A homologous laser-based spectral acoustic wave heterogeneous signal insulator detection method

Abstract

The application discloses a kind of based on homologous laser's spectral acoustic wave heterostructure signal insulator detection method, it belongs to insulator detection technical field, it solves the problem that existing insulator detection method cannot detect the defect inside and outside insulator simultaneously.The present application uses the same laser pulse single excitation and synchronously collects spectrum and acoustic wave signal, ensures the homology of signal, lays the physical foundation for synchronous detection of internal and external defects.By introducing equivalent dielectric response parameters and equivalent structure density parameters, the two types of features are physically modulated, which specifically represent the surface composition and internal structure state. After standardization and attention weighting fusion, the integrated representation of insulator comprehensive state is realized, and the rapid and non-contact identification of insulator defects is realized, which is especially suitable for online detection of power equipment.

Description

Technical Field

[0001] This invention relates to the field of insulator testing technology, specifically to an insulator testing method based on spectral acoustic heterogeneous signals from a homogeneous laser source. Background Technology

[0002] Insulators are key equipment for ensuring the safe operation of power grids. During long-term operation, the performance degradation of insulators is mainly manifested in external defects (such as surface dirt deposition, sheath aging, and surface cracks of porcelain insulators) and internal defects (such as delamination of the interface between the core rod and sheath of composite insulators, internal cracks and bubbles in porcelain insulators, and intrusion of metal accessories). These defects are major hidden dangers that may lead to mechanical fracture or electrical breakdown accidents.

[0003] Currently, the detection of internal defects in insulators mainly relies on traditional non-destructive testing techniques, such as ultrasonic testing, X-ray testing, and optical methods (such as infrared and ultraviolet). In recent years, the detection of external defects in insulators has primarily relied on laser probe technology, particularly laser-induced breakdown spectroscopy (LIBS), which has shown potential in assessing external defects in insulators due to its ability to provide information on the elemental composition of the material.

[0004] However, LIBS signals are generated at extremely shallow depths (micrometers), and therefore cannot directly reflect millimeter-scale physical structural defects within materials. While methods combining LIBS with external acoustic emission exist, these are typically simple superpositions of signals from different sources and locations. The lack of an intrinsic physical connection between the two signals results in insufficient information fusion depth, limiting the diagnostic accuracy and reliability for complex internal defects. A key challenge in achieving simultaneous detection of external and internal defects in insulators is how to simultaneously excite spectral signals containing chemical composition information and acoustic signals containing structural integrity information using the same excitation source.

[0005] In summary, existing insulator testing technologies lack a technique that can simultaneously detect internal and external defects in insulators. Summary of the Invention

[0006] To address the aforementioned shortcomings of existing technologies, this invention provides a method for detecting insulators using spectral acoustic heterogeneous signals based on co-source lasers. This method solves the problem that existing insulator detection methods cannot simultaneously detect internal and external defects in insulators.

[0007] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows:

[0008] A method for detecting heterogeneous spectral acoustic signals insulators based on homogeneous laser sources is provided, comprising the following steps:

[0009] S1. The same pulsed laser source is used to excite the surface of the insulator under test, and the spectral signal and acoustic signal based on the same laser excitation condition are collected and preprocessed simultaneously.

[0010] S2. Construct spectral feature vectors based on spectral signals and utilize the equivalent dielectric response parameters of insulators. The spectral eigenvectors are modulated; a short-time Fourier transform is performed on the acoustic signal to obtain its frequency domain amplitude function, and the energy of multiple frequency bands of the frequency domain amplitude function is calculated within multiple preset frequency bands. All frequency band energies are constructed into acoustic eigenvectors, and the compactness parameters of the insulator's equivalent structure are utilized. Modulate the sound wave feature vector;

[0011] S3. Perform Z-score normalization on the modulated spectral feature vector and acoustic feature vector respectively to obtain normalized spectral feature vector and normalized acoustic feature vector, and then perform attention-weighted fusion on the normalized spectral feature vector and normalized acoustic feature vector to obtain fused feature vector;

[0012] S4. Input the fused feature vector into the pre-trained insulator defect detection model to obtain the prediction probability vector that reflects the probability of the insulator defect category, and obtain the prediction probability of each preset defect category of the insulator.

