Transformer abnormality monitoring method and system based on acoustic signal and frequent item set mining

Abstract

The present invention discloses a transformer abnormality monitoring method and system based on acoustic signal and frequent item set mining. The method of the present invention comprises: obtaining the real-time acoustic signal of the transformer; extracting features of the real-time acoustic signal of the transformer to obtain multidimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters; discretizing the values ​​of the multidimensional features to obtain a discrete data table, and performing frequent item mining on the discrete data table using a preset frequent item set mining algorithm according to the discrete data table to obtain frequent item sets; if the number of frequent item sets exceeds the abnormal alarm line of the frequent item sets under the preset normal state, it is determined that the transformer is currently in an abnormal state, otherwise it is determined that the transformer is currently in a normal state. The present invention has the characteristics of not relying on expert experience and high diagnostic accuracy, and can be effectively applied to the diagnosis of transformer operation status in complex noise environments.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer abnormality monitoring, and in particular to a transformer abnormality monitoring method and system based on acoustic signal and frequent item set mining. Background Art

[0002] The transformer is one of the key equipment in the power system. Any failure of the transformer will bring safety hazards and significant economic losses, directly affecting the reliability and safety of the power grid. At the same time, the power system is becoming more complex, efficient and intelligent, while the operating environment of the transformer equipment is relatively harsh. There are problems such as limited space, complex wiring structure, and strong electromagnetic signal influence during operation, which makes it difficult to repair the transformer equipment. During long-term operation, the transformer will be subjected to voltage shock, thermal shock and mechanical vibration shock, resulting in frequent failures. Therefore, intelligent abnormal monitoring of the transformer operation status is of great significance to maintaining the safety of the power grid, reducing operation and maintenance costs, and protecting people's livelihoods. Summary of the invention

[0003] The technical problem to be solved by the present invention is as follows: In view of the above-mentioned problems in the prior art, a transformer abnormality monitoring method and system based on acoustic signals and frequent item set mining are provided. The present invention has the characteristics of being independent of expert experience and having high diagnostic accuracy, and can be effectively applied to the diagnosis of transformer operating status in complex noise environments.

[0004] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0005] A transformer anomaly monitoring method based on acoustic signals and frequent item set mining, comprising:

[0006] S101, obtaining a real-time acoustic signal of the transformer;

[0007] S102, extracting features of the real-time acoustic signal of the transformer to obtain multi-dimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters;

[0008] S103, discretizing the values ​​of the multidimensional features to obtain a discrete data table, and performing frequent item mining on the discrete data table using a preset frequent item set mining algorithm to obtain frequent item sets;

[0009] S104: If the number of frequent item sets exceeds the abnormal alarm line of the frequent item sets under the preset normal state, it is determined that the transformer is currently in an abnormal state; otherwise, it is determined that the transformer is currently in a normal state.

[0010] Optionally, before step S104, the process further includes determining an abnormal alarm line of a frequent item set under a normal state:

[0011] S201, obtaining a historical acoustic signal of the transformer;

[0012] S202, extracting features from the historical acoustic signal of the transformer to obtain multi-dimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters;

[0013] S203, discretizing the values ​​of the multidimensional features to obtain a discrete data table, and using a preset frequent item set mining algorithm to perform frequent item mining on the discrete data table to obtain frequent item sets, and determining the abnormal alarm line of the frequent item sets under a preset normal state according to the number of the frequent item sets obtained by mining.

[0014] Optionally, before extracting features from the historical acoustic signals of the transformer in step S101, a step of cleaning data of the historical acoustic signals of the transformer is also included.

[0015] Optionally, the data cleaning includes at least one of outlier discarding, missing value filling and data reduction.

[0016] Optionally, the multidimensional features include part or all of peak value, peak-to-peak value, root mean square value, standard deviation, slope, kurtosis, waveform index, margin index, and pulse index.

[0017] Optionally, the discretization of the values ​​of the multidimensional features refers to a ten-division binning operation, in which the value interval is divided into ten equal parts according to the maximum and minimum values ​​of each dimensional feature and corresponding labels are given, all values ​​within a certain value interval share the same label, and a discrete data table is established according to the labels based on the values ​​of the divided features.

