Method and system for predicting partial discharge source in transformer based on random forest

Abstract

The invention discloses a method and a system for predicting a partial discharge source in a transformer based on a random forest. The method comprises the following steps: acquiring signal data acquired by an acoustic sensor from the partial discharge source in a transformer model; extracting acoustic partial discharge data from the data acquired by each acoustic sensor; and selecting a feature set from the acoustic partial discharge data, inputting the feature set into the random forest classifier for training and testing, and identifying the type of the partial discharge source. According to the method, acoustic partial discharge data in an acoustic sensor are extracted through time sequence features based on extensible hypothesis testing, the features are selected by using a mutual information method and then imported into a random forest classifier, and independent partial discharge and coupled partial discharge are identified through data sample training and testing. The method improves the recognition capability of the partial discharge source, improves the prediction range of the partial discharge source, and greatly reduces the cost and the time required for training the classifier.

Description

Technical Field

[0001] The present invention relates to the technical field of partial discharge detection, and particularly to a prediction method, system, device and medium for internal partial discharge sources of transformers based on random forest. Background Art

[0002] Power transformers are important equipment to ensure the stable and reliable operation of the power supply system. Power transformers are one of the important equipment in the power system. Once a fault occurs, it will cause serious power outage accidents and have a significant impact on the economic operation of the power system. The existence of partial discharge sources is considered to be one of the main reasons for transformer failures. Therefore, the identification of internal partial discharge sources in power transformers is of great significance for the safe operation, maintenance and repair of power transformers.

[0003] The monitoring of PD (partial discharge) signals is based on various physical and chemical phenomena such as electricity, sound, light, temperature and gas generated during discharge, and is measured through these physical quantities that can represent PD. When PD occurs inside the transformer, a pulsed current will be generated, exciting ultra-high frequency, ultrasonic signals, etc., and accompanied by phenomena such as light, electricity and heat. By detecting these signals, the internal PD of the transformer can be monitored. PD detection methods mainly include electrical measurement methods and non-electrical detection methods. Among them, the pulsed current method and the ultra-high frequency method (Ultra High Frequency, abbreviated as UHF) belong to the electrical measurement methods. While the infrared spectroscopy detection method, ultraviolet spectroscopy detection method, optical measurement method, acoustic detection method, chemical detection method, etc. belong to the non-electrical measurement methods. In traditional PD detection methods, acoustic emission detection is more advantageous because it can locate PD activities in power transformers. A new type of integrated sensor for transformer PD identification based on acoustic emission (AE) and ultra-high frequency (UHF) methods inserts an AE sensor into the terminal part of the UHF probe, which is placed above the oil valve and inserted into the transformer, and simultaneously evaluates acoustic and electromagnetic PD signals. However, this method can only identify coupled partial discharge sources, cannot identify independent partial discharge sources, and has a relatively high cost. Once a partial discharge fault that it cannot identify occurs in the power transformer, it may cause serious safety accidents. Summary of the Invention

[0004] The purpose of the present invention is to provide a prediction method, system, device and medium for internal partial discharge sources of transformers based on random forest to solve the technical problems existing in the prior art.

[0005] The present invention is realized through the following technical solutions:

[0006] In the first aspect, a prediction method for internal partial discharge sources of transformers based on random forest provided by an embodiment of the present invention includes the following steps:

[0007] Obtain the signal data collected by the acoustic sensor from the partial discharge source in the transformer model;

[0008] Extract the acoustic partial discharge data from the data collected by each acoustic sensor;

[0009] Select a feature set from the acoustic partial discharge data, input the feature set into a random forest classifier for training and testing, and identify the type of partial discharge source.

[0010] Further, the specific method for extracting the acoustic partial discharge data from the data collected by each acoustic sensor includes:

[0011] Obtain the time series data collected by the acoustic sensor;

[0012] Use a time series feature extraction paradigm based on scalable hypothesis testing to extract features from the acoustic partial discharge data.

[0013] Further, the features include statistical features, frequency domain features, and time domain features.