[0013] Furthermore, the expression for calculating the spectral eigenvector is as follows:

[0014] ; ;

[0015] in, For the first Spectral eigenvectors under secondary excitation conditions; , , and The first Under the excitation condition, the first, second, and third rays in the spectral signal Article and Section The radiation intensity of the characteristic spectral lines of the bar; The acquisition response coefficients of the spectral signal; For the first The first excitation produces the second The particle number density of the element with the characteristic spectral lines; and The first The spectral transition probabilities and upper level energies corresponding to each characteristic spectral line; Boltzmann's constant; For the first The transient temperature of the plasma generated under secondary excitation.

[0016] Furthermore, the calculation expression for modulating the spectral eigenvector is as follows:

[0017] ; ;

[0018] in, For the first The modulated spectral feature vector under secondary excitation conditions; This is the reference total energy value obtained under standard insulator sample conditions.

[0019] Furthermore, the expression for calculating the acoustic wave eigenvector is as follows:

[0020] ; ;

[0021] ; ;

[0022] in, For the first Acoustic wave characteristic vector under secondary excitation conditions; In the first The frequency domain amplitude function of the acoustic signal under secondary excitation conditions; , , and They are the preset 1st, 2nd, and 3rd. The and the first Each frequency band boundary frequency; In the first The time-domain signal in the acoustic wave signal under secondary excitation conditions; For time window functions; Let be the complex exponential kernel function of the Fourier transform. , and These represent the imaginary unit, frequency, and time, respectively. For the first Peak sound pressure level of the sound wave signal under secondary excitation conditions; The attenuation coefficient of sound waves in the propagation medium; for The dominant frequency of the acoustic signal under secondary excitation conditions; The acoustic energy coupling coefficient; The single-pulse energy of the pulsed laser source; The distance between the acoustic sensor used to acquire sound wave signals and the laser excitation point in the pulsed laser source; This represents the pulse width of the pulsed laser source.

[0023] Furthermore, the calculation expression for modulating the acoustic wave eigenvector is as follows:

[0024] ; ;

[0025] in, For the first The characteristic vector of the modulated acoustic wave under secondary excitation conditions; and These are the preset lower and upper limits of the high-frequency band in the sound wave signal; It is the highest frequency in the sound wave signal.

[0026] Furthermore, the calculation expression for Z-score normalization of the modulated spectral feature vector and acoustic feature vector is as follows:

[0027] ;

[0028] in, and These are the standardized spectral eigenvectors and the standardized acoustic eigenvectors, respectively. and These are the acoustic eigenvectors and acoustic eigenvectors under all excitation conditions, respectively. and They are respectively and Standard deviation; and They are respectively and The mean.

[0029] Furthermore, the expression for the fused feature vectors is:

[0030] ;

[0031] in, To fuse feature vectors; and They are respectively and Attention weights.

[0032] Furthermore, the expressions for calculating the attention weights of the normalized spectral eigenvector and the normalized acoustic eigenvector are as follows:

[0033] , ;

[0034] in, and They are respectively and The preset nonlinear mapping function; This refers to the energy deposition efficiency on the insulator surface.

[0035] Furthermore, the spectral acoustic heterogeneous signal insulator detection method also includes:

[0036] S5. Based on the predicted probability of each preset defect category of the insulator, the physical consistency constraint verification is performed on each preset defect category in descending order of predicted probability. If the verified preset defect category satisfies all physical consistency constraint rules at the same time, the insulator is determined to have the currently verified preset defect category; otherwise, the insulator is determined not to have the currently verified preset defect category.

[0037] Furthermore, the physical consistency constraint rules include at least the following:

[0038] Dielectric response consistency constraint: equivalent dielectric response parameters The value is located within the preset range corresponding to the preset defect category being verified;

[0039] Structural compactness consistency constraints: equivalent structural compactness parameters The variation trend under multiple excitation conditions is consistent with the variation trend of the pre-set equivalent structure compactness parameter corresponding to the pre-set defect category verified.

[0040] Energy deposition consistency constraint: The ratio of spectral energy to acoustic energy of the pulsed laser source is matched with the preset surface energy deposition efficiency corresponding to the preset defect category.

