An online hydrological monitoring system and method for offshore wind farms based on wind power foundation fusion

Abstract

The application relates to the technical field of wind power monitoring schemes, and discloses an offshore wind farm hydrology online monitoring system and method based on wind power foundation fusion, which realizes two-way coupling of structure response and hydrology environment by collecting structure parameters and environmental hydrology parameters at each wind power foundation, combining forward calculation and reverse deduction, improves the spatial resolution and prediction accuracy of the whole-field hydrology parameter field, and outputs a graded early warning signal by comprehensively considering the structure response state and residual life, so that the early warning can reflect the dynamic safety margin of the foundation under the current degradation state, and the operation safety of the offshore wind farm under the complex marine environment and the scientificity of the maintenance decision are improved.

Description

Technical Field

[0001] This invention relates to the field of wind power monitoring technology, and in particular to an online hydrological monitoring system and method for offshore wind farms based on wind power infrastructure integration. Background Technology

[0002] Hydrological monitoring and structural safety monitoring of offshore wind farms are two crucial tasks for ensuring their safe operation. Currently, hydrological monitoring typically relies on a limited number of wave buoys, tide gauges, or remote sensing data, which can acquire environmental parameters with a certain level of accuracy. However, limitations in equipment cost and installation conditions result in a low density of monitoring points, leading to insufficient understanding of the spatial distribution of hydrological parameters across the entire wind farm, especially in the vicinity of each wind turbine foundation. Spatial resolution and temporal synchronization are also insufficient to meet the demands of refined monitoring. Structural safety monitoring, on the other hand, involves installing devices such as inclinometers on the foundation to collect foundation data for assessing its service condition.

[0003] However, in the existing monitoring system, hydrological data and structural data operate independently, failing to form effective information complementarity. This results in significant deficiencies in both the accuracy of hydrological monitoring and the effectiveness of structural safety early warning in complex or extreme marine environments, making it difficult to meet the needs of offshore wind farms for safe operation and intelligent maintenance. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide an online hydrological monitoring system and method for offshore wind farms based on wind power infrastructure integration, in order to solve the technical problem that the accuracy of hydrological monitoring and the effectiveness of structural safety early warning are significantly insufficient in complex or extreme marine environments, thus making it difficult to meet the needs of offshore wind farms for safe operation and intelligent maintenance.

[0005] The first aspect of this invention discloses an online hydrological monitoring system for offshore wind farms based on wind power foundation integration. The system includes a data acquisition module for collecting structural parameters and environmental hydrological parameters at the location of the wind power foundation.

[0006] The bidirectional calculation module includes a forward calculation unit and a reverse calculation unit; wherein, the forward calculation unit is used to calculate the current structural response state of the wind power foundation based on the structural parameters and environmental hydrological parameters, and to predict the remaining life of the wind power foundation based on the current structural response state.

[0007] The reverse calculation unit is used to reverse-engineer the local hydrological characteristics of the location of the wind power foundation based on the comprehensive structural response state of the wind power foundation; the comprehensive structural response state is determined based on the response state characterized by the structural parameters and the current structural response state.

[0008] The hydrological reconstruction module is used to perform spatiotemporal fusion of environmental hydrological parameters and local hydrological characteristics of each wind power foundation in the offshore wind farm, construct the hydrological parameter field of the wind farm based on the fusion results, and output the predicted values ​​of hydrological parameters based on the hydrological parameter field.

[0009] The joint early warning module is used to determine early warning signals based on the predicted values ​​of the hydrological parameters, the comprehensive structural response status of the wind power foundation, and the remaining lifespan.

[0010] Furthermore, the calculation process for the current structural response state specifically includes:

[0011] The environmental hydrological parameters and structural parameters are input into the forward extrapolation model, and the forward extrapolation model outputs the pile bending moment distribution and foundation yield depth of the wind power foundation as the current structural response state; wherein, the structural parameters include tilt angle, stress distribution and scour depth.

[0012] Furthermore, the prediction of the remaining life of the wind turbine foundation based on the current structural response state specifically includes:

[0013] The remaining fatigue life is predicted based on the bending moment distribution of the pile using the Miner linear accumulation method.

[0014] The remaining bearing capacity of the foundation is predicted based on the foundation yield depth using a gamma process.

[0015] The remaining life of the wind power foundation is obtained by simulating the impact of extreme load impacts on the fatigue remaining life and the foundation bearing capacity remaining life through the Poisson process.

[0016] Furthermore, the local hydrological characteristics include the dominant near-bottom flow direction and stability assessment value, significant wave height, and dominant wave direction; wherein, the process of determining the dominant near-bottom flow direction and stability assessment value specifically includes:

[0017] The spatial gradient of the scour depth is calculated based on the scour depth value in the integrated structural response state. The estimated value of the near-bottom dominant flow direction is determined based on the direction of the spatial gradient. The stability assessment value of the near-bottom dominant flow direction is determined by the gradient direction change rate in the time domain.

[0018] Furthermore, the process of determining the effective wave height and main wave direction specifically includes:

[0019] Based on the stress accumulation rate and pile bending moment distribution in the comprehensive structural response state, the effective wave height is inverted through the coupling relationship between the pre-calibrated stress, wave height, and bending moment.

