A power equipment sound diagnosis method

Abstract

The application discloses a kind of power equipment sound diagnosis methods, the technical field of power equipment fault diagnosis, including acquisition first signal;The first signal is preprocessed, and second signal is obtained;The second signal is decomposed, and third signal is obtained;Based on the acquisition time, the third signal is screened;The third signal is judged. When collecting transformer sound, the detection point on transformer is the same each time, the sound sensor is installed in the detection point, and the same model sound sensor is used, so that the influence of the collected transformer sound change caused by the model, erection angle, direction and distance of the sound sensor can be eliminated, the recognition accuracy is improved, the first signal is preprocessed to achieve the purpose of noise reduction, and then the second signal is decomposed into a single third signal, the interference of the sound change caused by various abnormal sounds on fault judgment is eliminated, which is beneficial to improve the recognition accuracy.

Description

Technical Field

[0001] This invention relates to the technical field of fault diagnosis of power equipment, and more particularly to a method for diagnosing power equipment by sound. Background Technology

[0002] With the rapid development of power technology and the continuous expansion of power grids, especially with the development of ultra-high voltage (UHV) and extra-high voltage (EHV) systems and the Global Energy Internet in my country, the future of the power industry has been clearly defined. Large power grids and EHV systems place higher demands on the reliability of high-voltage power equipment, which in turn requires more sophisticated testing and diagnostic capabilities. In addition to traditionally widely used signals for testing, the concept of using sound for power equipment testing has been proposed by some scholars due to advancements in sound acquisition and processing technologies. Sound is a mechanical wave radiating energy from vibration into the transmission medium of electrical equipment. Sound signals contain a wealth of vibration information and are an important indicator for analyzing equipment operating status. Furthermore, acquiring sound signals does not require stopping the equipment or physical contact; the devices are simple, signal acquisition is convenient, installation is flexible, and it does not interfere with the normal operation of the equipment.

[0003] Because there are numerous factors that can cause transformer malfunctions and fault conditions, such as structural deformation, damage, and abnormal electrothermal stress, actual operating transformer fault data can be affected. Furthermore, the influencing factors on transformer malfunctions and faults are diverse, including mechanical components that may malfunction, such as windings, core, clamps, leads, tap changers, and spacers. Abnormal sounds from transformers are unlikely to be caused by a single anomaly; in reality, multiple locations and conditions often cause changes in sound due to simultaneous anomalies. The variety of possible fault factors and the different combinations of anomalies make analysis extremely difficult. Moreover, the complex background of the transformer operating environment, with its significant noise levels, affects analysis. Furthermore, variations in the type, installation angle, orientation, and distance of the sound sensor can cause changes in the collected transformer sound, reducing the accuracy of identification. Summary of the Invention

[0004] The technical problem solved by this invention is that the abnormal sound of a transformer may not be caused by a single abnormal point. In reality, multiple parts and conditions often cause abnormalities at the same time, resulting in changes in the sound. Due to the wide variety of possible fault factors, the different combinations of abnormal points make the analysis extremely difficult. Moreover, the transformer operating environment is complex and contains a lot of noise, which affects the analysis. Furthermore, the transformer sound collected may vary due to the influence of the sound sensor model, installation angle, orientation, distance, etc., which reduces the accuracy of identification.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for sound diagnosis of power equipment, comprising: acquiring a first signal; preprocessing the first signal to obtain a second signal; decomposing the second signal to obtain a third signal; filtering the third signal based on the acquisition time; and judging the third signal.

[0006] As a preferred embodiment of the power equipment sound diagnosis method of the present invention, the acquisition of the first signal includes: pre-calibrating a detection point on the transformer; installing a sound sensor at the detection point; controlling the sound sensor to synchronously acquire the original sound signal of the transformer, and using the original sound signal of the transformer as the first signal; and recording the acquisition time corresponding to the first signal.

