An intelligent optimization method and system for vibration coupling of an offshore floating wind power platform

Abstract

The present invention relates to the field of offshore wind power technology, and discloses an intelligent optimization method and system for vibration coupling of an offshore floating wind power platform. The method comprises fusing vibration coupling data and environmental data of the offshore floating wind power platform to construct a vibration sample set, fusing data in the time domain, space domain, and environmental coupling of the vibration sample set through a spatiotemporal Kriging model, performing wavefront prediction, and generating prediction results. The cluster coordination degree of the damper is quantified, and a multi-objective function is constructed based on the quantification result and the wavefront prediction result, so as to solve the multi-objective function in the vibration coupling intelligent optimization simulation process of the offshore floating wind power platform and obtain vibration control parameters. The method comprises dynamically allocating the working mode of the damper according to the vibration control parameters, adjusting the dynamic allocation result by using a cluster consistency algorithm, and generating an intelligent optimization strategy for execution. The method realizes effective control of the offshore floating wind power platform and can adapt to different environmental conditions and vibration conditions in real time.

Description

Technical Field

[0001] The present invention relates to the technical field of offshore wind power, and in particular to an intelligent optimization method and system for vibration coupling of an offshore floating wind power platform. Background Art

[0002] Offshore floating wind turbine platforms are offshore structures and novel power generation devices. The platform's inherent vibration characteristics can affect the entire floating wind turbine's operating state, thereby impacting power generation and safety. Therefore, effectively controlling the vibration of offshore floating wind turbine platforms is becoming increasingly important. Currently, a common method involves using sensors installed on the offshore floating wind turbine platform to collect tower top displacement acceleration and platform pitch acceleration data. These data are then directly calculated to generate a cancellation signal, which is used to apply control force to the tuned mass damper in the wind turbine nacelle, thereby controlling the platform's stability. However, this method requires the calculation of a large amount of data in practice, resulting in low data processing accuracy, tedious processing, and a complex calculation process. Summary of the Invention

[0003] The present invention provides an intelligent optimization method and system for vibration coupling of an offshore floating wind power platform, which solves the problem of how to effectively control the vibration of the offshore floating wind power platform.

[0004] To solve the above technical problems, the first aspect of the present invention provides an intelligent optimization method for vibration coupling of an offshore floating wind power platform, comprising:

[0005] Acquire vibration coupling data between the offshore floating wind power platform and the damper and environmental data within a target range where the offshore floating wind power platform is located, and construct a vibration sample set based on the vibration coupling data and the environmental data;

[0006] Based on a pre-built spatiotemporal Kriging model, first data of the vibration sample set in the spatial domain is extracted, second data of the vibration sample set in the time domain is segmented, third data for environmental coupling modeling is determined from the vibration sample set, and the first data, the second data, and the third data are fused to obtain fused data for wavefront prediction, so as to generate a prediction result;

[0007] quantifying the cluster coordination degree of the dampers, constructing a multi-objective function based on the quantified cluster coordination degree and the prediction result, and solving the multi-objective function during the vibration coupling intelligent optimization simulation of the offshore floating wind power platform to obtain vibration control parameters;

[0008] The working mode of the damper is dynamically allocated according to the vibration control parameters, and the dynamic allocation result is adjusted using a cluster consistency algorithm to generate an intelligent optimization strategy for execution.

[0009] A second aspect of the present invention provides an intelligent optimization system for vibration coupling of an offshore floating wind power platform, comprising:

[0010] a sample set construction module, configured to obtain vibration coupling data between the offshore floating wind power platform and the damper and environmental data within a target range where the offshore floating wind power platform is located, and to construct a vibration sample set based on the vibration coupling data and the environmental data;

[0011] a model prediction module, configured to extract first data of the vibration sample set in the spatial domain, segment second data of the vibration sample set in the time domain, determine third data from the vibration sample set for environmental coupling modeling, and fuse the first data, the second data, and the third data to obtain fused data for wavefront prediction, thereby generating a prediction result;

[0012] a function solving module, configured to quantify the cluster coordination degree of the dampers, construct a multi-objective function based on the quantified cluster coordination degree and the prediction result, and solve the multi-objective function during the vibration coupling intelligent optimization simulation of the offshore floating wind power platform to obtain vibration control parameters;

[0013] A strategy execution module is used to dynamically allocate the working mode of the damper according to the vibration control parameters, and use a cluster consistency algorithm to adjust the dynamic allocation result to generate an intelligent optimization strategy for execution.

[0014] Compared with the prior art, the embodiments of the present invention have the following advantages:

[0015] This solution fully considers the impact of spatial, temporal and environmental factors on vibration through the space-time Kriging model, can more accurately predict wavefront vibration conditions, and provide a reliable basis for subsequent vibration control; by quantifying the cluster coordination of dampers and combining the prediction results to construct a multi-objective function, the optimal vibration control parameters are solved using an intelligent algorithm, which can achieve collaborative work between dampers, optimize the control effect, and improve the efficiency and effectiveness of vibration control; the working mode of the damper is dynamically allocated according to the vibration control parameters, and is adjusted using the cluster consistency algorithm, so that the system can adapt to different environmental conditions and vibration conditions in real time, improve the stability and reliability of the system, and further improve the computing efficiency and accuracy, thereby achieving effective control of the offshore floating wind power platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solution of the present invention, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 This is a flow chart of an intelligent optimization method for vibration coupling of an offshore floating wind power platform provided by a certain embodiment of the present invention;

[0018] Figure 2 This is a structural diagram of an intelligent optimization system for vibration coupling of an offshore floating wind power platform provided by one embodiment of the present invention;

[0019] Reference numerals:

[0020] Among them, 10, sample set construction module; 20, model prediction module; 30, function solving module; 40, strategy execution module. DETAILED DESCRIPTION

[0021] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings and embodiments. Obviously, the embodiments described are only some embodiments of the present invention, rather than all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] In the description of this application, the terms "first," "second," "third," etc. are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first," "second," "third," etc. may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0023] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be a communication between the two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the system or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to the specific circumstances.

[0024] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meanings as those commonly understood by those skilled in the art. The terms used in this specification are only for describing specific embodiments and are not intended to limit the present invention. Those skilled in the art will understand the specific meanings of the above terms in this application according to specific circumstances.

[0025] In one embodiment, if Figure 1 As shown, the first aspect of the present invention provides an intelligent optimization method for vibration coupling of an offshore floating wind power platform, comprising:

[0026] S1. Acquire vibration coupling data between an offshore floating wind power platform and a damper and environmental data within a target range where the offshore floating wind power platform is located, and construct a vibration sample set based on the vibration coupling data and the environmental data;

[0027] In one embodiment, step S1 includes:

[0028] Acquiring control parameters of the damper and structural parameters and vibration data of the offshore floating wind power platform as the vibration coupling data;

[0029] Obtaining wind speed, wind direction, wave direction and wave height within a target range where the offshore floating wind power platform is located as the environmental data;

[0030] The vibration coupling data and the environmental data are fused to construct a vibration sample set.

[0031] Specifically, the present invention sets a variety of sensors on the offshore floating wind power platform (also called floating wind power platform) and its wind turbine blades, and the corresponding positions of the dampers set on the platform, such as the platform main structure, the root and tip of the wind turbine blades, and the damper installation position to form a multi-source perception network to collect relevant data in real time, such as a three-dimensional scanner, a vibration sensor, an IMU attitude sensor, a distributed fiber Bragg grating sensor, a displacement sensor, a six-dimensional accelerometer, a laser anemometer, a wave meter, a current meter and a pressure wave height meter; then the aerodynamic shape parameters of the wind turbine blades (chord length distribution, torsion angle distribution, airfoil type, etc.), the control parameters of the damper (damping coefficient, operating frequency, phase difference, etc.), and the platform directly or indirectly obtained through the multi-source perception network are used to collect relevant data in real time. Structural parameters (column diameter, buoy draft, platform mass distribution, etc.) and vibration data (platform vibration acceleration, displacement, velocity, etc.) are used as vibration coupling data. Wind speed, wind direction, wave height, and wave direction within the platform's target range are used as the environmental dataset. Wind speed and direction are generally measured from the sea surface near the platform (usually starting at 10-20 meters above the sea surface) to the height of the wind turbine blade tips (reaching 150-250 meters or even higher). Wave height and direction are measured with a radius of 1-3 kilometers centered on the platform. When collecting environmental data, the target measurement range can be adjusted based on the data type and actual conditions. Finally, the vibration coupling data and environmental data are fused to construct a vibration sample set. The real-time data collected is extensive, covering not only the various types of data mentioned above, but also spatial location information of sensors and dampers, real-time ambient temperature data, and damper output. It should be noted that the "included data" listed here are only examples and are not exhaustive. The actual data types collected are more diverse.

