / or zero value is located.

[0049] More specifically, the filling step includes:

[0050] 1. Determine the position of the missing value: First, check which time points have missing or zero gas measurement values.

[0051] 2. Find adjacent known data points: For each missing value, find the two closest known data points before and after it. In this embodiment, preferably the points before and after in chronological order.

[0052] 3. Calculate the interpolation: Use the above linear interpolation formula to calculate the missing value. That is, based on the gas concentration values of the known data points before and after, infer the missing gas concentration.

[0053] 4. Replace the missing value: Use the interpolation result as the filling value for the missing value.

[0054] More specifically, calculate the Z-Score for each record using the following formula

[0055]

[0056] In Equation (2), x is the gas measurement value (such as the measurement values of hydrogen, carbon monoxide, etc.), μ is the mean (average value) of the gas measurement value, σ is the standard deviation (square root of the variance) of the gas measurement value, and Z-Score represents the standard deviation distance between the data point and the mean. It should be noted that when the Z-Score is large, it means that the data point deviates far from the mean and can be considered an outlier.

[0057] Specific steps for using Equation (2):

[0058] 1. Data preparation: Collect all gas measurement values (such as total hydrocarbons, hydrogen, carbon monoxide, carbon dioxide, etc.).

[0059] 2. Calculate the mean and standard deviation: For the data of each gas type, calculate the mean μ and standard deviation σ of the gas data.

[0060] 3. Calculate the Z-Score for each data point: Calculate the Z-Score for each measurement value using Equation (2).

[0061] 4. Set a threshold: Select a suitable Z-Score threshold (preferably 2 or 3 in this embodiment) to determine which data points are outliers.

[0062] 5. Mark outliers: If the Z-Score of a gas measurement value is greater than the set threshold (preferably 3 in this embodiment), then this measurement value is marked as an outlier E1.

[0063] Step 3: Construct three new features of CH4 / H2, C2H4 / C2H6, and C2H2 / C2H4 based on the IEC three-ratio method.

[0064] Step 4: Remove the abnormal data E1 after preprocessing, add the features constructed based on the IEC three-ratio, perform clustering using DBSCAN, and then conduct a detailed analysis for each cluster to identify the fault mode corresponding to each cluster.

[0065] Before performing clustering analysis, it is necessary to remove abnormal records and perform clustering in combination with the new features. It can be understood that DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is a density-based clustering method that can effectively discover clusters with different densities and can identify noise points (abnormal data points).

[0066] More specifically, the steps of performing clustering using DBSCAN include:

[0067] 1. Select clustering features: Use the columns containing the new features (such as CH4 / H2, C2H4 / C2H6, C2H2 / C2H4) as the input features for clustering.

[0068] 2. Set DBSCAN parameters: Neighborhood radius ∈ (determines the maximum distance between two sample points. If the distance is less than this value, it is considered that they belong to the same cluster) and minimum number of samples min_samples (the minimum number of samples required in a cluster to determine the definition of a core point).

[0069] 3. Execute clustering: Use the DBSCAN algorithm to perform clustering analysis on the cleaned data to obtain the cluster labels of each data point.

[0070] 4. After clustering, the data will be divided into multiple clusters. Analyze the fault mode for each cluster. Based on the gas ratios (e.g., CH4 / H2, C2H4 / C2H6, and C2H2 / C2H4) and the distribution of the clusters, the possible fault types corresponding to each cluster can be identified. The specific operations are as follows:

[0071] ① Analyze the gas ratio characteristics of clusters: For each cluster, analyze statistical characteristics such as the mean and standard deviation of the internal gas ratios to determine whether they conform to a certain typical fault mode.

[0072] ② Refer to IEC standards and historical data: According to IEC standards, combined with the gas ratios of clusters, identify possible fault modes. For example, some clusters may represent overheating faults, and others may represent discharge faults.

[0073] ③ Identify the fault mode of each cluster. Based on the statistical characteristics of gas ratios and historical experience, identify the possible fault modes corresponding to each cluster. In this embodiment, preferably, a high C2H2 / C2H4 ratio (possibly related to arc discharge), a high C2H4 / C2H6 ratio (possibly related to overheating faults), and a high CH4 / H2 ratio (possibly related to partial discharge).