[0007] In a preferred embodiment of the sound diagnosis method for power equipment according to the present invention, the preprocessing of the first signal to obtain the second signal includes:

[0008] The first signal is decomposed into multiple scales using an adaptive hierarchical threshold wavelet algorithm. Short-time high-frequency and low-frequency components are analyzed, and detail components are processed using a set threshold for denoising. The mathematical expression for the first signal is defined as follows:

[0009]

[0010] in, This refers to the first signal, which is the noisy signal. Represents the original signal. This indicates a noise signal.

[0011] As a preferred embodiment of the sound diagnosis method for power equipment described in this invention, wherein: for The mathematical expression for performing discrete wavelet transform is:

[0012]

[0013] in, Represents discrete wavelets, express wavelet transform coefficients, express The wavelet transform coefficients.

[0014] As a preferred embodiment of the sound diagnosis method for power equipment described in this invention, wherein: for Perform wavelet decomposition, using a threshold function. The mathematical expression for filtering noisy signals is:

[0015]

[0016]

[0017] in, Indicates the threshold. Indicates the upper threshold. Indicates the lower threshold. , , and thus ,Change The value changes At the same time, lower threshold The value of m is adjusted to shrink the size of the critical region where the upper and lower thresholds are located, thereby controlling the noise figure removal ratio and removing the noisy signal to the maximum extent to obtain the second signal.

[0018] In a preferred embodiment of the sound diagnosis method for power equipment according to the present invention, the decomposition of the second signal to obtain the third signal includes:

[0019] The second signal is decomposed into several components that form the third signal, and its mathematical expression is as follows:

[0020]

[0021]

[0022] Where k represents the number of decompositions, Indicates the center frequency. express The instantaneous amplitude, The instantaneous frequency is ,exist Within the interval range, It can be viewed as an amplitude value , frequency is Harmonic signals.

[0023] As a preferred embodiment of the power equipment sound diagnosis method of the present invention, the filtering of the third signal based on the acquisition time includes:

[0024] The start and stop times of the different third signals are compared;

[0025] If the time interval between the start and stop times of the two third signals is less than 0.1s, it indicates that the two third signals exist independently.

[0026] If the time interval between the start and stop times of the two third signals is less than 0.1s, it indicates that the two third signals are concurrent.

[0027] In a preferred embodiment of the power equipment sound diagnosis method of the present invention, the judgment of the third signal includes:

[0028] Collect audio information about defects in various components of the transformer;

[0029] Short-time spectrum analysis is performed on the audio information, and the audio signal is transformed into a first spectrogram that records the joint distribution information of the time domain and frequency domain using fast Fourier transform.

[0030] Classify and organize the first spectrograms corresponding to the defects of various transformer components, and establish an audio sample library;

[0031] Short-time spectrum analysis is performed on the third signal, and the audio signal is transformed into a second spectrogram using Fast Fourier Transform.

[0032] The second spectrogram is compared with the first spectrogram in the audio sample library;

[0033] If the time domain distribution information of the second spectrogram is the same as that of the first spectrogram in both the time and frequency domains, then the components on the transformer have defects corresponding to the first spectrogram.

[0034] The beneficial effects of this invention are as follows: When collecting transformer sound, the detection points on the transformer are pre-calibrated in the same way each time, and the sound sensor is installed at the detection point. The same type of sound sensor is used to eliminate the influence of the sound sensor model, installation angle, orientation and distance on the collected transformer sound, thereby improving the recognition accuracy. By preprocessing the first signal, the purpose of noise reduction is achieved. Then, the second signal is decomposed into a single third signal to eliminate the interference of various abnormal noises on the fault judgment, which is conducive to improving the recognition accuracy. Attached Figure Description

[0035] Figure 1 This is a basic flowchart illustrating a sound diagnosis method for power equipment provided in one embodiment of the present invention.

[0036] Figure 2 This is a schematic diagram of the original sound signal of a transformer before noise reduction in a sound diagnosis method for power equipment provided in an embodiment of the present invention.