[0032] The present invention comprehensively considers the vibration coupling data related to the offshore floating wind power platform, wind turbine blades and dampers, covering multiple aspects such as aerodynamic shape parameters, control parameters, structural parameters and vibration data, and can more comprehensively reflect the vibration characteristics of the system; at the same time, it obtains environmental data such as wind speed, wind direction, wave direction and wave height within the target range of the platform, which provides a basis for analyzing the impact of environmental factors on vibration and helps to improve the accuracy of the data; by fusing vibration coupling data and environmental data to construct a vibration sample set, the vibration state of the system under different environmental conditions can be more accurately characterized, providing a more reliable data basis for subsequent vibration analysis and control.

[0033] In one embodiment, fusing the vibration coupling data and the environmental data to construct a vibration sample set includes:

[0034] Processing the vibration coupling data and the environmental data by a dynamic time warping method and a spatial interpolation method to obtain aligned vibration data and aligned environmental data;

[0035] extracting time domain features, frequency domain features, and time-frequency domain features from the aligned vibration data as vibration features;

[0036] extracting statistical features, trend features, and spatial distribution features from the aligned environmental data as environmental features;

[0037] Inputting the vibration characteristics and the environmental characteristics into a deep neural network model for processing, and outputting a coupling index;

[0038] Based on the coupling index, the vibration coupling data and the environmental data are fused to generate vibration samples, and after sample enhancement, a vibration sample set is formed.

[0039] Specifically, because different sensors may have different sampling frequencies and data acquisition systems may have time errors, it is necessary to unify the time base of all data. This invention uses a dynamic time warping algorithm to time-align vibration coupling data and environmental data. DTW automatically adjusts the time axis mapping relationship based on the data characteristics, achieving optimal temporal matching of different sensor data. For example, for vibration acceleration data and ambient wind speed data, DTW can dynamically adjust the temporal correspondence between the two based on the start and end times of the vibration event and the changing trend of the wind speed. The alignment process for other data is similar and will not be elaborated on here. To achieve spatial alignment, the present invention adopts a spatial interpolation method combined with a finite element model of the platform: a finite element mesh model of the platform is constructed based on the installation location of the sensors and the geometric structure of the platform; for locations where no sensors are installed, a spatial interpolation method (such as inverse distance weighted interpolation, Kriging interpolation, etc.) is used to estimate the vibration and environmental parameters of the location based on the data from surrounding sensors. By interpolating the sensor data at different locations onto a unified finite element mesh node, spatial alignment and spatial continuity expression of the data are achieved; finally, the vibration data and environmental data that have undergone spatiotemporal alignment are used to generate aligned vibration data and aligned environmental data, respectively.

[0040] The time domain characteristics of the aligned vibration data, such as mean, variance, peak, kurtosis, etc., are calculated to reflect the amplitude, fluctuation and impact characteristics of the vibration; the aligned vibration data are converted to the frequency domain using fast Fourier transform, and the frequency domain characteristics such as the amplitude and power spectrum density of the spectrum are calculated to reveal the frequency components and energy distribution of the vibration; time-frequency analysis methods such as short-time Fourier transform and wavelet transform are used to extract the time-frequency domain characteristics of the aligned vibration data (such as wavelet packet energy, intrinsic mode function energy obtained by empirical mode decomposition, etc.) to simultaneously capture the changing patterns of vibration in time and frequency.

[0041] Calculate the statistical characteristics of the aligned environmental data, such as the mean, standard deviation, maximum value, minimum value, etc., to reflect the central tendency and dispersion of the environmental data; use time series analysis methods, such as the moving average method and the exponential smoothing method, to extract the trend characteristics of the aligned environmental data (such as linear trend, cyclical trend, etc.) to describe the changing trend of environmental data over time; for spatial environmental data, use existing methods to calculate its spatial autocorrelation coefficient, semivariogram function and other spatial distribution characteristics to reflect the spatial correlation and distribution of environmental data.

[0042] A deep neural network model is constructed, whose input layer is composed of vibration data-related features and environmental data-related features, and whose output layer is a coupling index. The coupling index can be a numerical value or a vector, representing the degree of correlation between vibration and the environment. The model also includes a convolutional neural network layer to process the time series features of the environmental data and a recurrent neural network layer to process the time-frequency features of the vibration data. Alternatively, an appropriate network structure can be selected based on the actual situation, such as a multi-layer perceptron (MLP), convolutional neural network (CNN), or recurrent neural network (RNN). Backpropagation and optimization algorithms, such as stochastic gradient descent (SGD) and Adam, can be used to adjust the model parameters so that the model can accurately learn the complex coupling relationship between vibration and environmental characteristics.

[0043] Finally, based on the coupling index, the vibration coupling data and environmental data are fused to generate vibration samples. Each sample contains vibration data (such as vibration acceleration, displacement, and frequency) from a specific time period, environmental data (such as wind speed, wind direction, wave height, and current velocity), and the coupling relationship index between them. For example, the coupling index can be normalized and used as a weight to perform a weighted fusion of the vibration coupling data and environmental data (the sum of the weights equals one) to generate a comprehensive vibration sample. Alternatively, vibration and environmental data from an hour can be used as a single sample. This sample contains not only the original sensor data but also information on the interaction strength and pattern between vibration and environmental factors, as determined through coupling analysis. To increase sample diversity and generalization, the present invention utilizes data enhancement techniques to process the samples. For example, for vibration data, new samples can be generated using methods such as noise addition, time stretching, and frequency modulation. For environmental data, random perturbation and data synthesis can be used for enhancement. The enhanced vibration samples are then combined with the original samples to form a vibration sample set.

[0044] Through data alignment, the present invention can more accurately analyze the correlation between vibration and environmental factors, avoiding analysis errors caused by data asynchrony or spatial incompleteness; by extracting multiple features from the aligned data, it provides more comprehensive and accurate information for the subsequent deep neural network model, which helps to improve the performance and prediction accuracy of the model; the vibration samples generated based on the coupling index can more realistically reflect the actual relationship between vibration and the environment, and the quality of the sample set is improved; and the sample enhancement technology can increase the diversity of samples, avoid model overfitting, and improve the generalization ability of the model.

[0045] The fusion of vibration coupling data and environmental data also includes the following steps:

[0046] Extract cross-modal common features and cross-modal non-common features from vibration coupling data and environmental data, and map them into a shared feature space;

[0047] Align the cross-modal common features to obtain aligned features;

[0048] By optimizing the encoder through contrastive learning, the cross-modal non-shared features are corrected to obtain the corrected features;

[0049] The correction features and alignment features are spliced ​​in the shared feature space to form splicing features to reconstruct them, and the reconstructed data is the fused data;

[0050] Alternatively, the dependency between the correction features and the alignment features in the shared feature space can be determined based on the attention mechanism, and the feature weights of the correction features and the alignment features can be dynamically assigned based on the dependency to weight them. The weighted results can be reconstructed, and the reconstructed data is the fused data.

[0051] Among them, cross-modal common features are the feature types that exist in both vibration coupling data and environmental data, which can reflect the coupling correlation between environmental loads and platform vibration; including numerical statistical features (that is, both contain quantifiable numerical parameters and statistics, such as mean and extreme values: the mean of environmental wind speed and the mean of platform vibration acceleration, wave height extreme values ​​and displacement extreme values ​​are all numerical statistical features; such as variance and discreteness: the variance of wind speed fluctuation and the variance of vibration velocity both describe the degree of data fluctuation), dynamic frequency features (that is, periodic frequency components, such as environmental wave frequency, that is, the wave height period is coupled with the natural frequency of platform vibration, resulting in The vibration data frequency domain has a peak that is consistent with the wave frequency, that is, the resonance frequency; for example: when the damper operating frequency matches the ambient wind / wave frequency, the frequency characteristics of the environment will be reflected in the vibration coupling data), directional correlation characteristics (that is, the correlation of vector directions, such as the angular parameters of wind direction and wave direction have a causal relationship with the direction of the platform vibration response, and the directional angle as a feature can be reflected by both) and time-varying trend characteristics (that is, the dynamic response in the time series, such as the sudden increase in ambient wind speed is synchronized with the sudden change trend of the platform vibration acceleration, both of which contain trend characteristics that change with time, that is, short-term fluctuations and long-term periodicity).

