Method and device for monitoring unbalanced stress of offshore wind turbine generator and electronic equipment

Abstract

The invention relates to the technical field of offshore wind power, and discloses an offshore wind turbine generator stress imbalance monitoring method and device and electronic equipment, and the method comprises the steps: obtaining a multi-dimensional signal reflecting the real-time stress state of a wind turbine generator; performing feature extraction on the multi-dimensional signals to obtain multi-dimensional stress feature vectors, and fusing the multi-dimensional stress feature vectors to generate a fusion index representing the stress balance of the wind turbine generator; based on the size relation between the fusion index and a preset threshold range, the stress balance of the wind turbine generator is judged; the operation parameters of the wind turbine generator with unbalanced stress are adjusted in real time through model predictive control until the stress balance of the wind turbine generator is restored to a preset safety range, the response speed and regulation and control precision of the system to the unbalanced stress state under the complex sea condition are remarkably improved, the safety and stability of unit operation are effectively enhanced, and the method is suitable for large-scale popularization and application. The problem that the whole stress balance state of the wind turbine generator cannot be accurately quantified due to limitation of traditional single parameter monitoring is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of offshore wind power, in particular to a method and device for monitoring force imbalance of an offshore wind turbine and an electronic device. BACKGROUND

[0002] Offshore wind turbines are long-term operated in complex and changeable marine environment, and the stability and safety of the structural force state become an important issue to ensure the reliable operation of the equipment and prolong the service life. However, the current monitoring and control method for the structural force of offshore wind turbines still has significant deficiencies. Many traditional schemes rely on static design or single sensor data, which is difficult to adapt to the dynamic changes of offshore wind conditions, and lacks real-time and systematic regulation and control capabilities, resulting in equipment failure or even damage under extreme conditions.

[0003] The limitations of the existing method mainly manifest as: the response lag to force imbalance and the singleness of the control means. Some monitoring technologies only stay at the data collection level and fail to effectively integrate analysis and feedback mechanisms, making it difficult to identify and resolve potential risks in a timely manner. In addition, the lack of early warning mechanisms or simplification also limits the system's ability to flexibly respond to different degrees of force abnormalities. These defects are particularly prominent when facing sudden strong wind and wave loads, directly threatening the safe operation of the wind turbine. SUMMARY

[0004] The present application provides a method and device for monitoring force imbalance of an offshore wind turbine and an electronic device to solve the limitations of traditional single parameter monitoring, which cannot accurately quantify the overall force balance state of the wind turbine.

[0005] In a first aspect, the present application provides a method for monitoring force imbalance of an offshore wind turbine, which comprises: obtaining a multi-dimensional signal reflecting the real-time force state of the wind turbine; extracting features from the multi-dimensional signal to obtain a multi-dimensional force feature vector, and fusing the multi-dimensional force feature vector to generate a fusion index representing the force balance of the wind turbine; judging the force balance of the wind turbine based on the size relationship between the fusion index and the preset threshold range; using model predictive control to adjust the operating parameters of the wind turbine with force imbalance in real time until the force balance of the wind turbine is restored to the preset safety range.

[0006] This invention provides a method for monitoring the stress imbalance of offshore wind turbines. By collecting multi-dimensional stress signals in real time and fusing them to generate a unified balance index, it achieves accurate quantitative perception and real-time judgment of the stress state of the offshore wind turbine structure. Furthermore, through model predictive control algorithms, it dynamically adjusts operating parameters based on index deviations, forming a closed-loop control of "monitoring-judgment-regulation". This significantly improves the system's response speed and control accuracy to stress imbalance under complex sea conditions, effectively enhancing the safety and stability of turbine operation. It solves the limitations of traditional single-parameter monitoring, which cannot achieve accurate quantification of the overall stress balance state of the wind turbine.

[0007] In one optional implementation, acquiring a multi-dimensional signal reflecting the real-time stress state of the wind turbine includes: By deploying multi-source sensors on multiple target components of the wind turbine, stress signals, vibration signals, and displacement signals of each target component are collected to obtain multi-dimensional signals reflecting the real-time stress state of the wind turbine.

[0008] This invention provides a method for monitoring the stress imbalance of offshore wind turbines. By deploying multi-source sensors on multiple core components of the turbine to synchronously collect stress, vibration, and displacement signals, a multi-dimensional data foundation that comprehensively reflects the real-time stress state of the structure is constructed. This overcomes the limitations of traditional single-parameter monitoring and provides key and reliable data input for subsequent accurate stress balance analysis and judgment.