[0041] Compared with the prior art, the present invention has the following significant advantages:

[0042] 1. This invention first employs a single laser pulse to excite and simultaneously acquire spectral and acoustic signals, ensuring that the two types of signals originate from the same spatiotemporal event and have an inherent physical correlation, thus laying the physical foundation for synchronous detection. Next, an equivalent dielectric response parameter is introduced to modulate the spectral features. This parameter is calculated from the overall energy of the spectrum, specifically characterizing external defects such as surface contamination and aging. Simultaneously, an equivalent structural compactness parameter is introduced to modulate the acoustic features. This parameter is calculated from the proportion of high-frequency energy in the acoustic signal and is specifically used to quantify the structural integrity of the insulator, such as microcracks, pores, and interface delamination, specifically characterizing internal structural defects such as cracks and delamination. This transforms the heterogeneous signals obtained from a single excitation into directional information carriers targeting external and internal defects. Then, these features carrying different physical meanings are standardized to eliminate dimensional differences and undergo attention-weighted fusion to form a unified fusion feature vector characterizing the overall state of the insulator. Finally, this fusion feature vector is input into a pre-trained insulator defect detection model, directly outputting predicted probabilities that simultaneously cover both internal and external defect categories. The entire solution constructs an integrated framework with a clear physical mechanism, from the source of data generation to the end of information processing and decision-making, and realizes the synchronous acquisition and collaborative diagnosis of internal and external defects of insulators.

[0043] 2. This invention uses insulators, widely used in power systems, as the detection object. It fully considers the characteristics of insulator materials: high dielectric strength, low conductivity, stable composition, but significant changes in service condition. By using a co-source laser excitation method to form a controllable plasma on the insulator surface, it achieves non-contact acquisition of material state information that is difficult to obtain directly using traditional electrical methods. During long-term operation, the performance degradation of insulators is mainly manifested in surface contamination, aging, microcracks, and decreased hydrophobicity. These changes directly affect the formation process and radiation characteristics of laser-induced plasma. This invention, by normalizing multiple characteristic spectral lines in the laser-induced breakdown spectrum, can effectively reflect changes in the composition and aging degree of the insulator surface, thereby achieving the detection of external defects in insulators. Considering the generally brittle nature of insulator materials and the high correlation between mechanical response and internal defects, this invention introduces a laser-induced acoustic signal, which is co-sourced with the spectral signal, as an auxiliary detection information source. The acoustic signal is highly sensitive to microcracks, changes in structural compactness, and energy deposition efficiency in insulators, compensating for the shortcomings of single spectral signals in characterizing structural defects.

[0044] 3. The present invention is based on the spectral signal and acoustic signal obtained under the same source laser conditions. They have consistency in time, space and excitation energy, so that the constructed spectral-acoustic heterogeneous features can simultaneously characterize the composition characteristics and mechanical response characteristics of the insulator material, which is more in line with the physical mechanism of the evolution of the insulator service state.

[0045] 4. This invention enables rapid, non-contact identification of insulator defects, aging, or contamination, avoiding the limitations imposed by environmental conditions, voltage levels, and safety distances on traditional manual inspections and contact testing methods. It is particularly suitable for online or near-online testing of insulators in transmission lines and substations, and has significant advantages in engineering applications.

[0046] 5. This invention introduces physical consistency constraints such as dielectric response characteristics, structural compactness, and energy deposition characteristics of insulators through physical consistency constraint rules, so that the discrimination results not only come from the output of the data-driven model, but also are consistent with the actual physical state of the insulator, thereby improving the reliability and engineering applicability of insulator defect detection. Attached Figure Description

[0047] Figure 1 This is a flowchart of a method for detecting insulators using spectral acoustic heterogeneous signals based on co-source lasers. Detailed Implementation

[0048] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0049] refer to Figure 1 This embodiment provides a method for detecting heterogeneous spectral acoustic signals insulators based on homogeneous laser sources, including the following steps:

[0050] S1. The same pulsed laser source is used to excite the surface of the insulator under test, and the spectral signal and acoustic signal based on the same laser excitation condition are collected and preprocessed simultaneously.