[0052] Cross-modal non-shared features are those unique to vibration coupling data, such as structural and shape parameters, control and dynamic parameters, and those unique to environmental data, such as spatial distribution characteristics. Shared features arise from the physical coupling of environmental loads and platform vibration (e.g., wind and wave excitation causing structural vibration), while non-shared features are determined by the intrinsic properties of the data.

[0053] Features from coupled vibration data and environmental data are extracted independently on a modal basis. For coupled vibration data, structural parameters (blade airfoil, damping coefficient) are directly extracted; for time series data, Fourier transform (frequency domain features) and wavelet transform (time-frequency features) are used; and statistical quantities such as mean, variance, and kurtosis are calculated. For environmental data, FFT is used to extract wave frequency and wind speed power spectrum; spatial features include wave direction angle distribution and spatial interpolation of wind energy density; and statistical features include wind speed extremes and wave height probability distribution. It should be noted that the extracted features listed here are only examples and not exhaustive; the actual types of features extracted are more diverse. Using modality-specific encoders (e.g., CNN for vibration data to extract frequency domain features and LSTM for environmental data to extract time series features), cross-modal shared and non-shared features are mapped into a shared feature space.

[0054] A twin neural network is constructed, and the extracted vibration-related and environmental-related features are fed into a shared-weight encoder. The cosine similarity of the feature vectors is calculated, and features with high similarity (e.g., cosθ > 0.7) are identified as cross-modal shared features. Cross-modal shared features can also be extracted using mutual information calculations, with thresholds set for screening. Cross-modal shared features are excluded, while the unique attributes of each data are retained. Features corresponding to these attributes are considered cross-modal non-shared features. The separated features are then verified for physical consistency: whether the cross-modal shared features conform to the coupling mechanism (e.g., whether a coupling relationship exists between the wave frequency and the platform vibration frequency) and whether the cross-modal non-shared features are independent of the other modal (e.g., whether the blade airfoil is not directly related to the environmental parameters). Cross-modal reconstruction verification is then performed on the verified features: cross-modal predictions are performed using the extracted shared features (e.g., using the environmental wave frequency to predict the vibration response frequency). A low prediction error (e.g., RMSE < 5%) indicates that the cross-modal shared features have been accurately extracted.

[0055] Contrastive learning or cycle consistency loss function is used to force the alignment of vector representations of cross-modal common features (such as minimizing the vector distance between vibration frequency and wave frequency) to form aligned features.

[0056] The cosine distance between each pair of cross-modal non-shared features is calculated to construct an initial distance matrix. Cross-modal non-shared features under the same operating condition (e.g., platform vibration acceleration characteristics corresponding to a wave height of 5m, i.e., physically correlated samples) are used as positive pairs, while cross-modal non-shared features under different operating conditions (e.g., vibration characteristics at a wave height of 2m and environmental characteristics at a wave height of 5m, i.e., samples without physical correlation) are used as negative pairs. A contrastive learning algorithm, commonly used in existing technologies, is used to drive the correction process. By minimizing the distance between positive pairs and maximizing the distance between negative pairs, the non-shared features are forced to converge in the shared space, ultimately yielding the corrected features. This correction process brings the cross-modal non-shared features of the vibration coupling data and the environmental data closer together in the shared feature space.

[0057] The corrected non-shared features and aligned shared features are concatenated into a complete feature vector, which is then fed into the decoder to reconstruct the fused data. Alternatively, the weights of each feature are calculated using an attention mechanism, and the data is reconstructed after dynamic weighting, which is then fed into the decoder to generate the fused data.

[0058] The present invention can eliminate semantic deviations in data from different modalities by aligning shared features across modalities, allowing shared features to form a unified representation in a shared space. Correction of non-shared features preserves the unique physical properties of each modality, preventing feature loss. In the process of optimizing encoders through comparative learning, the distance between non-shared features is used as the correction criterion, which can strengthen the physical association between data from different modalities and make the fused data more consistent with actual physical processes.

[0059] S2. Based on a pre-built spatiotemporal kriging model, extract first data of the vibration sample set in the spatial domain, segment second data of the vibration sample set in the time domain, determine third data from the vibration sample set for environmental coupling modeling, and fuse the first data, the second data, and the third data to obtain fused data for wavefront prediction, so as to generate a prediction result; wherein the kernel function of the spatiotemporal kriging model includes a spatial kernel, a temporal kernel, and an environmental coupling term;

[0060] Specifically, the present invention generates Delaunay triangulation based on the coordinates of each sensor to construct a spatial topological network, and then extracts the sensor and damper position coordinates and vibration data of the corresponding positions from the vibration sample set, and uses it as the first data of the vibration sample set in the spatial domain. This data is used to characterize the distribution characteristics of vibration in three-dimensional physical space and reflect the correlation of vibration energy in spatial position.

[0061] Centered on the arrival time of the wavefront, a time window (preferably 200s) is cut to extract the vibration-related data (vibration acceleration, displacement, and velocity of the platform) from the vibration sample set. The data is then segmented according to the time dimension, and the time-varying sequence of the vibration amplitude at each sampling time point and each position is recorded. This data is used as the second data of the vibration sample set in the time domain. This data is used to characterize the evolution of vibration over time and reflect the dynamic characteristics and timing laws of wavefront propagation.

[0062] Environmental factors related to vibration propagation in the vibration sample set, such as ambient temperature, wave height, wave direction, wave height, wave direction, etc., are extracted and converted into parameter forms that can be processed by the spatiotemporal Kriging model (which can be processed by Z-score standardization) to serve as the third data for environmental coupling modeling in the vibration sample set. This data is used to quantify the external environmental factors that affect vibration propagation and reflect the modulation effect of environmental conditions on wavefront propagation.

[0063] The first data, the second data, and the third data are combined to obtain observed vibration data. The pre-built spatiotemporal Kriging model is used to calculate the Kriging weight of each target spatiotemporal point. The first data, the second data, and the third data are then fused based on the weighted Kriging weight and the observed vibration data to obtain fused data. The calculation process of the fused data is expressed by the following formula:

[0064] λ T C=C(x,t)

[0065] z'(x,t)=λ T z

[0066] Where λ is the kriging weight of the target spacetime point; C is the spacetime covariance matrix, that is, the spacetime kriging model; x is the coordinate of the target spacetime point, which can also be understood as the location of the offshore floating wind power platform; t is the current time; z' is the fused data; and z is the observed vibration data.

[0067] Based on the fused data, a spatial contour map, that is, the vibration amplitude distribution, can be directly generated; the amplitude gradient of the fused data is calculated, and the wavefront propagation direction is the opposite direction of the gradient (from high amplitude to low amplitude). By normalizing the gradient vector, the unit direction vector of the wavefront propagation can be obtained; the time difference method is used to calculate the wavefront propagation velocity using the wavefront position predicted at the target time and space point at adjacent moments; finally, the vibration amplitude distribution, wavefront propagation direction and wavefront propagation velocity are output as prediction results.

[0068] The present invention describes the wavefront characteristics from the perspectives of static distribution and dynamic evolution through spatial domain data and time domain data respectively, while environmental coupling data is the bridge connecting the physical environment and vibration propagation. By quantifying the impact of environmental factors on the former two, the model can more realistically reflect the propagation law of the wavefront in actual scenarios. After the three are integrated, high-precision prediction of vibration wavefronts in complex environments can be achieved.

[0069] In one embodiment, the training process of the spatiotemporal kriging model includes:

[0070] constructing the spatial kernel using a Gaussian kernel function and an exponential kernel function, constructing the temporal kernel using a rational quadratic kernel function, and constructing the environmental coupling term based on the environmental data;

[0071] Constructing the space-time Kriging model according to the space kernel, the time kernel and the environmental coupling term;

[0072] A space-time cube data matrix is ​​constructed to optimize the parameters of the space-time Kriging model by maximizing the marginal likelihood function to obtain a trained space-time Kriging model.

