A Method for Extracting Time-Frequency Features and Nonlinear Characterization of Coupled Extreme Loads from Multiple Disasters

By integrating empirical mode decomposition and the construction of a feature database, the nonlinear characterization problem of multi-hazard coupled loads was solved, and the effective separation and quantitative description of typhoon and internal wave loads were achieved, thereby improving the safety assessment capability of marine engineering structures.

CN123115702APending Publication Date: 2026-11-17TIANJIN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611611954.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-19
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies are insufficient to accurately characterize the time-frequency characteristics and nonlinear features of extreme loads coupled with multiple disasters. Traditional methods are inadequate in describing non-stationary and time-varying loads and lack systematic quantitative characterization tools for typhoon and internal wave coupled loads, thus failing to meet the safety assessment needs of marine engineering structures.

Method used

The integrated empirical mode decomposition (EEMD) method is used to adaptively decompose the multi-hazard coupled load signal in time and frequency. By superimposing white noise and integrating averaging to eliminate mode aliasing, the peak factor and asymmetry index in the time domain and the Hilbert marginal spectrum energy features in the frequency domain are extracted. The coupling strength index between the characteristic components of the hazard source is calculated, and a multi-hazard coupled extreme load feature database is constructed.

Benefits of technology

It achieves effective separation of typhoon and internal wave loads, provides quantitative description from multiple perspectives, improves the interpretability and engineering practicality of time-frequency decomposition, and establishes a standardized feature database, providing a reliable basis for structural damage assessment and safety early warning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115702A_ABST
    Figure CN123115702A_ABST
Patent Text Reader

Abstract

This invention discloses a method for extracting time-frequency features and nonlinear characterizing multi-hazard coupled extreme loads, belonging to the field of intelligent analysis technology for extreme environmental loads in marine engineering. The method includes: collecting typhoon, internal wave, and wave time-history data and synthesizing them into a multi-hazard coupled load time-history; after preprocessing, using integrated empirical mode decomposition to eliminate mode aliasing and outputting intrinsic mode functions (EMFs) of each order; extracting the peak factor and asymmetry index in the time domain based on the standardized load signal; extracting the marginal spectral energy features in the frequency domain based on the EMFs of each order through Hilbert transform; and calculating the coupling strength index between different hazard source components; combining the above features into a feature vector, and constructing a multi-hazard coupled extreme load feature database after standardization. This invention achieves a systematic and quantitative characterization of the nonlinear characteristics of extreme loads under multi-hazard coupled scenarios, providing standardized feature inputs for damage assessment and intelligent prediction of marine engineering structures.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of intelligent analysis technology for extreme environmental loads in marine engineering, and in particular to a method for extracting time-frequency features and nonlinear characterizing multi-hazard coupled extreme loads. Background Technology

[0002] In deep-sea environments, floating wind turbines are subjected to the combined effects of marine environmental loads such as typhoons and internal waves. Typhoons, with their high-speed directional wind fields and extreme waves, exert strong aerodynamic and hydrodynamic loads on the floating platform; internal waves, with their low-frequency, long-period, large-amplitude shear flows, generate significant hydrodynamic excitation on the mooring system and platform. The loads generated by these two types of disasters exhibit complex nonlinear superposition in the time domain and broadband coupling in the frequency domain, forming multi-hazard coupled extreme loads with characteristics distinct from single-hazard loads, posing a severe threat to the structural safety of floating wind turbines.

[0003] Accurate characterization of the time-frequency and nonlinear properties of coupled extreme loads from multiple hazards is a crucial prerequisite for fatigue damage assessment, structural response prediction, and safety early warning. Existing extreme load analysis methods suffer from the following core shortcomings: First, traditional power spectral density (PSD) analysis is based on the assumption of a stationary random process, which is insufficient to describe the non-stationary and time-varying characteristics of loads during the passage of typhoons and internal waves, and makes it difficult to capture the transient characteristics of extreme events.

[0004] Second, although the Short Time Fourier Transform (STFT) can provide a joint time-frequency representation, its fixed window width makes it impossible to simultaneously achieve time resolution and frequency resolution; although the wavelet transform has multi-scale characteristics, its basis functions are fixed, making it difficult to adaptively match the non-stationary frequency variation characteristics of typhoon load and internal wave load.