[0020] The dominant wave direction is estimated in real time based on the changing trend of the tilt direction angle in the overall structural response state and the change in foundation stiffness reflected by the foundation yield depth using Kalman filtering.

[0021] Furthermore, the spatiotemporal fusion process in the construction of the hydrological parameter field includes spatial distribution fusion and temporal prediction fusion; wherein, the spatial distribution fusion process includes:

[0022] Using environmental hydrological parameters and local hydrological characteristics such as near-bottom dominant flow direction, significant wave height, and main wave direction as discrete measurement point data, an ensemble Kalman filter assimilation algorithm is used to map the discrete measurement point data to a spatial grid covering the entire wind farm, thereby obtaining the spatial distribution field of hydrological parameters at the current moment; wherein, the stability evaluation value is used to dynamically adjust the observation noise covariance matrix of the corresponding measurement point in the ensemble Kalman filter.

[0023] Furthermore, the temporal prediction fusion process includes:

[0024] The spatial distribution fields of hydrological parameters at multiple historical moments are used to construct a first time series. The first time series is then used to perform rolling predictions through a long short-term memory network to output the spatial prediction field of hydrological parameters at future moments. The input features of the long short-term memory network also include external covariates, which include tidal cycle prediction values ​​and meteorological forecast data.

[0025] Furthermore, the construction process of the hydrological parameter field also includes:

[0026] At each monitoring moment, the spatial distribution field of hydrological parameters at the current moment is taken as the latest element of the first time series, and the spatial prediction field of hydrological parameters at the next moment is obtained by rolling prediction through the long short-term memory network.

[0027] The consistency of the predicted hydrological parameter spatial field at the next moment with the actual environmental hydrological parameters collected by the data acquisition module at the next moment is verified. The assimilation weights of the discrete measurement point data in the ensemble Kalman filter are dynamically adjusted according to the verification error, and the hydrological parameter spatial distribution field at the next moment is updated, forming a closed-loop iterative hydrological parameter field construction process.

[0028] Furthermore, the process of determining the warning signal specifically includes:

[0029] The critical bearing capacity of the wind power foundation is determined based on the comprehensive structural response state. The critical bearing capacity is dynamically adjusted based on the remaining lifespan to obtain a dynamic safety threshold. The predicted values ​​of hydrological parameters are compared with the dynamic safety threshold. When the predicted values ​​of hydrological parameters exceed the dynamic safety threshold, a graded structural safety early warning signal is output.

[0030] The second aspect of this invention discloses an online hydrological monitoring method for offshore wind farms based on wind power infrastructure integration. This method is applied to the system disclosed in the first aspect, and includes:

[0031] Collect structural parameters and environmental hydrological parameters at the location of the wind turbine foundation;

[0032] The current structural response state of the wind power foundation is calculated based on the structural parameters and environmental hydrological parameters, and the remaining life of the wind power foundation is predicted based on the current structural response state.

[0033] The local hydrological characteristics of the wind power foundation's location are inferred from the comprehensive structural response state of the wind power foundation; the comprehensive structural response state is determined by combining the response state characterized by the structural parameters and the current structural response state.

[0034] The environmental hydrological parameters and local hydrological characteristics of each wind power foundation in the offshore wind farm are fused in time and space. Based on the fusion results, a hydrological parameter field of the wind farm is constructed, and the predicted values ​​of hydrological parameters are output based on the hydrological parameter field.

[0035] Based on the predicted hydrological parameters, the comprehensive structural response status of the wind power foundation, and the early warning signal for determining the remaining lifespan.

[0036] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0037] This invention collects structural parameters and environmental hydrological parameters at each wind turbine foundation, and combines forward and backward extrapolation to achieve bidirectional coupling between structural response and hydrological environment. This improves the spatial resolution and prediction accuracy of the hydrological parameter field across the entire site. Furthermore, it integrates structural response status and remaining lifespan to output graded early warning signals, enabling early warnings to reflect the dynamic safety margin of the foundation in its current degradation state. This enhances the operational safety of offshore wind farms in complex marine environments and the scientific basis of maintenance decisions. Attached Figure Description

[0038] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0039] Figure 1 This is a schematic diagram of the structure of an online hydrological monitoring system for offshore wind farms based on wind power infrastructure integration, as disclosed in Embodiment 1 of the present invention. Detailed Implementation

[0040] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0041] Example 1

[0042] The first aspect of this invention discloses an online hydrological monitoring system for offshore wind farms based on wind power infrastructure integration. Please refer to [link / reference]. Figure 1 , Figure 1 This is a schematic diagram of the structure of an online hydrological monitoring system for offshore wind farms based on wind power foundation integration, as disclosed in an embodiment of the present invention. The system includes a data acquisition module for collecting structural parameters and environmental hydrological parameters at the location of the wind power foundation.

[0043] The bidirectional calculation module includes a forward calculation unit and a reverse calculation unit; wherein, the forward calculation unit is used to calculate the current structural response state of the wind power foundation based on the structural parameters and environmental hydrological parameters, and to predict the remaining life of the wind power foundation based on the current structural response state.

[0044] The reverse calculation unit is used to reverse-engineer the local hydrological characteristics of the location of the wind power foundation based on the comprehensive structural response state of the wind power foundation; the comprehensive structural response state is determined based on the response state characterized by the structural parameters and the current structural response state.