[0074] Step Five: Find the data E2 corresponding to the outliers that do not belong to any cluster in the clustering results. Identify some clusters with unclear boundaries in the clustering, and find the data E3 corresponding to the points near the boundaries. It should be particularly noted that in DBSCAN clustering, outliers (Noise points) are usually not assigned to any cluster but are marked as -1 (the default mark in DBSCAN). In addition, DBSCAN may also generate some clusters with unclear boundaries, that is, there may be some sample points at the boundaries of the clusters. They have a lower similarity to other samples within the cluster but are still not marked as outliers. Through this step, this embodiment aims to identify these special points and determine whether they need further processing or verification.

[0075] 1. Find the outliers (E2) that do not belong to any cluster: In the results of DBSCAN clustering, for all outliers, the cluster labels will be marked as -1. Therefore, first, extract all the data points with cluster labels of -1 from the clustering results as outliers (E2), extract the data corresponding to these outliers, and mark it as E2.

[0076] 2. Identify clusters with unclear boundaries in the clustering results and find the data corresponding to the points near the boundaries, denoted as E3: It can be understood that in the clustering results, sometimes the boundaries of some clusters are not clear enough. The distances between the boundary points of these clusters and other points within the clusters are relatively large, but they are still classified into the same cluster. DBSCAN determines the boundaries of clusters based on density. However, in some cases, there may be fuzzy regions at the cluster boundaries, and the points in these regions may belong to this cluster or may belong to other clusters. One of the following methods can be selected to identify these boundary points: ① Calculate the neighborhood density of each data point: Determine whether it is located at the cluster boundary based on the neighborhood density of each data point (i.e., the distance between this point and its neighboring points). The points within the cluster generally have a higher density, while the density of the boundary points is relatively low. ② Identify boundary points: In DBSCAN clustering, the core points of a cluster usually have a higher density, while the density of the boundary points is lower. Boundary points can be identified based on the distance from a point to the cluster center or the density difference between a point and other points in its neighborhood. The points near the boundary may be the transitional points between the in-cluster points and the outliers, and they may be located in the edge region or the low-density region of the cluster. Select a certain threshold: Filter out the points near the boundary according to certain distance or density thresholds, and mark these points as E3.

[0077] Step 6. For the abnormal data E1, E2, and E3, use rule-based detection to check whether they are abnormal. In this embodiment, the Rogers ratio method is preferably used.

[0078] It can be understood that the Rogers ratio method is a classic gas chromatography analysis method, which is widely used for fault diagnosis of the results of dissolved gas analysis (DGA) in power transformer oil. It combines three ratios (CH4 / H2, C2H4 / C2H6, and C2H2 / C2H4) and combines with empirical rules to identify the fault types of transformers, such as overheating, discharge, etc. In addition, other rules (such as the IEC three-ratio method) can be introduced as a supplement.

[0079] 1. Data preparation: Aggregate the previously identified abnormal data E1, outliers E2, and boundary points E3 to form an abnormal data set E to be checked. total = E1 ∪ E2 ∪ E3. Specifically, summarize the data of E1, E2, and E3 into a unified table, including the original gas measurement values and the constructed ratio features.

[0080] 2. Rule detection based on the Rogers ratio method: According to the Rogers ratio method, calculate the ratios of CH4 / H2, C2H4 / C2H6, and C2H2 / C2H4, and determine whether they conform to the rules corresponding to certain fault types. Mark the fault types corresponding to the ratios in the data records according to the Rogers ratio method rule table.

[0081] 3. Verification using other rule methods (such as the IEC three-ratio method): Introduce the IEC three-ratio method as an auxiliary rule to further verify the data. The IEC three-ratio method classifies faults using the ratio ranges of CH4 / H2, C2H4 / C2H6, and C2H2 / C2H4. Specific operation: Calculate the IEC three-ratios for each record and determine whether the ratio conditions conform to a certain fault type. Cross-verify the results with the judgment results of the Rogers ratio method. If the two results are consistent, directly confirm the fault type of the data; if the results are inconsistent, mark it as "to be confirmed" and transfer it to an expert for further analysis.