[0005] Third, although Empirical Mode Decomposition (EMD) has the advantage of adaptively processing non-stationary signals, its inherent mode aliasing problem causes the load components of typhoons and internal waves to mix in the decomposition results, reducing the physical interpretability of intrinsic mode functions (IMFs) of each order.

[0006] Fourth, existing analysis methods mostly establish load characteristic description systems for single disaster types, lacking systematic quantitative characterization tools for the nonlinear characteristics of loads when typhoons and internal waves are coupled, and thus cannot provide standardized characteristic inputs for engineering design and safety assessment. Summary of the Invention

[0007] The purpose of this invention is to provide a method for extracting time-frequency features and nonlinear characterizing extreme loads coupled with multiple disasters. This method addresses the shortcomings of existing methods in describing the non-stationary characteristics of coupled loads with multiple disasters, the difficulty in separating physical components due to mode aliasing in EMD, and the lack of a standardized nonlinear characterization system. It achieves a systematic and quantitative characterization of the time-domain impact characteristics, frequency-domain energy distribution, and coupling strength between disasters of extreme loads in typhoon and internal wave coupled scenarios, providing standardized feature inputs for damage assessment and intelligent prediction of marine engineering structures.

[0008] To achieve the above objectives, this invention provides a method for extracting time-frequency features and nonlinear characterizing multi-hazard coupled extreme loads, comprising the following steps: S1. Collect the time history data of various disaster loads on the target structure, calculate the time history of multi-disaster coupled loads, and preprocess the time history of multi-disaster coupled loads to obtain standardized load signals; S2. The standardized load signal is adaptively decomposed in time and frequency using the Ensemble Empirical Mode Decomposition (EEMD) method. Mode aliasing is eliminated by multiple white noise superpositions and integrated averaging, and the Intrinsic Mode Function (IMF) and residual trend terms of each order are output. S3. Extract the peak factor and asymmetry index in the time domain based on the standardized load signal; extract the Hilbert marginal spectrum energy characteristics in the frequency domain based on the intrinsic mode functions of each order; and calculate the coupling strength index between characteristic components of different disaster sources based on the grouping of the intrinsic mode functions of each order. S4. The extracted peak factor, asymmetry index, Hilbert marginal spectrum energy characteristics, and coupling strength index are combined into a feature vector. After standardizing the feature vector, a multi-hazard coupled extreme load feature database is constructed.

[0009] Preferably, S1 specifically includes: S11. Collect typhoon wind field time history data, and calculate the aerodynamic thrust load on the target structure based on the typhoon wind field time history data. ; S12. Collect internal wave flow field time history data, and calculate the hydrodynamic load on the target structure based on the internal wave flow field time history data. ; S13. Collect wave field time history data and calculate the wave load on the target structure based on the wave field time history data. ; S14. Vector superposition of aerodynamic thrust load, hydrodynamic load, and wave load yields the multi-hazard coupled load time history: ; in, Represents a time variable; S15, Time history of multi-hazard coupled loads Trend term removal and mean normalization are performed, and sampling rates from different data sources are aligned to obtain the standardized load signal. .

[0010] Preferably, in step S2, the standardized load signal is adaptively decomposed in time and frequency using an integrated empirical mode decomposition method, specifically including: S21, in the standardized load signal Independent superposition The group mean is zero and the standard deviation is Gaussian white noise ,constitute Group of noisy signals: ; in, i Indicates the sequence number of the white noise group. ; S22, For each group of noisy signals Empirical Mode Decomposition (EMD) was performed to obtain the eigenmode functions of each order. ; S23, will The group decomposition results are used to take the integrated average of the eigenmode functions of each order, and the result is obtained as follows: eigenmode functions and output the residual trend term. And the decomposition result satisfies ,in, Indicates the order number of the intrinsic mode function. , This represents the total order of the intrinsic mode functions.

[0011] Preferably, in step S3, the extraction of the peak factor and asymmetry index in the time domain based on the standardized load signal specifically includes: Calculate the normalized load signal Peak factor : ; in, Indicates the total duration of the load signal. Represents standardized load signal The root mean square value; Calculate the normalized load signal Asymmetric index : ; in, This represents the mathematical expectation operator.