[0014] Further, the specific method for selecting a feature set from the acoustic partial discharge data includes:

[0015] Adopt the mutual information method to select the optimal feature set for the classification task of the acoustic partial discharge data.

[0016] In a second aspect, a prediction system for the internal partial discharge source of a transformer based on a random forest provided by an embodiment of the present invention includes: a data acquisition module, a data extraction module, and a classification and identification module.

[0017] The data acquisition module is used to obtain the signal data collected by the acoustic sensor from the partial discharge source in the transformer model;

[0018] The data extraction module is used to extract the acoustic partial discharge data from the data collected by each acoustic sensor;

[0019] The classification and identification module is used to select a feature set from the acoustic partial discharge data, input the feature set into a random forest classifier for classification, and identify the type of partial discharge source.

[0020] Further, the data extraction module includes a collection unit and a feature extraction unit, and the collection unit is used to obtain the time series data collected by the acoustic sensor;

[0021] The feature extraction unit is used to use a time series feature extraction paradigm based on scalable hypothesis testing to extract features from the acoustic partial discharge data.

[0022] Further, the features include statistical features, frequency domain features, and time domain features.

[0023] Furthermore, the classification and recognition module includes a selection unit, and the selection unit uses the mutual information method to select the optimal feature set for the classification task of the acoustic partial discharge data.

[0024] In a third aspect, an electronic device provided by an embodiment of the present invention includes a processor, an input device, an output device, and a memory. The processor is respectively connected to the input device, the output device, and the memory. The memory is used to store a computer program, and the computer program includes program instructions. The processor is configured to call the program instructions to execute the method described in the above embodiments.

[0025] In a fourth aspect, a computer-readable storage medium provided by an embodiment of the present invention stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, the processor is caused to execute the method described in the above embodiments.

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

[0027] A method for predicting internal partial discharge sources of a transformer based on random forest provided by an embodiment of the present invention extracts acoustic partial discharge data in an acoustic sensor through time series feature extraction based on an extensible hypothesis test, selects features using the mutual information method and then imports them into a random forest classifier. After training and testing data samples, independent partial discharges and coupled partial discharges are identified, improving the recognition ability of partial discharge sources, expanding the prediction range of partial discharge sources. Compared with other deep learning models, the cost and the time required to train the classifier are greatly reduced, providing a more sufficient margin for the structural safety assessment of power transformers.

[0028] A method for predicting internal partial discharge sources of a transformer based on random forest provided by an embodiment of the present invention has the following advantages: (1) This method belongs to a non-destructive testing method and will not cause destructive effects on the original power transformer structure, and has almost no impact on the normal operation of the power transformer;

[0029] (2) Acoustic PD data is obtained using an acoustic sensor based on the acoustic emission technology. The acoustic emission technology has advantages such as anti-electromagnetic interference, non-destructiveness, the ability to monitor equipment online, and low cost;

[0030] (3) It can not only identify coupled partial discharge sources but also independent partial discharge sources;

[0031] (4) Using machine learning methods and feature engineering, compared with other popular deep learning models, this method can execute faster, greatly reducing the cost and the time required to train the classifier.

[0032] The prediction system, device, and medium for local discharge sources in a transformer based on random forest provided by the embodiments of the present invention and the prediction method for local discharge sources in a transformer based on random forest are of the same inventive concept and have the same beneficial effects. Description of the Drawings

[0033] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings. In the drawings:

[0034] Figure 1 It is a flowchart of a prediction method for local discharge sources in a transformer based on random forest provided by the first embodiment of the present invention;

[0035] Figure 2 It is a schematic diagram of a transformer model;

[0036] Figure 3 It is a position map of the local discharge source;

[0037] Figure 4 It is a structural block diagram of a prediction system for local discharge sources in a transformer based on random forest provided by another embodiment of the present invention;

[0038] Figure 5 It is a structural block diagram of an electronic device provided by another embodiment of the present invention;

[0039] In the figure: 1. High-voltage electrode; 2. Fixing device; 3. Local discharge source; 4. Grounding electrode; 5. Insulating cover; 6. Acoustic sensor. Detailed Embodiments

[0040] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with the embodiments and the drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and do not limit the present invention.

