A method for detecting sound and light of a generator stator slot wedge based on laplace spectrum analysis

Abstract

The application discloses a kind of generator stator slot wedge sound-light detection methods based on Laplace spectrum analysis belonging to generator maintenance technical field.The target slot wedge is implemented to knock excitation, and optical displacement response signal and acoustic response signal are obtained;The optical displacement response signal and acoustic response signal are respectively preprocessed;The optical response signal and acoustic response signal after preprocessing are respectively subjected to Laplace transform, and the corresponding complex frequency domain response result is obtained;The complex frequency domain response result is subjected to spectrum analysis, and the corresponding optical characteristic parameter and acoustic characteristic parameter are extracted;The optical characteristic parameter and acoustic characteristic parameter are fused to construct joint feature vector;The joint feature vector is subjected to standardization processing and weighted fusion, and the comprehensive determination value of the current target slot wedge is calculated;The loose state of target slot wedge is output.The application improves the integrity, reliability and identification accuracy of detection result, and reduces the influence of artificial experience factor.

Description

Technical Field

[0001] This invention relates to the field of generator maintenance technology, and in particular to a generator stator slot wedge acoustic-optical detection method based on Laplace spectrum analysis. Background Technology

[0002] Generator stator slot wedges are used to press and fix stator bars, spacers, and related internal components within the stator slots. Their tightness directly affects the stability of the stator winding structure and the operational safety of the generator. During long-term generator operation, factors such as electromagnetic force, thermal stress, mechanical vibration, and material aging may cause some slot wedges to experience a decrease in clamping force or localized loosening.

[0003] When the slot wedges loosen, the constraint stiffness of the internal components decreases, making the stator bars more prone to additional vibration and friction during operation. This can lead to insulation wear, localized overheating, increased noise, and even equipment failure. Therefore, checking the tightness of the generator stator slot wedges is an important part of generator maintenance and condition assessment.

[0004] Currently, the tightness of generator stator slot wedges is mostly determined by manually tapping and judging the sound. While this method is simple to operate, the results rely heavily on the experience of maintenance personnel, are highly subjective, and differ among personnel in terms of tapping force and judgment criteria, leading to poor consistency in the results. Furthermore, given the limited space inside the generator, manual point-by-point inspection is inefficient, and the data is difficult to quantify and store, hindering subsequent condition analysis.

[0005] In recent years, some technical solutions have begun to utilize automatic tapping devices, sound acquisition devices, or optical inspection devices to detect the condition of slotted wedges. However, acoustic or optical inspection alone is still easily affected by factors such as environmental noise, surface reflection conditions, and installation location, and the stability and reliability of the detection still need to be improved.

[0006] Therefore, it is necessary to propose an acoustic-optical detection method for generator stator slot wedges based on Laplace spectrum analysis, enabling automated detection within the generator's internal environment. This method combines acoustic and optical signals for joint analysis to improve the accuracy and reliability of slot wedge tightness detection. Summary of the Invention

[0007] The purpose of this invention is to propose an acoustic-optical detection method for generator stator slot wedges based on Laplace spectral analysis. A maintenance robot is used to execute this method. The maintenance robot includes a robot body, a laser displacement detection module, a hammer striking module, and a sound acquisition module. The laser displacement detection module is located at the front of the robot and is used to acquire the optical displacement response signal of the target slot wedge after excitation. The hammer striking module is located in the middle of the robot and is used to apply standardized striking excitation to the target slot wedge. The sound acquisition module is located at the rear of the robot and is used to acquire the acoustic response signal of the target slot wedge after excitation.

[0008] Step 1: The maintenance robot enters the generator and moves to the location of the target slot wedge. It applies standardized tapping excitation to the target slot wedge using the hammer tapping module and simultaneously acquires the optical displacement response signal and acoustic response signal of the target slot wedge after being stimulated using the laser displacement detection module and the sound acquisition module.

[0009] Step 2 involves preprocessing the acquired optical displacement response signal and acoustic response signal, including removing DC components, bandpass filtering, time window truncation, amplitude normalization, and timing alignment, to improve the stability and comparability of subsequent signal analysis results.

[0010] Step 3: Perform Laplace transform on the preprocessed optical displacement response signal and acoustic response signal respectively to obtain the corresponding Laplace transform results, which are used to characterize the attenuation characteristics and frequency characteristics of the target slot wedge under the impact excitation.