[0018] Optionally, the preset frequent itemset mining algorithm is a Relim algorithm.

[0019] Optionally, the operation of determining the abnormal alarm line of the frequent itemsets under the normal state is a timed operation to implement a timed update of the abnormal alarm line of the frequent itemsets under the normal state.

[0020] In addition, the present invention also provides a transformer abnormality monitoring system based on acoustic signals and frequent item set mining, comprising a microprocessor and a memory connected to each other, wherein the microprocessor is configured to execute the transformer abnormality monitoring method based on acoustic signals and frequent item set mining.

[0021] In addition, the present invention also provides a computer-readable storage medium, in which a computer program is stored, and the computer program is used to be configured by a microprocessor to execute the transformer abnormality monitoring method based on acoustic signals and frequent item set mining.

[0022] Compared with the prior art, the present invention mainly has the following advantages: the method of the present invention includes obtaining the real-time acoustic signal of the transformer; extracting the features of the real-time acoustic signal of the transformer to obtain multidimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters; discretizing the values ​​of the multidimensional features to obtain a discrete data table, and using a preset frequent item set mining algorithm to mine the discrete data table for frequent items according to the discrete data table to obtain frequent item sets; if the number of frequent item sets exceeds the abnormal alarm line of the frequent item sets under the preset normal state, it is determined that the transformer is currently in an abnormal state, otherwise it is determined that the transformer is currently in a normal state. The present invention discretizes the values ​​of the multidimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters, and uses a preset frequent item set mining algorithm to mine the discrete data table for frequent items to obtain frequent item sets for abnormal state identification. It has the characteristics of not relying on expert experience and high diagnostic accuracy, and can be effectively applied to transformer operation state diagnosis in complex noise environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the basic flow of the method of the embodiment of the present invention.

[0024] Figure 2 It is a schematic diagram of model training and application in an embodiment of the present invention;.

[0025] Figure 3 It is a schematic diagram of discretized feature representation in an embodiment of the present invention.

[0026] Figure 4 Schematic diagram of frequent item set mining results in an embodiment of the present invention.

[0027] Figure 5 It is the monitoring result of the preferred embodiment in the embodiment of the present invention. DETAILED DESCRIPTION

[0028] like Figure 1 As shown, the transformer abnormality monitoring method based on acoustic signals and frequent item set mining in this embodiment includes:

[0029] S101, obtaining a real-time acoustic signal of the transformer;

[0030] S102, extracting features of the real-time acoustic signal of the transformer to obtain multi-dimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters;

[0031] S103, discretizing the values ​​of the multidimensional features to obtain a discrete data table, and performing frequent item mining on the discrete data table using a preset frequent item set mining algorithm to obtain frequent item sets;

[0032] S104: If the number of frequent item sets exceeds the abnormal alarm line of the frequent item sets under the preset normal state, it is determined that the transformer is currently in an abnormal state; otherwise, it is determined that the transformer is currently in a normal state.

[0033] In this embodiment, mining and obtaining frequent itemsets in step S103 can be regarded as the application process of the frequent itemset mining algorithm model, and the process of obtaining the abnormal alarm line of the frequent itemsets under the preset normal state in step S104 can be regarded as the training process of the frequent itemset mining algorithm model.

[0034] like Figure 2 As shown, before step S104, this embodiment also includes determining the abnormal alarm line of the frequent itemsets under normal conditions:

[0035] S201, obtaining a historical acoustic signal of the transformer;

[0036] S202, extracting features from the historical acoustic signal of the transformer to obtain multi-dimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters;

[0037] S203, discretizing the values ​​of the multidimensional features to obtain a discrete data table, and using a preset frequent item set mining algorithm to perform frequent item mining on the discrete data table to obtain frequent item sets, and determining the abnormal alarm line of the frequent item sets under a preset normal state according to the number of the frequent item sets obtained by mining.

[0038] In this embodiment, before extracting features from the historical acoustic signal of the transformer in step S101, a step of data cleaning of the historical acoustic signal of the transformer is also included. Generally speaking, data cleaning includes at least one of discarding abnormal values, filling missing values, and data reduction (normalizing / standardizing data according to a specified data interval).