[0045] The hydrological reconstruction module is used to perform spatiotemporal fusion of environmental hydrological parameters and local hydrological characteristics of each wind power foundation in the offshore wind farm, construct the hydrological parameter field of the wind farm based on the fusion results, and output the predicted values ​​of hydrological parameters based on the hydrological parameter field.

[0046] The joint early warning module is used to determine early warning signals based on the predicted values ​​of the hydrological parameters, the comprehensive structural response status of the wind power foundation, and the remaining lifespan.

[0047] The structural parameters include, but are not limited to, tilt angle, stress distribution, and scour depth.

[0048] Specifically, in this embodiment of the invention, the data acquisition module is deployed on each wind turbine foundation within the offshore wind farm. Each wind turbine foundation is equipped with at least one set of sensing nodes to collect structural parameters and environmental hydrological parameters at the foundation location in real time.

[0049] Structural parameters refer to raw data reflecting the mechanical state and geometric deformation of the foundation itself, obtained directly by sensors. Specifically, in this embodiment, structural parameters preferably include the foundation's tilt angle, stress distribution, and scour depth. The tilt angle can be obtained by single-axis or dual-axis tilt sensors installed at the top or multiple heights of the foundation to characterize the overall tilt direction and magnitude of the foundation. The stress distribution can be obtained by strain gauge stress sensors arranged in an array along the circumference and axial direction of the foundation to capture the stress response of the pile body under loads such as waves and ocean currents. The scour depth is calculated by the pressure difference at different heights of multiple pressure-type tide gauges arranged around the foundation to reflect the local scour pattern of the seabed around the foundation.

[0050] Furthermore, in this embodiment of the invention, environmental hydrological parameters mainly refer to static or quasi-static parameters that reflect the marine environmental state at the location of the foundation, obtained directly by sensors, such as water level, tide level, wave height, and wave period. These parameters are easy to measure directly and have strong spatial correlation at the wind farm scale, and can be provided by pressure tide gauges, wave buoys, or external meteorological and hydrological data sources.

[0051] In contrast to environmental hydrological parameters, in this invention, local hydrological characteristics refer to dynamic vector hydrological parameters that are difficult to measure directly on each wind turbine foundation using conventional sensors. These parameters include, but are not limited to, the dominant near-bottom flow direction and its stability assessment value, significant wave height, and dominant wave direction. These local hydrological characteristics characterize the refined flow field information of the interaction between the foundation and the surrounding water flow. They are characterized by dynamism, directionality, and uneven spatial distribution, and need to be obtained through back-calculation of the structural response.

[0052] Furthermore, the calculation process for the current structural response state specifically includes:

[0053] The environmental hydrological parameters and structural parameters are input into the forward calculation model, and the pile bending moment distribution and foundation yield depth of the wind power foundation are output by the forward calculation model as the current structural response state.

[0054] Preferably, the forward extrapolation model is set as a deep learning model based on a finite element proxy model to characterize the complex mapping relationship between environmental hydrological parameters, structural parameters, and foundation structural response. In the model building stage, a hydrological-structural coupled finite element model of the offshore wind power foundation is first established using numerical simulation software. This model covers physical mechanisms such as pile-soil interaction, wave load, ocean current load, and nonlinearity of foundation materials.

[0055] Then, using different environmental hydrological parameters and structural parameters as input variables, the distribution curves of pile bending moment along pile length and the yield depth of the foundation are calculated through a finite element model, generating a large number of training samples. The training dataset covers normal, extreme, and abnormal hydrological conditions, as well as structural responses under different scour degrees and tilt states, ensuring the model's generalization ability.

[0056] During model training, environmental hydrological parameters and structural parameters are used as input features, and pile bending moment distribution and foundation yield depth are used as output labels. A mean squared error loss function is employed for regression training. Furthermore, to improve the model's interpretability and extrapolation accuracy, this invention also employs physical information constraints, introducing a physical relationship term between bending moment and curvature into the loss function, ensuring that the model output conforms to the fundamental laws of structural mechanics.

[0057] The trained forward extrapolation model is deployed in the system's forward extrapolation unit. During actual operation, the data acquisition module acquires environmental hydrological and structural parameters of each wind turbine foundation in real time. After normalizing these data, they are input into the model. After forward propagation calculation, the model outputs the corresponding pile bending moment distribution and foundation yield depth. Because the model has learned the nonlinear mapping from environmental load and structural state to structural response, the forward extrapolation process can be completed quickly, significantly outperforming the computational efficiency of traditional finite element simulation. Furthermore, the model can integrate multi-source heterogeneous data. Even if some sensors experience temporary failures, the model can still provide a reasonable structural response estimate based on the remaining input parameters, thereby improving the system's real-time response capability in complex marine environments. Through these operations, the forward extrapolation unit achieves rapid and accurate extrapolation of the current structural response state of the wind turbine foundation, providing a reliable data foundation for remaining life prediction and joint early warning.

[0058] Furthermore, predicting the remaining life of a wind turbine foundation based on its current structural response state specifically includes:

[0059] The remaining fatigue life is predicted based on the bending moment distribution of the pile using the Miner linear accumulation method.

[0060] The remaining bearing capacity of the foundation is predicted based on the foundation yield depth using a gamma process.

