Transformer noise identification and reverse restoration method based on laser vibration measurement and medium

Through the noise recognition and reverse reduction method based on laser vibration measurement, the noise recognition problem of transformer is solved using the acoustic and vibration coupling model and neural network model, and the rapid and accurate identification and positioning in complex environments are achieved, and the operation complexity and cost are reduced. .

CN114091533BActive Publication Date: 2024-11-01STATE GRID SHANGHAI MUNICIPAL ELECTRIC POWER CO +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202111348818.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-11-15
Publication Date
2024-11-01
Estimated Expiration
2041-11-15

AI Technical Summary

Technical Problem

The prior art is difficult to effectively identify and locate noise sources in complex transformer environments. The traditional methods are complex in operation and are greatly affected by the environment. It is difficult to accurately measure transformer noise in high-strength magnetic, strong electric and high-voltage environments.

Method used

Using transformer noise recognition and reverse reduction methods based on laser vibration measurement, the acoustic and vibration coupling model is constructed, the measurement point layout is optimized, and the surface vibration signal is obtained using laser vibration measurement method, and the neural network model is transformed through pre-trained noise recognition and reduction. It is a sound field index and distribution to realize noise recognition and restoration.

Benefits of technology

Quickly and accurately identify transformer noise sources in complex scenarios, reduce the environmental impact of traditional contact measurement methods, improve the level of transformer vibration and noise detection, and reduce the complexity and cost of on-site operation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114091533B_ABST
    Figure CN114091533B_ABST
Patent Text Reader

Abstract

The present invention relates to a transformer noise identification and reverse restoration method and medium based on laser vibration measurement, the method comprising: constructing an acoustic-vibration coupling model of the transformer to be tested to obtain an optimized measurement point arrangement scheme; based on the optimized measurement point arrangement scheme, using laser vibration measurement to obtain the surface vibration signal of each measurement point; splitting the surface vibration signal by a set time length to obtain a plurality of surface vibration time domain signal segments; using the plurality of surface vibration time domain signal segments as inputs of a pre-trained noise identification and restoration neural network model to obtain corresponding equipment states and noise time domain signal estimation values; converting the noise time domain signal estimation values ​​into sound field indicators and sound field distribution to achieve transformer noise identification and restoration. Compared with the prior art, the present invention has the advantages of safety, convenience, reliability, etc., and can be applied to the testing of noise identification and reverse restoration of transformers of different models in various complex working scenarios such as strong electricity, strong magnetism, and high voltage.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of noise detection of power transformers, and in particular to a transformer noise identification and reverse restoration method and medium based on laser vibration measurement. Background Art

[0002] Substations are one of the most critical equipment to ensure the effective operation of power grid systems. They play the role of transforming voltage, receiving and distributing electric energy, controlling current and adjusting voltage. With the rapid economic development, in order to meet the needs of production and life, substations have gradually developed from suburban areas to urban centers. A large number of substations have been surrounded by residential communities, and the noise pollution they bring has a significant impact on the living environment of residents. The mechanism, monitoring and control of transformer noise is one of the important challenges currently facing this engineering field. How to effectively identify transformer noise and locate the sound source in its complex structure is a key basic technology to solve the transformer noise problem.

[0003] The mechanism and transmission path of the transformer body vibration noise are complex, there are many vibration excitation sources and influencing factors, and the on-site background noise is usually high and sudden. The harsh environment of strong magnetic field, strong electricity, high voltage, etc. makes it difficult to measure directly. The traditional sound pressure and sound intensity test methods can only obtain the noise after the mixed superposition of the body and the cooling system, while the vibration of the transformer box is affected by the body and the vibration of cooling equipment such as fans and oil pumps. Therefore, it is difficult to locate the noise source. On the other hand, the test requirements are high and the cost is high. The traditional vibration velocity method is mainly based on vibration velocity or acceleration sensors. The on-site operation is complicated, and the installation is greatly affected by the on-site environment, such as strong magnetic field, strong electricity, high voltage, position / distance, etc. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a safe, convenient and reliable transformer noise identification and reverse restoration method and medium based on laser vibration measurement, which can be applicable to the noise identification and reverse restoration of different types of transformers in various complex working scenarios such as strong electricity, strong magnetism and high voltage.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A method for transformer noise identification and reverse restoration based on laser vibration measurement includes the following steps:

[0007] Construct the acoustic-vibration coupling model of the transformer to be tested and obtain the optimized measurement point layout plan;

[0008] Based on the optimized measurement point arrangement scheme, a surface vibration signal of each measurement point is obtained by using a laser vibrometer method;

[0009] Splitting the surface vibration signal according to a set time length to obtain a plurality of surface vibration time domain signal segments;

[0010] Using the plurality of surface vibration time domain signal segments as inputs of a pre-trained noise identification and restoration neural network model, obtaining corresponding device states and noise time domain signal estimation values;

[0011] The noise time domain signal estimation value is converted into a sound field index and a sound field distribution to realize transformer noise identification and restoration.