[0082] 4. Output verification results: It can be understood that the data after rule detection is divided into three categories. That is, ① Confirmed abnormal: The data is judged to be abnormal by both the Rogers ratio method and the IEC three-ratio method. ② To be confirmed: The data fails the rule detection, or there are conflicts in the results of different rules. ③ Non-abnormal: The data passes the rule detection and is considered normal. Therefore, for the data confirmed to be abnormal, create a data table to record the ratio calculation results, fault type judgment results, and the final verification conclusion of each piece of data in E total ; for the data confirmed to be abnormal, continue to analyze the specific fault reasons. For the data to be confirmed, consider manual verification in combination with expert experience.

[0083] Example 2:

[0084] This example provides an on-line monitoring data cleaning and verification system for transformer oil chromatography, which is used to implement the cleaning and verification method described in Example 1.

[0085] Example 3:

[0086] As Figure 3 shown, this example provides an electronic device, which may include: at least one processor, at least one network interface, a user interface, a memory, and at least one communication bus.

[0087] Among them, the communication bus can be used to realize the connection and communication of the above-mentioned various components.

[0088] Among them, the user interface may include buttons, and the optional user interface may also include a standard wired interface and a wireless interface.

[0089] Among them, the network interface may but is not limited to include a Bluetooth module, an NFC module, a Wi-Fi module, etc.

[0090] Among them, the processor may include one or more processing cores. The processor uses various interfaces and circuits to connect various parts within the entire electronic device. By running or executing instructions, programs, code sets, or instruction sets stored in the memory, and by calling data stored in the memory, it executes various functions of the electronic device and processes data. Optionally, the processor may be implemented in at least one hardware form of DSP, FPGA, or PLA. The processor may integrate one or a combination of several of CPU, GPU, and modem, etc. Among them, the CPU mainly processes the operating system, user interface, application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communication. It can be understood that the above-mentioned modem may not be integrated into the processor and may be implemented separately by a single chip.

[0091] Among them, the memory may include RAM and may also include ROM. Optionally, the memory includes a non-transitory computer-readable medium. The memory can be used to store instructions, programs, code, code sets, or instruction sets. The memory may include a program storage area and a data storage area. Among them, the program storage area can store instructions for implementing the operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area can store the data involved in the above-mentioned various method embodiments. Optionally, the memory may also be at least one storage device located far from the aforementioned processor. As a computer storage medium, the memory may include an operating system, a network communication module, a user interface module, and a cleaning and verification application program. The processor can be used to call the cleaning and verification application program stored in the memory and execute the steps of the cleaning and verification method mentioned in the foregoing embodiments.

[0092] Embodiment 4:

[0093] This embodiment provides a computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When they run on a computer or a processor, they cause the computer or the processor to execute one or more of the steps in the above-mentioned Figure 1 illustrated embodiments. If each component module of the above-mentioned electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in the computer-readable storage medium.

[0094] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of this specification are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted through the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center in a wired manner (such as coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a Digital Versatile Disc (DVD)), or a semiconductor medium (for example, a Solid State Disk (SSD)), etc.

[0095] Those of ordinary skill in the art can understand that all or part of the processes in the method of the above-mentioned first embodiment can be completed by instructing relevant hardware through a computer program. This program can be stored in a computer-readable storage medium. When this program is executed, it can include the processes of the embodiments of the above-mentioned various methods. The foregoing storage medium includes: various media such as ROM, RAM, magnetic disk, or optical disc that can store program codes. Without conflict, the technical features in this embodiment and the implementation scheme can be combined arbitrarily.

[0096] Embodiment Five:

[0097] To verify the effectiveness of the method for cleaning and verifying on-line monitoring data of transformer oil chromatography described in this specification, in its actual application scenario, the following steps are executed:

[0098] Step 1, obtain the time series data of the dissolved gas concentration in the transformer oil as shown in Table 1.

[0099] Table 1

[0100]

[0101] Step 2, preprocess the oil chromatography data.

