Diagnostic method, system, device and medium for offshore photovoltaic support structural status

Abstract

The present invention relates to the technical field of civil engineering structure health monitoring, and in particular to a method, system, device, and medium for diagnosing the structural status of an offshore photovoltaic support. The method for diagnosing the structural status of an offshore photovoltaic support of the present invention comprises: generating acceleration response data of healthy and damaged working conditions based on a finite element model, performing time-frequency transformation on the acceleration signal data to construct time-frequency matrix data; calculating the autopower spectrum and cross-power spectrum of each acceleration response data, and calculating the coherence matrix data based on the time-frequency transformation through a two-dimensional convolution function; locating the damage by comparing the maximum position of the time-frequency coordinate change in the coherence matrix data before and after the damage, calculating the norm of the relative change of the eigenvalues ​​under healthy working conditions and damaged working conditions to quantify the degree of damage; and using a discrete wavelet neural operator to establish a mapping model between the acceleration response data and the damage location and quantification. This application significantly improves the accuracy and efficiency of health diagnosis of offshore photovoltaic supports.

Description

Technical Field

[0001] The present invention relates to the technical field of civil engineering structure health monitoring technology, and in particular to a method, system, device and computer-readable storage medium for diagnosing the structural status of an offshore photovoltaic support. Background Art

[0002] With the widespread adoption of offshore photovoltaic power generation systems, maintaining the safety and stability of photovoltaic mounts has become a critical issue. However, civil engineering structures such as offshore photovoltaic mounts are becoming increasingly complex and large-scale. During their service life, they are affected by multiple factors such as external loads, material aging, and environmental corrosion. Traditional detection methods (such as manual inspections and single-sensor monitoring) are inefficient, rely on experience, and have poor noise immunity, making it difficult to detect structural damage in complex environments in real time.

[0003] In existing technologies, although methods based on wavelet transform or modal parameters can partially identify damage, they have the following problems:

[0004] 1. Unable to effectively process the time-frequency dynamic characteristics of non-stationary vibration signals;

[0005] 2. Lack of sensitivity to minor injuries;

[0006] 3. Lack of quantitative analysis capability of damage extent.

[0007] Comparative documents (such as CN104458173B) propose damage indicators based on the rate of change of wavelet coefficients. However, these rely on single frequency domain features and lack deep learning methods, resulting in limited generalization capabilities. Therefore, an efficient diagnostic method that integrates signal processing and artificial intelligence is urgently needed. Summary of the Invention

[0008] Embodiments of the present invention provide a method, system, device and storage medium for diagnosing the structural status of an offshore photovoltaic support to address the problems of insufficient extraction of non-stationary signal features, low recognition rate of minor damage and lack of quantitative damage analysis in traditional methods.

[0009] In order to achieve the above object, on the one hand, a method for diagnosing the structural status of an offshore photovoltaic support is provided, comprising the following steps:

[0010] Generate acceleration response data of healthy and damaged working conditions based on the finite element model, and perform time-frequency transformation on the acceleration response data to construct time-frequency matrix data;

[0011] Calculating the auto-power spectrum and cross-power spectrum data of each acceleration response data according to the time-frequency matrix data;

[0012] Calculating coherence matrix data of each acceleration response data by a two-dimensional convolution function according to the auto-power spectrum and cross-power spectrum data;

[0013] Calculating the coherence matrix data of each acceleration response data under healthy working conditions and calculating the coherence matrix data of each acceleration response data under damaged working conditions. By comparing the coherence matrix data before and after damage, the location of the maximum change in time-frequency coordinates is determined as the damage location. The norm of the relative change of the coherence matrix data under the damaged working condition is calculated, and the norm is used to quantify the degree of damage.

[0014] According to the time-frequency coordinate change values ​​and the norm of the coherence matrix data before and after the injury, a predetermined discrete wavelet neural operator is used to establish mapping models between the acceleration response data and the damage location and damage quantification, and by inputting the measured acceleration response data into the mapping model, the damage location and damage quantification data are obtained.

[0015] In some embodiments, after the steps of establishing mapping models between acceleration response data and damage location and damage quantification using a discrete wavelet neural operator based on the time-frequency coordinate change values ​​of the coherence matrix data before and after the injury and the norm, and inputting the acceleration response data through the mapping model to obtain damage location and damage quantification data, the method further includes:

[0016] The model parameters are fine-tuned based on experimental data to verify the damage localization and quantitative results.

[0017] In some embodiments, the steps of generating acceleration response data for healthy and damaged working conditions based on a finite element model and performing time-frequency transformation on the acceleration response data to construct time-frequency matrix data include:

[0018] The damage condition is simulated by randomly reducing the node stiffness or removing elements;

[0019] The real environmental load is simulated by applying random white noise to the finite element model;

[0020] The damage conditions include local bolt loosening, beam end cracks, column bottom corrosion, support loosening and / or support fracture.

[0021] In some embodiments, the time-frequency transform includes an S-transform, a wavelet transform, and a short-time Fourier transform;

[0022] The expression of the S transform is:

[0023]

[0024] Where, is the acceleration response signal to be analyzed, The window width in the time domain is Gaussian window function, is a complex weight term, t For time, is the frequency, The time center.