[0061] The remaining life of the wind power foundation is obtained by simulating the impact of extreme load impacts on the fatigue remaining life and the foundation bearing capacity remaining life through the Poisson process.

[0062] First, it should be noted that in this embodiment of the invention, the current structural response state refers to the data output by the forward calculation unit, preferably the pile bending moment distribution and the foundation yield depth. It characterizes the mechanical state of the foundation under the current hydrological load and is the core basis for assessing whether the foundation has suffered immediate damage and the rate of cumulative damage. The comprehensive structural response state, on the other hand, is a more comprehensive state characterization obtained by further integrating structural parameters directly measured by the data acquisition module, based on the current structural response state. The comprehensive structural response state is mainly used for back-calculating local hydrological characteristics, because inferring hydrological information from the structural response requires simultaneously utilizing the model-calibrated response and the original sensor data to improve the accuracy of the inversion. It is understandable that the core input for remaining life prediction should be the current fatigue damage level and the degree of foundation degradation, which are fully reflected in the pile bending moment distribution and the foundation yield depth. Therefore, when performing remaining life prediction, this invention uses the current structural response state rather than the comprehensive structural response state, ensuring the clarity of the physical logic and avoiding data redundancy, thereby improving computational efficiency.

[0063] In the specific process of predicting the remaining life of wind power foundations, the fatigue remaining life and the foundation bearing capacity remaining life are predicted first, and then the comprehensive remaining life of the wind power foundation is determined based on these two data.

[0064] Specifically, the Miner linear accumulation method is used to predict the remaining fatigue life. This method calculates the stress amplitude of the pile section under each wave cycle based on the pile bending moment distribution, and then obtains the fatigue life consumption corresponding to this stress amplitude according to the material's SN curve. The fatigue damage caused by all historical and predicted wave cycles is linearly accumulated, and the fatigue life is considered to have ended when the accumulated damage reaches 1. The Miner linear accumulation method is simple to implement and has good engineering consistency with measured fatigue data, enabling it to quickly provide a basic estimate of the remaining fatigue life.

[0065] The gamma process is used to predict the remaining life of foundation bearing capacity. It is understandable that the yield depth of the foundation gradually increases with the repeated action of waves and ocean currents, as well as the plastic accumulation of the foundation soil; this is a monotonically increasing stochastic degradation process. The gamma process can effectively characterize this degradation trajectory with non-negative and independent increments. Using historical foundation yield depth monitoring data, the shape and scale parameters of the gamma process are fitted using maximum likelihood estimation, thereby predicting the remaining time required for the foundation yield depth to reach the critical threshold, which is the remaining life of the foundation bearing capacity. The gamma process naturally handles the randomness in degradation data, providing a probability distribution of the remaining life rather than a single deterministic value, thus improving the reliability of the prediction.

[0066] Furthermore, extreme load impacts (such as typhoon waves, earthquakes, or ship collisions) are low-probability but high-energy load events that significantly accelerate foundation fatigue damage and ground degradation. This embodiment employs a Poisson process to simulate the random arrival of such extreme loads. The Poisson process assumes that the frequency of extreme load events follows a Poisson distribution, with independent event intervals following an exponential distribution. The annual average occurrence rate of extreme loads is determined using historical disaster statistics. Then, within the framework of this Poisson process, each extreme load event, based on its intensity level, adds a corresponding increment to the current cumulative fatigue damage and foundation yield depth. This increment can be determined through finite element pre-calculation or empirical formulas. Monte Carlo simulation is then used to obtain the remaining life distribution that simultaneously considers normal cyclic loads and random extreme impacts. This coupling method makes the remaining life prediction results closer to the actual service environment, especially in typhoon-prone offshore wind farms. It effectively avoids optimistic estimates caused by ignoring extreme events, significantly improving the reliability and foresight of structural safety early warning. Finally, the smaller of the remaining fatigue life and the remaining foundation bearing capacity life is taken into account, or the overall remaining life of the wind power foundation is output using a joint probability method for use by the joint early warning module.

[0067] Furthermore, the local hydrological characteristics include the dominant near-bottom flow direction and stability assessment value, significant wave height, and dominant wave direction; wherein, the process of determining the dominant near-bottom flow direction and stability assessment value specifically includes:

[0068] The spatial gradient of the scour depth is calculated based on the scour depth value in the integrated structural response state. The estimated value of the near-bottom dominant flow direction is determined based on the direction of the spatial gradient. The stability assessment value of the near-bottom dominant flow direction is determined by the gradient direction change rate in the time domain.

[0069] Specifically, in the back-calculation unit, the estimation of the near-bottom dominant flow direction mainly relies on the scour depth value in the integrated structural response state. Since the scour pattern of the seabed around each wind turbine foundation is influenced by the near-bottom current, the spatial distribution of scour depth around the foundation usually exhibits a clear directionality, i.e., scour is heavier on the upstream side and lighter on the downstream side, and the contour lines of scour depth are elongated along the flow direction. Based on this physical law, this embodiment of the invention first acquires multiple scour depth sensors installed at different azimuth angles around the wind turbine foundation, forming a spatial discrete point set of scour depth around the foundation. The scour depth values ​​of these discrete points are mapped onto a polar coordinate grid centered on the foundation to obtain a continuous spatial distribution field of scour depth. Next, the spatial gradient of this distribution field is calculated, i.e., the rate of change of scour depth along the east-west and north-south directions on the horizontal plane. The magnitude of the gradient indicates the degree of most drastic change in scour depth, and the direction of the gradient points to the direction of the fastest increase in scour depth. Since the scour depth is greatest on the upstream side and decreases towards the downstream side, the opposite direction of the gradient direction is the direction of the incoming water flow. Therefore, rotating the direction of the spatial gradient by 180 degrees is used as an estimate of the dominant flow direction near the bottom.