[0012] Preferably, in step S3, extracting the Hilbert marginal spectral energy features in the frequency domain based on the eigenmode functions of each order specifically includes: For each order of eigenmode function Apply Hilbert transform Construct an analytical signal: ; in, The imaginary unit is used to construct analytic signals; Indicates the first Instantaneous phase and instantaneous amplitude of the first-order eigenmode function after Hilbert transform Instantaneous frequency Instantaneous frequency After end-point effect processing and low-pass filtering; Square the instantaneous amplitude Mapping to instantaneous frequency corresponding angular frequency At this point, construct the Hilbert energy spectrum on the time-frequency plane. The Hilbert marginal spectrum is obtained by integrating along the time axis: ; in, Angular frequency, angular frequency The instantaneous frequencies of each eigenmode function in the time domain The set that constitutes; In Hilbert's marginal spectrum Identify the characteristic frequency bands of each disaster and calculate the energy proportion of each frequency band. , , and main frequency .

[0013] Preferably, in the energy proportion of each frequency band, The proportion of energy within the main frequency band of the internal wave. This represents the proportion of energy within the main frequency band of the typhoon. This refers to the proportion of energy in the mid-frequency band between the main frequency band of internal waves and the main frequency band of typhoons.

[0014] Preferably, in step S3, the calculation of the coupling strength index between different disaster source characteristic components based on the grouping of the intrinsic mode functions of each order specifically includes: Eigenmode functions of each order Based on the dominant frequency range, it is classified into the first disaster dominant component set and the second disaster dominant component set; Calculate the Pearson correlation coefficient between the dominant component sets of the first hazard and the dominant component sets of the second hazard. and cross-power spectral density ; Based on the correlation coefficient and cross-power spectral density Determine the type of coupled operating condition when And cross power spectral density The maximum value within its domain exceeds its average value. When the value is doubled, it is determined to be a strongly coupled operating condition.

[0015] Preferably, S4 specifically includes: S41, Peak Factor Asymmetry index Energy proportions of each frequency band in Hilbert marginal spectrum energy characteristics , , and main frequency Combined into a six-dimensional feature vector: ; S42. Standardize the features of each dimension in the six-dimensional feature vector using the maximum-minimum normalization method: ; in, and These are the minimum and maximum values ​​of the current feature in the training scenario set, respectively; S43. Associate and store the standardized feature vectors with the working condition type labels to construct a multi-hazard coupled extreme load feature database.

[0016] Preferably, the target structure is a floating wind turbine, and the various disaster loads include typhoon loads and internal wave loads.

[0017] Compared with existing technologies, the present invention employs the above-mentioned method for extracting time-frequency features and nonlinear characterizing multi-hazard coupled extreme loads, and has the following technical advantages: (1) This invention uses integrated empirical mode decomposition (EEMD) to replace traditional EMD. Through the mechanism of multiple white noise superposition and integrated averaging, the mode mixing phenomenon is fundamentally eliminated, so that the typhoon dominant component and the internal wave dominant component are effectively separated in the intrinsic mode functions of each order. The intrinsic mode functions of each order have a clear physical correspondence, which significantly improves the interpretability and engineering applicability of the time-frequency decomposition results.

[0018] (2) The time-frequency dual-dimensional feature extraction system constructed in this invention uses peak factor to quantify load impact intensity, skewness index to characterize load asymmetry, and marginal spectral energy and dominant frequency to describe frequency distribution law. The four types of indicators complement each other to form a multi-angle quantitative description of the nonlinear characteristics of multi-hazard coupled extreme loads, overcoming the limitations of single time-domain or frequency-domain analysis in describing non-stationary loads.

[0019] (3) The coupling strength evaluation method proposed in this invention realizes the quantitative evaluation of the coupling effect between typhoon load and internal wave load in both time and frequency domains through the joint analysis of correlation coefficient and cross power spectral density, providing a reliable basis for the midstream damage assessment layer to implement differentiated safety factor correction under strong coupling conditions.

