Transformer voiceprint anomaly detection method

Abstract

The invention relates to the technical field of transformers, and discloses a transformer voiceprint anomaly detection method, which comprises the following steps of: acquiring voiceprint signals during operation of a transformer by adopting a plurality of high-sensitivity voiceprint sensors; the collected voiceprint signals are preprocessed; extracting a time domain feature, a frequency spectrum feature and a deep feature of the voiceprint signal; fusing the time domain feature, the frequency domain feature and the deep learning feature to form a comprehensive feature vector; an anomaly detection model is established by adopting a support vector machine, and an attention mechanism is introduced into the model. According to the transformer voiceprint anomaly detection method, high resolution and high signal-to-noise ratio of voiceprint signals are ensured through the multi-sensor layout and the high-sensitivity MEMS microphone, and the multi-sensor layout can cover different positions of the transformer, so that the running state of the transformer is more comprehensively captured, and the detection accuracy is improved. The extraction of time domain features and frequency spectrum features describes the characteristics of voiceprint signals from different angles.

Description

Technical Field

[0001] The present invention relates to the technical field of transformers, and in particular to a method for detecting abnormality in transformer voiceprints. Background Art

[0002] As a key equipment in the power system, the operating status of the transformer directly affects the stability and safety of the entire power system. However, the traditional transformer fault detection method mainly relies on regular maintenance and manual inspection, which is inefficient and difficult to detect potential faults in time. In recent years, with the development of voiceprint analysis technology, transformer anomaly detection methods based on voiceprints have gradually attracted attention. However, the existing methods still have shortcomings in feature extraction and anomaly detection, and it is difficult to meet the needs of practical applications. Therefore, a transformer voiceprint anomaly detection method based on deep learning and time-frequency domain joint analysis is needed. Through advanced feature extraction, time-frequency domain joint analysis and anomaly detection model optimization, efficient and accurate monitoring of the transformer operating status can be achieved. Summary of the invention

[0003] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0004] To achieve the above object, the present invention provides the following technical solutions: A transformer voiceprint anomaly detection method comprises: using multiple high-sensitivity voiceprint sensors to collect voiceprint signals when the transformer is running; preprocessing the collected voiceprint signals; extracting time domain features, spectrum features and deep-level features of the voiceprint signals; fusing the time domain features, frequency domain features and deep learning features to form a comprehensive feature vector; using a support vector machine to construct an anomaly detection model, and introducing an attention mechanism into the model; inputting the comprehensive feature vector into the anomaly detection model for detection to obtain a detection result; and recording each detection result.

[0005] As a further solution of the present invention: the use of multiple high-sensitivity voiceprint sensors to collect voiceprint signals when the transformer is running, the specific steps are: arranging multiple high-sensitivity voiceprint sensors around the transformer, arranging at least 4 sensors, distributed at the four corners of the transformer, the expression is: ; in, is the optimization degree of sensor layout, is the number of sensors, For the The weight of each sensor, For the The sensitivity of the sensor; Select a high-sensitivity MEMS microphone as the voiceprint sensor, with a frequency response range between 20Hz and 20kHz; use a high-sampling rate data acquisition device with a sampling rate of no less than 20kHz.

[0006] As a further solution of the present invention: the collected voiceprint signal is preprocessed, and the specific steps are to use an adaptive filter to remove background noise from the voiceprint signal, and the expression is: ; in, is the denoised voiceprint signal, is the original voiceprint signal, is the smoothing factor, is the estimated value of the denoised signal at the previous moment; The collected voiceprint signal is divided into segments of fixed length, and the length of each segment is 1 second to 5 seconds.

[0007] As a further solution of the present invention: the extraction of time domain features, spectrum features and deep-level features of voiceprint signals comprises the following specific steps: calculating the time domain features of each voiceprint segment, including mean, variance, zero-crossing rate and energy; using short-time Fourier transform to convert the voiceprint signal into frequency domain representation and extract spectrum features; inputting the time domain features and spectrum features into convolutional neural networks and recurrent neural networks to automatically extract deep-level features, and the expression is: ; ; in, is the time domain feature vector, is the spectral feature vector, is a convolutional neural network, is a recurrent neural network, A deep-level feature.

[0008] As a further solution of the present invention: the time domain features, frequency domain features and deep learning features are fused to form a comprehensive feature vector. The specific steps are: the extracted time domain features, frequency spectrum features and deep learning features are fused to form a comprehensive feature vector, and the expression is: ; in, is the comprehensive feature vector, and is the weight factor.