[0041] Embodiment 1

[0042] A prediction method for local discharge sources inside a transformer based on random forest provided by the first embodiment of the present invention includes two stages. The first stage is the collection stage of acoustic local discharge data, which is obtained by establishing a transformer model and placing acoustic sensors on its surface. The second stage is the selection and identification stage of acoustic local discharge data (i.e., the data analysis and processing stage), in which the acoustic local discharge data obtained in the first stage is extracted by time series features based on scalable hypothesis testing, and after feature selection by the mutual information method, it is imported into a random forest classifier to identify the types of local discharge sources.

[0043] Specifically, as Figure 1 shown, a prediction method for local discharge sources inside a transformer based on random forest provided by an embodiment of the present invention includes the following steps:

[0044] S10, establish a transformer model of an oil-filled cube;

[0045] S20, place artificially created local discharge sources at different positions inside the transformer model;

[0046] S30, use acoustic sensors to capture signal data emitted from the local discharge source, and the signal is amplified by an integrated preamplifier;

[0047] S40, for each sensor reading corresponding to each sample, features are extracted using a time series feature extraction paradigm (tsfresh) based on scalable hypothesis testing;

[0048] S50, perform feature selection using the mutual information method, use a random forest classifier for training and testing, and identify independent partial discharge (IPD) and coupled partial discharge (CPD). Independent partial discharge (IPD) is a discharge at a single position, and coupled partial discharge (CPD) is a discharge at two or more positions.

[0049] As Figure 2 shown, a schematic structural diagram of the transformer model is shown. An insulating cover 5 is provided at the top of the transformer model, and a fixing device 2 is provided on the upper part of the insulating cover 5 for fixing the high-voltage electrode. A local discharge source 3 is placed inside the transformer, and the grounding electrode 4 is connected to the local discharge source 3 and grounded. Five acoustic sensors 6 are placed outside the transformer model. These five acoustic sensors are respectively placed at the midpoint positions of the other five outer surfaces except the top surface.

[0050] A prediction method for local discharge sources inside a transformer based on random forest provided by an embodiment of the present invention extracts acoustic PD data through time series feature extraction based on scalable hypothesis testing, selects it through the mutual information method and then imports it into a random forest classifier to accurately and efficiently identify independent partial discharge (IPD) sources and coupled partial discharge (CPD) sources in power transformers, improves the identification ability of local discharge sources, expands the prediction range of local discharge sources, and compared with other deep learning models, greatly reduces the cost and the time required to train the classifier, providing a more sufficient margin for the structural safety assessment of power transformers.

[0051] Among them, step S10 includes two sub-steps:

[0052] S101: Select a transformer model of an oil-filled cube with a capacity of 0.32 m 3 ~0.4 m 3 and a thickness of 5 mm to 7 mm.

[0053] S102: The transformer model uses acrylic insulation material, and a high-voltage electrode is placed above it.

[0054] Step S20 includes the following two sub-steps:

[0055] S201: Place a pair of acrylic insulation materials with a thickness of 0.1 mm between two point-plane electrodes inside the transformer model to form an acoustic local discharge source. Specifically, place a pair of acrylic insulation materials with a thickness of 0.1 mm between two point-plane electrodes inside the transformer model to form an acoustic PD source. The high-voltage (HV) electrode of the discharge source is excited by a 0.23 / 50 kV, 10 kVA HV transformer.

[0056] S202: Place the acoustic local discharge source at different positions inside the transformer model to obtain different acoustic local discharge data. Specifically, the acoustic PD source is stored at 27 different positions in the transformer model grid. The acoustic sensor identifies several PD patterns at single and multiple local discharge positions. The data consists of 27 different single-PD fault positions and 20 different multi-PD fault positions, a total of 47 different positions. Each category has 5 samples, and each sample consists of 5 sensor values at 2500 time instants.