[0009] In one optional implementation, feature extraction is performed on the multidimensional signal to obtain a multidimensional force feature vector, and the multidimensional force feature vector is fused to generate a fusion index characterizing the force balance of the wind turbine, including: A signal decomposition algorithm is used to denoise and extract features from the multidimensional signal to obtain multiple independent force feature vectors. After parameter standardization of the multiple independent force feature vectors, multiple parameter-standardized independent force feature vectors are obtained. A pre-defined multi-source data fusion model is used to weight and fuse multiple independent force feature vectors after parameter standardization to generate a fusion index characterizing the force balance of the wind turbine.

[0010] This invention provides a method for monitoring the force imbalance of offshore wind turbines. By denoising, extracting features and standardizing multidimensional raw signals, and generating a unified fusion index using a weighted fusion model, it achieves a precise transformation of the complex force state of offshore wind turbines from multidimensional and multi-component to a single quantitative representation, significantly improving the accuracy and comprehensiveness of force balance judgment.

[0011] In one optional implementation, a pre-defined multi-source data fusion model is used to weight and fuse multiple independent force feature vectors after parameter standardization to generate a fusion index characterizing the force balance of the wind turbine, including: Assign component weights to each target component based on its importance. Based on the importance of each independent force feature vector within each target, parameter weights are assigned to each independent force feature vector; Based on component weights and parameter weights, a pre-set multi-source data fusion model is used to perform weighted fusion calculations on all force feature vectors to generate a fusion index characterizing the force balance of the wind turbine.

[0012] This invention provides a method for monitoring the force imbalance of offshore wind turbines. By introducing a two-level weighted fusion mechanism based on the importance of components and parameters, it achieves the scientific integration of force data from different sources and with different physical meanings under a unified quantitative framework.

[0013] In one optional implementation, based on component weights and parameter weights, a preset multi-source data fusion model is used to perform weighted fusion calculations on all force feature vectors to generate a fusion index characterizing the force balance of the wind turbine, including: Based on parameter weights, a pre-set multi-source data fusion model is used to perform weighted fusion calculations on each force feature vector within each target component to obtain the comprehensive state value of all feature vectors of each component. Based on component weights, a pre-set multi-source data fusion model is used to perform weighted fusion calculations on the comprehensive state values ​​of all independent force characteristic vectors of each component, thereby obtaining a fusion index characterizing the force balance of the wind turbine.

[0014] The present invention provides a method for monitoring the force imbalance of offshore wind turbines. By adopting a two-level structured weighted calculation that first integrates the internal parameters of components and then integrates the components of the whole machine, the fused index can not only accurately reflect the internal state of each component, but also scientifically integrate the force balance of the whole machine.

[0015] In one optional implementation, the force balance of the wind turbine is determined based on the relationship between the fusion index and a preset threshold range, including: If the value of the fusion index is within the preset threshold range, it is determined that the wind turbine does not have a force imbalance. If the value of the fusion index exceeds the preset threshold range, it is determined that the wind turbine has a force imbalance.

[0016] This invention provides a method for monitoring the force imbalance of offshore wind turbines. By setting clear and quantified threshold ranges and directly comparing fusion indicators, it achieves rapid, objective, and unambiguous automatic judgment of the force balance state of wind turbines, providing a clear and reliable decision-making basis for subsequent precise control.

[0017] In one optional implementation, model predictive control is used to adjust the operating parameters of the wind turbine with stress imbalance in real time until the stress balance of the wind turbine is restored to a preset safe range, including: Obtain the current independent force feature vector in the fusion index, and based on the current independent force feature vector and the corresponding historical independent force feature vector, determine the degree and duration of force imbalance. Analysis of the changing trend of force imbalance in wind turbines based on the degree and duration of anomalies; Based on the changing trend, model predictive control is used to dynamically optimize the operating parameters of the wind turbine under preset constraints, and the operating status of the wind turbine is adjusted based on the dynamically optimized operating parameters until the force balance of the wind turbine is restored to the preset safe range.