[0020] (4) The standardized multi-hazard coupled extreme load feature database established by this invention forms a unified data interface with a six-dimensional normalized feature vector and four types of working condition labels. It can directly provide high-quality input for midstream fatigue damage assessment and response extreme value statistics, while providing load prior knowledge for downstream intelligent prediction and safety early warning, realizing the efficient transformation of upstream load characterization results into engineering applications.

[0021] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the method for extracting time-frequency features and nonlinear characterizing multi-hazard coupled extreme loads according to the present invention. Figure 2 This is a schematic diagram of the overall technical architecture of the multi-hazard coupled extreme load time-frequency feature extraction and nonlinear characterization method of the present invention. Detailed Implementation

[0023] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] Example like Figure 1 As shown, a method for extracting time-frequency features and nonlinear characterizing multi-hazard coupled extreme loads includes the following steps: S1. Collect the time history data of various disaster loads (typhoon load and internal wave load) on the target structure (i.e., floating wind turbine), calculate the time history of multi-disaster coupled load, and preprocess the time history of multi-disaster coupled load to obtain standardized load signals; This invention is applicable to the analysis of extreme loads caused by the combined effects of typhoons and internal waves on deep-sea floating wind turbines. In one specific embodiment, the IEA 15MW semi-submersible floating wind turbine is used as the research object, and the combined effects of typhoons and internal waves are taken as a typical multi-hazard coupling scenario. The typhoon duration is preset to be 24 hours, the internal wave period is 8-15 minutes, and the sampling interval is 0.5 seconds.

[0026] S1 specifically includes: S11. Collect typhoon wind field time history data, and calculate the aerodynamic thrust load on the target structure based on the typhoon wind field time history data. ; Typhoon wind field data were simulated using the WRF numerical weather prediction model or collected from field observation stations during the typhoon's passage. Wind direction calendar The system outputs wind speed and direction data at the hub height of the floating wind turbine, with a uniform sampling interval of 0.5 seconds. The aerodynamic thrust load on the floating wind turbine is then calculated using blade element momentum theory (BEM). .

[0027] S12. Collect internal wave flow field time history data, and calculate the hydrodynamic load on the target structure based on the internal wave flow field time history data. ; Internal wave current field data were calculated using the MITgcm ocean circulation model, or the horizontal velocity distribution induced by internal waves was obtained using an acoustic Doppler current profiler (ADCP). and vertical shear rate Calculate the hydrodynamic load of the internal wave flow field on the floating platform and mooring system. In one embodiment, the modified Morison equation is used to account for vertical shear effects in calculating the internal wave load.

[0028] S13. Collect wave field time history data and calculate the wave load on the target structure based on the wave field time history data. , calculated using the Morison equation; S14. Vector superposition of aerodynamic thrust load, hydrodynamic load, and wave load yields the multi-hazard coupled load time history: ; in, Represents a time variable; Typhoons, as one of the most destructive meteorological disasters, exert intense aerodynamic and hydrodynamic loads on floating platforms with their high-speed, directional wind fields and extreme ocean waves. Internal waves, as nonlinear wave phenomena in the deep sea caused by density stratification, generate significant hydrodynamic excitation on mooring systems and platforms through low-frequency, long-period, large-amplitude shear flows. In deep-sea engineering scenarios, the passage of typhoons is often accompanied by an intensification of internal wave activity. The loads generated by these two types of disasters exhibit complex nonlinear superposition in the time domain and broadband coupling in the frequency domain, forming multi-hazard coupled extreme loads with characteristics distinct from single-hazard events. The aforementioned superposition method fully considers the combined effects of the three types of loads: typhoon wind field, wave field, and internal wave flow field.

[0029] S15, Time history of multi-hazard coupled loads Trend term removal and mean normalization are performed, and sampling rates from different data sources are aligned to obtain the standardized load signal. .

[0030] Specifically, this includes: fitting and removing the time history of multi-hazard coupled loads using the least squares method. The low-frequency trend term is removed; the load signal after removing the trend term is normalized to zero mean; data from different models or devices are standardized to the same sampling rate (e.g., 0.5 seconds) using linear interpolation to form a standardized load signal with a unified time axis. .