[0057] In step S30, the sound waves emitted by the discharge source are recorded by all sensors fixed at the midpoints of the five outer surfaces of the transformer model, and the signal data emitted from the local discharge source is captured by the acoustic sensor and amplified by an integrated preamplifier.

[0058] In step S40, each sensor reading can be regarded as a waveform (time series data), and features are extracted from the acoustic PD data using a time series feature extraction paradigm (tsfresh) based on scalable hypothesis testing. The so-called time series based on scalable hypothesis testing implements a machine learning related program for classifying the location of partial discharge (PD) sources based on sensor readings in the embodiments of the present invention. Since the measurements are made at regular time intervals, this can be considered a time series problem, and intuitively the task becomes calculating the waveform corresponding to a specific PD. For each sensor reading corresponding to each sample, many features are extracted using a time series feature extraction paradigm based on scalable hypothesis testing (tsfresh). Tsfresh provides a rich set of features, covering various statistical features, frequency domain features, time domain features, etc. Using the feature extraction functions provided by the tsfresh library, feature extraction is performed on the time series data.

[0059] In step S50, feature selection is performed using the mutual information method, and a random forest classifier is used for training and testing to identify independent partial discharge (IPD) and coupled partial discharge (CPD), which are divided into two sub-steps S501 and S502.

[0060] S501: Select the optimal feature set for the required classification task using the mutual information method.

[0061] The mutual information (Mutual Information Method) is used to select the optimal feature set for the required classification task. The mutual information method is a feature selection method commonly used to evaluate the correlation between features and the target variable. This method is based on the concept of information theory and measures the degree of mutual dependence between features and the target variable by calculating the mutual information between them.

[0062] S502: Input the optimal feature set into the random forest classifier for classification, train and test the samples for each category, and identify the types of partial discharge sources.

[0063] Random forest first uses the Bootstrap sampling method to randomly draw multiple sample sets from the original dataset for training. The size of each sample set is the same as that of the original dataset, but each sample may be drawn repeatedly. Multiple decision trees are constructed for training. Each decision tree is a binary tree structure, where each node represents a splitting condition of a feature, and each leaf node represents the final classification result. At each split, the decision tree selects the feature that can most effectively divide the data. Once multiple decision trees are trained, the random forest will make predictions on new samples. Each decision tree will give its own prediction result (category), and the final classification result is determined by the prediction results of all decision trees, usually using the majority voting rule. The random forest classifier randomly selects a part of the features from the original features for training. This can increase the difference between decision trees, improve the generalization ability of the integrated model, and avoid overfitting; delete redundant and unimportant features to improve the performance of the classifier. The mutual information between two random variables X and Y is evaluated by the reduction in the uncertainty of the result of a random variable X, which is based on the known result of another random variable Y, and the similarity ratio is determined by the product of the joint probability distribution P(X,Y) and the marginal probability distribution P(X)·P(Y).

[0064] Referring to Table 1 and Table 2, in S502, a summary of the dataset and the positions and symbols of several independent partial discharges (IPDs) and coupled partial discharges (CPDs) are given. The entire dataset is divided into a training set and a test set in a ratio of 3:2. For each class, 3 samples are used for training and 2 samples are used for testing. Therefore, a total of 141 samples are used for training and 94 samples are used for testing. A total of 794 features are extracted from each sensor, resulting in a total of 3970 features. After feature extraction, the mutual information method is used for feature selection, and the top 300 features are selected. Then they are input into a random forest classifier with 1500 trees to obtain the accuracy rate on the test dataset.

[0065] Table 1

[0066] Number of sensors 5 Number of fault sources 27 (independent) + 20 (coupled) Number of data sets in each category 5 Number of independent partial discharge data sets 135 Number of coupled partial discharge data sets 100 Total number of data sets 235 Number of independent partial discharge data sets for training 81 Number of data sets for testing 54 (independent) + 100 (coupled)

[0067] Table 2

[0068]

[0069] A prediction method for the local discharge source of a transformer based on random forest provided by an embodiment of the present invention extracts acoustic local discharge data in an acoustic sensor through time series feature extraction based on an extensible hypothesis test, selects features using the mutual information method and then imports them into a random forest classifier. After training and testing on data samples, independent local discharges and coupled local discharges are identified, improving the recognition ability of local discharge sources, expanding the prediction range of local discharge sources. Compared with other deep learning models, it greatly reduces the cost and the time required to train the classifier, providing a more sufficient margin for the structural safety assessment of power transformers.