[0011] Step four: Perform spectral analysis on the Laplace transform results of the optical displacement response signal and the acoustic response signal respectively, and extract the corresponding acousto-optic characteristic parameters; the acousto-optic characteristic parameters include optical characteristic parameters and acoustic characteristic parameters;

[0012] Step 5: Construct a joint feature vector from the optical and acoustic feature parameters, standardize and weight the joint feature vector to obtain a comprehensive judgment value; output the tightness state of the target wedge based on the correspondence between the comprehensive judgment value of the current target wedge and the judgment threshold; wherein, the judgment threshold is predetermined by calibration test of wedge samples with known tightness states.

[0013] Furthermore, the laser displacement detection module is arranged along the normal direction of the target wedge surface, and the detection area of ​​the target wedge is irradiated with laser. The displacement change of the target wedge surface is obtained according to the change of the reflected optical path to form an optical displacement response signal; the sound acquisition module acquires the airborne sound signal generated by the target wedge under the impact.

[0014] Furthermore, the hammer striking module, laser displacement detection module, and sound acquisition module are controlled by the same control module and perform synchronous actions according to a unified triggering sequence. At the same time as issuing the striking trigger command, the control module issues a synchronous acquisition trigger command to the laser displacement detection module and the sound acquisition module, and uses the moment when the hammer actually contacts the target slot wedge as a unified time reference to ensure that the optical displacement response signal and the acoustic response signal correspond to the same striking excitation process.

[0015] Furthermore, amplitude normalization in preprocessing is performed according to the following formula:

[0016] ;

[0017] in, The normalized discrete detection signal The discrete detection signal before normalization. This represents the mean of the discrete detected signals within the current sampling window. The standard deviation of the discrete detection signal within the current sampling window. , This represents the number of sampling points.

[0018] Furthermore, the Laplace transform in step three is specifically as follows:

[0019] Let the preprocessed optical displacement response signal be The preprocessed acoustic response signal is Then, by performing Laplace transforms on the optical displacement response signal and the acoustic response signal respectively, we obtain:

[0020] ;

[0021] ;

[0022] in, , As the attenuation factor, Angular frequency, It is the imaginary unit.

[0023] Furthermore, the response of the target slot wedge after the impact excitation satisfies the damped vibration model:

[0024] ;

[0025] in, The initial amplitude, The attenuation coefficient is... The dominant oscillation frequency, This is the initial phase;

[0026] Under discrete sampling conditions, if the sampling period is Then the discrete Laplace expressions for the optical displacement response signal and the acoustic response signal are respectively:

[0027] ;

[0028] ;

[0029] in, It is a discrete optical displacement response signal. It is a discrete acoustic response signal.

[0030] Furthermore, optical characteristic parameters include the optical root mean square value. Optical main frequency Optical spectral centroid Optical bandwidth energy ratio and optical attenuation coefficient Acoustic characteristic parameters include the acoustic root mean square value. Acoustic main frequency acoustic spectrum centroid Acoustic frequency band energy ratio Acoustic attenuation coefficient .

[0031] Furthermore, the optical root mean square value Root mean square value of acoustics Optical main frequency Acoustic main frequency Optical spectral centroid acoustic spectrum centroid Optical bandwidth energy ratio and acoustic frequency band energy ratio Calculate according to the following formulas respectively:

[0032] ;

[0033] ;

[0034] ;

[0035] Where R includes the optical root mean square value Root mean square value of acoustics f includes optical main frequency Acoustic main frequency C includes the optical spectrum centroid. acoustic spectrum centroid ;

[0036] Divide the spectrum into low-frequency bands High frequency band and satisfy ,but:

[0037] ;

[0038] ;

[0039] in, For the optical displacement response signal at the 1st Frequency points Spectral energy distribution at that location For the acoustic response signal at the 1st Frequency points Spectral energy distribution at that location This represents the total number of frequency points.

[0040] Furthermore, the optical attenuation coefficient Acoustic attenuation coefficient Obtained through envelope fitting; let the envelope function of the stimulated damping vibration model be:

[0041] ;

[0042] Taking the natural logarithm of both sides of the envelope function, we get:

[0043] ;

[0044] right With time A linear fit is performed, and the negative of the slope of the fitted line is used as the attenuation coefficient. Including optical attenuation coefficient Acoustic attenuation coefficient .