[0073] Specifically, the space-time kriging model in the present invention introduces the time dimension and environmental coupling term on the basis of traditional kriging interpolation to form a four-dimensional prediction framework, including the spatial kernel, the temporal kernel and the environmental coupling term:

[0074] Among them, the spatial kernel is constructed by Gaussian kernel function and exponential kernel function, which is expressed by the following formula:

[0075]

[0076] Where Ks is the spatial kernel, which is used to describe the vibration correlation between points with similar spatial positions; Δr is the spatial distance vector, specifically the Euclidean distance between different dampers, different sensors, or different vibration monitoring locations on an offshore floating wind turbine platform; σs is the weight coefficient (ranging from 0.6 to 0.8), which determines the contribution ratio of the Gaussian kernel and the exponential kernel to the overall spatial kernel, reflecting the relative importance of the vibration data at different distances in space on the vibration at the current location; s1 is the spatial correlation length (optimization range 5-50m), which characterizes the propagation and attenuation characteristics of vibration energy in space and reflects the vibration transmission efficiency of the platform structure.

[0077] The time kernel is constructed by a rational quadratic kernel function, which is expressed as follows:

[0078]

[0079] Where Kt is the time kernel, which is used to characterize the correlation of vibration in the time dimension; Δt is the time interval, preferably 0.1s; τ is the time decay constant (optimization range 0.1-10s), which is used to describe the residence characteristics of vibration energy in the time dimension and quantify the duration of vibration; α is the shape parameter (range 1.2-2.0), which is used to control the decay rate of the time kernel function with the time interval, reflecting the change of the correlation degree between vibration data at different times over time.

[0080] The environmental coupling term is constructed from environmental data and is expressed as follows:

[0081]

[0082] Where Ke is the environmental coupling term, which is used to quantify the impact of environmental factors on vibration propagation; γ and δ are environmental coupling coefficients (optimization range 0.01-1.0), among which γ is used to characterize the intensity of the influence of wind-induced vibration on platform vibration and quantify wind-structure interaction; δ is used to characterize the coupling intensity between wave excitation and platform vibration and quantify wave-structure interaction; vw is the wind speed; Hs is the wave height; Δθ is the angle between the wave direction and the position vector.

[0083] The constructed spatial kernel, temporal kernel and environmental coupling terms are combined to form a spatiotemporal Kriging model, that is, a spatiotemporal covariance matrix, which characterizes the correlation of vibration data between any spatiotemporal points and is expressed as follows:

[0084]

[0085] Where Cov is the space-time Kriging model.

[0086] The data in the vibration sample set are organized according to the spatial, temporal, and environmental dimensions to construct a space-time cube data matrix. Each element of the matrix corresponds to the vibration data at a specific spatial location, time point, and environmental condition. The parameters of the space-time kriging model are optimized by maximizing the marginal likelihood function, which can be expressed as:

[0087]

[0088] Where L is the marginal likelihood function; theta is the model parameter, namely the spatial correlation length, time decay constant, environmental coupling coefficient, time kernel shape parameter, and weight coefficient σs; y is the vibration data vector; T is the transpose operation; K is the kernel matrix constructed based on Cov; and n is the number of samples.

[0089] Finally, an optimization algorithm, such as the conjugate gradient method and the quasi-Newton method, is used to maximize the marginal likelihood function to obtain the optimal model parameters, thereby obtaining a trained spatiotemporal kriging model. The spatiotemporal kriging model in the present invention comprehensively considers the influence of space, time, and environmental factors on vibration, and can more accurately describe the propagation law of vibration in space and time and the mechanism of action of environmental factors on vibration, thereby outputting more accurate prediction results such as vibration amplitude, wavefront propagation direction, and speed; wherein, the introduction of the environmental coupling term fully considers the influence of environmental data on vibration, enables the model to be adaptively adjusted according to different environmental conditions, and improves the accuracy and reliability of the prediction results in practical applications; by constructing a spatiotemporal cube data matrix and maximizing the marginal likelihood function to optimize the model parameters, the model can be better adapted to different data scenarios, improve the robustness and generalization ability of the model, and reduce the occurrence of overfitting and underfitting phenomena.

[0090] S3. quantifying the cluster coordination degree of the dampers, constructing a multi-objective function based on the quantified cluster coordination degree and the prediction result, and solving the multi-objective function during the vibration coupling intelligent optimization simulation of the offshore floating wind power platform to obtain vibration control parameters;

[0091] In one embodiment, quantifying the cluster coordination degree of the damper includes:

[0092] quantifying the distance from the damper to the vibration hotspot and combining it with the coordination radius to calculate the spatial weight of the damper;

[0093] Obtaining the actual output and theoretical target output of the damper to quantify the coordination deviation of the damper;

[0094] quantifying the instantaneous degree of cooperation of the damper based on the spatial weight and the cooperation deviation;

[0095] The instantaneous synergy is subjected to time window integration and time smoothing processing to obtain the cluster synergy.

[0096] Specifically, cluster coordination is the core indicator for measuring the effect of cluster control. Its calculation combines spatial coordination accuracy and temporal stability, and is used to reflect the consistency, coordination and effectiveness of the overall action of the damper cluster, that is, whether each damper can cooperate with each other in the expected manner and work together when suppressing vibration. High cluster coordination means that the actions of the dampers are highly consistent and can effectively work together to suppress vibration; while low cluster coordination indicates that there are differences in the actions of the dampers, which may lead to poor vibration suppression effects or even mutual interference. The present invention calculates the spatial weight of each damper by quantifying the distance from each damper to the vibration hotspot and combining it with the coordination radius, which is calculated by the following formula:

[0097]

[0098] Where w i (t) is the spatial weight of the i-th damper, which is used to characterize the spatial influence of the damper on the vibration hotspot; d i (t) is the Euclidean distance from the i-th damper to the vibration hotspot; σ is the cooperative radius; k is the number of dampers.

[0099] The present invention determines areas on offshore floating wind turbines with more intense vibrations by analyzing a set of vibration samples, and marks them as vibration hotspots. The determination of vibration hotspots can be based on indicators such as vibration amplitude and vibration energy, or a threshold value of vibration amplitude (such as displacement, velocity, acceleration) can be set, and areas exceeding the threshold value are determined as areas of intense vibration. The threshold value is determined based on historical data or actual needs, and no specific numerical limit is given here. The coordination radius, as one of the vibration control parameters, can be dynamically adjusted according to the solution of the objective function. The spatial weight is used to characterize that the closer the damper is to the vibration hotspot, the greater the weight and the more significant the impact on the coordination degree.

[0100] The actual output and theoretical target output of each damper are obtained to quantify the coordination deviation of each damper, which is calculated by the following formula:

[0101] Ei=∣Fact,i-Fopt,i∣ / (Fopt,i+ε)

[0102] where Ei is the coordination deviation of the i-th damper, that is, the relative deviation between the actual output of the i-th damper and the theoretical target output; Fact,i and Fopt,i are the actual output and theoretical target output of the i-th damper, respectively; and ε is the zero-prevention constant (1e-5).

[0103] Among them, the actual output of each damper can be obtained by real-time monitoring of the damper through sensors. The theoretical target output is also one of the vibration control parameters and can be dynamically adjusted in combination with the solution results of the objective function.

[0104] The instantaneous coordination degree of the damper is quantified based on the spatial weight and coordination deviation, which is calculated as follows:

[0105] J 3,t =Σw i *exp(-β*Ei)

[0106] Where, J 3,t is the instantaneous coordination degree, which is used to measure the space-output coordination effectiveness of the damper cluster at time t; β is the sensitivity coefficient, which is preferably 2.0.

[0107] Finally, a suitable time window (preferably 300s) is selected, and the instantaneous coordination of the cluster is integrated within the time window to obtain the average coordination within the time window; the length of the time window can also be determined according to the vibration period and the response speed of the system; the average coordination after the time window integration is time-smoothed to eliminate the influence of short-term fluctuations and noise, so as to obtain a more stable cluster coordination; the time smoothing process can adopt a moving average, that is, the average value of the average coordination of several recent time windows is calculated as the current cluster coordination.

[0108] By quantifying spatial weights, the present invention can more accurately determine the importance of each damper in vibration control, achieve reasonable allocation of damper resources, avoid resource waste, and improve the efficiency and accuracy of vibration control; by quantifying coordination deviations, the gap between the performance of the damper in actual work and the ideal state can be clarified, providing a direct basis for optimizing the control strategy of the damper, which is helpful to improve the working performance of the damper; based on the spatial weights and coordination deviations, the instantaneous coordination degree is quantified, and the instantaneous coordination degree is time-window integrated and time-smoothed to obtain the cluster coordination degree, which can evaluate the overall coordinated working state of the damper cluster in real time and dynamically, discover coordination problems in time, and provide a basis for adjusting control parameters and optimizing cluster working modes.