[0070] A prediction method for the local discharge source inside a transformer based on random forest provided by an embodiment of the present invention has the following advantages: (1) This method belongs to a non-destructive testing method and will not cause destructive effects on the original power transformer structure, and has almost no impact on the normal operation of the power transformer.

[0071] (2) Using an acoustic sensor to obtain acoustic PD data based on the acoustic emission technology, the acoustic emission technology has the advantages of anti-electromagnetic interference, non-destructiveness, the ability to monitor equipment online, and low cost.

[0072] (3) It can not only identify the sources of coupled partial discharges (CPD), but also identify the sources of independent partial discharges (IPD).

[0073] (4) The machine learning method and feature engineering used can be executed faster compared with other popular deep learning models, greatly reducing the cost and the time required to train the classifier.

[0074] As Figure 2 shown, a prediction system for the local discharge source inside a transformer based on random forest provided by another embodiment of the present invention includes: a data acquisition module, a data extraction module, and a classification and recognition module. The data acquisition module is used to acquire the signal data collected by the acoustic sensor from the local discharge source in the transformer model; the data extraction module is used to extract the acoustic local discharge data from the data collected by each acoustic sensor; the classification and recognition module is used to select a feature set from the acoustic local discharge data, input the feature set into a random forest classifier for classification, and identify the types of local discharge sources.

[0075] Among them, the data extraction module includes a collection unit and a feature extraction unit. The collection unit is used to acquire the time series data collected by the acoustic sensor; the feature extraction unit is used to extract features from the acoustic local discharge data using a time series feature extraction paradigm based on an extensible hypothesis test.

[0076] Among them, the features include statistical features, frequency-domain features, and time-domain features. The classification and recognition module includes a selection unit, and the selection unit uses the mutual information method to select the optimal feature set for the classification task of acoustic partial discharge data.

[0077] A prediction system for transformer partial discharge sources based on random forest provided by an embodiment of the present invention extracts acoustic partial discharge data in an acoustic sensor through time series features based on scalable hypothesis testing, selects features using the mutual information method and then imports them into a random forest classifier. After training and testing the data samples, independent partial discharges and coupled partial discharges are identified, improving the recognition ability of partial discharge sources, expanding the prediction range of partial discharge sources, and greatly reducing the cost and the time required to train the classifier compared with other deep learning models, providing a more sufficient margin for the structural safety assessment of power transformers.

[0078] The advantages of a prediction system for in-transformer partial discharge sources based on random forest provided by an embodiment of the present invention are as follows: (5) This method belongs to a non-destructive testing method and will not cause destructive effects on the original power transformer structure, and has almost no impact on the normal operation of the power transformer.

[0079] (6) Using an acoustic sensor to obtain acoustic partial discharge data based on the acoustic emission technology, and the acoustic emission technology has advantages such as anti-electromagnetic interference, non-destructiveness, the ability to monitor equipment online, and low cost.

[0080] (7) It can not only identify coupled partial discharge sources but also independent partial discharge sources.

[0081] (8) Using machine learning methods and feature engineering, compared with other popular deep learning models, this method can execute faster, greatly reducing the cost and the time required to train the classifier.

[0082] Embodiment 3

[0083] As Figure 5 shown, a structural block diagram of an electronic device according to another embodiment of the present invention. The device includes a processor, an input device, an output device, and a memory. The processor is respectively connected to the input device, the output device, and the memory. The memory is used to store a computer program, and the computer program includes program instructions. The processor is configured to call the program instructions to execute the method described in the above first embodiment.

[0084] It should be understood that in the embodiments of the present invention, the so-called processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0085] The input device may include a touchpad, a fingerprint sensor (for collecting the fingerprint information and the direction information of the fingerprint of the user), a microphone, etc., and the output device may include a display (such as an LCD), a speaker, etc.