[0045] Furthermore, the joint feature vector is constructed as follows:

[0046] ;

[0047] For the first in the joint eigenvector Feature parameters Standardization is performed to obtain the standardized feature parameters. The calculation formula is as follows:

[0048] ;

[0049] in, For the first The mean of each feature parameter in the calibration sample For the first The standard deviation of each characteristic parameter in the calibration sample;

[0050] Comprehensive judgment value Calculate using the following formula:

[0051] ;

[0052] in, For the first The weight coefficients corresponding to each feature parameter satisfy the following conditions:

[0053] ;

[0054] Based on the comprehensive judgment value Compared with the preset judgment threshold , , The relationship between the output target slot wedges determines their tightness and satisfies the following conditions: ;

[0055] The judgment rule is as follows: .

[0056] The beneficial effects of this invention are as follows:

[0057] This invention addresses the problems of generator stator slot wedge tightness detection relying primarily on manual tapping and listening, resulting in subjective, poorly repeatable, and inefficient detection, as well as difficulty in quantitative analysis. It achieves quantitative grading of target slot wedge tightness, improving the completeness, reliability, and accuracy of detection results, reducing the influence of human experience, and facilitating the recording, storage, subsequent re-inspection, comparison, and trend analysis of detection data. By applying standardized tapping excitation to the target slot wedge and simultaneously acquiring its optical displacement and acoustic response signals, combined with Laplace transform, feature extraction, standardization processing, and weighted fusion, this invention solves the problems of generator stator slot wedge tightness detection relying mainly on manual tapping and listening, resulting in highly subjective test results, poor repeatability, low detection efficiency, and difficulty in quantitative analysis. Attached Figure Description

[0058] Figure 1 This is a schematic flowchart of the generator stator slot wedge acousto-optic detection method based on Laplace spectrum analysis according to the present invention.

[0059] Figure 2 This is a structural diagram of the maintenance robot;

[0060] Figure 3 This is a schematic diagram of the maintenance robot working inside the generator. Detailed Implementation

[0061] This invention proposes a generator stator slot wedge acoustic-optic detection method based on Laplace spectrum analysis. The invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0062] Figure 1This is a flowchart illustrating the acoustic-optical detection method for generator stator slot wedges based on Laplace spectral analysis, applicable to detecting the tightness of slot wedges inside generator stators. The method uses a maintenance robot as a platform, applying tapping excitation to the target slot wedge while simultaneously acquiring its optical displacement and acoustic response signals. These signals are then subjected to Laplace transform, spectral analysis, feature extraction, and fusion determination to identify the tightness of the target slot wedge. Details are as follows:

[0063] Figure 2 This is a schematic diagram of the maintenance robot. In this embodiment, the maintenance robot is installed in the internal detection channel of the generator stator and is used to move along the stator to reach the position of the wedge to be inspected. The maintenance robot is equipped with a sound acquisition module 1, a hammer module 2, and a laser displacement detection module 3. The sound acquisition module 1 is located at the rear of the maintenance robot and is used to acquire the acoustic response signal generated by the target wedge under the impact excitation. The hammer module 2 is located in the middle of the maintenance robot and is used to apply mechanical impact excitation to the target wedge. The laser displacement detection module 3 is located at the front of the maintenance robot and is used to irradiate the surface of the target wedge with a laser and acquire the optical displacement response signal of the target wedge.

[0064] 1. The maintenance robot enters the generator and moves to the location of the target slot wedge. It applies standardized tapping excitation to the target slot wedge using the hammer tapping module, and simultaneously acquires the optical displacement response signal and acoustic response signal of the target slot wedge after being stimulated using the laser displacement detection module and the sound acquisition module.

[0065] Figure 3 This is a schematic diagram of the maintenance robot operating inside the generator. In this embodiment, after entering the generator stator, the maintenance robot first moves to the location of the target slot wedge and aligns the laser displacement detection module 3 with the predetermined detection area on the surface of the target slot wedge. Subsequently, the hammer module 2 performs a controlled tap on the target slot wedge, causing the target slot wedge to generate transient mechanical vibration and corresponding acoustic response after being stimulated. At the same time, the laser displacement detection module 3 collects the minute displacement changes on the surface of the target slot wedge, and the sound acquisition module 1 collects the acoustic response of the target slot wedge after being tapped, thereby obtaining the original acoustic and optical detection signal corresponding to the tightness state of the target slot wedge.