[0039] In this embodiment, the multidimensional features include some or all of the peak value, peak-to-peak value, RMS value, standard deviation, slope, kurtosis, waveform index (ratio of RMS value to absolute average value), margin index (ratio of peak value to square root amplitude), and pulse index (ratio of peak value to absolute average value). It should be noted that the peak value, peak-to-peak value, RMS value, standard deviation, slope, kurtosis, waveform index, margin index, and pulse index are all existing known parameters. This embodiment only involves the application of the above-mentioned known parameters, so the detailed calculation method thereof will not be described in detail here.

[0040] In this embodiment, discretization of the values ​​of multi-dimensional features refers to the use of a ten-division binning operation, where the value interval is divided into ten equal parts according to the maximum and minimum values ​​of each dimensional feature and given corresponding labels, all values ​​within a certain value interval share the same label, and a discrete data table is established according to the labels based on the values ​​of the divided features. Undoubtedly, other feasible discretization methods can also be used as needed.

[0041] In this embodiment, the preset frequent item set mining algorithm is the Relim algorithm. The Relim algorithm is used to obtain deep correlation information between features, which can ensure the accuracy of its abnormal state identification. The Relim algorithm is a well-known frequent item set mining algorithm. This embodiment only involves the application of this method, and does not involve the improvement of this method, so its implementation details are not described in detail here.

[0042] In this embodiment, the operation of determining the abnormal alarm line of the frequent item set under the normal state is performed as a timed operation to realize the timed update of the abnormal alarm line of the frequent item set under the normal state, thereby ensuring that during the loop execution of steps S101 to S104, the abnormal alarm line of the frequent item set under the normal state is always updated regularly to improve the monitoring accuracy.

[0043] In order to verify the actual effect of the transformer abnormality monitoring method based on acoustic signals and frequent item set mining in this embodiment on the intelligent recognition of transformer acoustic signals, a large amount of historical operating acoustic signal data was collected from Changsha 110kv Mawangdui Substation, Changsha 110kv Hefeng Substation, Changsha 110kv Sports New City Field Substation and Loudi Jiulun Substation. The results of data discretization and frequent item set mining for one of the signals are shown in the figure below. Figure 3 and Figure 4 shown. Figure 3 The discrete data table shown contains 10 feature dimensions, each of which contains 10 numerical partitions. Each feature dimension corresponds to a numerical partition and shares a discrete label. For example Figure 3 The discrete label j2 pointed by the middle arrow is the discrete label of the 10th dimension and the 4th numerical partition. Figure 4 In the results of frequent item set mining shown in the figure, the number of frequent item sets changes continuously with the change of time. The number of frequent item sets is directly related to the abnormal state of the transformer. Therefore, the abnormal alarm line of the frequent item sets under the preset normal state can be determined according to the number of frequent item sets obtained by mining. The abnormal alarm line of the frequent item sets under the preset normal state should be greater than the number of frequent item sets under the normal state, and the required margin can be increased as needed, so as to accurately distinguish the operating state of the transformer. Figure 5The black circle is the alarm point triggered by the captured abnormal state. It can be found that the transformer abnormality monitoring method based on acoustic signal and frequent item set mining in this embodiment can effectively monitor the operation of the equipment.

[0044] In summary, the transformer is one of the key equipment in the power system. Any failure of the transformer will bring safety hazards and significant economic losses, directly affecting the reliability and safety of the power grid. In the actual operation process, the acoustic signal of the transformer contains strong noise, its frequency components are complex, there are many signal interference sources, and the state monitoring is difficult. The transformer abnormality monitoring method based on acoustic signals and frequent item set mining in this embodiment overcomes the problem of poor diagnostic performance caused by the difficulty of transformer fault feature extraction in strong noise conditions in the existing methods, and improves the state monitoring capability and practical application value. First, based on the transformer historical data, multi-perspective feature parameters are extracted, and then the deep connection of feature parameters is analyzed by frequent item sets, and the transformer health state mapping space is constructed to realize intelligent monitoring and analysis of the transformer operation state. The transformer abnormality monitoring method based on acoustic signals and frequent item set mining in this embodiment realizes accurate abnormality monitoring of the transformer in the absence of fault data by learning the correlation between the transformer operation state characteristics. Compared with the traditional abnormality monitoring method, it has better monitoring performance in the face of strong noise environment and is more in line with the actual scenario of industrial application.