[0051] The specific process is as follows: The same pulsed laser source is used to excite the surface of the insulator under test; when the laser power density... When the laser-induced breakdown threshold of the insulator material is greater than or equal to the threshold, laser breakdown occurs on the insulator surface, causing rapid vaporization and ionization of the local material, forming laser-induced plasma. During the formation and expansion of the plasma, on the one hand, the plasma radiates a laser-induced breakdown spectral signal containing information about the composition of the insulator material during cooling; on the other hand, due to the transient high-temperature and high-pressure expansion of the plasma, it impacts the surrounding medium, thereby exciting acoustic signals. By synchronously triggering the spectral acquisition device and the acoustic sensor, spectral and acoustic signals are acquired simultaneously under the same laser excitation condition, thus forming spectral-acoustic heterogeneous signal data under the same laser source condition.

[0052] S2. Construct spectral feature vectors based on spectral signals and utilize the equivalent dielectric response parameters of insulators. Modulate the spectral eigenvectors.

[0053] Specifically, the expression for calculating the spectral eigenvector is as follows:

[0054] ; ;

[0055] in, For the first Spectral eigenvectors under secondary excitation conditions; , , and The first Under the excitation condition, the first, second, and third rays in the spectral signal Article and Section The radiation intensity of the characteristic spectral lines of the bar; The acquisition response coefficients of the spectral signal; For the first The first excitation produces the second The particle number density of the element with the characteristic spectral lines; and The first The spectral transition probabilities and upper level energies corresponding to each characteristic spectral line; Boltzmann's constant; For the first The transient temperature of the plasma generated under secondary excitation.

[0056] To address the aging, contamination, and hydrophobic degradation phenomena that occur in insulator materials during long-term service, an equivalent dielectric response parameter for insulators is introduced. This is used to characterize the overall response capability of the insulator surface to the laser electromagnetic field. Based on the overall energy distribution characteristics of the spectral signal, the equivalent dielectric response parameters are constructed: ,in, This is the reference total energy value of the spectrum obtained under standard insulator sample conditions. The calculation expression for modulating the spectral eigenvector is: ;in, For the first The modulated spectral feature vector under secondary excitation conditions. The electrical state (such as contamination and aging) of the insulator surface can be directly quantified by calculating the total spectral energy, thereby enabling the modulated spectral feature vector to specifically characterize external defects.

[0057] The sound wave signal is subjected to a short-time Fourier transform to obtain its frequency domain amplitude function. The energy of multiple frequency bands of the frequency domain amplitude function is calculated within multiple preset frequency bands. All frequency band energies are constructed into a sound wave feature vector, and the compactness parameter of the insulator's equivalent structure is utilized. Modulate the sound wave feature vector.

[0058] Specifically, the expression for calculating the acoustic wave eigenvector is as follows:

[0059] ; ;

[0060] ; ;

[0061] in, For the first The acoustic wave characteristic vector under secondary excitation conditions is composed of acoustic wave energy characteristics in multiple frequency bands and is used to characterize the mechanical response characteristics, structural compactness and microcrack state of the insulator under laser excitation conditions. In the first The frequency domain amplitude function of the acoustic signal under secondary excitation conditions; , , and They are the preset 1st, 2nd, and 3rd. The and the first Each frequency band has a boundary frequency, and two adjacent boundary frequencies constitute a frequency band. This band is used to calculate the sound wave energy characteristics within that band. By reasonably selecting the number of frequency bands, a balance can be achieved between feature resolution and computational complexity. In the first Under secondary excitation conditions, the time-domain signal in the acoustic signal reflects the transient impact response of laser-induced plasma expansion to the surrounding medium. For time window functions; Let be the complex exponential kernel function of the Fourier transform. , and These represent the imaginary unit, frequency, and time, respectively. For the first Peak sound pressure level of the sound wave signal under secondary excitation conditions; The attenuation coefficient of sound waves in the propagation medium; For the first The dominant frequency of the acoustic signal under secondary excitation conditions; The acoustic energy coupling coefficient is a core proportionality coefficient that correlates the input laser energy with the output acoustic signal, reflecting the efficiency of the insulator in converting laser energy into acoustic energy. The system characteristic parameter can be calculated by using a standard insulator sample (known to be defect-free or in good condition) and the current detection system (pulsed laser source and acoustic sensor) to back-calculate the system parameter based on the peak sound pressure of the acoustic signal. The single-pulse energy of the pulsed laser source; The distance between the acoustic sensor used to acquire sound wave signals and the laser excitation point in the pulsed laser source; This represents the pulse width of the pulsed laser source.