[0066] The hammer module 2 performs controlled hammering on the target wedge, with controlled parameters including hammering position, hammering force, and hammering timing to ensure consistency of excitation conditions between different detection points and improve the comparability of detection results. The laser displacement detection module 3 is preferably arranged along the normal or approximately normal direction of the target wedge surface and irradiates the detection area of ​​the target wedge surface with laser light. It obtains the minute displacement changes of the target wedge surface by detecting changes in the reflected optical path, forming an optical displacement response signal. The sound acquisition module 1 is used to acquire the airborne sound signal generated by the target wedge after hammering excitation.

[0067] The hammer module 2, laser displacement detection module 3, and sound acquisition module 1 are controlled by the same control module and operate synchronously according to a unified trigger sequence. Simultaneously with issuing the strike trigger command, the control module sends a synchronous acquisition trigger command to both the laser displacement detection module 3 and the sound acquisition module 1. Using the strike trigger moment as a unified time reference, the starting sampling moments of the optical displacement response signal and the acoustic response signal are recorded respectively, ensuring that both types of signals correspond to the same strike excitation process.

[0068] Second, the acquired optical displacement response signal and acoustic response signal are preprocessed, including DC component removal, bandpass filtering, time window truncation, amplitude normalization, and timing alignment, to improve the stability and comparability of subsequent signal analysis results.

[0069] After acquiring the raw optical displacement response signal and the raw acoustic response signal, the two types of signals were preprocessed. Preprocessing included DC component removal, bandpass filtering, time window truncation, amplitude normalization, and timing alignment. DC component removal was used to eliminate signal baseline drift; bandpass filtering was used to suppress environmental noise and high-frequency random interference; time window truncation was used to preserve the effective attenuation response segment of the target slot wedge after impact; amplitude normalization was used to eliminate amplitude scale differences between different detection points; and timing alignment was used to ensure that the optical displacement response signal and the acoustic response signal corresponded to the same impact excitation process.

[0070] In this embodiment, the preprocessed optical displacement response signal is assumed to be The preprocessed acoustic response signal is The amplitude normalization can be performed according to the following formula:

[0071] ;

[0072] in, The normalized discrete detection signal The discrete detection signal before normalization. This represents the mean of the discrete detected signals within the current sampling window. The standard deviation of the discrete detection signal within the current sampling window. , This represents the number of sampling points.

[0073] Third, the preprocessed optical displacement response signal and acoustic response signal are subjected to Laplace transform to obtain the corresponding Laplace transform results, which are used to characterize the attenuation characteristics and frequency characteristics of the target slot wedge under the impact excitation.

[0074] After preprocessing, Laplace transforms are performed on both the optical displacement response signal and the acoustic response signal. Let the preprocessed optical displacement response signal be... The preprocessed acoustic response signal is Then their Laplace transforms are as follows:

[0075] ;

[0076] ;

[0077] in, , As the attenuation factor, Angular frequency, It is the imaginary unit.

[0078] The response of a slotted wedge under the impact of a hammer is a typical transient damped vibration process. Observing the signal amplitude change only in the time domain is insufficient to fully reflect the dynamic characteristics of the slotted wedge. While ordinary Fourier spectrum analysis can reflect the frequency distribution, it cannot directly represent the damping changes during the damping process. Through Laplace transform, the damping and frequency characteristics of the response signal can be simultaneously reflected in the complex frequency domain, thus more accurately characterizing the differences in the dynamic response of the slotted wedge under different tension states. The excited response of the slotted wedge can be described using the following damped vibration model:

[0079] ;

[0080] in, The initial amplitude, The attenuation coefficient is... The dominant oscillation frequency, This represents the initial phase. For slotted wedges in different tightness states, due to differences in their boundary constraints and contact stiffness, the corresponding dominant oscillation frequency, attenuation coefficient, and spectral distribution all differ. Therefore, their states can be distinguished using the characteristics of the Laplace domain and the spectral domain.

[0081] Under discrete sampling conditions, if the sampling period is Then the discrete Laplace expressions for the optical displacement response signal and the acoustic response signal are respectively:

[0082] ;

[0083] ;

[0084] in, It is a discrete optical displacement response signal. It is a discrete acoustic response signal.