[0102] (1) Identify the missing values and zero values and remove them, and use linear interpolation to fill the data. As can be seen from Table 1, the carbon monoxide measurement value of Data No. 6 is 0, and the carbon dioxide measurement value is missing. In this embodiment, the linear interpolation method is used to combine Data No. 5 and Data No. 7 to obtain a carbon monoxide measurement value of 489.34 and a carbon dioxide measurement value of 652.925, which are filled into Data No. 6 to complete it. In addition, as can also be seen from Table 1, the acetylene measurement values of all data are 0, which is within the normal range, so no processing is done.

[0103] (2) Detect significantly abnormal values for the remaining values based on Z-Score. For the hydrogen measurement values, according to the formula The calculated mean is 11.16, the variance is 7.44, and the Z-Score (the standard deviation distance between the data point and the mean. When the Z-Score is large, it means that the data point deviates far from the mean and can be considered an outlier) is 1.1531, 1.1921, 1.0172, 0.6796, 0.7146, 0.6823, 0.3313, -1.0917, -1.1387, -1.1697, -1.1616, -1.2087. No significantly abnormal data is found. Calculations are performed on other gases, and no significantly abnormal data is found either. Therefore, E1 is empty, and no abnormal data is recorded in this step.

[0104] Step 3: Based on the IEC three-ratio method, construct three new features: CH4 / H2, C2H4 / C2H6, and C2H2 / C2H4. According to the calculation, the values of these three features are shown in Table 2.

[0105] Table 2

[0106] Serial number Time <![CDATA[CH4 / H2]]> <![CDATA[C2H4 / C2H6]]> <![CDATA[C2H2 / C2H4]]> 1 2018 / 3 / 5 0 0 0 2 2018 / 3 / 6 0 1 0 3 2018 / 3 / 7 0 0 0 4 2018 / 3 / 8 0 0 0 5 2018 / 3 / 9 0 0 0 6 2018 / 3 / 10 0 0 0 7 2018 / 3 / 11 0 0 0 8 2018 / 3 / 12 0 0 0 9 2018 / 3 / 13 0 0 0 10 2018 / 3 / 14 0 0 0 11 2018 / 3 / 15 0 0 0 12 2018 / 3 / 16 0 0 0

[0107] Step 4: Remove the abnormal data E1 after preprocessing, add the features constructed based on the IEC three-ratio, and use DBSCAN for clustering. Then, perform a detailed analysis on each cluster to identify the fault mode corresponding to each cluster. Set the neighborhood radius of DBSCAN to 0.5 and the minimum number of samples to 5. The clustering result is as Figure 2 shown. It can be seen that the data is clearly clustered into two categories. By taking representative data from these two categories for discrimination, it can be concluded that cluster0 is normal data and cluster1 is a low-temperature overheating fault.

[0108] Step 5: Find the data E2 corresponding to the outliers that do not belong to any cluster in the clustering result. Identify some clusters with unclear boundaries in the clustering and find the data E3 corresponding to the points near the boundaries. Since the data in this embodiment is small and the points near the boundaries cover a large area, in this embodiment, the representative Data No. 2 is taken and recorded as E3.

[0109] Step 6: For the abnormal data E1, E2, and E3, use rule-based detection (such as the Rogers ratio method) to verify whether they are abnormal. In this embodiment, only one abnormal data, i.e., the No. 2 data, is obtained (as shown in Table 3). For the No. 2 data, calculate its Rogers ratios as: 0.098, 0.862, and 0. According to the common fault modes and ratio relationships of the IEC 60599 standard, it can be judged that the No. 2 data may indicate a slight overload state, but there is no obvious fault occurring.

[0110] Table 3

[0111]

[0112] Based on the above, it can be seen that by applying the method for cleaning and verifying on-line monitoring data of transformer oil chromatography according to the present invention, on the basis of a small amount of historical data and using a small amount of computing resources, a high recognition accuracy can be achieved. That is, this embodiment verifies the effectiveness of the method for cleaning and verifying on-line monitoring data of transformer oil chromatography described in this specification.