[0031] S2. The standardized load signal is adaptively decomposed in time and frequency using the Ensemble Empirical Mode Decomposition (EEMD) method. Mode aliasing is eliminated by multiple white noise superpositions and integrated averaging, and the intrinsic mode functions and residual trend terms of each order are output.

[0032] The adaptive time-frequency decomposition of the standardized load signal using the integrated empirical mode decomposition method specifically includes: S21, in the standardized load signal Independent superposition The group mean is zero and the standard deviation is Gaussian white noise ,constitute Group of noisy signals: ; in, i Indicates the sequence number of the white noise group. ; In a preferred embodiment, the number of white noise superpositions Take 100-300 times, white noise standard deviation Take standardized load signal 0.1 to 0.2 times the standard deviation. In this embodiment, Take 200, Take 0.15 ( (Standard deviation of the standardized load signal).

[0033] S22, For each group of noisy signals Empirical Mode Decomposition (EMD) was performed to obtain the eigenmode functions of each order. ; The specific process of performing EMD on each group of noisy signals is as follows: successively filter the extreme points of the signal, construct the upper and lower envelopes through cubic spline interpolation, and iteratively extract the intrinsic mode functions of each order using the envelope mean curve as the screening criterion.

[0034] S23, will The group decomposition results are used to take the integrated average of the eigenmode functions of each order, and the result is obtained as follows: eigenmode functions and output the residual trend term. And the decomposition result satisfies ,in, Indicates the order number of the intrinsic mode function. , This represents the total order of the intrinsic mode functions.

[0035] EEMD effectively eliminates mode aliasing in traditional EMD through integrated averaging, giving each order of intrinsic mode function a clear physical meaning. In one embodiment, eight orders of intrinsic mode functions and one residual trend term are obtained: low-order intrinsic mode functions ( The dominant frequency (0.02–0.1 Hz) corresponds to the typhoon-wave dominant component (including typhoon wind field and short-period wave excitation); the intermediate-order intrinsic mode function ( The dominant internal wave component (dominant frequency range 0.001~0.01Hz) corresponds to the dominant internal wave component; higher-order intrinsic mode functions ( ) and residuals This corresponds to a low-frequency drift trend.

[0036] S3. Extract the peak factor and asymmetry index in the time domain based on the standardized load signal; extract the Hilbert marginal spectrum energy characteristics in the frequency domain based on the intrinsic mode functions of each order; and calculate the coupling strength index between characteristic components of different disaster sources based on the grouping of the intrinsic mode functions of each order. The peak factor and asymmetric index in the time domain are extracted based on the standardized load signal, specifically including: Calculate the normalized load signal Peak factor : ; in, Indicates the total duration of the load signal. Represents standardized load signal The root mean square value; Calculate the normalized load signal Asymmetric index : ; in, This represents the mathematical expectation operator.

[0037] Asymmetry Index Defined as the ratio of the third moment to the second moment raised to the power of 3 / 2 (i.e., skewness), it is used to characterize the degree of asymmetry in load distribution. A positive value indicates that the positive peak load is dominant (a characteristic of typhoons). A value close to zero indicates the positive and negative symmetrical reciprocating load characteristics induced by internal waves. Combining these two factors can distinguish between typhoon-dominated and internal wave-dominated conditions, providing a characteristic basis for the automatic identification of multiple disaster types.

[0038] Hilbert marginal spectrum energy features in the frequency domain are extracted based on eigenmode functions of each order, specifically including: For each order of eigenmode function Apply Hilbert transform Construct an analytical signal: ; in, The imaginary unit is used to construct analytic signals; Indicates the first Instantaneous phase and instantaneous amplitude of the first-order eigenmode function after Hilbert transform Instantaneous frequency .

[0039] In a preferred embodiment, instantaneous frequency The feature extraction process involves endpoint effect processing and low-pass filtering. The endpoint effect processing employs the mirror continuation method to eliminate the impact of endpoint oscillations on feature extraction accuracy. The low-pass filtering uses a 5th-order Butterworth low-pass filter with a cutoff frequency of 80% of the Nyquist frequency to suppress high-frequency noise introduced by endpoint oscillations.