[0085] Fourth, perform spectral analysis on the Laplace transform results of the optical displacement response signal and the acoustic response signal respectively, and extract the corresponding acousto-optic characteristic parameters; the acousto-optic characteristic parameters include optical characteristic parameters and acoustic characteristic parameters.

[0086] After obtaining the Laplace transform results of the optical displacement response signal and the acoustic response signal, spectral analysis is performed on both, and the corresponding characteristic parameters are extracted. In this embodiment, the optical characteristic parameters include the optical root mean square value. Optical main frequency Optical spectral centroid Optical bandwidth energy ratio and optical attenuation coefficient Acoustic characteristic parameters include the acoustic root mean square value. Acoustic main frequency acoustic spectrum centroid Acoustic frequency band energy ratio Acoustic attenuation coefficient .

[0087] The optical root mean square (RMS) value and the acoustic RMS value are calculated according to the following formulas:

[0088] ;

[0089] ;

[0090] The dominant frequency is defined as the frequency point at which the spectral energy distribution of the optical displacement response signal and the acoustic response signal reaches its maximum value, respectively.

[0091] ;

[0092] ;

[0093] in, For the optical displacement response signal at the 1st Frequency points Spectral energy distribution at that location For the acoustic response signal at the 1st Frequency points The spectral energy distribution at that location.

[0094] The centroids of the optical and acoustic spectra are calculated using the following formulas:

[0095] ;

[0096] ;

[0097] in, This represents the total number of frequency points. The spectrum is divided into low-frequency bands. High frequency band and satisfy Then the optical band energy ratio and the acoustic band energy ratio can be expressed as follows:

[0098] ;

[0099] ;

[0100] For the extraction of the damping coefficient, let the envelope function of the excited damping vibration model be:

[0101] ;

[0102] Taking the natural logarithm of both sides of the envelope function, we get:

[0103] ;

[0104] Through the With time Perform a linear fit, and the negative of the slope of the fitted line is used as the attenuation coefficient; the attenuation coefficient obtained from the optical displacement response signal is denoted as... The attenuation coefficient obtained from the acoustic response signal is denoted as .

[0105] 5. Construct a joint feature vector from the optical and acoustic feature parameters, standardize and weight the joint feature vector to obtain a comprehensive judgment value; output the tightness state of the target wedge based on the correspondence between the comprehensive judgment value of the current target wedge and the judgment threshold; wherein, the judgment threshold is predetermined by calibration test of wedge samples with known tightness states.

[0106] After obtaining the five optical feature parameters and the five acoustic feature parameters respectively, they are constructed into a unified joint feature vector:

[0107] .

[0108] To eliminate the influence of differences in the dimensions of different feature parameters on the judgment results, the feature parameters in the joint feature vector are standardized. Let the first... The original values ​​of the feature parameters are: Its standardized value is ,but:

[0109] ;

[0110] in, For the first The mean of each feature parameter in the calibration sample For the first The standard deviation of each characteristic parameter in the calibration sample.

[0111] The weighting coefficients and judgment thresholds corresponding to the ten characteristic parameters are predetermined through calibration tests. In the calibration tests, slot wedge samples with known tightness, known moderate tightness, known slight looseness, and known obvious looseness are selected respectively. The maintenance robot is used to repeatedly test the slot wedge samples in each state, collect the corresponding optical displacement response signals and acoustic response signals, and calculate the ten characteristic parameters corresponding to each sample.

[0112] For the The first under similar tightness state Let the standardized joint feature vector of n samples be denoted as:

[0113] ;

[0114] in, Indicates the type of wedge tightness. This indicates the sample number within that category.

[0115] For each type of slotted wedge state sample, calculate the mean vector of its ten-dimensional standardized feature vector, which serves as the standard feature center for that type. Let the first... The standard feature center of the groove wedge-like state is:

[0116] ;

[0117] in: ;

[0118] For the first Number of samples in the slot wedge state.

[0119] During the calibration process, the discrimination degree of each feature parameter between different slot wedge state categories is calculated, and the weight coefficient corresponding to each feature parameter is determined based on the discrimination degree. For the first... There are _ ... , defined as the ratio of the range of the feature parameter among the standard feature centers of each category to the average fluctuation within the category, i.e.:

[0120] ;

[0121] in, For the first In the class of samples, the first The standard deviation of each characteristic parameter.