[0025] In some embodiments, the step of calculating the coherence matrix data of each acceleration response data by a two-dimensional convolution function based on the autopower spectrum and the cross-power spectrum data includes:

[0026]

[0027] in, is the temporal operator of the two-dimensional convolution, is the scale operator of two-dimensional convolution, is the convolution operator, Acceleration response data The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The time-frequency matrix of the acceleration response data The cross power spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data The autopower spectrum of the time-frequency matrix Translation operations in the time and frequency domains, Acceleration response data j The autopower spectrum of the time-frequency matrix Translation operations in the time and frequency domains,

[0028] Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix Translation operations in the time and frequency domains, Acceleration response data j The time-frequency matrix of the acceleration response data The cross power spectrum of the time-frequency matrix Translation operations in the time and frequency domains, For a two-dimensional convolution mask, the definition formula is as follows:

[0029]

[0030] Where, n is an integer, and 2≤ n ≤5;

[0031] The coherence matrix data is as follows:

[0032] .

[0033] Where, is the coherence matrix data, Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix The square of the absolute value of the mean unbiased estimator of Acceleration response data The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The autopower spectrum of the time-frequency matrix The average unbiased estimate of .

[0034] In some embodiments, the norm includes a Frobenius norm, a spectral norm, and a nuclear norm.

[0035] In some embodiments, the steps of calculating coherence matrix data of each acceleration response data under healthy working conditions and calculating coherence matrix data of each acceleration response data under damaged working conditions, comparing the coherence matrix data before and after damage, determining the location of the damage as the maximum change in time-frequency coordinates, and calculating the norm of the relative change of the coherence matrix data under the damaged working conditions, wherein the norm is used in the step of quantifying the degree of damage, include:

[0036] Calculate the average value of the coherence matrix data of the acceleration response data under healthy working conditions, and convert the obtained average coherence matrix data into As the positioning benchmark, subscript 0 indicates healthy working conditions;

[0037] The eigenvalue decomposition of the average coherence matrix data is performed using the following formula:

[0038]

[0039] Where, is the coherence matrix data, for The matrix composed of the eigenvectors corresponding to each eigenvalue in , for The transposed matrix of for The eigenvalue matrix obtained after eigenvalue decomposition;

[0040] The eigenvalues ​​obtained by decomposition as a quantitative benchmark;

[0041] By calculating the coherence matrix data under each damage condition , where the subscript d In the damage condition, the damage location is performed using the following formula:

[0042]

[0043] Where, represents the difference matrix of the coherence matrix before and after damage, For positioning reference, is the coherence matrix data under damage conditions;

[0044] The coherence matrix data under each damage condition is subjected to eigenvalue decomposition to obtain the corresponding eigenvalues. The relative change in the eigenvalues ​​under healthy and damaged conditions is used to represent the overall damage degree of the structure, as shown in the following formula:

[0045]

[0046] in, is the Frobenius norm, that is, the square root of the sum of the squares of all time-frequency coordinates in the matrix. DI represents the overall damage degree of the structure, and 0≤DI≤1. The closer DI is to 1, the greater the damage degree. and are the eigenvalue matrices under quantitative benchmark and damage conditions, respectively.

[0047] In some embodiments, the step of establishing a mapping model between acceleration response and damage area using a discrete wavelet neural operator includes:

[0048] Will Input to the discrete wavelet neural operator, is the acceleration response data, t For time;

[0049] Use local transformations Increase the dimension of the input to get , Built by convolutional neural networks;

[0050] Will Passed to a series of wavelet kernel integration layers for multi-layer wavelet decomposition: Parameterize and obtain horizontal, vertical and diagonal coefficients at different levels to complete discrete wavelet decomposition

[0051] , for The kernel function, is the convolution operator;

[0052] Reconstructing the convolution input , is the convolution operator;

[0053] Use a convolutional neural network W to perform a linear transformation: , is the convolution operator;

[0054] Will and Add and activate, and pass local transformation Reduce the dimension to complete Corresponding to The mapping between DI and damage location mapping model and damage quantification mapping model are obtained respectively.

[0055] In another aspect, a diagnostic system for the structural status of an offshore photovoltaic support is provided, comprising:

[0056] The time-frequency matrix module is used to generate acceleration response data of healthy and damaged working conditions based on the finite element model, and perform time-frequency transformation on the acceleration response data to construct time-frequency matrix data;

[0057] A power spectrum module, used to calculate the auto-power spectrum and cross-power spectrum data of each acceleration response data according to the time-frequency matrix data;

[0058] A coherence data module, configured to calculate coherence matrix data of each acceleration response data by a two-dimensional convolution function based on the auto-power spectrum and cross-power spectrum data;

[0059] A damage location quantification module is used to calculate the coherence matrix data of each acceleration response data under healthy working conditions and under damaged working conditions. By comparing the coherence matrix data before and after damage, the location of the maximum change in time-frequency coordinates is determined as the damage location. The norm of the relative change of the coherence matrix data under the damaged working condition is calculated, and the norm is used to quantify the degree of damage.

[0060] The discrete wavelet module is used to establish mapping models between acceleration response data and damage location and damage quantification based on the time-frequency coordinate change values ​​and the norm of the coherence matrix data before and after damage, using discrete wavelet neural operators. The acceleration response data is input through the mapping model to obtain damage location and damage quantification data.