[0040] Constructing the Hilbert energy spectrum on the time-frequency plane The Hilbert marginal spectrum is obtained by integrating along the time axis: ; in, Angular frequency; In Hilbert's marginal spectrum Identify the characteristic frequency bands of each disaster and calculate the energy proportion of each frequency band. , , and main frequency .

[0041] In terms of the energy proportion of each frequency band, The proportion of energy within the main frequency band of the internal wave. This represents the proportion of energy within the main frequency band of the typhoon. This refers to the energy proportion in the mid-frequency band between the main frequency band of internal waves and the main frequency band of typhoons. Main frequency Hilbert marginal spectrum The frequency value corresponding to the maximum amplitude is used to characterize the energy concentration frequency of the coupled load.

[0042] In one embodiment, the typhoon main frequency band is set to 0.02–0.1 Hz, the intermediate frequency transition band is set to 0.01–0.02 Hz, and the internal wave main frequency band is set to 0.001–0.01 Hz; The proportion of energy within the main frequency band of the internal wave. This represents the energy percentage within the mid-frequency transition band. This represents the energy proportion within the typhoon's main frequency band. The boundaries of each frequency band can be adaptively adjusted based on the actual sea wave frequency and typhoon wind speed parameters. Frequency band energy distribution and main frequency combination serve as frequency domain characteristic indicators to distinguish the coupling type and intensity of multiple disasters.

[0043] The coupling strength index between characteristic components of different disaster sources is calculated based on the grouping of eigenmode functions of each order, specifically including: Eigenmode functions of each order Based on the dominant frequency range, it is classified into the first disaster dominant component set and the second disaster dominant component set; in one embodiment, Classified into the dominant components of typhoons It is classified as the set of dominant internal wave components.

[0044] Calculate the Pearson correlation coefficient between the dominant component sets of the first hazard and the dominant component sets of the second hazard. and cross-power spectral density Pearson correlation coefficient The linear coupling degree of loads from different disaster sources in the time domain was quantified, with values ​​ranging from [-1, 1]. The closer the absolute value is to 1, the stronger the linear correlation between the two component sets. Cross-power spectral density. Used to identify frequency-domain coupled resonance frequencies, when the two component sets have a significant coherent energy peak at a certain frequency, it indicates that the two sources of disasters have strong coupled resonance at that frequency.

[0045] Based on the correlation coefficient and cross-power spectral density Determine the type of coupled operating condition when And cross power spectral density There is a distinct resonance peak (i.e., the amplitude at a certain frequency within its defined domain exceeds its average value). times, In one embodiment, the preset multiple is used. When the value is 2-3, it is determined to be a strongly coupled working condition. This determination result is used to mark multi-hazard strongly coupled scenarios in the feature database, so that the midstream damage assessment layer can increase the safety factor in a targeted manner during fatigue calculation.

[0046] S4. The extracted peak factor, asymmetry index, Hilbert marginal spectrum energy characteristics, and coupling strength index are combined into a feature vector. After standardizing the feature vector, a multi-hazard coupled extreme load feature database is constructed.

[0047] S41, Peak Factor Asymmetry index Energy proportions of each frequency band in Hilbert marginal spectrum energy characteristics , , and main frequency Combined into a six-dimensional feature vector: ; S42. Standardize the features of each dimension in the six-dimensional feature vector using the maximum-minimum normalization method: ; in, and These are the minimum and maximum values ​​of the current feature in the training scenario set, respectively; S43. Associate and store the standardized feature vectors with the working condition type labels to construct a multi-hazard coupled extreme load feature database.

[0048] In one embodiment, the load condition type label includes at least four categories: typhoon-dominated, internal wave-dominated, strongly coupled, and weakly coupled. The load condition type label serves as a classification identifier for each analyzed load condition and is derived from the peak factor extracted based on S3. Asymmetry index Energy proportion of each frequency band , , The determination is based on a combination of the results of the coupling strength assessment.