[0037] Figure 3 This is a schematic diagram of the original sound signal of a transformer after noise reduction, provided as an embodiment of the present invention for a sound diagnosis method for power equipment. Detailed Implementation

[0038] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0039] Example 1

[0040] Reference Figures 1 to 3 As an embodiment of the present invention, a method for sound diagnosis of power equipment is provided, comprising:

[0041] S1: Acquiring the first signal includes:

[0042] Pre-calibrate the test points on the transformer;

[0043] Install the sound sensor at the detection point;

[0044] The sound sensor is controlled to synchronously acquire the original sound signal of the transformer, and the original sound signal of the transformer is used as the first signal.

[0045] Record the acquisition time corresponding to the first signal meter.

[0046] Because the sound sensor model, installation angle, orientation, and distance can affect the collected transformer sound, detection points are pre-calibrated on the transformer. These detection points are identical for each test, and the same model of sound sensor is used. This eliminates the influence of sound sensor model, installation angle, orientation, and distance on the collected transformer sound, improving identification accuracy. Since abnormal transformer sounds are not from a single point of failure, but often involve multiple locations and conditions simultaneously causing sound changes, and given the numerous possible fault factors, the acquisition time corresponding to the first signal meter is recorded in real time.

[0047] S2: Preprocessing the first signal to obtain the second signal includes:

[0048] The first signal is decomposed into multiple scales using an adaptive hierarchical threshold wavelet algorithm. Short-time high-frequency and low-frequency components are analyzed, and detail components are processed using a set threshold for denoising. The mathematical expression for the first signal is defined as follows:

[0049]

[0050] in, This refers to the first signal, which is the noisy signal. Represents the original signal. This indicates a noise signal.

[0051] right The mathematical expression for performing discrete wavelet transform is:

[0052]

[0053] in, Represents discrete wavelets. express wavelet transform coefficients, express The wavelet transform coefficients are denoted by dt, where dt represents the differential.

[0054] right Perform wavelet decomposition, using a threshold function. The mathematical expression for filtering noisy signals is:

[0055]

[0056]

[0057] in, Indicates the threshold. Indicates the upper threshold. Indicates the lower threshold. , , and thus ,Change The value changes At the same time, lower threshold Consequently, the adjustable threshold function differs from the classic soft and hard threshold functions. This function tends to fall between the two, possessing the advantages of both soft and hard threshold functions. In the critical region where the upper and lower thresholds are located, by adjusting the value of m, the size of the critical region is reduced, thereby controlling the noise figure removal ratio and maximizing the removal of noisy signals to obtain the second signal.

[0058] Transformer installation environments are diverse and complex. The audible sound band is greatly affected by external factors, so noise reduction is performed on the original transformer sound collected through preprocessing.

[0059] S3: Decompose the second signal to obtain the third signal, including:

[0060] The second signal is decomposed into several intrinsic mode functions (EMFs). These EMFs are then used as the third signal, and their mathematical expression is as follows:

[0061]

[0062]

[0063] Where k represents the number of decompositions, Indicates the center frequency. express The instantaneous amplitude, The instantaneous frequency is ,exist Within the interval range, It can be viewed as an amplitude value , frequency is The harmonic signals. Assume each third signal... It is a finite bandwidth around the center frequency. The signal is adaptively decomposed by searching for the optimal solution of the constraint variational model. The center frequency and bandwidth of each third signal are continuously updated in the process of iteratively solving the optimal solution of the variational model. Based on the frequency domain characteristics of the actual signal, the second signal is adaptively decomposed in the frequency domain to obtain several narrowband third signals.

[0064] S4: Filtering the third signal based on the acquisition time includes:

[0065] The start and stop times of the different third signals are compared;

[0066] If the time interval between the start and stop times of the two third signals is less than 0.1s, it indicates that the two third signals exist independently.

[0067] Based on the timing of abnormal noises from different components of the transformer, the cause of the fault corresponding to the third signal is analyzed. A single third signal can be regarded as an abnormal noise corresponding to a single component on the transformer, and the cause of the fault can be directly determined.