[0061] On the other hand, a device for diagnosing the structural status of an offshore photovoltaic support is provided, comprising a memory and a processor, wherein the memory stores at least one program, and the at least one program is executed by the processor to implement the above-mentioned method for diagnosing the structural status of an offshore photovoltaic support.

[0062] On the other hand, a computer-readable storage medium is provided, in which at least one program is stored. The at least one program is executed by a processor to implement the above-mentioned method for diagnosing the structural status of the offshore photovoltaic support.

[0063] The above technical solution has the following technical effects:

[0064] This application uses the adaptive time-frequency resolution of time-frequency transformation to accurately capture the local mutation characteristics of non-stationary vibration signals, and can detect minor stiffness damage (such as loose bolts and crack initiation). Compared with traditional wavelet methods, it significantly improves sensitivity; based on the norm change of the eigenvalues ​​of the coherence matrix data, it quantifies the degree of damage, solving the problem that traditional methods can only make qualitative judgments; through discrete wavelet neural operators, end-to-end learning of the mapping relationship between time-frequency characteristics and damage is achieved, without the need for manual design of indicators, and is suitable for multi-type damage diagnosis in complex environments (such as waves and wind loads), reducing maintenance costs to a certain extent; this application significantly improves the accuracy and efficiency of health diagnosis of offshore photovoltaic brackets. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 Schematic diagram of a flow chart of a method for diagnosing the structural status of an offshore photovoltaic support according to an embodiment of the present invention;

[0066] Figure 2 A schematic diagram of a finite element model according to an embodiment of the present invention;

[0067] Figure 3 This is a flow chart of a method for establishing a mapping model between acceleration response and damage area using a discrete wavelet neural operator according to another embodiment of the present invention;

[0068] Figure 4 This is a structural schematic diagram of a diagnostic system for the structural status of an offshore photovoltaic support according to another embodiment of the present invention. DETAILED DESCRIPTION

[0069] To further illustrate various embodiments, the present invention is provided with accompanying drawings. These drawings form part of the present disclosure and are primarily used to illustrate the embodiments and, in conjunction with the relevant description in the specification, to explain the operating principles of the embodiments. By referring to these drawings, one of ordinary skill in the art will understand other possible embodiments and the advantages of the present invention. The components in the figures are not drawn to scale, and similar reference numerals are generally used to represent similar components.

[0070] Structural diagnosis of offshore photovoltaic mounting systems is crucial, primarily due to their unique offshore application scenarios and harsh environmental challenges. The offshore environment is characterized by high humidity, severe salt spray corrosion, constant wave impact, strong wind loads, and tidal fluctuations. These factors can accelerate the accumulation of damage such as mounting material fatigue, bolt loosening, weld cracking, and supporting structure corrosion. Failure to promptly detect these issues can lead to localized failure or even complete collapse, resulting in not only reduced power generation efficiency and economic losses, but also environmental pollution and potential safety incidents. Therefore, structural health monitoring is a core requirement for ensuring the long-term stable operation of offshore photovoltaic systems.

[0071] The randomness of waves and wind loads causes vibration signals to exhibit non-stationary characteristics, making it difficult for traditional time-domain or frequency-domain methods to capture dynamic changes. At the same time, the interference from marine environmental noise (such as wave impact and equipment vibration) is significant. Effectively suppressing high-frequency noise and improving the signal-to-noise ratio are both challenging issues in diagnosing offshore photovoltaic support structures. By integrating time-frequency analysis with deep learning, this application not only provides a high-precision, interference-resistant damage diagnosis solution for offshore photovoltaic supports, but its technical framework also has the potential for cross-scenario migration. By simply optimizing parameters based on the load characteristics, damage type, and noise environment of a specific scenario, it can be widely used in intelligent health monitoring of onshore infrastructure, aerospace, and industrial equipment.

[0072] The technical solution of this application, including time-frequency analysis, coherence eigenvalue decomposition, and discrete wavelet neural operator (DWNO), is universal and can be extended to other scenarios, such as:

[0073] Wind turbine tower: The tower is affected by wind loads and mechanical vibrations. This application can detect bolt preload loss or tower tilt.

[0074] Aircraft wing / fuselage: The aerodynamic loads and vibration signals during flight are complex. The S-transform can extract the time-frequency characteristics of fatigue cracks, and DWNO is suitable for real-time analysis of high-speed data streams.

[0075] The present invention will now be further described with reference to the accompanying drawings and specific embodiments.

[0076] Reference Figure 1 The present invention provides a method for diagnosing the structural status of an offshore photovoltaic support, comprising the following steps:

[0077] Step S101: Generate acceleration response data of healthy and damaged working conditions based on the finite element model, and perform time-frequency transformation on the acceleration response data to construct time-frequency matrix data;

[0078] Step S102, calculating the auto-power spectrum and cross-power spectrum data of each acceleration response data according to the time-frequency matrix data;

[0079] Step S103, calculating coherence matrix data of each acceleration response data by a two-dimensional convolution function according to the auto-power spectrum and cross-power spectrum data;

[0080] Step S104: Calculate the coherence matrix data of each acceleration response data under the healthy working condition, calculate the coherence matrix data of each acceleration response data under the damaged working condition, compare the coherence matrix data before and after the damage, determine the location of the maximum change in time-frequency coordinates as the damage location, and calculate the norm of the relative change of the coherence matrix data under the damaged working condition. The norm is used to quantify the degree of damage.