[0049] Load case type labels are used to identify the load dominance type and coupling strength of each analysis load case, specifically including: Typhoon dominance: The typhoon component dominates the total load, manifested as the peak factor. Larger, asymmetric index Positive value, high frequency band energy ratio Significant; Internal wave dominance: The internal wave component dominates the total load, exhibiting an asymmetric index. Near-zero, low-frequency energy ratio Significant; Strong coupling: There is a strong correlation between the typhoon component and the internal wave component, which manifests as follows: Furthermore, the cross-power spectral density exhibits resonance peaks; Weak coupling: The correlation between the typhoon component and the internal wave component is weak, exhibiting the following characteristics: Furthermore, the cross-power spectral density does not show obvious resonance peaks.

[0050] The four operating condition type labels are mutually exclusive categories, and are determined in the following order of priority: first, determine whether it belongs to strong coupling or weak coupling based on the coupling strength index; if it does not belong to the coupling category, then determine whether it is typhoon-dominated or internal wave-dominated based on the frequency domain energy ratio and time domain asymmetry index.

[0051] By operating condition type label, normalized feature vector and nonlinear characterization index ( , Coupling strength A database of multi-hazard coupled extreme load characteristics is constructed using three types of information. This database is stored in a structured format and provides a standardized data interface for direct access by the midstream response assessment layer (fatigue damage calculation, response extreme value statistics) and the downstream prediction and early warning layer (intelligent prediction model, hierarchical early warning system), enabling efficient integration of upstream load characterization results into engineering applications.

[0052] like Figure 2 As shown, the above-mentioned multidimensional nonlinear feature index extraction system and the multi-hazard coupled extreme load feature database constitute a complete interface structure: the time-domain features extracted by S3 (peak factor) Asymmetry index Frequency domain characteristics (energy proportion of each frequency band) , , and main frequency ) and coupling strength indices (correlation coefficient) Cross-power spectral density The results of the strong coupling judgment are written into the feature database after being standardized by S4, forming a standardized nonlinear characterization index system that can be used for midstream damage assessment and downstream intelligent prediction.

[0053] Therefore, this invention employs the aforementioned method for extracting time-frequency features and nonlinear characterizing multi-hazard coupled extreme loads. It replaces traditional EMD with EEMD, fundamentally eliminating mode aliasing through white noise superposition and integrated averaging. This effectively separates typhoon and internal wave components in each order of IMF, ensuring clear physical correspondence between IMFs and significantly improving the interpretability of time-frequency decomposition. The constructed time-domain-frequency domain dual-dimensional feature extraction system quantifies impact intensity using peak factor, characterizes asymmetry using skewness index, and describes frequency domain distribution using marginal spectral energy, achieving a multi-angle quantitative description of nonlinear loads. The proposed coupling strength evaluation method achieves joint quantification of the coupling effect between typhoon and internal wave loads in both the time and frequency domains, providing a basis for differentiated safety corrections under strongly coupled conditions. The established standardized feature database forms a unified data interface, enabling efficient transformation of upstream load characterization results into midstream damage assessment and downstream intelligent prediction.

[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for extracting time-frequency features and nonlinear characterizing multi-hazard coupled extreme loads, characterized in that, Includes the following steps: S1. Collect the time history data of various disaster loads on the target structure, and vectorically superimpose the aerodynamic thrust load, hydrodynamic load and wave load to obtain the time history of multi-disaster coupled loads. Then, preprocess the time history of multi-disaster coupled loads to obtain standardized load signals. S2. Adaptive time-frequency decomposition of the standardized load signal is performed by the integrated empirical mode decomposition method. Mode aliasing is eliminated by multiple white noise superpositions and integrated averaging. The intrinsic mode functions of each order and residual trend terms are output. S3. Peak factor and asymmetry index in the time domain extracted based on standardized load signal; Hilbert marginal spectrum energy features in the frequency domain are extracted based on the intrinsic mode functions of each order; And based on the grouping of eigenmode functions of each order, the coupling strength index between characteristic components of different disaster sources is calculated; S4. The extracted peak factor, asymmetry index, Hilbert marginal spectrum energy characteristics, and coupling strength index are combined into a feature vector. After standardizing the feature vector, a multi-hazard coupled extreme load feature database is constructed.