[0122] Based on the discriminative power of each feature parameter The size determines the weighting coefficients. for:

[0123] ;

[0124] In this embodiment, the weighting coefficients corresponding to each feature parameter are predetermined through calibration tests. During the calibration tests, slotted wedge samples with known tightness, known moderate tightness, known slight looseness, and known significant looseness are selected, and their corresponding optical displacement response signals and acoustic response signals are collected. The feature parameters corresponding to each sample are then calculated. Based on the ability of different feature parameters to distinguish between the tightness and looseness of the slotted wedge, the weighting coefficients corresponding to each feature parameter are determined, and then the comprehensive judgment value is calculated.

[0125] ;

[0126] in, For the first The weight coefficients corresponding to each feature parameter satisfy the following conditions:

[0127] .

[0128] Based on the comprehensive judgment value Compared with the preset judgment threshold , , The relationship between these factors is used to classify and determine the tightness of the target slot wedge. The judgment rules are as follows:

[0129] ;

[0130] After the maintenance robot completes one inspection of a target slot wedge, it sequentially performs acoustic and optical signal acquisition, preprocessing, Laplace transform, spectrum analysis, feature parameter extraction, feature parameter standardization, weighted fusion, and state determination. It then establishes a correspondence between the inspection result of the target slot wedge and the current slot position information, thereby outputting the tightness level of the target slot wedge.

[0131] In this embodiment, the maintenance robot can inspect multiple target slot wedges one by one along the internal channel of the generator stator, and output the corresponding detection results for each target slot wedge. By employing a combined acoustic and optical detection method, and performing Laplace transform and spectral analysis on the optical displacement response signal and acoustic response signal respectively, this invention can more fully reflect the attenuation characteristics, frequency characteristics, and energy distribution characteristics of the target slot wedge under different tightness states, thereby improving the accuracy and reliability of slot wedge tightness detection.

Claims

1. A method for optical detection of a generator stator slot wedge sound based on Laplace spectral analysis, characterized in that, A maintenance robot is used to perform an acoustic-optical detection method for generator stator slot wedges based on Laplace spectrum analysis. The maintenance robot includes a robot body, a laser displacement detection module, a hammer striking module, and a sound acquisition module. The laser displacement detection module is located at the front of the maintenance robot and is used to acquire the optical displacement response signal of the target slot wedge after excitation. The hammer striking module is located in the middle of the maintenance robot and is used to apply standardized striking excitation to the target slot wedge. The sound acquisition module is located at the rear of the maintenance robot and is used to acquire the acoustic response signal of the target slot wedge after excitation. Step 1: The maintenance robot enters the generator and moves to the location of the target slot wedge. It applies standardized tapping excitation to the target slot wedge using the hammer tapping module and simultaneously acquires the optical displacement response signal and acoustic response signal of the target slot wedge after being stimulated using the laser displacement detection module and the sound acquisition module. Step 2 involves preprocessing the acquired optical displacement response signal and acoustic response signal, including removing DC components, bandpass filtering, time window truncation, amplitude normalization, and timing alignment, to improve the stability and comparability of subsequent signal analysis results. Step 3: Perform Laplace transform on the preprocessed optical displacement response signal and acoustic response signal respectively to obtain the corresponding Laplace transform results, which are used to characterize the attenuation characteristics and frequency characteristics of the target slot wedge under the impact excitation. Step four: Perform spectral analysis on the Laplace transform results of the optical displacement response signal and the acoustic response signal respectively, and extract the corresponding acousto-optic characteristic parameters; the acousto-optic characteristic parameters include optical characteristic parameters and acoustic characteristic parameters; Step 5: Construct a joint feature vector from the optical and acoustic feature parameters, standardize and weight the joint feature vector to obtain a comprehensive judgment value; output the tightness state of the target wedge based on the correspondence between the comprehensive judgment value of the current target wedge and the judgment threshold; wherein, the judgment threshold is predetermined by calibration test of wedge samples with known tightness states.

2. The method of claim 1, wherein the method is based on Laplacian spectral analysis of the stator slot wedge sound. The laser displacement detection module is arranged along the normal direction of the target wedge surface, and irradiates the detection area of ​​the target wedge with laser. The displacement change of the target wedge surface is obtained according to the change of reflected optical path to form an optical displacement response signal; the sound acquisition module acquires the airborne sound signal generated by the target wedge under the impact.