[0012] Furthermore, the measuring point arrangement scheme includes measuring point locations and the number of measuring points.

[0013] Furthermore, the acoustic-vibration coupling model is constructed according to the geometric model and operating conditions of the transformer, and based on the acoustic-vibration coupling model, a harmonic response sound field simulation of the transformer is performed to obtain the optimized measurement point arrangement scheme.

[0014] Furthermore, the set time length is an integer multiple of a period corresponding to the transformer operating frequency.

[0015] Furthermore, the training process of the noise recognition and restoration neural network model is specifically as follows:

[0016] Obtaining surface vibration signals and sound pressure signals of multiple groups of transformers;

[0017] Splitting the surface vibration signal and the sound pressure signal according to a set time length to obtain a plurality of surface vibration time domain signal segments and sound pressure time domain signal segments, and calculating the noise radiation index of each segment;

[0018] The noise radiation indexes of multiple segments are combined into a noise radiation index feature matrix, and the mapping relationship between the device type, device status, sound pressure and the noise radiation index feature matrix is ​​added to form a noise radiation index feature library;

[0019] The noise radiation index feature library is used as a sample set to train a noise recognition and restoration neural network model.

[0020] Furthermore, the noise radiation index of each segment is calculated as follows:

[0021] The noise radiation index identification algorithm is used to calculate the noise radiation index of each frequency band in the segment, and the average noise radiation index of multiple frequency bands is taken as the noise radiation index of the segment.

[0022] Furthermore, the noise radiation indexes of the multiple segments with the same working conditions and the same equipment type are synthesized into a noise radiation index feature matrix.

[0023] Furthermore, the sound pressure signal is obtained through a microphone.

[0024] Furthermore, the surface vibration signal is obtained by a laser vibrometer.

[0025] The present invention also provides a computer-readable storage medium, comprising one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs include instructions for executing the transformer noise identification and reverse restoration method based on laser vibration measurement as described above.

[0026] Compared with the prior art, the present invention combines laser vibration measurement and transformer noise feature analysis to identify transformer noise, restore transformer noise distribution in complex scenarios, improve transformer vibration and noise detection level, and effectively solve the problems of complex on-site operation and installation being greatly affected by the on-site environment in traditional transformer vibration and noise detection methods. It has the following beneficial effects:

[0027] (1) A neural network model for noise source identification and restoration is used. By taking the noise radiation index feature library consisting of the noise radiation index of each frequency band in multiple time segments as the input data set, the noise sources of transformers of different types and working conditions can be effectively distinguished, facilitating fast and accurate noise source identification and noise distribution restoration on site.

[0028] (2) The traditional contact measurement method is changed to a laser vibration measurement method based on the Doppler effect, which effectively reduces the impact of the transformer working environment such as strong electricity, strong magnetic field and high voltage on the measurement. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a basic flow chart of the present invention;

[0030] Figure 2 for Figure 1 Schematic diagram of the specific implementation steps of the acoustic-vibration coupling model and measurement point arrangement;

[0031] Figure 3 for Figure 1 Schematic diagram of the specific implementation steps of the noise characteristic model;

[0032] Figure 4 for Figure 1 Schematic diagram of the specific implementation steps of noise identification and restoration;

[0033] Figure 5 Schematic diagram of the equipment composition for establishing noise characteristic model;

[0034] Figure 6 Schematic diagram of the equipment composition for transformer noise identification and restoration;

[0035] Among them, 1 is a transformer, 2 is a database server for storing transformer noise radiation index characteristic matrix data, 3 is a model calculation unit, which is a computer here, which contains a program for noise characteristic model generation and noise identification and restoration, 4 is a data collector, 5 is a laser vibrometer, and 6 is a microphone. DETAILED DESCRIPTION

[0036] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0037] The present embodiment provides a transformer noise identification and reverse restoration method based on laser vibration measurement. The method is based on a pre-built and trained sound identification and restoration neural network model, and the surface vibration signal of each measuring point, to obtain the corresponding standby state and noise time domain signal estimation value, and convert the noise time domain signal estimation value into a sound field index and a sound field distribution, so as to realize transformer noise identification and restoration.

[0038] refer to Figure 1 As shown, the method specifically includes three parts: acoustic-vibration coupling model and measurement point arrangement, noise characteristic model construction, and noise identification and restoration.