[0018] This invention provides a method for monitoring the force imbalance of offshore wind turbines. By integrating trend analysis (degree of abnormality and duration) with model predictive control dynamic optimization, it realizes intelligent feedforward adjustment of the force imbalance state, significantly improves the control response speed and parameter adjustment accuracy, thereby ensuring that the wind turbine can safely and efficiently return to a stable equilibrium state.

[0019] In one alternative implementation, the method further includes: When the dynamically optimized wind turbine parameter set fails to restore the wind turbine's force balance to the preset safe range, the operating parameters of the dynamically optimized wind turbine are analyzed based on a preset deep learning model to determine the optimized values ​​of the wind turbine's operating parameters. The optimal operating parameters of the wind turbine are iteratively calculated using a preset optimization algorithm to obtain the optimal operating parameters of the wind turbine, and the operating status of the wind turbine is updated based on the optimal operating parameters.

[0020] This invention provides a method for monitoring the force imbalance of offshore wind turbines. When the initial control fails to achieve the expected results, by introducing a deep learning-based secondary diagnosis and global optimization algorithm, the system can escape local optima and make more refined and intelligent optimization adjustments to the operating parameters. This not only significantly enhances the system's adaptive control capability and problem-solving depth under extreme or complex sea conditions, but also ensures that the turbine can eventually return to a stable and efficient operating state through a closed-loop iterative mechanism, thereby comprehensively improving the robustness and reliability of the entire monitoring and control system.

[0021] Secondly, the present invention provides a device for monitoring the force imbalance of offshore wind turbines, the device comprising: The multi-dimensional signal acquisition module is used to acquire multi-dimensional signals that reflect the real-time stress state of the wind turbine. The fusion index generation module is used to extract features from multi-dimensional signals to obtain multi-dimensional force feature vectors, and to fuse the multi-dimensional force feature vectors to generate fusion indexes that characterize the force balance of wind turbine units. The force balance judgment module is used to judge the force balance of the wind turbine based on the relationship between the fusion index and the preset threshold range. The predictive control module is used to adjust the operating parameters of wind turbines with force imbalance in real time using model predictive control until the force balance of the wind turbine is restored to the preset safe range.

[0022] Thirdly, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the offshore wind turbine force imbalance monitoring method described in the first aspect or any corresponding embodiment.

[0023] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the offshore wind turbine force imbalance monitoring method of the first aspect or any corresponding embodiment described above.

[0024] Fifthly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute the offshore wind turbine force imbalance monitoring method described in the first aspect or any corresponding embodiment. Attached Figure Description

[0025] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0026] Figure 1 This is a schematic diagram of an application scenario according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the first process of the method for monitoring the force imbalance of offshore wind turbines according to an embodiment of the present invention; Figure 3This is a schematic diagram of the second process of the method for monitoring the force imbalance of offshore wind turbines according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the third process of the method for monitoring the force imbalance of offshore wind turbines according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the fourth process of the offshore wind turbine force imbalance monitoring method according to an embodiment of the present invention; Figure 6 This is a structural block diagram of a force imbalance monitoring device for offshore wind turbines according to an embodiment of the present invention; Figure 7 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0029] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0030] As an optional application scenario of this invention, such as Figure 1 As shown, application 101 is installed in terminal device 110, and user 130 can interact with application 101 through terminal device 110 and / or access device of terminal device 110.

[0031] For example, application 101 can be any application that provides question-and-answer related services. For instance, application 101 could be a question-and-answer interactive application, such as a text-to-text application, an image-to-text application, etc. Figure 1In the application scenario shown, if application 101 is active, the terminal device 110 can display the interface 102 of application 101. The interface 102 may include various pages that application 101 can provide, such as interactive pages, settings pages, query pages, etc.

[0032] In some embodiments, terminal device 110 is communicatively connected to server 120 to provide services to application 101. Terminal device 110 may be a mobile terminal, fixed terminal, or portable terminal, etc., including but not limited to mobile phones, desktop computers, laptop computers, multimedia tablets, e-book devices, gaming devices, or any combination thereof, including accessories and peripherals of these devices or any combination thereof. In some embodiments, terminal device 110 may also support any type of interface, and server 120 may be various types of computing systems or servers capable of providing computing power, including but not limited to mainframes, edge computing nodes, computing devices in cloud environments, etc.

[0033] It should be noted that, Figure 1 This is merely an example of an application scenario and does not limit the scope of protection of this invention.