[0009] As a further solution of the present invention: the anomaly detection model is constructed by using a support vector machine, and an attention mechanism is introduced into the model. The specific steps are: selecting a support vector machine as the anomaly detection model, introducing an attention mechanism into the SVM model, so that the model can focus on the feature part that is most helpful for anomaly detection, and the expression is: ; ; in, is the decision function, is the number of support vectors, is the Lagrange multiplier, For the The labels of samples, is the kernel function, Bias term, is the attention weight, is the feature dimension, For the The weight of the feature, For the The value of a feature.

[0010] As a further solution of the present invention: the comprehensive feature vector is input into the anomaly detection model for detection to obtain the detection result. The specific steps are as follows: Input into the trained support vector machine model; determine whether the result is abnormal based on the abnormality score, the expression is: ; in, For the test results, is the anomaly threshold. If the anomaly score is greater than the threshold , it is judged as abnormal, otherwise it is judged as normal.

[0011] As a further solution of the present invention: the specific steps of recording each test result are: creating a structured data table for recording the result of each test; using a CSV file or a database table to store the test results; after each test is completed, recording the test results in the data table.

[0012] The present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of a transformer voiceprint anomaly detection method as described in the first aspect of the present invention is implemented.

[0013] The present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: the computer program is processed.

[0014] Compared with the prior art, the present invention has the following beneficial effects: The high resolution and high signal-to-noise ratio of the voiceprint signal are ensured by the multi-sensor layout and high-sensitivity MEMS microphone. The multi-sensor layout can cover different positions of the transformer, so as to capture the operating status of the transformer more comprehensively. The extraction of time domain features and spectral features describes the characteristics of the voiceprint signal from different angles, while the convolutional neural network and recurrent neural network can automatically extract deep features and capture the complex patterns of the signal. The multi-dimensional feature extraction method can comprehensively describe the characteristics of the voiceprint signal and improve the richness and diversity of the features. The extraction of deep features can capture the implicit patterns of the signal and provide rich feature support for subsequent anomaly detection. The feature fusion step ensures the comprehensiveness and representativeness of the comprehensive feature vector and improves the input data quality of the model. The comprehensive feature vector contains information of time domain, frequency domain and deep features, which can better reflect the operating status of the transformer. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The figure is a flow chart of a transformer voiceprint anomaly detection method. DETAILED DESCRIPTION

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, the specific implementation modes of the present invention are described in detail below in conjunction with the accompanying drawings.

[0017] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0018] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0019] Example See also Figure 1 , which is the first embodiment of the present invention, and provides a transformer voiceprint anomaly detection method, comprising: S1. Use multiple high-sensitivity voiceprint sensors to collect voiceprint signals when the transformer is running; Specifically, four high-sensitivity MEMS microphones are arranged around the transformer, respectively located at the four corners of the transformer. The frequency response range of the sensor is 20 Hz to 20 kHz, and the sampling rate is not less than 20 kHz.

[0020] It should be noted that the arrangement of sensors can cover different positions of the transformer, so as to capture the operating status of the transformer more comprehensively. The highly sensitive MEMS microphone and high sampling rate data acquisition equipment ensure the high resolution and high signal-to-noise ratio of the voiceprint signal, providing a high-quality data foundation for subsequent feature extraction and anomaly detection.

[0021] S2. Preprocessing the collected voiceprint signal; Specifically, an adaptive filter is used to remove background noise from the voiceprint signal, and the collected voiceprint signal is divided into segments of fixed length, each segment having a length of 2 seconds.

[0022] It should be noted that the adaptive filter can effectively remove background noise and improve the signal-to-noise ratio of the voiceprint signal. Signal segmentation divides the continuous voiceprint signal into segments of fixed length to facilitate subsequent feature extraction and processing, while balancing the computational complexity and feature integrity to ensure that each segment contains sufficient information.

[0023] S3, extracting the time domain features, spectrum features and deep features of the voiceprint signal; Specifically, the time domain features of each voiceprint fragment are calculated, including mean, variance, zero-crossing rate and energy. The voiceprint signal is converted into frequency domain representation using short-time Fourier transform, and the spectral features are extracted. The time domain features and spectral features are input into the convolutional neural network (CNN) and recurrent neural network (RNN) to automatically extract deep-level features.

[0024] It should be noted that the extraction of time domain features and spectral features describes the characteristics of voiceprint signals from different perspectives, while convolutional neural networks and recurrent neural networks can automatically extract deep-level features and capture complex patterns of signals. Multi-dimensional feature extraction methods can comprehensively describe the characteristics of voiceprint signals and improve the richness and diversity of features.

[0025] S4, integrating time domain features, frequency domain features and deep learning features to form a comprehensive feature vector; Specifically, the extracted time domain features, spectrum features and deep learning features are fused to form a comprehensive feature vector, which is expressed as: ; in, is the comprehensive feature vector, and is the weight factor.