[0039] (a) Acoustic-vibration coupling model and measurement point arrangement

[0040] In this part, by establishing an acoustic-vibration coupling model, the results of the acoustic field simulation feature changes are obtained. According to the measurement point layout specifications and strategies, the noise and vibration measurement points are optimized, including the arrangement of the measurement point positions and numbers. This part is the basis for model construction and noise identification and restoration. Figure 2 As shown, the specific implementation steps of this part are as follows:

[0041] ① According to the geometric model and operating conditions of transformer 1, a transformer acoustic-vibration coupling model is constructed;

[0042] ② Based on the acoustic-vibration coupling model, the harmonic response acoustic field of transformer 1 is simulated to obtain the surface vibration of transformer 1 and the one-meter far-field sound pressure level distribution;

[0043] ③ According to the measurement point layout specifications and optimization layout strategies, arrange more measurement points at strong vibration locations and optimize the location and number of measurement points.

[0044] (b) Noise feature model construction

[0045] In this part, the surface vibration signal and noise sound pressure signal of the transformer are tested and a neural network model for noise recognition and restoration is established.

[0046] According to different actual working scenarios, the transformer surface vibration signal and noise sound pressure signal are obtained through laser vibrometer and microphone testing; the obtained test signal is divided into multiple time segments according to a specific time length, the noise radiation index of each frequency band in each time segment is calculated, and the corresponding equipment type, working condition and sound pressure information are marked to build a noise radiation index feature library. Based on the noise radiation index feature library, a neural network model for noise source identification and restoration is established. This part adopts the following method: Figure 5 The specific implementation steps are as follows: Figure 3 As shown, including:

[0047] ① Arrange the laser vibrometer 5 and the microphone 6 according to the steps described in (a) to obtain the position and number of measuring points, and obtain the surface vibration signal and sound pressure signal of the measuring point;

[0048] ② According to the working condition of the tested transformer, an integer multiple of the period corresponding to the transformer working frequency is selected as the reference segment length, and the time domain signals of the measured surface vibration and sound pressure are split into multiple signal segments equal to the reference segment length through the processing program in the model calculation unit 3;

[0049] ③ For each split time domain signal segment of the surface vibration and sound pressure, the noise radiation index identification algorithm in the model calculation unit 3 is applied to solve the noise radiation index of each segment; the specific steps of the noise radiation index identification algorithm are as follows:

[0050] Step 1. Convert the time domain signal fragment into the frequency domain signal fragment through fast Fourier transform;

[0051] Step 2. Construct a 1 / 3 octave filter to perform octave filtering on the frequency domain signal fragment;

[0052] Step 3. Perform inverse fast Fourier transform on the filtered frequency domain signal fragments to convert them into time domain signal fragments;

[0053] Step 4. According to the acoustic theory formula, obtain the vibration velocity level and sound power level of each frequency band; if the effective value of the vibration velocity corresponding to a certain measuring point on the transformer surface is v, and the effective value of the sound pressure at the measuring point is p, then the vibration velocity level L V Harmonic power level L W It can be expressed by the following formula:

[0054] Vibration speed level:

[0055]

[0056] Sound power level:

[0057]

[0058] Where:

[0059] v 0 is the reference vibration velocity, v 0 =10 -9 m / s;

[0060] ρc is the acoustic impedance of the current medium;

[0061] ρ 0 c 0 is the reference acoustic impedance, ρ 0 c 0 =400kg / (m 2 s);

[0062] p 0 is the reference sound pressure, p 0 =2×10 -5 Pa;

[0063] W is the sound power at the measuring point, and A is the wavefront area corresponding to the measuring point;

[0064] W 0 is the reference sound power, A 0 is the unit area, W 0 =10 -12 Pa.

[0065] Step 5. According to the noise radiation index calculation formula, solve and obtain the noise radiation index of each segment.

[0066] Noise radiation index σ and vibration speed level L in each frequency band V Harmonic power level L W The relationship is:

[0067]

[0068] Where:

[0069] S 0 is the reference structural sound radiation area, S 0 =1m 2 ;

[0070] S i is the structural sound radiation area of ​​the surface where the measuring point is located.

[0071] ④ The program in the model calculation unit 3 combines the noise radiation indexes of multiple segments into a noise radiation index feature matrix, adds the mapping relationship between the current working condition, operating status and sound pressure and the noise radiation index feature matrix, and adds it to the noise radiation index feature library; after the test, it is stored in the database server 2.

[0072] ⑤ After the data is stored, the program in the model calculation unit 3 splits the data set in the noise radiation index feature library into a training set and a test set in a ratio of 4:1, and generates a neural network model for transformer noise recognition and restoration.

[0073] (c) Noise Identification and Restoration

[0074] In this part, the vibration velocity of the sound source is measured by the laser vibration velocity method, and the impact of the sound source on the surrounding sound field is reversed. The vibration signal is converted into a sound waveform, the digital characteristics of the waveform are recorded, and the noise distribution of the actual transformer during operation is restored.