[0068] If the time interval between the start and stop times of the two third signals is less than 0.1s, it indicates that the two third signals are concurrent.

[0069] Concurrent abnormal noises can occur under multiple conditions simultaneously. For example, a loose connection may produce a vibration sound, while a current discharge sound may also occur. By understanding the concurrent relationships between different abnormal noises, the accuracy of fault diagnosis can be improved.

[0070] S5: Judging the third signal includes:

[0071] Audio information of defects in various components of the transformer is collected. When collecting the transformer sound, the detection points on the transformer are pre-calibrated in the same way each time. The sound sensor is installed at the detection point and the same model of sound sensor is used to eliminate the influence of the sound sensor model, installation angle, orientation and distance on the collected transformer sound, thereby improving the recognition accuracy.

[0072] Short-time spectrum analysis is performed on the audio information, and the audio signal is transformed into a first spectrogram that records the joint distribution information of the time domain and frequency domain using fast Fourier transform.

[0073] Classify and organize the first spectrograms corresponding to the defects of various transformer components, and establish an audio sample library;

[0074] Short-time spectrum analysis is performed on the third signal, and the audio signal is transformed into a second spectrogram using Fast Fourier Transform.

[0075] The second spectrogram is compared with the first spectrogram in the audio sample library;

[0076] If the time domain distribution information of the second spectrogram is the same as that of the first spectrogram in both the time and frequency domains, then the components on the transformer have defects corresponding to the first spectrogram.

[0077] When collecting transformer sound, the same detection points are pre-calibrated on the transformer each time. The sound sensor is installed at the detection point, and the same model of sound sensor is used to eliminate the influence of sound sensor model, installation angle, orientation and distance on the collected transformer sound, thereby improving the accuracy of identification. The first signal is pre-processed to achieve the purpose of noise reduction. Then, the second signal is decomposed into individual third signals to eliminate the interference of various abnormal noises on the fault diagnosis, which is conducive to improving the accuracy of identification.

[0078] Example 2

[0079] In another embodiment of the present invention, which differs from the first embodiment, an experimental verification of a sound diagnosis method for power equipment is provided. To verify and explain the technical effect of the method, this embodiment uses a traditional technical solution to compare and test with the method of the present invention, and compares the test results with scientific demonstration methods to verify the real effect of the method.

[0080] To verify the accuracy of this method compared to existing methods for transformer sound detection, 10 faulty transformers were used as test samples. Five types of single transformer abnormal noises were paired to form 25 abnormal noises that combined two types of faults. The traditional manual judgment method and this method were used for judgment.

[0081] The five types of abnormal sounds from a single transformer are overvoltage, noise, discharge sound, cracking sound, and boiling water sound.

[0082] Overvoltage refers to the increase in transformer noise when a single-phase ground fault or resonant overvoltage occurs in the power grid. In such cases, a comprehensive judgment can be made by combining the readings of the voltmeter.

[0083] When a transformer is overloaded, it will emit a heavy "humming" sound. If the transformer load is found to exceed the allowable normal overload value, the transformer load should be reduced in accordance with the on-site regulations.

[0084] Noise indicates vibration caused by loose components on the transformer. If the transformer noise is significantly increased, but the current and voltage are not obviously abnormal, it may be due to loose internal clamps or screws tightening the core, causing increased vibration of the silicon steel sheets. Solution: If it does not affect transformer operation, no action is necessary for now. Record the incident, strengthen monitoring, and report to dispatch and relevant personnel to request a power outage for inspection and handling.

[0085] A crackling or popping sound indicates a discharge from the transformer. If blue corona or sparks are seen near the transformer bushings at night or in rainy weather, it indicates severe contamination of the ceramic components or poor contact of the equipment wiring. Internal discharge within the transformer is caused by electrostatic discharge from ungrounded components, inter-turn discharge in the coils, or poor contact in the tap changer. The solution is to report this to dispatch and relevant personnel and request a power outage for inspection and repair of the transformer.