[0109] Alternatively, the cluster synergy between dampers can be determined by the following method: The buoyancy, tower, and blades of an offshore floating wind turbine platform are considered rigid bodies, with the dampers serving as flexible connection units, and multibody dynamic equations are established. Potential flow theory is applied to the WAMIT software to calculate wave forces acting on the platform, including wave excitation and diffraction forces, which are then input as external loads into the multibody dynamic model. Radiation damping and hydrodynamic added mass are also considered to establish a complete water-structure coupling mathematical model. A wave load time history representing typical sea conditions (e.g., a significant wave height of 4 m and a period of 10 s) is input into the mathematical model to solve the time-domain response of the damper cluster over 600 s. The cross-power spectrum between the dampers is calculated, the energy transfer efficiency is analyzed, and the force-displacement phase difference is calculated using the Hilbert transform method to derive the time-domain synergy. The damper responses are Fourier transformed to obtain the platform's principal heave and roll vibration frequencies. The coherence function of each damper at these two frequencies is then calculated and averaged with the time-domain synergy to obtain the cluster synergy of the dampers.

[0110] In one embodiment, constructing a multi-objective function based on the quantified cluster synergy and the prediction result includes:

[0111] Determining input energy based on the prediction result, and quantifying dissipated energy according to the working state of the damper to determine energy dissipation efficiency with the input energy;

[0112] Obtaining a stress spectrum of the damper and determining fatigue damage in combination with material fatigue performance data of the damper;

[0113] The multi-objective function is constructed with the goal of minimizing the energy dissipation efficiency and the fatigue damage, and maximizing the cluster coordination degree.

[0114] Specifically, the vibration amplitude reflects the intensity of the vibration, and the energy input to the system is proportional to the square of the vibration amplitude. The present invention is based on the vibration prediction results and quantifies the input energy by combining the vibration amplitude with the mass and natural frequency of the vibration device; the input energy can also be obtained by integrating the energy of the vibration field; the working state of the damper is monitored, and its dissipated energy can be calculated by integrating the work done by the damping force; the energy dissipation efficiency is the ratio of the energy dissipated by the damper to the system input energy, thereby obtaining the energy dissipation efficiency.

[0115] Strain gauges are placed at key locations on the damper to measure its real-time strain spectrum. The stress spectrum reflects the stress changes experienced by the damper during operation and can be obtained through finite element analysis or actual measurement. Combined with the damper's material fatigue performance data, such as the SN curve (stress-life curve), the stress spectrum is processed using methods such as the rain flow counting method to calculate the damper's fatigue damage. The fatigue damage can be cumulatively calculated based on the Palmgren-Mainer law.

[0116] A multi-objective function is constructed with the goals of minimizing energy dissipation efficiency, minimizing fatigue damage, and maximizing cluster coordination. The weight coefficient of each objective is 1. In addition, the weight coefficient of each objective can be determined according to actual needs and system characteristics, or through expert experience, sensitivity analysis, or optimization algorithms to determine the appropriate weight coefficient to achieve a balance between different objectives and needs.

[0117] The multi-objective function in the present invention comprehensively considers three key indicators: energy dissipation efficiency, fatigue damage and cluster coordination, and can achieve comprehensive optimization of the vibration control system of the offshore floating wind power platform; by optimizing the cluster coordination, the damper cluster can work better together, improving the stability and reliability of the system.

[0118] In one embodiment, solving the multi-objective function to obtain vibration control parameters includes:

[0119] The decision variables are encoded using real number encoding method, and a set of initial solutions are randomly generated as the initial population of the genetic algorithm;

[0120] Performing selection, crossover, and mutation operations based on the initial population to construct a mixed population, quantifying the fitness value of each individual in the mixed population according to the multi-objective function, and performing non-dominated sorting and crowding calculation based on the quantified results to select a number of elite individuals from the mixed population as a new population;

[0121] The new population is used as a particle swarm, so that the position of each particle corresponds to the code of the individual in the genetic algorithm, and the speed of each particle in the particle swarm is initialized to zero;

[0122] updating the velocity of each particle according to the individual extreme value and the group extreme value corresponding to each particle, and updating the position of each particle according to the updated velocity to update the particle group;

[0123] performing non-dominated sorting and crowding calculation on the updated particle swarm based on the multi-objective function, and selecting a number of particles as the swarm for the next iteration according to the calculation results;

[0124] The generation process of the new population and the generation process of the next iterative population are iteratively executed based on the next iterative population until a preset number of iterations is reached, and a set of solutions is selected from the current population to be output as vibration control parameters.

[0125] Specifically, in the vibration coupling intelligent optimization simulation process of the offshore floating wind power platform, the present invention adopts a hybrid algorithm formed by the genetic algorithm and the particle swarm algorithm to solve the multi-objective function: first set the algorithm parameters, such as the population size N, the maximum number of iterations, the crossover probability, the mutation probability, the inertia weight, the learning factor, etc.; then use the coordination radius of the damper cluster, the consistency gain coefficient, the damper working mode switching threshold and the damper parameters, such as the theoretical target output, as decision variables, use real number coding to encode the decision variables, and randomly generate a set of initial solutions as the initial population of the genetic algorithm. Each individual in the population represents a set of vibration control parameters, that is, the decision variables.

[0126] Using the tournament selection method, excellent individuals are selected from the initial population and placed in the mating pool. The individuals in the mating pool are subjected to simulated binary crossover to generate the offspring population, and polynomial mutation is performed on the individuals in the offspring population to increase population diversity. The initial population and the offspring population are merged to form a mixed population of size 2N. The fitness value of each individual in the mixed population is quantified by a multi-objective function, and a fast non-dominated sorting is performed based on the fitness quantification results to divide the population into different levels. The crowding degree of individuals in the same non-dominated layer is calculated to maintain the distribution of the solution set. According to the non-dominated level and crowding degree, the top N individuals from the mixed population are selected as elite individuals to form a new population.

[0127] The new population is regarded as a particle swarm. The position of each particle corresponds to the encoding of the individual in the genetic algorithm, that is, the value of the vibration control parameter. The speed of each particle is initialized to zero. According to the individual extreme value (the particle's own historical optimal position) and the group extreme value (the entire particle swarm's historical optimal position) corresponding to each particle, the speed and position of each particle are updated to update the particle swarm.

[0128] The fitness value of each particle in the updated particle swarm is calculated according to the multi-objective function, and the non-dominated sorting and crowding calculation are performed on the updated particle swarm based on the fitness calculation results to evaluate the quality and diversity of each particle; according to the crowding calculation results, several particles are selected as the population for the next iteration; wherein, the selection strategy can be similar to the selection strategy in the genetic algorithm, giving priority to individuals in a better non-dominated layer and with a larger crowding degree.

[0129] Based on the population from the next iteration, the process of generating a new population (genetic algorithm operation) and the process of generating the next population (particle swarm optimization operation) are iteratively executed until the preset number of iterations is reached. Finally, a set of solutions is selected from the current population as the vibration control parameter output. Individuals in the optimal non-dominated layer with good overall performance can be selected as the final vibration control parameters. Alternatively, the NSGA-III algorithm or other optimization algorithms can be used to solve multi-objective functions. The specific process will not be described in detail here.

[0130] The present invention combines the global search capabilities of the genetic algorithm with the local fine-tuning capabilities of the particle swarm algorithm, creating complementary advantages. The genetic algorithm can extensively explore the solution space through selection, crossover, and mutation operations, avoiding being trapped in local optima. The particle swarm algorithm utilizes historical optimal information of individuals and groups to guide the search direction to converge to more optimal areas. The combination of the two can improve the efficiency and accuracy of solving multi-objective functions, thereby obtaining more optimal vibration control parameters. In addition, the hybrid algorithm combining the genetic algorithm and the particle swarm algorithm adopted in the present invention is compared with the single algorithm. The comparison results are shown in the following table:

[0131] Table 1 Comparison of optimization effects Table 1

[0132]

[0133] As can be seen from Table 1 above, compared with the single genetic algorithm and particle swarm algorithm, the hybrid algorithm adopted by the present invention has a higher proportion of non-dominated solutions, better convergence, better overall performance of the solution set, fewer iterations required to achieve the same accuracy, faster approach to the optimal solution, and stronger stability. It can be seen that the hybrid algorithm is significantly superior to traditional GA and PSO in optimization effect (optimal solution quality, average solution quality) and convergence speed (convergence degree, number of iterations), and is more suitable for high-precision and high-efficiency vibration control parameter optimization scenarios.