[0034] The embodiments of the present invention will now be described with reference to the accompanying drawings. It should be understood that the pages shown in the drawings are merely examples, and various page designs are possible in practice. The various graphic elements on the page may have different arrangements and different visual representations; one or more elements may be omitted or replaced, and one or more other elements may also be present, without any limitation in the embodiments of the present invention. Furthermore, the embodiments described below primarily pertain to terminal device 110. It should be understood that the actions described relative to terminal device 110 can be performed by application 101 on terminal device 110, or can be performed by application 101 in conjunction with its server (e.g., server 120).

[0035] The core challenges in this field lie in how to achieve real-time analysis of stress data, accurate detection of stress imbalances, and effective execution of safety control mechanisms. Real-time stress data requires high-frequency acquisition and processing capabilities, while imbalance detection relies on the fusion and accurate identification of multi-source data. The effectiveness of the control mechanism involves the dynamic optimization and adjustment of operating parameters. Because these technical factors have not yet been fully resolved, wind turbines in complex sea conditions are prone to structural fatigue due to the accumulation of abnormal stresses, potentially leading to significant safety hazards and posing a major technical challenge. Therefore, how to accurately detect imbalances by analyzing the stress data of various components of a wind turbine in real time and automatically adjust operating parameters to ensure safe operation has become a critical issue that urgently needs to be addressed.

[0036] This invention provides a method for monitoring the force imbalance of offshore wind turbines. By using multi-dimensional signals to continuously monitor and optimize the force balance of wind turbines, the method effectively improves the force stability and operating efficiency of wind turbines under dynamic sea conditions.

[0037] According to an embodiment of the present invention, an embodiment of a method for monitoring the force imbalance of an offshore wind turbine is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0038] This embodiment provides a method for monitoring the force imbalance of offshore wind turbines, which can be used in the aforementioned electronic equipment or terminal equipment. Figure 2 This is a flowchart of a method for monitoring the force imbalance of offshore wind turbines according to an embodiment of the present invention, as shown below. Figure 2 As shown, the process includes the following steps: Step S201: Obtain multi-dimensional signals reflecting the real-time stress state of the wind turbine.

[0039] Specifically, wind turbines include core components such as blades, towers, and gearboxes. Stress signals, vibration signals, and displacement signals of these core components are collected to form multidimensional signals.

[0040] Step S202: Extract features from the multidimensional signal to obtain multidimensional force feature vectors, and fuse the multidimensional force feature vectors to generate a fusion index characterizing the force balance of the wind turbine.

[0041] Specifically, the multidimensional signal is denoised and feature vectors are extracted to determine the independent components of the stress changes in each component. A multi-source data fusion model is used to integrate the feature vectors with weights of 50% for the blades, 30% for the tower, and 20% for the gearbox, generating a fusion index. For example, given the inherent axial stress of the blade (100 MPa), raw data of different magnitudes (such as stress in MPa, vibration in mm / s², and displacement in mm) are uniformly mapped to the 0-1 range to obtain standardized parameter values. These standardized parameter values ​​are then used as input in the subsequent calculation of the fusion index, which is calculated as "component weight × Σ(parameter weight × standardized parameter value)," ensuring that parameters with different units and numerical ranges can be weighted and integrated, avoiding distortion of the fusion results due to differences in magnitude.

[0042] Step S203: Based on the relationship between the fusion index and the preset threshold range, determine the force balance of the wind turbine.

[0043] Specifically, this step is to determine whether the wind turbine has an unbalanced state that deviates from the normal range: if the fusion index is within the preset threshold range, it is determined that the wind turbine does not have a force imbalance; if it is not within the preset threshold range, it is determined that the wind turbine has a force imbalance.

[0044] Step S204: Model predictive control is used to adjust the operating parameters of the wind turbine in real time if there is a force imbalance, until the force balance of the wind turbine is restored to the preset safe range.

[0045] Specifically, model predictive control is adopted to optimize and adjust the rotational speed and blade angle in real time, generate an optimized parameter set (i.e., the wind turbine operating parameter set), and drive actuators such as pitch and speed control to adjust the operating state, so that the force balance is restored to the preset safe range.