[0070] Based on this, to further evaluate the reliability of the estimated flow direction, this embodiment of the invention introduces gradient direction change rate analysis in the time domain. Specifically, the system continuously records the time series of the estimated near-bottom dominant flow direction, calculates the change in flow direction angle between adjacent sampling times, and then obtains the gradient direction change rate. If the change rate is small and remains stable, it indicates that the near-bottom flow direction is relatively constant, and the stability assessment value is high, representing high reliability of the inverted flow direction. If the change rate fluctuates drastically, it indicates that the flow direction is unstable or the spatial distribution of scour depth is disturbed by other factors such as large-scale eddies or tidal turning periods, and the stability assessment value is low, representing significant uncertainty in the flow direction estimation result. The stability assessment value is quantified as a continuous value between 0 and 1, or divided into three levels: high, medium, and low. This stability assessment value is then passed to the hydrological reconstruction module to dynamically adjust the observation noise covariance matrix of the corresponding measuring points in the ensemble Kalman filter during the spatial fusion process.

[0071] Through the above operations, the present invention can adaptively suppress the negative impact of unstable flow direction inversion results on the reconstruction of the entire hydrological field, while retaining the effective information of high-confidence measuring points, thereby significantly improving the reconstruction accuracy and anti-interference ability of the hydrological parameter field, and is especially suitable for environmental conditions with rapid changes in flow direction such as tidal reciprocating flow and typhoon passage.

[0072] Furthermore, the process of determining the effective wave height and main wave direction specifically includes:

[0073] Based on the stress accumulation rate and pile bending moment distribution in the comprehensive structural response state, the effective wave height is inverted through the coupling relationship between the pre-calibrated stress, wave height, and bending moment.

[0074] The dominant wave direction is estimated in real time based on the changing trend of the tilt direction angle in the overall structural response state and the change in foundation stiffness reflected by the foundation yield depth using Kalman filtering.

[0075] Specifically, regarding the real-time estimation of the main wave direction, this embodiment of the invention employs Kalman filtering technology, using the trend of the tilt direction angle change in the overall structural response state as the observed value, and the change in foundation stiffness reflected by the foundation yield depth as part of the state transition. Specifically, the foundation will undergo periodic tilting under the action of external waves, and there is a clear phase relationship between the instantaneous change of its tilt direction angle and the wave propagation direction. That is, when the time history curve of the foundation tilt direction angle exhibits regular oscillations, the principal axis direction of the oscillation is usually perpendicular to the main wave direction or has a certain offset angle. This offset angle can be preliminarily estimated by performing principal component analysis or frequency domain analysis on the tilt direction angle time series.

[0076] The Kalman filter's prediction step utilizes the estimated wave direction angle from the previous moment and the current stiffness state of the foundation (the greater the yield depth, the lower the stiffness, and the greater the phase lag in the foundation's response to waves) to establish a state transition model and predict the wave direction angle at the current moment. The update step uses the measured trend of tilt direction angle changes (e.g., the direction of the principal axis of oscillation extracted from the time series collected by tilt sensors) as observations, combined with the observation noise covariance, and fuses the predicted and observed values ​​through Kalman gain weighting to obtain the optimal wave direction angle estimate. The foundation stiffness change reflected by the yield depth is used to adjust the parameters in the state transition matrix, enabling the filter model to adaptively track changes in the dynamic characteristics of the foundation caused by scour or foundation degradation, thus maintaining the accuracy of wave direction estimation at different service stages.

[0077] In this embodiment of the invention, by employing a recursive estimation method using Kalman filtering, the temporal dynamic information of the tilt direction angle and the information of foundation stiffness degradation are fully utilized, effectively suppressing the interference of high-frequency noise and short-term fluctuations on wave direction estimation. Even when the wave direction changes rapidly or the foundation structure undergoes gradual degradation, it can still maintain high estimation accuracy and stability.

[0078] Furthermore, the spatiotemporal fusion process in the construction of the hydrological parameter field includes spatial distribution fusion and temporal prediction fusion; wherein, the spatial distribution fusion process includes:

[0079] Using environmental hydrological parameters and local hydrological characteristics such as near-bottom dominant flow direction, significant wave height, and main wave direction as discrete measurement point data, an ensemble Kalman filter assimilation algorithm is used to map the discrete measurement point data to a spatial grid covering the entire wind farm, thereby obtaining the spatial distribution field of hydrological parameters at the current moment; wherein, the stability evaluation value is used to dynamically adjust the observation noise covariance matrix of the corresponding measurement point in the ensemble Kalman filter.