[0134] In one embodiment, in addition to using the elite individuals selected by the genetic algorithm as a particle swarm for optimization, this embodiment also proposes a dynamic hierarchical co-evolution framework that deeply integrates the global exploration capability of the genetic algorithm (GA) with the local development capability of the particle swarm algorithm (PSO). Through an adaptive weight mechanism and a multi-modal solution retention strategy, efficient multi-objective optimization of vibration control parameters is achieved.

[0135] Step S4 further includes:

[0136] 1. Coding and Initialization: A real number coding strategy is used to map the vibration control parameters to chromosomes and particle position vectors. Initialize the mixed population: the first 50% of individuals are generated through crossover mutation of GA, and the remaining 50% are initialized through random distribution of PSO. The particle velocity is initialized to zero to ensure initial diversity. For setting the algorithm parameters, please refer to the above content and will not be elaborated here.

[0137] 2. Dynamic hierarchical search mechanism:

[0138] 1) Global Exploration Layer (GA-dominated): Performs selection, crossover, and mutation operations to generate offspring populations. It uses a multi-objective function to quantify the fitness values ​​of individuals in the population. Based on the quantified results, it uses the NSGA-II algorithm to perform non-dominated sorting and crowding calculations. Based on the calculation results, it selects outstanding individuals to enter the elite pool to update the convergence archive (CA).

[0139] 2) Local development layer (PSO-dominated): Adaptive weights are introduced to replace the inertia weights in the velocity update formula of the original PSO algorithm, thereby forming an improved velocity update formula, which is used to update the particle velocity. The position update still uses the position update formula of the original PSO algorithm; the adaptive weight ω is expressed as: ω=ωmax-g / G(ωmax-ωmin); where ωmax and ωmin are the maximum value of the inertia weight (usually set to 0.9 or 1.0) and the minimum value of the inertia weight (usually set to 0.4 or 0.2), respectively; G and g are the maximum number of iterations and the number of iterations executed by the current algorithm, respectively; sparse regional solutions are retained through the hypervolume contribution (calculated by comparing the dominance relationship between each solution in the solution set and other solutions) to prevent premature convergence, so as to build a diverse archive (DA) based on the updated particles;

[0140] After each iteration, the environment selection is performed on CA and DA: the individuals in CA and DA are merged to form a merged population, and the merged population is sorted non-dominatedly, divided into different frontier layers, and individuals are selected in sequence starting from the optimal frontier layer until the preset population size is reached. If the number of individuals in a certain frontier layer exceeds the remaining selectable number, further screening is performed through crowding comparison or super volume contribution;

[0141] 3) Introduce the synergy factor u to dynamically adjust the operation ratio of the genetic algorithm (GA) and the particle swarm optimization (PSO) in the hybrid algorithm: u=1 / (1+e -h(g / G-0.5) ), where h is a parameter that controls the rate of change of the synergy factor, and is usually set to 10 or greater to allow the synergy factor to change rapidly during the iteration process. In the early stage (u<0.5), GA dominates the global search; in the later stage (u≥0.5), PSO dominates the fine optimization.

[0142] 3. Iterate step 2 until the maximum number of iterations is reached, and extract the Pareto frontier from the CA. Use the fuzzy TOPSIS method and the decision maker's preference to select the final parameter combination as the vibration control parameter output.

[0143] This method adopts a dynamic hierarchical collaboration mechanism, which balances global exploration and local development through the hierarchical collaboration of GA and PSO, avoiding the parameter sensitivity problem of traditional hybrid algorithms; adopts a dual archiving mechanism to simultaneously ensure the convergence and diversity of the solution set; introduces adaptive weights to enable the algorithm to automatically switch behavior modes according to the search progress, thereby improving robustness.

[0144] S5. Dynamically assigning the operating mode of the damper according to the vibration control parameters, and adjusting the dynamic assignment result using a cluster consensus algorithm to generate an intelligent optimization strategy for execution;

[0145] In one embodiment, step S5 includes:

[0146] dynamically assigning the operating mode of the damper based on the vibration control parameters and the environmental data to obtain an adaptive mode switching strategy;

[0147] quantifying the dynamic weight according to the adaptive mode switching strategy, and weighting the consistency protocol of the cluster consistency coordination algorithm by the dynamic weight to obtain a weighted protocol;

[0148] The adaptive mode switching strategy is adjusted through the weighted protocol to generate an intelligent optimization strategy for execution.

[0149] Specifically, the present invention divides the working modes of the damper into four modes, including: standby mode, air pressure energy storage mode, magnetorheological mode and mixed mode; when the damper working mode switching threshold is greater than the first threshold and the wave height is less than two meters, the damper is controlled to switch to standby mode; when the damper working mode switching threshold is greater than the second threshold and less than or equal to the first threshold, and the wave height is not less than two meters and not more than four meters, the damper is controlled to switch to air pressure energy storage mode; when the damper working mode switching threshold is not greater than the second threshold and the wave height is greater than four meters, the damper is controlled to switch to magnetorheological mode; when the damper working mode switching threshold is other values ​​other than the above three and the wave height is not limited, the damper is controlled to switch to mixed mode; among them, the first and second thresholds are preferably 500 and 200 respectively, and the two thresholds can also be adjusted according to actual needs. Based on the above switching rules, combined with the damper operating mode switching threshold in the vibration control parameters and the wave height in the environmental parameters, the damper operating mode is dynamically allocated in real time to send mode switching instructions to the damper. For example, an adaptive mode switching strategy can be obtained by controlling the current of the pneumatic valve or magnetorheological fluid through an electrical signal.

[0150] In one embodiment, quantizing the dynamic weight according to the adaptive mode switching strategy includes:

[0151] determining a basic weight based on the distance between the dampers and the vibration control parameter;

[0152] Determine the mode matching degree according to the adaptive mode switching strategy at the current moment;

[0153] Obtaining a historical number of failures of the damper to determine a health factor;

[0154] The dynamic weight is determined by the basic weight, the pattern matching degree and the health factor.

[0155] The present invention quantifies the basic weight based on the distance and cooperation radius between each damper in the damper cluster, which is used to quantify the synergistic influence strength between the dampers and is expressed by the following formula:

[0156]

[0157] Where a is the basic weight; d is the distance between dampers, which can be calculated by Euclidean distance.

[0158] In addition, the above embodiment uses a formula to calculate the basic weights. Alternatively, this method can be implemented using a corresponding AI (artificial intelligence) algorithm model: a sample dataset containing the distances between dampers and the coordination radius is constructed, and the sample data is labeled with data result labels. The data result labels are used to represent the basic weights corresponding to the distances between dampers and the coordination radius. Then, based on a learning algorithm, the AI ​​algorithm model is trained using the sample dataset. During the training process, the convolutional neural network model can be retrained or fine-tuned, referring to existing methods, to improve the model's generalization capability. Finally, a trained convolutional neural network model is obtained. In practical applications, the distances between dampers and the coordination radius are input into the convolutional neural network model, which analyzes and processes the data and outputs the relevant data results (i.e., the corresponding basic weights). Through the processing of the above algorithm model, the convolutional neural network model is capable of inputting the distances between dampers and the coordination radius, logically deducing, and outputting the basic weights. It should be noted that the above training method is merely an example, and those skilled in the art may also select other appropriate methods based on the scenario, such as reinforcement learning, federated learning, transfer learning, or other common learning paradigms. This is not specifically limited in the present embodiment. In addition, other data calculated using formulas can also be implemented using this AI algorithm model, and there is no specific limitation here.

[0159] The mode matching degree is determined according to the adaptive mode switching strategy at the current moment, which is expressed by the following formula:

[0160]

[0161] Where M is the mode matching degree; mi and mj are the operating modes of the damper at the previous moment and the current moment, respectively; m corresponds to the standby mode, air pressure energy storage mode, magnetorheological mode, and mixed mode, respectively, 0.25, 0.5, 0.75, and 1.

[0162] The health factor of the damper is calculated based on the number of historical failures, which is expressed as follows:

[0163]

[0164] Where Q is the health factor and P is the number of historical failures.