[0046] The offshore wind turbine stress imbalance monitoring method provided in this embodiment innovatively adopts a multi-source sensor collaborative acquisition and weighted fusion strategy at the sensing level: by deploying stress, vibration, and displacement sensors to cover the core components, and combining signal denoising and feature extraction techniques to separate the independent stress components of each component, a fusion index is generated according to the safety priority weights of 50% for blades, 30% for the tower, and 20% for the gearbox. This overcomes the limitations of traditional single-parameter monitoring and achieves accurate quantification of the overall stress balance state. At the control and optimization level, a model predictive control algorithm is used to optimize the rotational speed and blade angle in real time to quickly respond to stress imbalances.

[0047] This embodiment provides a method for monitoring the force imbalance of offshore wind turbines, which can be used in the aforementioned electronic equipment or terminal equipment. Figure 3 This is a flowchart of a method for monitoring the force imbalance of offshore wind turbines according to an embodiment of the present invention, as shown below. Figure 3 As shown, the process includes the following steps: Step S301: Obtain multi-dimensional signals reflecting the real-time stress state of the wind turbine.

[0048] Specifically, step S301 includes: Step a: By deploying multi-source sensors on multiple target components of the wind turbine, stress signals, vibration signals, and displacement signals of each target component are collected to obtain multi-dimensional signals reflecting the real-time stress state of the wind turbine.

[0049] Multi-source sensors (strain gauges for stress measurement, accelerometers for vibration measurement, and displacement sensors for displacement measurement) are deployed for each core component of the wind turbine (such as blades, towers, and gearboxes). Force data is collected at high frequency (e.g., 1000 times per second) and transmitted to a processing unit (a core computing device or system deployed on or near the offshore wind turbine for centralized reception, processing, analysis, and decision-making; its physical entity can be an electronic device, such as an industrial-grade computer). The raw dataset reflecting the real-time stress state is obtained, resulting in a multi-dimensional signal containing stress, vibration, and displacement.

[0050] Step S302: Extract features from the multidimensional signal to obtain a multidimensional force feature vector, and fuse the multidimensional force feature vector to generate a fusion index characterizing the force balance of the wind turbine.

[0051] Specifically, step S302 includes: Step S3021: The signal decomposition algorithm is used to denoise and extract features from the multidimensional signal to obtain multiple independent force feature vectors. After parameter standardization of the multiple independent force feature vectors, multiple parameter-standardized independent force feature vectors are obtained.

[0052] Based on the original dataset of the multidimensional signal, signal decomposition algorithms (such as Empirical Mode Decomposition (EMD) and Independent Component Analysis (ICA) are used to denoise (filter out interference from ocean waves, electromagnetic fields, etc.) and extract features from the multidimensional signal to obtain the separated force feature vectors (such as vibration frequency, peak stress; axial force, radial force, axial vibration, and radial vibration). Independent components reflecting the force changes of each component (blade, tower, gearbox, etc.) are then determined (such as the axial force of the blade and the radial vibration of the tower).

[0053] Parameter standardization: Calculate the ratio between measured values ​​and inherent values ​​to map the raw data of different components at different magnitudes (units, numerical ranges) to a unified, standardized numerical range. The inherent value refers to the design rated stress value of each component of the wind turbine model (blade axial stress inherent value 100MPa, tower radial vibration inherent value 5mm / s², determined according to GB / T19073-2008 "Wind Turbine Generator Gearbox" and the manufacturer's design manual). The data is standardized to the 0-1 range by "measured value ÷ inherent value".

[0054] For example, the specific data items and standardized calculation methods corresponding to the secondary parameters of the blade are as follows: the stress parameter (50% weight) corresponds to the peak stress of the blade's axial force, and its standardized value is the measured value of the peak stress ÷ 100 MPa; the vibration parameter (30% weight) corresponds to the vibration frequency of the blade's axial vibration, and its standardized value is the measured value of the vibration frequency ÷ the blade's inherent vibration value; the displacement parameter (20% weight) corresponds to the peak displacement of the blade, and its standardized value is the measured value of the peak displacement ÷ the blade's inherent displacement value. When calculating the fusion index, the blade part first calculates the sum of "parameter weight × parameter standardized value" for each secondary parameter, then multiplies it by the 50% component weight of the blade, and subsequently adds the contributions of the tower and gearbox (the logic is the same as the blade, only the parameter items, weights, and inherent values ​​are different), finally mapping it to the 0-100 range. Simultaneously, each component contains 3 secondary parameters, corresponding to 3 parameter standardized values, for a total of 9 parameter standardized values ​​participating in the subsequent fusion calculation.