[0080] In this embodiment of the invention, the reason for using the stability assessment value to dynamically adjust the observation noise covariance matrix of the corresponding measuring point in the ensemble Kalman filter is that, during the ensemble Kalman filter assimilation process, the size of the observation noise covariance matrix determines the contribution weight of the measuring point's data to the final assimilation result. That is, the smaller the covariance, the more reliable the observation is considered, and the greater its weight in the assimilation; the larger the covariance, the lower the reliability of the observation, and its weight decreases accordingly. Since the local hydrological characteristics such as the near-bottom dominant flow direction, significant wave height, and main wave direction obtained by back-derived model are not directly measured values, but estimated values ​​obtained through structural response inversion, their estimation accuracy varies under different operating conditions. For example, under stable flow conditions with a relatively constant flow direction, the spatial gradient direction of the scour depth is clear, and the reliability of the inverted flow direction is high. At this time, the stability assessment value is high. By lowering the coefficient of the observation noise covariance matrix of the measuring point, the measuring point can play a greater role in the hydrological field reconstruction. During tidal transitions or in environments with strong turbulence, the direction of flow velocity changes rapidly, and the gradient direction of scour depth fluctuates dramatically, reducing the reliability of the inverted flow direction. In such cases, the stability assessment value is low. By increasing the coefficient of the observation noise covariance matrix at this measurement point, the negative impact of unreliable data on the overall assimilation results can be suppressed.

[0081] Furthermore, due to the complex influence of various factors such as the foundation shape, scour pattern, and tidal phase on the local flow field around the wind turbine foundation, the inversion results at a single moment inevitably contain errors or uncertainties. If all discrete measurement points are treated equally without distinguishing data quality, some unreliable inversion results may propagate to the entire grid through the assimilation process, leading to spurious structures or outliers in the subsequently reconstructed hydrological parameter field. By introducing a stability assessment value to dynamically adjust the observation noise covariance matrix, ensemble Kalman filtering can softly reduce the weight of unreliable measurement points, preserving potentially useful information in the data while effectively suppressing the interference of outliers. This maintains the continuity and physical rationality of the reconstructed hydrological field under various hydrological conditions.

[0082] Furthermore, the temporal prediction fusion process includes:

[0083] The spatial distribution fields of hydrological parameters at multiple historical moments are used to construct a first time series. The first time series is then used to perform rolling predictions through a long short-term memory network to output the spatial prediction field of hydrological parameters at future moments. The input features of the long short-term memory network also include external covariates, which include tidal cycle prediction values ​​and meteorological forecast data.

[0084] Furthermore, the construction process of the hydrological parameter field also includes:

[0085] At each monitoring moment, the spatial distribution field of hydrological parameters at the current moment is taken as the latest element of the first time series, and the spatial prediction field of hydrological parameters at the next moment is obtained by rolling prediction through the long short-term memory network.

[0086] The consistency of the predicted hydrological parameter spatial field at the next moment with the actual environmental hydrological parameters collected by the data acquisition module at the next moment is verified. The assimilation weights of the discrete measurement point data in the ensemble Kalman filter are dynamically adjusted according to the verification error, and the hydrological parameter spatial distribution field at the next moment is updated, forming a closed-loop iterative hydrological parameter field construction process.

[0087] It is understood that, in this embodiment of the invention, the spatial distribution field of hydrological parameters at the current moment represents the optimal estimate of the distribution of hydrological parameters across the entire wind farm at the current moment, but it can only reflect the instantaneous state and does not have predictive capabilities. The spatial prediction field of hydrological parameters for future moments is the result output after rolling prediction using a Long Short-Term Memory (LSTM) network, based on a time series composed of distribution sites from multiple historical moments, and has the ability to predict future changes in hydrological conditions. The final constructed hydrological parameter field organically unifies the two through a closed-loop iterative mechanism. That is, at each monitoring moment, the spatial distribution field of hydrological parameters at the current moment is input into the LSM network as the latest element of the time series to predict the spatial prediction field of hydrological parameters for the next moment. When the next moment is reached, the data acquisition module acquires the actual environmental hydrological parameters, then uses the actual values ​​to perform consistency verification on the prediction field, and dynamically adjusts the assimilation weights of discrete measurement point data in the ensemble Kalman filter according to the verification error, thereby generating an updated spatial distribution field of hydrological parameters for the next moment.

[0088] This process enables continuous mutual calibration between the predicted field and the measured distribution field. That is, the hydrological parameter field construction process of this invention introduces a dynamic correction operation of the prediction results on the spatial fusion weights. The prediction error is used to adjust the confidence of the observation data during the assimilation process, so that the spatial distribution field of the next round not only integrates the current measured data, but also remembers the prediction deviation of the previous round.

[0089] Through the above operations, the continuity of the hydrological parameter field in the time dimension and the interpolation accuracy in the spatial dimension of the entire field are significantly improved. Especially in the offshore wind farm environment where hydrological measuring points are sparse and data noise is large, it can effectively suppress the interference of measurement anomalies at a single moment on the reconstruction of the entire field. At the same time, the ability of the long short-term memory network to capture periodic hydrological changes is enhanced by using external covariates such as tidal cycle and weather forecast. Finally, a set of spatiotemporally continuous, self-consistent and dynamically adaptable hydrological parameter fields is output, providing a reliable data foundation for subsequent joint early warning.