2. The method according to claim 1, characterized in that, S1 specifically includes: S11. Collect typhoon wind field time history data, and calculate the aerodynamic thrust load on the target structure based on the typhoon wind field time history data. ; S12. Collect internal wave flow field time history data, and calculate the hydrodynamic load on the target structure based on the internal wave flow field time history data. ; S13. Collect wave field time history data and calculate the wave load on the target structure based on the wave field time history data. ; S14. Vector superposition of aerodynamic thrust load, hydrodynamic load, and wave load yields the multi-hazard coupled load time history: ; in, Represents a time variable; S15, Time history of multi-hazard coupled loads Trend term removal and mean normalization are performed, and sampling rates from different data sources are aligned to obtain the standardized load signal. .

3. The method according to claim 2, characterized in that, In step S2, an integrated empirical mode decomposition method is used to adaptively decompose the standardized load signal in time and frequency, specifically including: S21, in the standardized load signal Independent superposition The group mean is zero and the standard deviation is Gaussian white noise ,constitute Group of noisy signals: ; in, i Indicates the sequence number of the white noise group. ; S22, For each group of noisy signals Empirical mode decomposition was performed separately to obtain the eigenmode functions of each order. ; S23, will The group decomposition results are used to take the integrated average of the eigenmode functions of each order, and the result is obtained as follows: eigenmode functions and output the residual trend term. And the decomposition result satisfies ,in, Indicates the order number of the intrinsic mode function. , This represents the total order of the intrinsic mode functions.

4. The method according to claim 3, characterized in that, In S3, the peak factor and asymmetric index in the time domain are extracted based on the standardized load signal, specifically including: Calculate the normalized load signal Peak factor : ; in, Indicates the total duration of the load signal. Represents standardized load signal The root mean square value; Calculate the normalized load signal Asymmetric index : ; in, This represents the mathematical expectation operator.

5. The method according to claim 4, characterized in that, In step S3, the extraction of Hilbert marginal spectral energy features in the frequency domain based on the eigenmode functions of each order specifically includes: For each order of eigenmode function Apply Hilbert transform Construct an analytical signal: ; in, The imaginary unit is used to construct analytic signals; Indicates the first Instantaneous phase and instantaneous amplitude of the first-order eigenmode function after Hilbert transform Instantaneous frequency Instantaneous frequency After end-point effect processing and low-pass filtering; Square the instantaneous amplitude Mapping to instantaneous frequency corresponding angular frequency At this point, construct the Hilbert energy spectrum on the time-frequency plane. The Hilbert marginal spectrum is obtained by integrating along the time axis: ; in, Angular frequency, angular frequency The instantaneous frequencies of each eigenmode function in the time domain The set that constitutes; In Hilbert's marginal spectrum Identify the characteristic frequency bands of each disaster and calculate the energy proportion of each frequency band. , , and main frequency .

6. The method according to claim 5, characterized in that, Of the energy percentages of each frequency band... The proportion of energy within the main frequency band of the internal wave. This represents the proportion of energy within the main frequency band of the typhoon. This refers to the proportion of energy in the mid-frequency band between the main frequency band of internal waves and the main frequency band of typhoons.

7. The method according to claim 6, characterized in that, In step S3, the coupling strength index between different disaster source characteristic components is calculated based on the grouping of the intrinsic mode functions of each order, specifically including: Eigenmode functions of each order Based on the dominant frequency range, it is classified into the first disaster dominant component set and the second disaster dominant component set; Calculate the Pearson correlation coefficient between the dominant component sets of the first hazard and the dominant component sets of the second hazard. and cross-power spectral density Cross-power spectral density Used to identify frequency-domain coupled resonance frequencies; Based on the correlation coefficient and cross-power spectral density Determine the type of coupled operating condition when And cross power spectral density The maximum value within its domain exceeds its average value. When the value is doubled, it is determined to be a strongly coupled operating condition.

8. The method according to claim 7, characterized in that, S4 specifically includes: S41, Peak Factor Asymmetry index Energy proportions of each frequency band in Hilbert marginal spectrum energy characteristics , , and main frequency Combined into a six-dimensional feature vector: ; S42. Standardize the features of each dimension in the six-dimensional feature vector using the maximum-minimum normalization method: ; in, and These are the minimum and maximum values ​​of the current feature in the training scenario set, respectively; S43. Associate and store the standardized feature vectors with the working condition type labels to construct a multi-hazard coupled extreme load feature database.