[0026] It should be noted that the feature fusion step ensures the comprehensiveness and representativeness of the comprehensive feature vector, improves the input data quality of the model, and the comprehensive feature vector contains information in the time domain, frequency domain and deep-level features, which can better reflect the operating status of the transformer and enhance the detection accuracy and robustness of the model.

[0027] S5. Use support vector machine to build an anomaly detection model and introduce attention mechanism into the model; Specifically, a support vector machine (SVM) is selected as the anomaly detection model, and an attention mechanism is introduced into the SVM model so that the model can focus on the feature parts that are most helpful for anomaly detection. The comprehensive feature vector is input into the trained support vector machine model, and whether the result is abnormal is determined according to the anomaly score.

[0028] It should be noted that the support vector machine (SVM) is an effective binary classification model suitable for processing high-dimensional data. After the introduction of the attention mechanism, the model can pay more attention to the feature parts that are most helpful for anomaly detection, improve the accuracy and robustness of detection, and more reliably identify transformer anomalies by determining the anomaly score.

[0029] S6. Input the comprehensive feature vector into the anomaly detection model for detection, obtain the detection result, and record each detection result; Specifically, create a structured data table to record the results of each test, use a CSV file to store the test results, and record the test results in the data table after each test is completed.

[0030] It should be noted that the steps of recording the test results ensure the integrity and traceability of the data, which is convenient for subsequent data analysis and fault diagnosis. The use of structured data tables and CSV files makes data management and query more convenient, improving the practicality and reliability of the system.

[0031] This embodiment also provides a computer device, which is suitable for the transformer voiceprint anomaly detection method, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the transformer voiceprint anomaly detection method proposed in the above embodiment.

[0032] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0033] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for detecting transformer voiceprint anomaly proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0034] Experimental data table: Table 1: Comparison of experimental data on transformer voiceprint anomaly detection Parameter name Traditional method 1 Traditional method 2 Method of the present invention unit Number of sensors 2 3 4 indivual Sensor layout optimization 0.75 0.80 0.90 Sampling rate 10kHz 15kHz 20kHz Hz Signal-to-noise ratio after noise removal 20dB 25dB 30dB dB Feature extraction time 10s 8s 5s s Anomaly detection accuracy 80% 85% 95% % Anomaly detection missed rate 20% 15% 5% % Anomaly detection false positive rate 10% 8% 2% % Comprehensive feature vector dimension 100 150 200 Model training time 30min 25min 20min min From the above experimental data table, it can be clearly seen that the method of the present invention has significant advantages over the traditional method in many aspects: The method of the present invention uses 4 sensors, while traditional methods 1 and 2 use 2 and 3 sensors respectively. A larger number of sensors can more comprehensively cover the operating status of the transformer and reduce the possibility of missed detection. The sensor layout optimization degree of the method of the present invention is 0.90, which is much higher than 0.75 of traditional method 1 and 0.80 of traditional method 2. The optimized sensor layout can more effectively capture the voiceprint signal of the transformer and improve the accuracy of detection.

[0035] The sampling rate of the method of the present invention is 20kHz, which is higher than 10kHz of traditional method 1 and 15kHz of traditional method 2. The higher sampling rate can capture more signal details and improve the resolution of the signal. The signal-to-noise ratio of the method of the present invention after noise removal is 30dB, which is much higher than 20dB of traditional method 1 and 25dB of traditional method 2. Effective noise removal improves the purity of the signal and reduces the influence of background noise on the detection result.

[0036] The feature extraction time of the method of the present invention is 5 seconds, which is much lower than 10 seconds of traditional method 1 and 8 seconds of traditional method 2. The fast feature extraction improves the real-time performance of the system and can respond to the operating status of the transformer more quickly. The abnormality detection accuracy of the method of the present invention is 95%, which is much higher than 80% of traditional method 1 and 85% of traditional method 2. The higher accuracy means that the method of the present invention can more reliably identify abnormal conditions of the transformer and reduce the possibility of misjudgment.

[0037] The abnormality detection missed detection rate of the method of the present invention is 5%, which is much lower than 20% of traditional method 1 and 15% of traditional method 2. The lower missed detection rate means that the method of the present invention can more comprehensively capture the abnormal conditions of the transformer and reduce the risk of missed detection. The abnormality detection false detection rate of the method of the present invention is 2%, which is much lower than 10% of traditional method 1 and 8% of traditional method 2. The lower false detection rate means that the method of the present invention can more accurately judge the normal operating status of the transformer and reduce the possibility of false alarms.