[0081] In step S105, based on the time-frequency coordinate change value and the norm of the coherence matrix data before and after the injury, a discrete wavelet neural operator is used to establish mapping models between the acceleration response data and the damage location and damage quantification, and the damage location and damage quantification data are obtained by inputting the measured acceleration response data into the mapping model.

[0082] In the above embodiment, the time-frequency transform includes the S transform, wavelet transform, and short-time Fourier transform. The S transform, also known as the Stockwell transform, is a time-frequency analysis tool that combines the advantages of the short-time Fourier transform (STFT) and wavelet transform. It can adaptively adjust the time-frequency resolution while preserving phase information and is widely used in nonstationary signal analysis.

[0083] In some embodiments, based on the acceleration response data, an S-transform is performed on the acceleration signal according to the following formula (1) to obtain an acceleration response time-frequency matrix database:

[0084] (1)

[0085] Where, is the acceleration response time-frequency matrix data, is the acceleration response signal to be analyzed, The window width in the time domain is Gaussian window function, is a complex weight term, t For time, is the frequency, The time center.

[0086] In the above embodiment, the finite element model (FEM) is a numerical analysis tool used to discretize complex physical systems such as mechanical structures, fluids, electromagnetic fields, etc. into multiple simple small units such as triangles, quadrilaterals, hexahedrons, etc., and finally combine them to obtain an approximate solution for the entire system by solving the mathematical equations of each unit. Its core idea is "divide and conquer", which converts continuous problems into discrete problems. It is suitable for analyzing complex geometric shapes and nonlinear behaviors. In the present application, the continuous structure of the offshore photovoltaic support is divided into a finite number of simple units such as beam units, shell units, solid units, etc., and the units are connected by nodes, thereby converting the continuous problem into a discrete set of equations. The finite element model is suitable for the structural stability assessment of the offshore photovoltaic support of the present application. When applied to the structural mechanics analysis of the offshore photovoltaic support, it can simulate the dynamic response of the support under healthy working conditions and different damage working conditions.

[0087] refer to Figure 2 As an example of a finite element model, a MATLAB-driven 4-story steel frame finite element model provided by the ASCE benchmark model Phase Ⅰ is selected. Of course, other finite element models such as ABAQUS and OpenSees can also be selected. The finite element model generates a large amount of acceleration response data under different working conditions, including healthy working conditions and multiple customized damage working conditions. In some embodiments, the damage working conditions include local bolt loosening, beam end cracks, column base corrosion, support loosening, and support fracture. Each damage working condition is simulated and implemented in the following ways:

[0088] Local bolt loosening is achieved by randomly reducing the local stiffness matrix of specific nodes. This is used to simulate the possible loosening of bolts caused by the long-term effects of wind and waves on offshore photovoltaic brackets in marine environments.

[0089] Beam end cracks are achieved by randomly reducing the local element stiffness at both ends of the beam, which is used to simulate the possibility that long-term wave impact or wind loads on offshore photovoltaic supports may cause cracks at the beam ends.

[0090] The column bottom corrosion is achieved by randomly reducing the column bottom stiffness, which is used to simulate the column bottom corrosion caused by seawater erosion of offshore photovoltaic brackets;

[0091] Support loosening is achieved by randomly reducing the stiffness of the support rod connection nodes, and support fracture is achieved by randomly removing the support rod elements completely.

[0092] By applying random white noise to the finite element model to simulate the real environmental load, acceleration sensors are arranged on each floor to collect structural x 、 yBidirectional vibration response signal, sampling frequency ≥ 500 Hz, single sampling time 30 seconds, generates an acceleration response database including healthy conditions and multiple custom damage conditions.

[0093] In some embodiments, after the step of using a discrete wavelet neural operator to establish a mapping model between acceleration response and damage area, the method further includes: fine-tuning model parameters based on experimental data to verify damage location and damage quantification results.

[0094] For example, the acceleration response dataset of a laboratory four-story steel frame model provided by ASCE benchmark model Phase Ⅱ is selected to simulate the real structural acceleration response of the offshore photovoltaic bracket. Only the acceleration response under healthy conditions is used for fine-tuning, and the acceleration response under damaged conditions is used to verify the damage location DWNO and damage quantification DWNO.

[0095] After training the DWNO model, transfer learning was performed using laboratory measured data (ASCE Phase II). The convolution kernel parameters were fine-tuned to optimize the model's adaptability to real-world noise. Fine-tuning the model parameters based on experimental data improved the generalization of the mapping model, increasing the accuracy of damage-related conditions and verifying a good linear fit between the damage results and the actual damage extent.

[0096] In addition to choosing the laboratory four-story steel frame model provided by ASCE benchmark model Phase Ⅱ, you can also choose to build a refined finite element model of the offshore photovoltaic bracket based on models such as ABAQUS and OpenSees, which will not be discussed here.