[0055] The calculation of standardized values ​​for blade parameters needs to be clarified as follows: Among the secondary parameters of the blade, the stress parameter has a weight of 50% (i.e., 0.5). The corresponding measured data item is the peak stress value under axial force on the blade, not a product of "0.5 (axial force × a certain value)". The calculation logic for the standardized value is "measured value of the parameter ÷ corresponding inherent value". The inherent value of the blade's axial stress is explicitly defined as 100 MPa. For example, if the measured peak stress value under axial force on the blade is 50 MPa, its standardized value is 50 ÷ 100 = 0.5. Then, multiplying this standardized value by the stress parameter weight of 0.5 yields the contribution of the stress parameter to the weighted sum of the blade's internal parameters. Similarly, other blade parameters (vibration, displacement) are first standardized to the 0-1 range by "measured value ÷ corresponding inherent value" (the inherent values ​​of vibration and displacement need to be determined according to GB / T19073-2008 and the manufacturer's design manual), and then multiplied by 30% and 20% of the parameter weights respectively. Finally, the weighted sum of all secondary parameters of the blade is multiplied by 50% of the component weight and used to participate in the calculation of the overall fusion index.

[0056] Step S3022: Using a preset multi-source data fusion model, multiple independent force feature vectors after parameter standardization are weighted and fused to generate a fusion index characterizing the force balance of the wind turbine.

[0057] By using a pre-defined multi-source data fusion model, the force feature vectors are weighted and integrated (e.g., weights are assigned according to the importance of components: blades 50%, tower 30%, gearbox 20%), to obtain a fusion index that characterizes the overall force balance.

[0058] In some optional implementations, step S3022 above includes: Step a1: Assign component weights to each target component based on its importance.

[0059] Step a2: Assign parameter weights to each independent force feature vector based on the importance of each independent force feature vector within each target.

[0060] Step a3: Based on component weights and parameter weights, a preset multi-source data fusion model is used to perform weighted fusion calculations on all force feature vectors to generate a fusion index characterizing the force balance of the wind turbine.

[0061] In some alternative implementations, step a3 above includes: Step a31: Based on parameter weights, a preset multi-source data fusion model is used to perform weighted fusion calculations on each force feature vector within each target component to obtain the comprehensive state value of all feature vectors of each component.

[0062] Step a32: Based on component weights, a pre-set multi-source data fusion model is used to perform weighted fusion calculation on the comprehensive state values ​​of all independent force characteristic vectors of each component to obtain a fusion index characterizing the force balance of the wind turbine.

[0063] For example: using a two-level fusion of 'component weights + parameter weights': Weighting of primary components: blades 50%, tower 30%, gearbox 20%; Secondary parameter weights (within each component): Blade: Stress parameter 50% × axial force × peak stress, vibration parameter 30% × axial vibration / vibration frequency, displacement parameter 20% × peak displacement; Tower: Vibration parameters 60%, stress parameters 30%, displacement parameters 10%; Gearbox: Displacement parameter 40%, vibration parameter 35%, stress parameter 25%; The fusion index = Σ[component weight × Σ(parameter weight × parameter standardized value)], which is ultimately mapped to the 0-100 range.

[0064] Among them, the parameter standardization value refers to the measured value of a parameter divided by the inherent value of the parameter (the inherent value is the design rated stress value of the corresponding model of wind turbine component, such as the inherent value of blade axial stress of 100MPa, and the inherent values ​​of other parameters are determined according to GB / T19073-2008 "Wind Turbine Gearbox" and the manufacturer's design manual), and the result is mapped to the 0-1 range.

[0065] Step S303: Based on the relationship between the fusion index and the preset threshold range, determine the force balance of the wind turbine. For details, please refer to [link to relevant documentation]. Figure 2 Step S203 of the illustrated embodiment will not be described again here.

[0066] Step S304: Model predictive control is used to adjust the operating parameters of the wind turbine with stress imbalance in real time until the stress balance of the wind turbine is restored to the preset safe range. For details, please refer to [link to relevant documentation]. Figure 2 Step S204 of the illustrated embodiment will not be described again here.

[0067] The offshore wind turbine stress imbalance monitoring method provided in this embodiment achieves a precise transformation of the complex stress state of offshore wind turbines from multi-dimensional and multi-component to a single quantitative representation by denoising, feature extraction and standardization of multi-dimensional raw signals and generating a unified fusion index using a weighted fusion model. This significantly improves the accuracy and comprehensiveness of stress balance judgment.