[0165] Finally, the product of the basic weight, pattern matching degree and health factor of each damper is taken as its dynamic weight.

[0166] The present invention determines the basic weight by considering the distance between dampers, which can enable adjacent or spatially important dampers to play a greater role in cluster control, optimize spatial coordination effects, and improve the vibration control performance of local areas; determine the mode matching degree according to the adaptive mode switching strategy to ensure that the current damper's working mode is highly matched with the optimal strategy, improve the system's response speed and control efficiency, and make vibration control more accurate and timely; use the health factor based on the number of historical failures to evaluate the health status and reliability of the damper, avoid the stability of the entire system affected by equipment aging or failure, and extend the service life of the equipment; determine the dynamic weight by combining the basic weight, mode matching degree and health factor, so that the system can adjust the weight in real time according to the spatial position, control strategy and equipment health status, and dynamically adapt to the complex and changeable marine environment faced by offshore floating wind power platforms.

[0167] The dynamic weights are normalized so that their values ​​range from 0 to 1, which facilitates the subsequent weighted protocol generation and policy adjustment. The normalized dynamic weights are used to weight the consistency protocol of the cluster consistency coordination algorithm to obtain a weighted protocol. The weighted protocol is expressed as follows:

[0168] ui(k+1)=ui(k)+γ ∑{j∈Ni} wi [(uj(k) - ui(k))+α (xj(k) - xi(k)) ]

[0169] where ui(k) and ui(k+1) are the control instructions for the i-th damper at time k and k+1, respectively. These instructions are generated based on the adaptive mode switching strategy at the corresponding time, combined with the obtained damper working mode switching threshold and theoretical target output of each damper; γ is the consistency gain coefficient, which ranges from [0.1 to 1.0] and is used to control the convergence speed. It increases when the sea conditions are severe; wi is the normalized dynamic weight of the i-th damper, which is used to reflect the importance of inter-node collaboration; Ni is the neighbor set of the i-th damper; α is the position difference gain, which ranges from [0.2 to 0.8] and is used to enhance the spatial collaboration effect; xj(k) and xi(k) are the real-time position states of the j-th and i-th dampers at time k, respectively, including position, attitude, and vibration phase; and uj(k) is the control instruction for the j-th damper at time k.

[0170] Based on the weighted protocol and cluster consensus algorithm, the adaptive mode switching strategies of each damper are coordinated and adjusted: each damper sends the current control instruction ui(k) and position state xi(k) to the neighboring node, and the received information is weighted according to the dynamic weight wj of the neighboring node, and the information of the node with high weight accounts for a larger proportion; the control instruction is updated through the following steps: 1. Calculate the control instruction difference and position difference between the neighboring node and itself; 2. Multiply the weighted sum by the gain coefficient γ to obtain the adjustment amount; 3. Update the control instruction ui(k+1) at the next moment based on the adjustment amount; repeat the iterative update process of the control instruction until the control instructions of all dampers tend to be consistent, forming a collaborative action strategy, which is the intelligent optimization strategy and is sent to each damper for execution.

[0171] In an offshore floating wind power platform in a certain area, when the wave height in a certain area suddenly rises from 3m to 5m: the vibration control parameter triggers the threshold condition, and the original strategy plans to switch the dampers in this area to magnetorheological mode; the cluster consensus algorithm detects through a weighted protocol that the nodes with high health among the adjacent dampers have switched to magnetorheological mode (high mode matching), so the dynamic weight of the dampers in this area is increased; the weighted protocol forces the acceleration of the mode switching of the dampers in this area to ensure coordinated vibration suppression and avoid resonance caused by delayed switching of individual dampers. The present invention dynamically allocates the working mode of the damper according to vibration control parameters and environmental data, which can ensure that the damper is always in the optimal working state under different working conditions, give full play to its vibration control ability, thereby improving vibration control efficiency and reducing unnecessary energy consumption; through the adaptive mode switching strategy, the working mode of the damper can be adjusted in real time according to the vibration conditions and environmental changes, so that the system has stronger adaptability and can better cope with the complex and changeable marine environment faced by offshore floating wind power platforms; the cluster consistency algorithm is used to adjust the dynamic allocation results, which can coordinate the work between each damper, make the action of the entire damper cluster more consistent and coordinated, avoid the situation where the abnormal operation of individual dampers affects the overall performance, and thus optimize the overall performance of the cluster; the intelligent optimization strategy comprehensively considers multiple aspects such as vibration control, environmental adaptation and cluster coordination, and can enable the system to maintain stable and reliable operation under various working conditions, reduce the occurrence of system failures, and extend the service life of the equipment.

[0172] In the embodiment of the present application, based on the problem of how to effectively control the vibration of an offshore floating wind power platform, an intelligent optimization method for vibration coupling of an offshore floating wind power platform is designed. The method fully considers the influence of spatial, temporal and environmental factors on vibration through a pre-constructed space-time Kriging model, and can more accurately predict the wavefront vibration situation, providing a reliable basis for subsequent vibration control; by quantifying the cluster coordination degree of the dampers, and combining the prediction results to construct a multi-objective function, and using an intelligent algorithm to solve the optimal vibration control parameters, it is possible to achieve collaborative work between the dampers, optimize the control effect, and improve the efficiency and effect of vibration control; the working mode of the dampers is dynamically allocated according to the vibration control parameters, and is adjusted using a cluster consistency algorithm, so that the system can adapt to different environmental conditions and vibration conditions in real time, improve the stability and reliability of the system, and further improve the calculation efficiency and accuracy, thereby achieving effective control of the offshore floating wind power platform.

[0173] It should be noted that although the steps in the above flowchart are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders.

[0174] In another embodiment, Figure 2 As shown, the second aspect of the present invention provides an intelligent optimization system for vibration coupling of an offshore floating wind power platform, comprising:

[0175] A sample set construction module 10 is configured to obtain vibration coupling data between the offshore floating wind power platform and the damper and environmental data within a target range where the offshore floating wind power platform is located, and to construct a vibration sample set based on the vibration coupling data and the environmental data;

[0176] The model prediction module 20 is configured to extract first data of the vibration sample set in the spatial domain based on a pre-built spatiotemporal Kriging model, segment second data of the vibration sample set in the time domain, determine third data from the vibration sample set for environmental coupling modeling, and fuse the first data, the second data, and the third data to obtain fused data for wavefront prediction, thereby generating a prediction result.

[0177] a function solving module 30 for quantifying the cluster coordination degree of the dampers, constructing a multi-objective function based on the quantified cluster coordination degree and the prediction result, and solving the multi-objective function to obtain vibration control parameters during the vibration coupling intelligent optimization simulation of the offshore floating wind power platform;

[0178] The strategy execution module 40 is used to dynamically allocate the working mode of the damper according to the vibration control parameters, and adjust the dynamic allocation result using a cluster consistency algorithm to generate an intelligent optimization strategy for execution.

[0179] It should be noted that each module in the above-mentioned intelligent optimization system for vibration coupling of an offshore floating wind power platform can be implemented in whole or in part through software, hardware, and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned modules. For the specific definition of an intelligent optimization system for vibration coupling of an offshore floating wind power platform, please refer to the definition of an intelligent optimization method for vibration coupling of an offshore floating wind power platform above. The two have the same functions and effects and will not be repeated here.

[0180] In summary, the present invention relates to the field of offshore wind power technology, and discloses an intelligent optimization method and system for vibration coupling of an offshore floating wind power platform, which fuses the vibration coupling data and environmental data of the offshore floating wind power platform to construct a vibration sample set, and fuses the data in the time domain, spatial domain, and environmental coupling of the vibration sample set through a space-time Kriging model to perform wavefront prediction and generate prediction results; quantifies the cluster coordination degree of the damper, and constructs a multi-objective function based on the quantification result and the wavefront prediction result, so as to solve it in the vibration coupling intelligent optimization simulation process of the offshore floating wind power platform and obtain vibration control parameters; dynamically allocates the working mode of the damper according to the vibration control parameters, and uses the cluster consistency algorithm to adjust the dynamic allocation result to generate an intelligent optimization strategy for execution; realizes effective control of the offshore floating wind power platform, and can adapt to different environmental conditions and vibration conditions in real time.

[0181] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the various technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the various technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0182] The above-described embodiments merely represent several preferred implementations of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent. It should be noted that a person skilled in the art could make several improvements and substitutions without departing from the technical principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be based on the scope of protection of the claims.