[0097] In some embodiments, based on generating acceleration response data of healthy and damaged working conditions based on a finite element model, performing time-frequency transformation on the acceleration response data to construct time-frequency matrix data, the autopower spectrum and cross-power spectrum data of each acceleration response data are calculated according to the time-frequency matrix data; wherein the autopower spectrum is calculated as shown in equations (2) to (3):

[0098] (2)

[0099] (3)

[0100] Where, Acceleration response data The autopower spectrum of the time-frequency matrix is Acceleration response data j The autopower spectrum of the time-frequency matrix is Acceleration response data The acceleration response time-frequency matrix, Acceleration response data j The acceleration response time-frequency matrix of , * represents the complex conjugate, that is for The complex conjugate matrix of for The complex conjugate matrix of is the frequency, The time center.

[0101] Calculate the cross power spectrum as shown in equations (4) to (5):

[0102] (4)

[0103] (5)

[0104] Where, Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix is Acceleration response data j The time-frequency matrix of the acceleration response data The cross power spectrum of the time-frequency matrix is Acceleration response data The acceleration response time-frequency matrix, Acceleration response data j The acceleration response time-frequency matrix of , * represents the complex conjugate, that is for The complex conjugate matrix of for The complex conjugate matrix of is the frequency, The time center.

[0105] Power spectral density is the density function of the signal power distribution with frequency. It is used to describe the power contained in each frequency component of the signal in the frequency domain. It is an important tool for measuring signal strength in the frequency domain.

[0106] The coherence based on S transform needs to be averaged to obtain a suitable estimate. Lack of averaging will lead to biased estimates. In order to obtain an average unbiased estimate of the coherence based on S transform, the process of calculating the coherence based on S transform using a two-dimensional convolution function is shown in formulas (6)-(9):

[0107] (6)

[0108] (7)

[0109] (8)

[0110] (9)

[0111] in, is the temporal operator of the two-dimensional convolution, is the scale operator of two-dimensional convolution, is the convolution operator, Acceleration response data The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The time-frequency matrix of the acceleration response data The cross power spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data The autopower spectrum of the time-frequency matrix Translation operations in the time and frequency domains, Acceleration response data j The autopower spectrum of the time-frequency matrix Translation operations in the time and frequency domains, Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix Translation operations in the time and frequency domains, Acceleration response data j The time-frequency matrix of the acceleration response data The cross power spectrum of the time-frequency matrix Translation operations in the time and frequency domains, For a two-dimensional convolution mask, the definition formula is as follows:

[0112] (10)

[0113] Where, n is an integer, and 2≤ n ≤5.

[0114] The coherence matrix data calculation process is as follows:

[0115] (11)

[0116] Where, is the coherence matrix data, Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix The square of the absolute value of the mean unbiased estimator of Acceleration response data The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The autopower spectrum of the time-frequency matrix The average unbiased estimate of .

[0117] Coherence is a normalized metric that examines the relationship between two analysis responses. Smoothing the coherence matrix using a two-dimensional convolution mask helps suppress noise interference. Coherence analysis can improve the sensitivity of identifying subtle stiffness damage, such as loose bolts.

[0118] In some embodiments, the norm includes a Frobenius norm, a spectral norm, and a nuclear norm, wherein the Frobenius norm is the square root of the sum of the squares of all values ​​in the matrix, the spectral norm is the maximum value among the singular values ​​of the matrix, and the nuclear norm is the sum of the singular values ​​of the matrix.

[0119] In some embodiments, the steps of calculating coherence matrix data of each acceleration response data under healthy working conditions and calculating coherence matrix data of each acceleration response data under damaged working conditions, comparing the coherence matrix data before and after damage, determining the location of the damage as the maximum change in time-frequency coordinates, and calculating the norm of the relative change of the coherence matrix data under the damaged working conditions, wherein the norm is used in the step of quantifying the degree of damage, include:

[0120] Calculate the average value of the coherence matrix data of the acceleration response data under healthy working conditions, and convert the obtained average coherence matrix data into As the positioning benchmark, subscript 0 indicates healthy working conditions;

[0121] The eigenvalue decomposition of the average coherence matrix data is performed using the following formula:

[0122] (12)

[0123] Where, is the coherence matrix data, for The matrix composed of the eigenvectors corresponding to each eigenvalue in , for The transposed matrix of for The eigenvalue matrix obtained after eigenvalue decomposition.

[0124] Will Substitute into formula (12) and decompose to get the eigenvalue ,Will as a quantitative benchmark;

[0125] By calculating the coherence matrix data under each damage condition , where the subscript d In the damage condition, the damage location is performed using the following formula:

[0126] (13)

[0127] Where, represents the difference matrix of the coherence matrix before and after damage, For positioning reference, is the coherence matrix data under damage condition.

[0128] The coherence matrix data under each damage condition is subjected to eigenvalue decomposition to obtain the corresponding eigenvalues. The relative change in the eigenvalues ​​under healthy and damaged conditions is used to represent the overall damage degree of the structure, as shown in the following formula:

[0129] (14)

[0130] Where, is the Frobenius norm, which is the square root of the sum of the squares of all the time-frequency coordinates in the matrix. The DI value is a The closer the number is to 1, the greater the degree of damage; As a quantitative benchmark, is the eigenvalue matrix under damage condition.