[0113] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0114] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0115] The above are only the exemplary embodiments of the present invention, and the scope of the present invention cannot be limited thereby. That is, all equivalent changes and modifications made according to the teachings of the present invention still fall within the scope covered by the present invention. Those skilled in the art will easily think of other implementation manners of the present invention after considering the specification and practicing the disclosure herein. The present invention is intended to cover any variations, uses, or adaptive changes of the present invention, which follow the general principles of the present invention and include the common general knowledge or conventional technical means in the technical field not recorded in the present invention. The specification and the embodiments are only regarded as exemplary, and the scope and spirit of the present invention are defined by the claims.

Claims

1. A method for cleaning and verifying on-line monitoring data of transformer oil chromatography, characterized in that Including the steps: S1. Obtain the concentrations of dissolved gases in transformer oil within a preset time period, and construct gas concentration time series data, where the gases include hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide, and carbon dioxide; S2. For the gas concentration time series data, identify and remove the missing values and / or zero values therein, use the linear interpolation method to fill in the removed data, then use the Z-Score method to detect the remaining values, determine the significantly abnormal values and mark them as abnormal data E1; S3. Remove the data marked as E1 from the gas concentration time series data to obtain the preprocessed gas concentration data; S4. Construct ratio characteristic data according to the IEC three-ratio method, and add it to the preprocessed gas concentration data. Subsequently, use the DBSCAN clustering algorithm to identify the points that do not belong to the data clusters obtained by clustering as outliers. At the same time, identify the data clusters with fuzzy boundaries after clustering and find the points near their boundaries and mark them as boundary points, thereby obtaining outlier data E2 and boundary point data E3; S5. Aggregate the data E1, E2, and E3 to form an abnormal data set E to be verified total , and use the Rogers ratio method to process the abnormal data set E total for verification to determine whether it is abnormal.

2. A method for cleaning and verifying on-line monitoring data of transformer oil chromatography according to claim 1, characterized in that: The calculation formula for filling in the removed data by the linear interpolation method is Where y1 and y2 are the gas measurement values of two known data points before and after the missing value and / or zero value, x1 and x2 are the corresponding times of the two known data points before and after the missing value and / or zero value, and x missing is the time where the missing value and / or zero value is located.

3. A method for cleaning and verifying on-line monitoring data of transformer oil chromatography according to claim 1, characterized in that: The threshold value detected by the Z-Score method described in step S2 is set to 3, and its calculation formula is In the formula, x is a certain remaining value, μ is the mean of all remaining values, and σ is the standard deviation of all remaining values.

4. A method for cleaning and verifying on-line monitoring data of transformer oil chromatography according to claim 1, characterized in that: The parameters of the DBSCAN clustering algorithm in step S4 include the neighborhood radius ∈ and the minimum number of samples min_samples.

5. A method for cleaning and verifying on-line monitoring data of transformer oil chromatography according to claim 1, characterized in that: The ratio characteristic data constructed according to the IEC three-ratio method in step S4 includes CH4 / H2, C2H4 / C2H6, and C2H2 / C2H4.

6. A method for cleaning and verifying on-line monitoring data of transformer oil chromatography according to claim 1, characterized in that The steps for identifying the data clusters with fuzzy boundaries after clustering and finding the points near their boundaries and marking them as boundary points in step S4 are: Calculate the distance between the data point and its neighboring points to obtain the neighborhood density, and determine whether the data point is located on the boundary of the data cluster in combination with the density threshold; Or calculate the distance between the data point and the center of the data cluster, and determine whether the data point is located on the boundary of the data cluster in combination with the distance threshold.

7. A method for cleaning and verifying on-line monitoring data of transformer oil chromatography according to claim 1, characterized in that: The abnormal data set E total is expressed as E total = E1 ∪ E2 ∪ E3.

8. An on-line monitoring data cleaning and verification system for transformer oil chromatography, characterized in that, For implementing the cleaning and verification method according to any one of claims 1 to 7.

9. A computer device, the computer device comprising a memory, a processor, and a computer program, characterized in that, When the computer program is executed by a processor, it implements the cleaning and verification method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the cleaning and verification method according to any one of claims 1 to 7.

Images