[0090] Furthermore, the process of determining the warning signal specifically includes:

[0091] The critical bearing capacity of the wind power foundation is determined based on the comprehensive structural response state. The critical bearing capacity is dynamically adjusted based on the remaining lifespan to obtain a dynamic safety threshold. The predicted values ​​of hydrological parameters are compared with the dynamic safety threshold. When the predicted values ​​of hydrological parameters exceed the dynamic safety threshold, a graded structural safety early warning signal is output.

[0092] The aforementioned critical bearing capacity reflects the ultimate hydrological load that a wind turbine foundation can withstand without failure. Specifically, to compare this with predicted hydrological parameters (such as wave height and flow velocity), this invention uses a pre-defined hydrological and structural load transfer relationship to inversely convert this mechanical limit value into the ultimate hydrological parameter value that the foundation can withstand under its current condition—the dynamic safety threshold. For example, based on the foundation's pile cross-sectional characteristics, ground stiffness, and wave-foundation interaction model, it is calculated that when the effective wave height reaches a certain critical value, the foundation will generate a bending moment response equal to the critical bearing capacity. Therefore, the dynamic safety threshold is essentially a threshold value expressed in units of hydrological parameters, reflecting the foundation's current actual bearing capacity.

[0093] In the process of dynamically adjusting the critical bearing capacity based on the predicted remaining life, the longer the remaining life, the greater the safety margin of the foundation, and the more relaxed the dynamic safety threshold can be; conversely, the shorter the remaining life, the smaller the safety margin of the foundation, and the more tightened the dynamic safety threshold should be. This adjustment makes the safety threshold no longer a fixed constant, but a dynamic value that adapts to the aging process of the foundation.

[0094] Next, the predicted hydrological parameters are compared with the dynamic safety threshold. When the predicted value exceeds the threshold, a graded safety warning signal is output according to the degree of exceedance, such as a level one warning, a level two warning, or a level three warning, and it may be recommended to shut down the equipment or evacuate personnel.

[0095] This invention integrates comprehensive structural response status and remaining life information, enabling early warning to truly reflect the actual load-bearing capacity of wind power foundations during their current service phase. This avoids the problems of missed dangers due to excessively high thresholds or frequent false alarms due to excessively low thresholds, significantly improving the scientific nature and accuracy of early warning and providing effective decision support for the safe operation and maintenance of offshore wind farms in complex marine environments.

[0096] Example 2

[0097] The second aspect of this invention discloses an online hydrological monitoring method for offshore wind farms based on wind power infrastructure integration, the method comprising:

[0098] Collect structural parameters and environmental hydrological parameters at the location of the wind turbine foundation;

[0099] The current structural response state of the wind power foundation is calculated based on the structural parameters and environmental hydrological parameters, and the remaining life of the wind power foundation is predicted based on the current structural response state.

[0100] The local hydrological characteristics of the wind power foundation's location are inferred from the comprehensive structural response state of the wind power foundation; the comprehensive structural response state is determined by combining the response state characterized by the structural parameters and the current structural response state.

[0101] The environmental hydrological parameters and local hydrological characteristics of each wind power foundation in the offshore wind farm are fused in time and space. Based on the fusion results, a hydrological parameter field of the wind farm is constructed, and the predicted values ​​of hydrological parameters are output based on the hydrological parameter field.

[0102] Based on the predicted hydrological parameters, the comprehensive structural response status of the wind power foundation, and the early warning signal for determining the remaining lifespan.

[0103] It should be noted that the specific implementation process of Embodiment 2 is similar to that of Embodiment 1, and will not be repeated in this embodiment.

[0104] Finally, it should be noted that the above-described embodiments include multiple parallel implementations of the present invention. Deleting or otherwise adjusting one or more implementations will not affect the implementation of the solution. Furthermore, the hydrological online monitoring system and method for offshore wind farms based on wind power infrastructure integration disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An online hydrological monitoring system for offshore wind farms based on wind power foundation integration, comprising a data acquisition module for collecting structural parameters and environmental hydrological parameters at the location of the wind power foundation; characterized in that, The system also includes: The bidirectional calculation module includes a forward calculation unit and a reverse calculation unit; wherein, the forward calculation unit is used to calculate the current structural response state of the wind power foundation based on the structural parameters and environmental hydrological parameters, and to predict the remaining life of the wind power foundation based on the current structural response state. The reverse calculation unit is used to reverse-engineer the local hydrological characteristics of the location of the wind power foundation based on the comprehensive structural response state of the wind power foundation; the comprehensive structural response state is determined based on the response state characterized by the structural parameters and the current structural response state. The hydrological reconstruction module is used to perform spatiotemporal fusion of environmental hydrological parameters and local hydrological characteristics of each wind power foundation in the offshore wind farm, construct the hydrological parameter field of the wind farm based on the fusion results, and output the predicted values ​​of hydrological parameters based on the hydrological parameter field. The joint early warning module is used to determine early warning signals based on the predicted values ​​of the hydrological parameters, the comprehensive structural response status of the wind power foundation, and the remaining lifespan.

2. The offshore wind farm hydrological online monitoring system based on wind power infrastructure integration as described in claim 1, characterized in that, The calculation process for the current structural response state specifically includes: The environmental hydrological parameters and structural parameters are input into the forward extrapolation model, and the forward extrapolation model outputs the pile bending moment distribution and foundation yield depth of the wind power foundation as the current structural response state; wherein, the structural parameters include tilt angle, stress distribution and scour depth.