[0131] In the above embodiment, by calculating the average coherence matrix under healthy conditions And perform eigenvalue decomposition to obtain the benchmark eigenvalue ; Under damage conditions, calculate the damage coherence matrix And perform eigenvalue decomposition to obtain . By comparing the Frobenius norm of the eigenvalue matrix before and after damage, the overall damage degree of the structure is quantified as a normalized index (DI value) ranging from 0 to 1. Eigenvalue decomposition extracts the principal component information of the coherence matrix, and the Frobenius norm comprehensively characterizes the degree of damage through the global square sum change of the matrix time-frequency coordinates, avoiding the limitations of a single frequency domain feature. Directly quantifying the severity of damage (such as the torque loss rate of bolt loosening) through the DI value solves the problem that traditional methods can only qualitatively judge the existence of damage; principal component analysis based on eigenvalues ​​suppresses noise interference (such as random vibration of waves); the Frobenius norm comprehensively reflects the overall change of the matrix, greatly improving the sensitivity to distributed damage (such as column bottom corrosion); the algorithm has high computational efficiency and is suitable for embedded devices to meet the real-time monitoring needs of offshore photovoltaic brackets.

[0132] refer to Figure 3 In some embodiments, the step of using a discrete wavelet neural operator to establish a mapping model between acceleration response and damage area includes:

[0133] Step S1051: Input to DWNO, is the acceleration response data, t For time;

[0134] Step S1052: Use local transformation Increase the dimension of the input to get , It is a dimensional convolutional neural network. The convolution kernel size of the convolutional neural network can be 1 1, 2 2, 3 3, etc., the dimensions n, n 2;

[0135] Step S1053: Passed to a series of wavelet kernel integration layers for multi-layer wavelet decomposition, that is, the wavelet kernel Parameterize and obtain horizontal, vertical and diagonal coefficients at different levels to complete discrete wavelet decomposition , for The kernel function, is the convolution operator;

[0136] Step S1054, reconstruct the convolution input , is the convolution operator;

[0137] Step S1055, use the convolutional neural network W to perform linear transformation: , is the convolution operator;

[0138] Step S1056: and Add and activate, and finally transform locally Reduce the dimension to the same as the input acceleration response data, It is a dimensionality reduction convolutional neural network, where the convolution kernel size of the convolutional neural network can be 1 1, 2 2, 3 3, etc., further completed Corresponding to Mapping between DI and DI, through local transformation Reduce the dimension to complete Corresponding to The mapping between DI and damage location mapping model and damage quantification mapping model are obtained.

[0139] The Discrete Wavelet Neural Operator (DWNO) is a novel operator learning algorithm in deep learning that learns mappings between infinite-dimensional function spaces. This algorithm directly learns nonlinear mappings between two function spaces, resulting in stronger generalization capabilities than traditional deep learning. DWNO leverages the superiority of wavelets in localizing functions in time and frequency, enabling accurate pattern tracking and efficient learning of function mappings in the spatial domain. Because wavelets are localized in time, space, and frequency, DWNO offers high spatial and frequency resolution. DWNO first decomposes the input function space into high- and low-frequency components using the Discrete Wavelet Transform (DWT). In the DWT, the input function space is decomposed using wavelets at different scales. Typically, lower-level components of the function space contain primarily noise, potentially obscuring critical information at these scales. Higher-level wavelet coefficients, however, contain more valuable feature information. Since the goal is to extract the most relevant features from the function space, only higher-level wavelet coefficients from the discrete wavelet transform are selected to construct the informative subset. DWNO leverages the advantages of wavelet transform to effectively learn solution operators for highly nonlinear partial differential equations (PDEs) with irregular domains and boundary conditions.

[0140] The above method is used to establish a mapping model between acceleration response and damage area. Wavelet decomposition can accurately capture local mutations of non-stationary signals such as crack impact at the beam end and frequency domain characteristics such as structural resonance offset, greatly improving the accuracy of damage location. Through high-frequency noise suppression, wavelet threshold denoising and parameterized kernel adaptive learning, the model still has good recognition accuracy for untrained damage types such as support fracture. Since the original signal is directly mapped to the damage result, artificial feature design is avoided, and the training efficiency is greatly improved compared with the traditional method. By integrating time-frequency analysis and deep learning, the diagnosis error of multiple types of damage such as bolt loosening and column base corrosion under interference such as waves and salt spray is small, which is significantly better than the single modal method.

[0141] On the other hand, the present invention also provides a diagnostic system for the structural status of an offshore photovoltaic support, comprising:

[0142] The time-frequency matrix module is used to generate acceleration response data of healthy and damaged working conditions based on the finite element model, and perform S-transformation on the acceleration response data to construct the time-frequency matrix data;

[0143] A power spectrum module, used to calculate the auto-power spectrum and cross-power spectrum data of each acceleration response data according to the time-frequency matrix data;

[0144] A coherence data module, configured to calculate coherence matrix data of each acceleration response data by a two-dimensional convolution function based on the auto-power spectrum and cross-power spectrum data;

[0145] A damage location quantification module is used to calculate the coherence matrix data of each acceleration response data under healthy working conditions and under damaged working conditions. By comparing the coherence matrix data before and after damage, the location of the maximum change in time-frequency coordinates is determined as the damage location. The norm of the relative change of the coherence matrix data under the damaged working condition is calculated, and the norm is used to quantify the degree of damage.

[0146] The discrete wavelet module is used to establish mapping models between acceleration response data and damage location and damage quantification based on the time-frequency coordinate change values ​​and the norm of the coherence matrix data before and after damage, using discrete wavelet neural operators. The acceleration response data is input through the mapping model to obtain damage location and damage quantification data.