[0062] The expression for calculating the modulation of the acoustic wave eigenvector is as follows:

[0063] ; ;

[0064] in, For the first The characteristic vector of the modulated acoustic wave under secondary excitation conditions; and These are the preset lower and upper limits of the high-frequency band in the sound wave signal; It is the highest frequency in the sound wave signal. This refers to the energy in the high-frequency band of the sound wave signal; The total energy of the sound wave signal; By calculating the proportion of high-frequency energy in the acoustic signal, the structural integrity (such as cracks, pores, and delamination) inside the material can be directly quantified, so that the modulated acoustic feature vector can specifically characterize the internal defects.

[0065] S3. Perform Z-score normalization on the modulated spectral feature vector and acoustic feature vector respectively to obtain normalized spectral feature vector and normalized acoustic feature vector. Then, perform attention-weighted fusion on the normalized spectral feature vector and normalized acoustic feature vector to obtain fused feature vector.

[0066] Specifically, the calculation expression for Z-score normalization of the modulated spectral feature vector and acoustic feature vector is as follows:

[0067] ;

[0068] in, and These are the standardized spectral eigenvectors and the standardized acoustic eigenvectors, respectively. and These are the acoustic eigenvectors and acoustic eigenvectors under all excitation conditions, respectively. and They are respectively and Standard deviation; and They are respectively and The mean.

[0069] The expression for the fused feature vectors is:

[0070] ; , ;

[0071] in, To fuse feature vectors; and They are respectively and Attention weights. and They are respectively and The preset nonlinear mapping function can include Sigmoid, Tanh or exponential functions to achieve nonlinear adjustment of weights between 0 and 1; This refers to the surface energy deposition efficiency of the insulator.

[0072] In this embodiment, the surface energy deposition efficiency The effective absorption capacity of laser energy on the surface of an insulator is used to characterize the spectral energy inversion method, and its calculation expression is as follows:

[0073] ; ; ;

[0074] in, In the first Under the same laser excitation conditions, the total spectral energy of the laser-induced plasma radiation spectrum and the material absorption energy satisfy a monotonic correspondence. To perform on standard insulators under the same testing conditions The normalized index obtained from the second measurement; For the first Single laser pulse energy.

[0075] To better understand, and The method for obtaining the data is as follows: First, a calibration dataset is constructed, which includes standard insulator samples and defective insulator samples. The defective insulator samples include various different defects. Then, the equivalent dielectric response parameters of each sample in the calibration dataset are collected. Surface energy deposition efficiency and equivalent structural compactness parameters The corresponding defect labels are recorded; with the goal of minimizing the detection error of the fused feature vectors, the optimal attention weights are obtained. and Then, a mapping function from the input parameters to the weights is established by fitting, such as the Sigmoid, Tanh, or exponential function, with the Sigmoid function being preferred. The expression is as follows: ; ;in, All are coefficients, obtained through calibration by minimizing detection error. The mapping relationship is as follows: and The higher the value, the more reliable the spectral signal. The higher; The higher the level, the stronger the sound wave characterization ability. The larger.

[0076] S4. Input the fused feature vector into the pre-trained insulator defect detection model to obtain the prediction probability vector that reflects the probability of the insulator defect category, and obtain the prediction probability of each preset defect category of the insulator.