[0038] In summary, the high resolution and high signal-to-noise ratio of the voiceprint signal are ensured through the multi-sensor layout and high-sensitivity MEMS microphone. The multi-sensor layout can cover different positions of the transformer, so as to capture the operating status of the transformer more comprehensively. The extraction of time domain features and spectral features describes the characteristics of the voiceprint signal from different angles, while convolutional neural networks and recurrent neural networks can automatically extract deep features and capture complex patterns of signals. The multi-dimensional feature extraction method can comprehensively describe the characteristics of the voiceprint signal and improve the richness and diversity of features. The extraction of deep features can capture the implicit patterns of signals and provide rich feature support for subsequent anomaly detection. The feature fusion step ensures the comprehensiveness and representativeness of the comprehensive feature vector and improves the input data quality of the model. The comprehensive feature vector contains information of time domain, frequency domain and deep features, which can better reflect the operating status of the transformer.

[0039] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A transformer voiceprint anomaly detection method, characterized by: include, Use multiple high-sensitivity voiceprint sensors to collect voiceprint signals when the transformer is running; Preprocessing the collected voiceprint signal; Extract the time domain features, spectrum features and deep features of voiceprint signals; Fuse time domain features, frequency domain features and deep learning features to form a comprehensive feature vector; Use support vector machine to build an anomaly detection model and introduce attention mechanism into the model; The comprehensive feature vector is input into the anomaly detection model for detection to obtain the detection result; Record each test result.

2. A transformer voiceprint anomaly detection method according to claim 1, characterized in that: The method uses multiple high-sensitivity voiceprint sensors to collect voiceprint signals when the transformer is running. The specific steps are: Arrange multiple high-sensitivity voiceprint sensors around the transformer, at least 4 sensors are arranged, distributed at the four corners of the transformer, the expression is: ; in, is the optimization degree of sensor layout, is the number of sensors, For the The weight of each sensor, For the The sensitivity of the sensor; Choose a highly sensitive MEMS microphone as the voiceprint sensor, with a frequency response range between 20Hz and 20kHz; Use high sampling rate data acquisition equipment, the sampling rate should not be less than 20kHz.

3. A transformer voiceprint anomaly detection method according to claim 1, characterized in that: The collected voiceprint signal is preprocessed, and the specific steps are: Use an adaptive filter to remove background noise from the voiceprint signal. The expression is: ; in, is the denoised voiceprint signal, is the original voiceprint signal, is the smoothing factor, is the estimated value of the denoised signal at the previous moment; The collected voiceprint signal is divided into segments of fixed length, and the length of each segment is 1 second to 5 seconds.

4. A transformer voiceprint anomaly detection method according to claim 1, characterized in that: The specific steps of extracting the time domain features, spectrum features and deep-level features of the voiceprint signal are as follows: Calculate the time domain features of each voiceprint segment, including mean, variance, zero-crossing rate, and energy; Use short-time Fourier transform to convert the voiceprint signal into frequency domain representation and extract spectrum features; The time domain features and spectrum features are input into the convolutional neural network and the recurrent neural network to automatically extract the deep features. The expression is: ； ； in, is the time domain feature vector, is the spectral feature vector, is a convolutional neural network, is a recurrent neural network, A deep-level feature.

5. The transformer voiceprint anomaly detection method according to claim 1, characterized in that: The time domain features, frequency domain features and deep learning features are integrated to form a comprehensive feature vector. The specific steps are: The extracted time domain features, spectrum features and deep learning features are integrated to form a comprehensive feature vector, which is expressed as: ; in, is the comprehensive feature vector, and is the weight factor.

6. A transformer voiceprint anomaly detection method according to claim 1, characterized in that: The support vector machine is used to build an anomaly detection model, and the attention mechanism is introduced into the model. The specific steps are: The support vector machine is selected as the anomaly detection model. The attention mechanism is introduced into the SVM model so that the model can focus on the features that are most helpful for anomaly detection. The expression is: ; ; in, is the decision function, is the number of support vectors, is the Lagrange multiplier, For the The labels of samples, is the kernel function, Bias term, is the attention weight, is the feature dimension, For the The weight of the feature, For the The value of a feature.

7. A transformer voiceprint anomaly detection method according to claim 1, characterized in that: The comprehensive feature vector is input into the anomaly detection model for detection to obtain the detection result. The specific steps are: The comprehensive feature vector Input into the trained support vector machine model; Determine whether the result is abnormal based on the abnormal score. The expression is: ; in, For the test results, is the anomaly threshold. If the anomaly score is greater than the threshold , it is judged as abnormal, otherwise it is judged as normal.

8. The transformer voiceprint anomaly detection method according to claim 1, characterized in that: The specific steps of recording each test result are as follows: Create a structured data table to record the results of each test; Use CSV files or database tables to store test results; After each test is completed, the test results are recorded in the data table.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a transformer voiceprint anomaly detection method as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a transformer voiceprint anomaly detection method as described in any one of claims 1 to 8 are implemented.

Images