[0147] In some embodiments, the above system further includes a fine-tuning verification module for fine-tuning model parameters based on experimental data and verifying damage location and damage quantification results.

[0148] Specific fine-tuning can be achieved by gradually increasing the number of layers from 2 when performing multi-layer wavelet decomposition, choosing different wavelet bases such as db6 (Daubechies 6), sym4 (Symlet 4), and coif5 (Coiflet 5), and choosing different convolution kernel sizes and dimensions when using dimensionality-enhanced convolutional neural networks. and When adding and activating, you can use different activation functions, such as ReLU, LeakyReLU, Sigmoid, or Tanh.

[0149] The present invention also provides a diagnostic device for the structural status of an offshore photovoltaic support, such as Figure 4 As shown, the device includes a processor 401, a memory 402, a bus 403, and a computer program stored in the memory 402 and executable on the processor 401. The processor 401 includes one or more processing cores. The memory 402 is connected to the processor 401 via the bus 403. The memory 402 is used to store program instructions. When the processor executes the computer program, the steps in the above-mentioned method embodiment of the present invention are implemented.

[0150] Furthermore, as an executable solution, the diagnostic device for the offshore photovoltaic support structure status can be a computer unit, which can be a computing device such as a desktop computer, laptop, PDA, or cloud server. The computer unit may include, but is not limited to, a processor and a memory. Those skilled in the art will understand that the above-mentioned computer unit structure is merely an example of a computer unit and does not constitute a limitation on the computer unit. The computer unit may include more or fewer components than those described above, or a combination of certain components, or different components. For example, the computer unit may also include input and output devices, network access devices, buses, etc., which are not limited in the embodiments of the present invention.

[0151] Furthermore, as an executable solution, the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer unit and connects various parts of the entire computer unit using various interfaces and lines.

[0152] The memory can be used to store the computer programs and / or modules. The processor implements the various functions of the computer unit by running or executing the computer programs and / or modules stored in the memory and accessing the data stored in the memory. The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function; the data storage area may store data generated based on the use of the mobile phone. Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0153] In some embodiments, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned method of the embodiment of the present invention are implemented.

[0154] If the modules / units integrated into the computer unit are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the present invention can also implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include any entity or device capable of carrying the computer program code, recording medium, USB flash drive, removable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased based on the requirements of legislation and patent practice within a jurisdiction.

[0155] In some embodiments, the present invention further provides a computer program product, comprising a computer program, which implements the steps of the method described above when executed by a processor.

[0156] Although the present invention has been particularly shown and described in conjunction with preferred embodiments, it will be understood by those skilled in the art that various changes in form and details may be made to the present invention without departing from the spirit and scope of the invention as defined in the appended claims, and all such changes are within the scope of protection of the present invention.

Claims

1. A method for diagnosing the structural status of an offshore photovoltaic support, characterized in that: The following steps are involved: S101, generating acceleration response data of healthy and damaged working conditions based on the finite element model, and performing time-frequency transformation on the acceleration response data to construct time-frequency matrix data; S102, calculating the auto-power spectrum and cross-power spectrum data of each acceleration response data according to the time-frequency matrix data; S103, calculating coherence matrix data of each acceleration response data by a two-dimensional convolution function according to the auto-power spectrum and cross-power spectrum data; S104, calculating coherence matrix data of each acceleration response data under a healthy working condition, calculating coherence matrix data of each acceleration response data under a damaged working condition, comparing the coherence matrix data before and after the damage, determining the location of the damage at the location where the time-frequency coordinates of the coherence matrix data have the largest change, and calculating the norm of the relative change of the coherence matrix data under the damaged working condition, wherein the norm is used to quantify the degree of damage; S105. Based on the time-frequency coordinate change values ​​and the norm of the coherence matrix data before and after the injury, a discrete wavelet neural operator is used to establish mapping models between the acceleration response data and the damage location and damage quantification, respectively. The measured acceleration response data is input into the mapping models to obtain the damage location and damage quantification data. The time-frequency transform in S101 includes S transform, wavelet transform and short-time Fourier transform; The expression of the S transform is: Where, is the time-frequency matrix data, is the acceleration response data, The window width in the time domain is Gaussian window function, is a complex weight term, t For time, f is the frequency, For the time center; The S104 includes: Calculate the average value of the coherence matrix data of the acceleration response data under healthy working conditions, and convert the obtained average coherence matrix data into As the positioning benchmark, subscript 0 indicates healthy working conditions; The eigenvalue decomposition of the average coherence matrix data is performed using the following formula: Where, is the coherence matrix data, for The matrix composed of the eigenvectors corresponding to each eigenvalue in , for The transposed matrix of for The eigenvalue matrix obtained after eigenvalue decomposition; Will Substitute into the formula and decompose to get the eigenvalue ,Will as a quantitative benchmark; By calculating the coherence matrix data under each damage condition , where the subscript d In the damage condition, the damage location is performed using the following formula: Where, represents the difference matrix of the coherence matrix before and after damage, For positioning reference, is the coherence matrix data under damage conditions; The coherence matrix data under each damage condition is subjected to eigenvalue decomposition to obtain the corresponding eigenvalues. The relative change in the eigenvalues ​​under healthy and damaged conditions is used to represent the overall damage degree of the structure, as shown in the following formula: in, is the Frobenius norm, which is the square root of the sum of the squares of all time-frequency coordinates in the matrix. DI represents the overall damage degree of the structure, and 0≤DI≤1. The closer DI is to 1, the greater the damage degree. and are the eigenvalue matrices under quantitative benchmark and damage conditions, respectively.