[0045] In addition, this embodiment also provides a transformer abnormality monitoring system based on acoustic signals and frequent item set mining, including a microprocessor and a memory connected to each other, and the microprocessor is configured to execute the aforementioned transformer abnormality monitoring method based on acoustic signals and frequent item set mining.

[0046] In addition, this embodiment also provides a computer-readable storage medium, in which a computer program is stored. The computer program is configured by a microprocessor to execute the aforementioned transformer abnormality monitoring method based on acoustic signals and frequent item set mining.

[0047] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, may be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0048] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A transformer anomaly monitoring method based on acoustic signals and frequent item set mining, characterized in that: include: S101, obtaining a real-time acoustic signal of the transformer; S102, extracting features of the real-time acoustic signal of the transformer to obtain multi-dimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters; S103, discretizing the values ​​of the multidimensional features to obtain a discrete data table, and performing frequent item mining on the discrete data table using a preset frequent item set mining algorithm according to the discrete data table to obtain frequent item sets; the multidimensional features include part or all of peak value, peak-to-peak value, root mean square value, standard deviation, slope, kurtosis, waveform index, margin index, and pulse index; the discretization of the values ​​of the multidimensional features refers to using a ten-division binning operation, dividing the value interval into ten equal parts according to the maximum and minimum values ​​of each dimensional feature and giving corresponding labels, all values ​​in a certain value interval share the same label, and establishing a discrete data table according to the labels based on the values ​​of the divided features; S104: If the number of frequent item sets exceeds the abnormal alarm line of the frequent item sets under the preset normal state, it is determined that the transformer is currently in an abnormal state; otherwise, it is determined that the transformer is currently in a normal state.

2. The transformer abnormality monitoring method based on acoustic signal and frequent item set mining according to claim 1 is characterized in that: Before step S104, it also includes determining the abnormal alarm line of the frequent item set in the normal state: S201, obtaining a historical acoustic signal of the transformer; S202, extracting features from the historical acoustic signal of the transformer to obtain multi-dimensional features including dimensioned amplitude domain parameters and dimensionless amplitude domain parameters; S203, discretizing the values ​​of the multidimensional features to obtain a discrete data table, and using a preset frequent item set mining algorithm to perform frequent item mining on the discrete data table to obtain frequent item sets, and determining the abnormal alarm line of the frequent item sets under a preset normal state according to the number of the frequent item sets obtained by mining.

3. The transformer abnormality monitoring method based on acoustic signal and frequent item set mining according to claim 2 is characterized in that: Before extracting features from the historical acoustic signals of the transformer in step S101, a step of cleaning data of the historical acoustic signals of the transformer is also included.

4. The transformer abnormality monitoring method based on acoustic signal and frequent item set mining according to claim 3 is characterized in that: The data cleaning includes at least one of outlier discarding, missing value filling and data reduction.

5. The transformer abnormality monitoring method based on acoustic signal and frequent item set mining according to claim 1 is characterized in that: The preset frequent itemset mining algorithm is the Relim algorithm.

6. The transformer abnormality monitoring method based on acoustic signal and frequent item set mining according to claim 2 is characterized in that: The operation of determining the abnormal alarm line of the frequent itemsets under the normal state is a timed operation to implement a timed update of the abnormal alarm line of the frequent itemsets under the normal state.

7. A transformer anomaly monitoring system based on acoustic signals and frequent item set mining, comprising a microprocessor and a memory connected to each other, characterized in that: The microprocessor is configured to execute the transformer abnormality monitoring method based on acoustic signals and frequent item set mining as described in any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program stored therein, characterized in that: The computer program is used to be configured by a microprocessor to execute the transformer abnormality monitoring method based on acoustic signals and frequent item set mining as described in any one of claims 1 to 6.

Images