3. The offshore wind farm hydrological online monitoring system based on wind power infrastructure integration as described in claim 2, characterized in that, The prediction of the remaining life of the wind turbine foundation based on the current structural response state specifically includes: The remaining fatigue life is predicted based on the bending moment distribution of the pile using the Miner linear accumulation method. The remaining bearing capacity of the foundation is predicted based on the foundation yield depth using a gamma process. The remaining life of the wind power foundation is obtained by simulating the impact of extreme load impacts on the fatigue remaining life and the foundation bearing capacity remaining life through the Poisson process.

4. The offshore wind farm hydrological online monitoring system based on wind power infrastructure integration as described in claim 1, characterized in that, The local hydrological characteristics include the dominant near-bottom flow direction and stability assessment value, significant wave height, and dominant wave direction; wherein, the process of determining the dominant near-bottom flow direction and stability assessment value specifically includes: The spatial gradient of the scour depth is calculated based on the scour depth value in the integrated structural response state. The estimated value of the near-bottom dominant flow direction is determined based on the direction of the spatial gradient. The stability assessment value of the near-bottom dominant flow direction is determined by the gradient direction change rate in the time domain.

5. The offshore wind farm hydrological online monitoring system based on wind power infrastructure integration as described in claim 4, characterized in that, The process of determining the effective wave height and main wave direction specifically includes: Based on the stress accumulation rate and pile bending moment distribution in the comprehensive structural response state, the effective wave height is inverted through the coupling relationship between the pre-calibrated stress, wave height, and bending moment. The dominant wave direction is estimated in real time based on the changing trend of the tilt direction angle in the overall structural response state and the change in foundation stiffness reflected by the foundation yield depth using Kalman filtering.

6. The offshore wind farm hydrological online monitoring system based on wind power infrastructure integration as described in claim 1, characterized in that, The spatiotemporal fusion process in the construction of the hydrological parameter field includes spatial distribution fusion and temporal prediction fusion; wherein, the spatial distribution fusion process includes: Using environmental hydrological parameters and local hydrological characteristics such as near-bottom dominant flow direction, significant wave height, and main wave direction as discrete measurement point data, an ensemble Kalman filter assimilation algorithm is used to map the discrete measurement point data to a spatial grid covering the entire wind farm, thereby obtaining the spatial distribution field of hydrological parameters at the current moment; wherein, the stability evaluation value is used to dynamically adjust the observation noise covariance matrix of the corresponding measurement point in the ensemble Kalman filter.

7. The offshore wind farm hydrological online monitoring system based on wind power infrastructure integration as described in claim 6, characterized in that, The temporal prediction fusion process includes: The spatial distribution fields of hydrological parameters at multiple historical moments are used to construct a first time series. The first time series is then used to perform rolling predictions through a long short-term memory network to output the spatial prediction field of hydrological parameters at future moments. The input features of the long short-term memory network also include external covariates, which include tidal cycle prediction values ​​and meteorological forecast data.

8. The offshore wind farm hydrological online monitoring system based on wind power infrastructure integration as described in claim 7, characterized in that, The process of constructing the hydrological parameter field also includes: At each monitoring moment, the spatial distribution field of hydrological parameters at the current moment is taken as the latest element of the first time series, and the spatial prediction field of hydrological parameters at the next moment is obtained by rolling prediction through the long short-term memory network. The consistency of the predicted hydrological parameter spatial field at the next moment with the actual environmental hydrological parameters collected by the data acquisition module at the next moment is verified. The assimilation weights of the discrete measurement point data in the ensemble Kalman filter are dynamically adjusted according to the verification error, and the hydrological parameter spatial distribution field at the next moment is updated, forming a closed-loop iterative hydrological parameter field construction process.

9. The offshore wind farm hydrological online monitoring system based on wind power infrastructure integration as described in claim 1, characterized in that, The process of determining the warning signal specifically includes: The critical bearing capacity of the wind power foundation is determined based on the comprehensive structural response state. The critical bearing capacity is dynamically adjusted based on the remaining lifespan to obtain a dynamic safety threshold. The predicted values ​​of hydrological parameters are compared with the dynamic safety threshold. When the predicted values ​​of hydrological parameters exceed the dynamic safety threshold, a graded structural safety early warning signal is output.

10. A method for online hydrological monitoring of offshore wind farms based on wind power infrastructure integration, applied to the system described in any one of claims 1-9, the method comprising: Collect structural parameters and environmental hydrological parameters at the location of the wind turbine foundation; The method is characterized in that it further includes: The current structural response state of the wind power foundation is calculated based on the structural parameters and environmental hydrological parameters, and the remaining life of the wind power foundation is predicted based on the current structural response state. The local hydrological characteristics of the wind power foundation's location are inferred from the comprehensive structural response state of the wind power foundation; the comprehensive structural response state is determined by combining the response state characterized by the structural parameters and the current structural response state. The environmental hydrological parameters and local hydrological characteristics of each wind power foundation in the offshore wind farm are fused in time and space. Based on the fusion results, a hydrological parameter field of the wind farm is constructed, and the predicted values ​​of hydrological parameters are output based on the hydrological parameter field. Based on the predicted hydrological parameters, the comprehensive structural response status of the wind power foundation, and the early warning signal for determining the remaining lifespan.

Images