3. The method of claim 1, wherein the method further comprises: The hammer striking module, laser displacement detection module, and sound acquisition module are controlled by the same control module and perform synchronous actions according to a unified triggering sequence. The control module sends a synchronous acquisition triggering command to the laser displacement detection module and the sound acquisition module at the same time as issuing the striking trigger command, and uses the moment when the hammer actually contacts the target slot wedge as a unified time reference to ensure that the optical displacement response signal and the acoustic response signal correspond to the same striking excitation process.

4. The method of claim 1, wherein the method further comprises: Amplitude normalization in preprocessing is performed according to the following formula: ； in, The normalized discrete detection signal The discrete detection signal before normalization. This represents the mean of the discrete detected signals within the current sampling window. The standard deviation of the discrete detection signal within the current sampling window. , This represents the number of sampling points.

5. The generator stator slot wedge acoustic-optic detection method based on Laplace spectrum analysis according to claim 1, characterized in that, The Laplace transform in step three specifically involves: Let the preprocessed optical displacement response signal be The preprocessed acoustic response signal is Then, by performing Laplace transforms on the optical displacement response signal and the acoustic response signal respectively, we obtain: ； ； in, , As the attenuation factor, Angular frequency, It is the imaginary unit.

6. The generator stator slot wedge acoustic-optic detection method based on Laplace spectrum analysis according to claim 5, characterized in that, The response of the target slot wedge after impact excitation satisfies the damped vibration model: ； in, The initial amplitude, The attenuation coefficient is... The dominant oscillation frequency, This is the initial phase; Under discrete sampling conditions, if the sampling period is Then the discrete Laplace expressions for the optical displacement response signal and the acoustic response signal are respectively: ； ； in, It is a discrete optical displacement response signal. It is a discrete acoustic response signal.

7. The generator stator slot wedge acoustic-optic detection method based on Laplace spectrum analysis according to claim 1, characterized in that, The optical characteristic parameters include the optical root mean square value. Optical main frequency Optical spectral centroid Optical band energy ratio and optical attenuation coefficient The acoustic characteristic parameters include the acoustic root mean square value. Acoustic main frequency acoustic spectrum center of gravity Acoustic frequency band energy ratio Acoustic attenuation coefficient .

8. The generator stator slot wedge acoustic-optic detection method based on Laplace spectrum analysis according to claim 7, characterized in that, Optical root mean square value Root mean square value of acoustics Optical main frequency Acoustic main frequency Optical spectral centroid acoustic spectrum center of gravity Optical band energy ratio and acoustic frequency band energy ratio Calculate according to the following formulas respectively: ； ； ； Where R includes the optical root mean square value Root mean square value of acoustics f includes optical main frequency Acoustic main frequency C includes the optical spectrum centroid. acoustic spectrum center of gravity ; Divide the spectrum into low-frequency bands High frequency band and satisfy ,but: ； ； in, For the optical displacement response signal at the 1st Frequency points Spectral energy distribution at that location For the acoustic response signal at the 1st Frequency points Spectral energy distribution at that location This represents the total number of frequency points.

9. The generator stator slot wedge acousto-optic detection method based on Laplace spectrum analysis according to claim 7, characterized in that, Optical attenuation coefficient Acoustic attenuation coefficient Obtained through envelope fitting; let the envelope function of the stimulated damping vibration model be: ； Taking the natural logarithm of both sides of the envelope function, we get: ； right With time A linear fit is performed, and the negative of the slope of the fitted line is used as the attenuation coefficient. Including optical attenuation coefficient Acoustic attenuation coefficient .

10. The generator stator slot wedge acousto-optic detection method based on Laplace spectrum analysis according to claim 1, characterized in that, The joint feature vector is constructed as follows: ； For the first in the joint eigenvector One feature parameter Standardization is performed to obtain the standardized feature parameters. The calculation formula is as follows: ； in, For the first The mean of each feature parameter in the calibration sample For the first The standard deviation of each characteristic parameter in the calibrated sample; Comprehensive judgment value Calculate using the following formula: ； in, For the first The weight coefficients corresponding to each feature parameter satisfy the following conditions: ； Based on the comprehensive judgment value Compared with the preset judgment threshold , , The relationship between the output target slot wedges determines their tightness and satisfies the following conditions: ; The judgment rule is as follows: .

Images