[0086] A popping sound indicates that the transformer's internal or surface insulation has broken down, and the transformer should be shut down immediately for inspection.

[0087] The sound of boiling water indicates a short circuit in the transformer windings or severe overheating caused by poor contact in the tap changer.

[0088] Table 1: Comparison of Experimental Results

[0089]

[0090] As can be seen from the table above, the method of the present invention has a higher recognition accuracy and a faster recognition speed compared with the traditional experimental method. Due to the influence of environmental noise, it is more difficult for manual identification to distinguish the reverberation of the two abnormal sounds. The present invention has a higher accuracy, which demonstrates the effectiveness of the present invention.

[0091] It should be recognized that embodiments of the present invention can be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium. The method can be implemented using standard programming techniques—including a non-transitory computer-readable storage medium configured with a computer program, wherein such a storage medium causes the computer to operate in a specific and predefined manner—according to the methods and drawings described in the specific embodiments. Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system. However, if desired, the program can be implemented in assembly or machine language. In any case, the language can be a compiled or interpreted language. Furthermore, for this purpose, the program can run on a programmed application-specific integrated circuit (ASIC).

[0092] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for sound diagnosis of electrical equipment, characterized in that, include: Acquire the first signal; The first signal is preprocessed to obtain the second signal; The second signal is decomposed to obtain the third signal; The third signal is filtered based on the acquisition time; The third signal is judged; Pre-calibrate the test points on the transformer; Install the sound sensor at the detection point; The sound sensor is controlled to synchronously acquire the original sound signal of the transformer, and the original sound signal of the transformer is used as the first signal. Record the acquisition time corresponding to the first signal meter; Preprocessing the first signal to obtain the second signal includes: The first signal is decomposed into multiple scales using an adaptive hierarchical threshold wavelet algorithm. Short-time high-frequency and low-frequency components are analyzed, and detail components are processed using a set threshold for denoising. The mathematical expression for the first signal is defined as follows: in, This refers to the first signal, which is the noisy signal. Represents the original signal. Indicates a noise signal; right The mathematical expression for performing discrete wavelet transform is: in, Represents discrete wavelets. express wavelet transform coefficients, express wavelet transform coefficients; right Perform wavelet decomposition, using a threshold function. The mathematical expression for filtering noisy signals is: in, Indicates the threshold. Indicates the upper threshold. Indicates the lower threshold. , , and thus ,Change The value changes At the same time, lower threshold The value of m is adjusted to shrink the size of the critical region where the upper and lower thresholds are located, thereby controlling the noise figure removal ratio and removing the noisy signal to the maximum extent to obtain the second signal. The second signal is decomposed into several components that form the third signal, and its mathematical expression is as follows: Where k represents the number of decompositions, Indicates the center frequency. express The instantaneous amplitude, The instantaneous frequency is ,exist Within the interval range, It can be viewed as an amplitude value , frequency is Harmonic signals; Filtering the third signal based on the acquisition time includes: The start and stop times of the different third signals are compared; If the time interval between the start and stop times of the two third signals is less than 0.1s, it indicates that the two third signals exist independently. If the time interval between the start and stop times of the two third signals is less than 0.1s, it indicates that the two third signals are concurrent. The determination of the third signal includes: Collect audio information about defects in various components of the transformer; Short-time spectrum analysis is performed on the audio information, and the audio signal is transformed into a first spectrogram that records the joint distribution information of the time domain and frequency domain using fast Fourier transform. Classify and organize the first spectrograms corresponding to the defects of various transformer components, and establish an audio sample library; Short-time spectrum analysis is performed on the third signal, and the audio signal is transformed into a second spectrogram using Fast Fourier Transform. The second spectrogram is compared with the first spectrogram in the audio sample library; If the time domain distribution information of the second spectrogram is the same as that of the first spectrogram in both the time and frequency domains, then the components on the transformer have defects corresponding to the first spectrogram.

Images