[0077] In this implementation, the insulator defect detection model and its pre-training method are a conventional neural network and training steps, respectively. Specifically, insulator samples covering various defects and normal states (standardized) are collected, and steps S1 and S3 are used to detect all insulator samples, generating corresponding fused feature vectors. The defect category of each sample is accurately labeled, forming a dataset with paired "feature vector - defect label vector". The generated dataset is then divided into training, validation, and test sets. A deep learning model (e.g., support vector machine, random forest, fully connected neural network, or convolutional neural network) is selected as the infrastructure. The training set data is input into the model for training. The model's internal parameters are continuously adjusted using optimization algorithms (e.g., gradient descent) to achieve convergence. The validation set is used to periodically evaluate the model's performance during training, monitoring for overfitting or underfitting, and adjusting model hyperparameters (e.g., learning rate, number of network layers) or applying regularization techniques accordingly. Finally, an independent test set is used to evaluate the trained model, calculating metrics such as accuracy, recall, and F1 score to measure its generalization ability. Once the performance meets the requirements, the final model structure and parameters are saved, thus obtaining the pre-trained insulator defect detection model.

[0078] As a further embodiment, the method for detecting heterogeneous signals from spectral acoustic waves insulators also includes:

[0079] S5. Based on the predicted probability of each preset defect category of the insulator, the physical consistency constraint verification is performed on each preset defect category in descending order of predicted probability. If the verified preset defect category satisfies all physical consistency constraint rules at the same time, the insulator is determined to have the currently verified preset defect category; otherwise, the insulator is determined not to have the currently verified preset defect category.

[0080] In this embodiment, the physical consistency constraint rules include at least the following:

[0081] Dielectric response consistency constraint: equivalent dielectric response parameters The value is located within the preset range corresponding to the preset defect category being verified;

[0082] Structural compactness consistency constraints: equivalent structural compactness parameters The variation trend under multiple excitation conditions is consistent with the variation trend of the pre-set equivalent structure compactness parameter corresponding to the pre-set defect category verified.

[0083] Energy deposition consistency constraint: The ratio of spectral energy to acoustic energy of the pulsed laser source is matched with the preset surface energy deposition efficiency corresponding to the preset defect category.

[0084] In summary, this scheme first employs a single laser pulse to excite and simultaneously acquire spectral and acoustic signals, ensuring that the two types of signals originate from the same spatiotemporal event and have an inherent physical correlation, thus laying the physical foundation for synchronous detection. Next, an equivalent dielectric response parameter is introduced to modulate the spectral features. This parameter, calculated from the overall spectral energy, specifically characterizes external defects such as surface contamination and aging. Simultaneously, an equivalent structural compactness parameter is introduced to modulate the acoustic features. This parameter, calculated from the proportion of high-frequency energy in the acoustic signal, is specifically used to quantify the structural integrity of the insulator, such as microcracks, pores, and interface delamination, specifically characterizing internal structural defects like cracks and delamination. This transforms the heterogeneous signals obtained from a single excitation into directional information carriers targeting external and internal defects. Then, these features carrying different physical meanings are standardized to eliminate dimensional differences and fused using attention-weighted fusion to form a unified fusion feature vector characterizing the overall state of the insulator. Finally, this fusion feature vector is input into a pre-trained insulator defect detection model, directly outputting predicted probabilities that simultaneously cover both internal and external defect categories. The entire solution constructs an integrated framework with a clear physical mechanism, from the source of data generation to the end of information processing and decision-making, and realizes the synchronous acquisition and collaborative diagnosis of internal and external defects of insulators.

Claims

1. A method for detecting heterogeneous spectral acoustic signals insulators based on co-source lasers, characterized in that, Including the following steps: S1. The same pulsed laser source is used to excite the surface of the insulator under test, and the spectral signal and acoustic signal based on the same laser excitation condition are collected and preprocessed simultaneously. S2. Construct spectral feature vectors based on spectral signals and utilize the equivalent dielectric response parameters of insulators. Modulate the spectral feature vector; The sound wave signal is subjected to a short-time Fourier transform to obtain its frequency domain amplitude function. The energy of multiple frequency bands of the frequency domain amplitude function is calculated within multiple preset frequency bands. All frequency band energies are constructed into a sound wave feature vector, and the compactness parameter of the insulator's equivalent structure is utilized. Modulate the sound wave feature vector; S3. Perform Z-score normalization on the modulated spectral feature vector and acoustic feature vector respectively to obtain normalized spectral feature vector and normalized acoustic feature vector, and then perform attention-weighted fusion on the normalized spectral feature vector and normalized acoustic feature vector to obtain fused feature vector; S4. Input the fused feature vector into the pre-trained insulator defect detection model to obtain the prediction probability vector that reflects the probability of the insulator defect category, and obtain the prediction probability of each preset defect category of the insulator.

2. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 1, characterized in that, The expression for calculating the spectral eigenvector is: ； ； in, For the first Spectral eigenvectors under secondary excitation conditions; , , and The first Under the excitation condition, the first, second, and third rays in the spectral signal Article and Section The radiation intensity of the characteristic spectral lines of the bar; The acquisition response coefficients of the spectral signal; For the first The first excitation produces the second The particle number density of the element with the characteristic spectral lines; and The first The spectral transition probabilities and upper level energies corresponding to each characteristic spectral line; Boltzmann's constant; For the first The transient temperature of the plasma generated under secondary excitation.

3. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 2, characterized in that, The calculation expression for modulating the spectral eigenvector is as follows: ； ； in, For the first The modulated spectral feature vector under secondary excitation conditions; This is the reference total energy value obtained under standard insulator sample conditions.

4. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 3, characterized in that, The expression for calculating the eigenvector of a sound wave is: ； ； ； ； in, For the first Acoustic wave characteristic vector under secondary excitation conditions; In the first The frequency domain amplitude function of the acoustic signal under secondary excitation conditions; , , and They are the preset 1st, 2nd, and 3rd. The and the first Each frequency band boundary frequency; In the first The time-domain signal in the acoustic wave signal under secondary excitation conditions; For time window functions; Let be the complex exponential kernel function of the Fourier transform. , and These represent the imaginary unit, frequency, and time, respectively. For the first Peak sound pressure level of the sound wave signal under secondary excitation conditions; The attenuation coefficient of sound waves in the propagation medium; for The dominant frequency of the acoustic signal under secondary excitation conditions; The acoustic energy coupling coefficient; The single-pulse energy of the pulsed laser source; The distance between the acoustic sensor used to acquire sound wave signals and the laser excitation point in the pulsed laser source; This represents the pulse width of the pulsed laser source.

5. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 4, characterized in that, The expression for calculating the modulation of the acoustic wave eigenvector is as follows: ； ； in, For the first The characteristic vector of the modulated acoustic wave under secondary excitation conditions; and These are the preset lower and upper limits of the high-frequency band in the sound wave signal; It is the highest frequency in the sound wave signal.

6. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 5, characterized in that, The Z-score normalization expression for the modulated spectral eigenvector and acoustic eigenvector is as follows: ； in, and These are the standardized spectral eigenvectors and the standardized acoustic eigenvectors, respectively. and These are the acoustic eigenvectors and acoustic eigenvectors under all excitation conditions, respectively. and They are respectively and Standard deviation; and They are respectively and The mean.

7. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 6, characterized in that, The expression for the fused feature vectors is: ； in, To fuse feature vectors; and They are respectively and Attention weights.

8. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 6, characterized in that, The expressions for calculating the attention weights of the normalized spectral eigenvector and the normalized acoustic eigenvector are as follows: ， ； in, and They are respectively and The preset nonlinear mapping function; This refers to the surface energy deposition efficiency of the insulator.

9. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 1, characterized in that, The spectral acoustic heterogeneous signal insulator detection method also includes: S5. Based on the predicted probability of each preset defect category of the insulator, the physical consistency constraint verification is performed on each preset defect category in descending order of predicted probability. If the verified preset defect category satisfies all physical consistency constraint rules at the same time, the insulator is determined to have the currently verified preset defect category; otherwise, the insulator is determined not to have the currently verified preset defect category.

10. The method for detecting heterogeneous spectral acoustic signals of insulators based on homogeneous lasers according to claim 8, characterized in that, Physical consistency constraints must include at least the following: Dielectric response consistency constraint: equivalent dielectric response parameters The value is located within the preset range corresponding to the preset defect category being verified; Structural compactness consistency constraints: equivalent structural compactness parameters The variation trend under multiple excitation conditions is consistent with the variation trend of the pre-set equivalent structure compactness parameter corresponding to the pre-set defect category verified. Energy deposition consistency constraint: The ratio of spectral energy to acoustic energy of the pulsed laser source is matched with the preset surface energy deposition efficiency corresponding to the preset defect category.

Images