[0075] Based on the steps in (a), an optimized measurement point arrangement scheme of the transformer to be tested is obtained; based on the optimized measurement point arrangement scheme, based on the laser vibration measurement and vibration velocity method noise testing principles, the surface vibration signal of each measurement point is obtained; the surface vibration signal is split according to a set time length to obtain multiple surface vibration time domain signal segments; the multiple surface vibration time domain signal segments are used as inputs of a pre-trained noise recognition and restoration neural network model, and the neural network model is obtained based on the steps in (b), to obtain the corresponding equipment status and noise time domain signal estimation value; the noise time domain signal estimation value is converted into a sound field index and a sound field distribution to realize transformer noise recognition and restoration. This part adopts the following methods: Figure 6 The specific implementation steps are as follows: Figure 4 As shown, including:

[0076] ① Arrange the measurement points and the number of the laser vibrometer 5 according to the steps described in (a) to obtain the surface vibration signals of the measurement points;

[0077] ② According to the actual transformer type, select the integer multiple of the period corresponding to the transformer operating frequency as the reference segment length, and split the measured surface vibration time domain signal into multiple signal segments equal to the reference segment length;

[0078] ③ Taking the measured surface vibration time domain signal as input, the equipment working condition and operation status and the estimated value of the noise time domain signal are obtained through the generated neural network model;

[0079] ④ Based on the basic theory of acoustics, the program in the model calculation unit 3 expresses the estimated value of the noise time domain signal as sound field indicators and distribution such as sound pressure level and sound power level, so as to realize the identification and restoration of transformer noise.

[0080] If the above method is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0081] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for transformer noise identification and reverse restoration based on laser vibration measurement, characterized in that: The following steps are involved: Construct the acoustic-vibration coupling model of the transformer to be tested and obtain the optimized measurement point layout plan; Based on the optimized measurement point arrangement scheme, a surface vibration signal of each measurement point is obtained by using a laser vibrometer method; Splitting the surface vibration signal according to a set time length to obtain a plurality of surface vibration time domain signal segments; Using the plurality of surface vibration time domain signal segments as inputs of a pre-trained noise identification and restoration neural network model, obtaining corresponding device states and noise time domain signal estimation values; Convert the noise time domain signal estimation value into sound field index and sound field distribution to realize transformer noise identification and restoration; The training process of the noise recognition and restoration neural network model is specifically as follows: Obtaining surface vibration signals and sound pressure signals of multiple groups of transformers; Splitting the surface vibration signal and the sound pressure signal according to a set time length to obtain a plurality of surface vibration time domain signal segments and sound pressure time domain signal segments, and calculating the noise radiation index of each segment; The noise radiation indexes of multiple segments are combined into a noise radiation index feature matrix, and the mapping relationship between the device type, device status, sound pressure and the noise radiation index feature matrix is ​​added to form a noise radiation index feature library; The noise radiation index feature library is used as a sample set to train a noise recognition and restoration neural network model.

2. The method for transformer noise identification and reverse restoration based on laser vibration measurement according to claim 1 is characterized in that: The measuring point arrangement scheme includes measuring point locations and the number of measuring points.

3. The method for transformer noise identification and reverse restoration based on laser vibration measurement according to claim 1 is characterized in that: According to the geometric model and operating conditions of the transformer, the acoustic-vibration coupling model is constructed, and based on the acoustic-vibration coupling model, the harmonic response sound field simulation of the transformer is performed to obtain the optimized measurement point arrangement scheme.

4. The method for transformer noise identification and reverse restoration based on laser vibration measurement according to claim 1 is characterized in that: The set time length is an integer multiple of the period corresponding to the transformer operating frequency.

5. The method for transformer noise identification and reverse restoration based on laser vibration measurement according to claim 1 is characterized in that: The noise radiation index of each segment is calculated as follows: The noise radiation index identification algorithm is used to calculate the noise radiation index of each frequency band in the segment, and the average noise radiation index of multiple frequency bands is taken as the noise radiation index of the segment.

6. The method for transformer noise identification and reverse restoration based on laser vibration measurement according to claim 1 is characterized in that: The noise radiation indexes of the multiple segments with the same working conditions and the same equipment type are synthesized into a noise radiation index feature matrix.

7. The method for transformer noise identification and reverse restoration based on laser vibration measurement according to claim 1 is characterized in that: The sound pressure signal is obtained through a microphone.

8. The method for transformer noise identification and reverse restoration based on laser vibration measurement according to claim 1 is characterized in that: The surface vibration signal is obtained by a laser vibrometer.

9. A computer-readable storage medium, characterized in that: The method comprises one or more programs for execution by one or more processors of an electronic device, wherein the one or more programs comprise instructions for executing the transformer noise identification and reverse restoration method based on laser vibration measurement as claimed in any one of claims 1 to 8.