[0086] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A part of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0087] In a specific implementation, the processor, input device, and output device described in the embodiments of the present invention may implement the implementation manners described in the method embodiments provided by the embodiments of the present invention, or may also implement the implementation manners of the system embodiments described in the embodiments of the present invention, which will not be elaborated herein.

[0088] Embodiment 4

[0089] In another embodiment of the present invention, there is also provided an embodiment of a computer-readable storage medium. The computer-readable storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, the processor is caused to execute the method described in the above first embodiment.

[0090] The computer-readable storage medium may be an internal storage unit of the terminal described in the foregoing embodiments, such as the hard disk or memory of the terminal. The computer-readable storage medium may also be an external storage device of the terminal, such as a plug-in hard disk equipped on the terminal, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the computer-readable storage medium may also include both the internal storage unit and the external storage device of the terminal. The computer-readable storage medium is used to store the computer program and other programs and data required by the terminal. The computer-readable storage medium may also be used to temporarily store data that has been output or is to be output.

[0091] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of the examples have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0092] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the terminal and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated here.

[0093] In several embodiments provided in the present application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed couplings, direct couplings, or communication connections to each other may be indirect couplings or communication connections through some interfaces, devices, or units, and may also be electrical, mechanical, or other forms of connection.

[0094] The specific embodiments described above further elaborate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for predicting local discharge source in a transformer based on random forest, characterized in that: The following steps are involved: Acquiring acoustic sensors to collect signal data from local discharge sources in the transformer model; extracting acoustic partial discharge data from the data collected by each acoustic sensor; Feature sets are selected from acoustic partial discharge data and input into random forest classifier for training and testing to identify the types of partial discharge sources.

2. The method for predicting partial discharge source in a transformer based on random forest according to claim 1, characterized in that: The specific method for extracting acoustic partial discharge data from the data collected by each acoustic sensor includes: Obtain time series data collected by acoustic sensors; Features are extracted from acoustic partial discharge data using a scalable hypothesis testing based time series feature extraction paradigm.

3. The method for predicting partial discharge source in a transformer based on random forest according to claim 2, characterized in that: The features include statistical features, frequency domain features and time domain features.

4. The method for predicting partial discharge source in a transformer based on random forest according to claim 3, characterized in that: The specific method of selecting a feature set from acoustic partial discharge data includes: The mutual information method is used to select the optimal feature set for the classification task of acoustic partial discharge data.

5. A prediction system for partial discharge source in transformer based on random forest, characterized in that: include: Data acquisition module, data extraction module and classification recognition module, The data acquisition module is used to acquire signal data collected by the acoustic sensor from the local discharge source in the transformer model; The data extraction module is used to extract acoustic partial discharge data from the data collected by each acoustic sensor; The classification and recognition module is used to select a feature set from the acoustic partial discharge data, input the feature set into a random forest classifier for classification, and identify the type of partial discharge source.

6. The prediction system for partial discharge source in transformer based on random forest according to claim 5, characterized in that: The data extraction module includes a collection unit and a feature extraction unit, wherein the collection unit is used to obtain time series data collected by the acoustic sensor; The feature extraction unit is used to extract features from acoustic partial discharge data using a time series feature extraction paradigm based on scalable hypothesis testing.

7. The prediction system for partial discharge source in transformer based on random forest according to claim 6, characterized in that: The features include statistical features, frequency domain features and time domain features.

8. The prediction system for partial discharge source in transformer based on random forest according to claim 7, characterized in that: The classification and recognition module includes a selection unit, which selects an optimal feature set for a classification task of acoustic partial discharge data using a mutual information method.

9. An electronic device, characterized in that: It includes a processor, an input device, an output device and a memory, the processor is connected to the input device, the output device and the memory respectively, the memory is used to store a computer program, the computer program includes program instructions, and is characterized in that the processor is configured to call the program instructions to execute the method according to any one of claims 1 to 4.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a processor, the processor is caused to perform the method according to any one of claims 1 to 4.

Images