Claims

1. An intelligent optimization method for vibration coupling of an offshore floating wind power platform, characterized in that: include: Acquire vibration coupling data between the offshore floating wind power platform and the damper and environmental data within a target range where the offshore floating wind power platform is located, and construct a vibration sample set based on the vibration coupling data and the environmental data; Based on a pre-built spatiotemporal Kriging model, first data of the vibration sample set in the spatial domain is extracted, second data of the vibration sample set in the time domain is segmented, third data for environmental coupling modeling is determined from the vibration sample set, and the first data, the second data, and the third data are fused to obtain fused data for wavefront prediction, so as to generate a prediction result; quantifying the cluster coordination degree of the dampers, constructing a multi-objective function based on the quantified cluster coordination degree and the prediction result, and solving the multi-objective function during the vibration coupling intelligent optimization simulation of the offshore floating wind power platform to obtain vibration control parameters; The quantifying the cluster coordination degree of the damper includes: quantifying the distance from the damper to the vibration hotspot and combining it with the coordination radius to calculate the spatial weight of the damper; Obtaining the actual output and theoretical target output of the damper to quantify the coordination deviation of the damper; quantifying the instantaneous degree of cooperation of the damper based on the spatial weight and the cooperation deviation; Performing time window integration and time smoothing processing on the instantaneous synergy to obtain the cluster synergy; The constructing of the multi-objective function includes: Determining input energy based on the prediction result, and quantifying dissipated energy according to the working state of the damper to determine energy dissipation efficiency with the input energy; Obtaining a stress spectrum of the damper and determining fatigue damage in combination with material fatigue performance data of the damper; Constructing the multi-objective function with the goal of minimizing the energy dissipation efficiency and the fatigue damage and maximizing the cluster coordination degree; The working mode of the damper is dynamically allocated according to the vibration control parameters, and the dynamic allocation result is adjusted using a cluster consistency algorithm to generate an intelligent optimization strategy for execution.

2. The intelligent optimization method for vibration coupling of an offshore floating wind power platform according to claim 1, characterized in that: The obtaining of vibration coupling data between the offshore floating wind power platform and the damper and environmental data within a target range of the offshore floating wind power platform, and constructing a vibration sample set based on the vibration coupling data and the environmental data, includes: Acquiring control parameters of the damper and structural parameters and vibration data of the offshore floating wind power platform as the vibration coupling data; Obtaining wind speed, wind direction, wave direction and wave height within a target range where the offshore floating wind power platform is located as the environmental data; The vibration coupling data and the environmental data are fused to construct a vibration sample set.

3. The intelligent optimization method for vibration coupling of an offshore floating wind power platform according to claim 2, characterized in that: The fusing the vibration coupling data and the environmental data to construct a vibration sample set includes: Processing the vibration coupling data and the environmental data by a dynamic time warping method and a spatial interpolation method to obtain aligned vibration data and aligned environmental data; extracting time domain features, frequency domain features, and time-frequency domain features from the aligned vibration data as vibration features; extracting statistical features, trend features, and spatial distribution features from the aligned environmental data as environmental features; Inputting the vibration characteristics and the environmental characteristics into a deep neural network model for processing and outputting a coupling index; the coupling index represents the degree of correlation between the vibration and the environment; Based on the coupling index, the vibration coupling data and the environmental data are fused to generate vibration samples, and after sample enhancement, a vibration sample set is formed.

4. The intelligent optimization method for vibration coupling of an offshore floating wind power platform according to claim 1, characterized in that: The kernel function of the space-time Kriging model includes a spatial kernel, a temporal kernel and an environmental coupling term; wherein, The training process of the space-time Kriging model includes: constructing the spatial kernel using a Gaussian kernel function and an exponential kernel function, constructing the temporal kernel using a rational quadratic kernel function, and constructing the environmental coupling term based on the environmental data; Constructing the space-time Kriging model according to the space kernel, the time kernel and the environmental coupling term; A space-time cube data matrix is ​​constructed to optimize the parameters of the space-time Kriging model by maximizing the marginal likelihood function to obtain a trained space-time Kriging model.

5. The intelligent optimization method for vibration coupling of an offshore floating wind power platform according to claim 1, characterized in that: Solving the multi-objective function to obtain vibration control parameters includes: The decision variables are encoded using real number encoding method, and a set of initial solutions are randomly generated as the initial population of the genetic algorithm; Performing selection, crossover, and mutation operations based on the initial population to construct a mixed population, quantifying the fitness value of each individual in the mixed population according to the multi-objective function, and performing non-dominated sorting and crowding calculation based on the quantified results to select a number of elite individuals from the mixed population as a new population; The new population is used as a particle swarm, so that the position of each particle corresponds to the code of the individual in the genetic algorithm, and the speed of each particle in the particle swarm is initialized to zero; updating the velocity of each particle according to the individual extreme value and the group extreme value corresponding to each particle, and updating the position of each particle according to the updated velocity to update the particle group; performing non-dominated sorting and crowding calculation on the updated particle swarm based on the multi-objective function, and selecting a number of particles as the swarm for the next iteration according to the calculation results; The generation process of the new population and the generation process of the next iterative population are iteratively executed based on the next iterative population until a preset number of iterations is reached, and a set of solutions is selected from the current population to be output as vibration control parameters.

6. The intelligent optimization method for vibration coupling of an offshore floating wind power platform according to claim 1, characterized in that: The dynamically allocating the working mode of the damper according to the vibration control parameters, and adjusting the dynamic allocation result by using a cluster consistency algorithm to generate an intelligent optimization strategy for execution, including: dynamically assigning the operating mode of the damper based on the vibration control parameters and the environmental data to obtain an adaptive mode switching strategy; quantifying the dynamic weight according to the adaptive mode switching strategy, and weighting the consistency protocol of the cluster consistency coordination algorithm by the dynamic weight to obtain a weighted protocol; The adaptive mode switching strategy is adjusted through the weighted protocol to generate an intelligent optimization strategy for execution.

7. The intelligent optimization method for vibration coupling of an offshore floating wind power platform according to claim 6, characterized in that: The quantizing the dynamic weight according to the adaptive mode switching strategy includes: determining a basic weight based on the distance between the dampers and the vibration control parameter; Determine the mode matching degree according to the adaptive mode switching strategy at the current moment; Obtaining a historical number of failures of the damper to determine a health factor; The dynamic weight is determined by the basic weight, the pattern matching degree and the health factor.

8. An intelligent optimization system for vibration coupling of an offshore floating wind power platform, characterized in that: include: a sample set construction module, configured to obtain vibration coupling data between the offshore floating wind power platform and the damper and environmental data within a target range where the offshore floating wind power platform is located, and to construct a vibration sample set based on the vibration coupling data and the environmental data; a model prediction module, configured to extract first data of the vibration sample set in the spatial domain, segment second data of the vibration sample set in the time domain, determine third data from the vibration sample set for environmental coupling modeling, and fuse the first data, the second data, and the third data to obtain fused data for wavefront prediction, thereby generating a prediction result; a function solving module, configured to quantify the cluster coordination degree of the dampers, construct a multi-objective function based on the quantified cluster coordination degree and the prediction result, and solve the multi-objective function during the vibration coupling intelligent optimization simulation of the offshore floating wind power platform to obtain vibration control parameters; The quantifying the cluster coordination degree of the damper includes: quantifying the distance from the damper to the vibration hotspot and combining it with the coordination radius to calculate the spatial weight of the damper; Obtaining the actual output and theoretical target output of the damper to quantify the coordination deviation of the damper; quantifying the instantaneous degree of cooperation of the damper based on the spatial weight and the cooperation deviation; Performing time window integration and time smoothing processing on the instantaneous synergy to obtain the cluster synergy; The constructing of the multi-objective function includes: Determining input energy based on the prediction result, and quantifying dissipated energy according to the working state of the damper to determine energy dissipation efficiency with the input energy; Obtaining a stress spectrum of the damper and determining fatigue damage in combination with material fatigue performance data of the damper; Constructing the multi-objective function with the goal of minimizing the energy dissipation efficiency and the fatigue damage and maximizing the cluster coordination degree; A strategy execution module is used to dynamically allocate the working mode of the damper according to the vibration control parameters, and use a cluster consistency algorithm to adjust the dynamic allocation result to generate an intelligent optimization strategy for execution.

Images