2. The method for diagnosing the structural status of an offshore photovoltaic support according to claim 1, characterized in that: After the step of establishing a mapping model between acceleration response and damage area using a discrete wavelet neural operator, the method further includes: The model parameters are fine-tuned based on experimental data to verify the damage location and damage quantification results.

3. The method for diagnosing the structural status of an offshore photovoltaic support according to claim 1, characterized in that: The steps of generating acceleration response data of healthy and damaged working conditions based on the finite element model and performing time-frequency transformation on the acceleration response data to construct time-frequency matrix data include: The damage condition is simulated by randomly reducing the node stiffness or removing elements; The real environmental load is simulated by applying random white noise to the finite element model; The damage conditions include local bolt loosening, beam end cracks, column bottom corrosion, support loosening and / or support fracture.

4. The method according to claim 1, wherein The step of calculating the coherence matrix data of each acceleration response data by a two-dimensional convolution function based on the autopower spectrum and the cross-power spectrum data includes: Where, is the temporal operator of the two-dimensional convolution, is the scaling operator of the two-dimensional convolution, is the convolution operator, Acceleration response data The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The time-frequency matrix of the acceleration response data The cross power spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data The autopower spectrum of the time-frequency matrix Translation operations in the time and frequency domains, Acceleration response data j The autopower spectrum of the time-frequency matrix Translation operations in the time and frequency domains, Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix Translation operations in the time and frequency domains, Acceleration response data j The time-frequency matrix of the acceleration response data The cross power spectrum of the time-frequency matrix Translation operations in the time and frequency domains, is a two-dimensional convolution mask, The definition formula is as follows: Where, is the temporal operator of the two-dimensional convolution, is the scaling operator of the two-dimensional convolution, n is an integer, and 2≤ n ≤5; The coherence matrix data is as follows: ； Where, is the coherence matrix data, Acceleration response data The time-frequency matrix of the acceleration response data j The cross power spectrum of the time-frequency matrix The square of the absolute value of the mean unbiased estimator of Acceleration response data The autopower spectrum of the time-frequency matrix The average unbiased estimate of Acceleration response data j The autopower spectrum of the time-frequency matrix The average unbiased estimate of .

5. The method according to claim 1, wherein The norms include Frobenius norm, spectral norm and nuclear norm.

6. The method according to claim 1, characterized in that The method of using a discrete wavelet neural operator to establish mapping models between acceleration response data and damage location and damage quantification based on the time-frequency coordinate change value and the norm of the coherence matrix data before and after the damage includes: Will Input to the predetermined discrete wavelet neural operator, is the acceleration response data, t For time; Use local transformations Increase the dimension of the input to get , Built by convolutional neural networks; Will Passed to a series of wavelet kernel integration layers for multi-layer wavelet decomposition: Parameterize and obtain horizontal, vertical and diagonal coefficients at different levels to complete discrete wavelet decomposition , for The kernel function, is the convolution operator; Reconstructing the convolution input , is the convolution operator; Use a convolutional neural network W to perform a linear transformation: , is the convolution operator; Will and Add and activate, and pass local transformation Reduce the dimension to complete Corresponding to The mapping between DI and damage location mapping model and damage quantification mapping model are obtained respectively.

7. A diagnostic system using the method for diagnosing the structural status of an offshore photovoltaic support according to any one of claims 1 to 6, characterized in that: include: The time-frequency matrix module is used to generate acceleration response data of healthy and damaged working conditions based on the finite element model, and perform time-frequency transformation on the acceleration response data to construct time-frequency matrix data; A power spectrum module, used to calculate the auto-power spectrum and cross-power spectrum data of each acceleration response data according to the time-frequency matrix data; A coherence data module, configured to calculate coherence matrix data of each acceleration response data by a two-dimensional convolution function based on the auto-power spectrum and cross-power spectrum data; A damage location quantification module is used to calculate the coherence matrix data of each acceleration response data under healthy working conditions and under damaged working conditions. By comparing the coherence matrix data before and after damage, the location of the maximum change in time-frequency coordinates is determined as the damage location. The norm of the relative change of the coherence matrix data under the damaged working condition is calculated, and the norm is used to quantify the degree of damage. The discrete wavelet module is used to establish a mapping model between acceleration response data and damage location and damage quantification based on the time-frequency coordinate change value and the norm of the coherence matrix data before and after the damage, and input the acceleration response data through the mapping model to obtain damage location and damage quantification data.

8. A diagnostic device for the structural status of an offshore photovoltaic support, characterized in that: It comprises a memory and a processor, the memory stores at least one program, and the at least one program is executed by the processor to implement the method for diagnosing the structural status of an offshore photovoltaic support as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The storage medium stores at least one program, and the at least one program is executed by a processor to implement the method for diagnosing the structural status of an offshore photovoltaic support as described in any one of claims 1 to 6.

Images