[0068] This embodiment provides a method for monitoring the force imbalance of offshore wind turbines, which can be used in the aforementioned electronic equipment or terminal equipment. Figure 4 This is a flowchart of a method for monitoring the force imbalance of offshore wind turbines according to an embodiment of the present invention, as shown below. Figure 4 As shown, the process includes the following steps: Step S401: Obtain multi-dimensional signals reflecting the real-time stress state of the wind turbine. For details, please refer to [link to relevant documentation]. Figure 3 Step S301 of the illustrated embodiment will not be described again here.

[0069] Step S402 involves extracting features from the multidimensional signal to obtain a multidimensional force feature vector, and then fusing these vectors to generate a fusion index characterizing the force balance of the wind turbine. For details, please refer to [link to relevant documentation]. Figure 3 Step S302 of the illustrated embodiment will not be described again here.

[0070] Step S403: Based on the relationship between the fusion index and the preset threshold range, determine the force balance of the wind turbine.

[0071] Specifically, step S403 includes: Step b: If the value of the fusion index is within the preset threshold range, it is determined that the wind turbine unit is not under stress imbalance; if the value of the fusion index exceeds the preset threshold range, it is determined that the wind turbine unit is under stress imbalance.

[0072] For example, in this embodiment, the preset threshold range of the fusion index is consistent with the preset safety range, both being 70-90. If the fusion index value is within 70-90, it is determined that the wind turbine is not under stress imbalance and monitoring continues. If the fusion index value is <70 or the fusion index value is >90, it is determined that the wind turbine is under stress imbalance and subsequent control procedures are triggered.

[0073] Step S404: Model predictive control is used to adjust the operating parameters of the wind turbine in real time if there is a force imbalance, until the force balance of the wind turbine is restored to the preset safe range.

[0074] Specifically, step S404 includes: Step S4041: Obtain the current independent force feature vector in the fusion index, and based on the current independent force feature vector and the corresponding historical data independent force feature vector, determine the degree of abnormality and duration of force imbalance.

[0075] Specifically, when there is a stress imbalance in the wind turbine, the degree of abnormality (e.g., stress exceeds the normal value by 20%) and duration (e.g., has lasted for 3 hours) are derived based on the current independent components and the independent components of historical data.

[0076] Step S4042: Analyze the changing trend of the force imbalance of the wind turbine based on the degree of anomaly and duration.

[0077] Specifically, based on the degree and duration of the anomaly, further judgment is made on the trend of the imbalance, including the dynamic evolution of the stress anomaly (e.g., "as the wind speed increases, the stress increases at a rate of 5% per hour").

[0078] Step S4043: Based on the changing trend, model predictive control is used to dynamically optimize the wind turbine operating parameters under preset constraints, and the operating state of the wind turbine is adjusted based on the dynamically optimized wind turbine operating parameters until the force balance of the wind turbine is restored to the preset safe range.

[0079] Specifically, based on the changing trends, real-time optimization algorithms (such as model predictive control) are used to dynamically adjust the parameters of the operational control strategy. Specifically, the prediction time domain is 10 seconds (based on a 1000Hz sampling rate, with a total of 10,000 data points), and the control time domain is 3 steps, with constraints of blade rotation speed ≤ 20 rpm and blade angle ≤ 30°. The adjusted rotation speed (controlling rotational force) and blade angle (controlling the windward area, affecting the magnitude of the force) are obtained to acquire an optimized parameter set that reduces force imbalance. By optimizing the parameter set to drive the actuator (such as the pitch system and speed control system) to adjust the operating status of the wind turbine, and obtaining the adjusted force data (such as real-time feedback of stress and vibration) feedback, it is determined whether the force balance after regulation has been restored to a safe range (70-90).

[0080] Step S405: When the dynamically optimized wind turbine parameter set fails to restore the force balance of the wind turbine to the preset safe range, the dynamically optimized wind turbine operating parameters are analyzed based on the preset deep learning model to determine the optimized values ​​of the wind turbine operating parameters; the optimized values ​​of the wind turbine operating parameters are iteratively calculated using the preset optimization algorithm to obtain the optimal wind turbine operating parameters, and the operating status of the wind turbine is updated based on the optimal wind turbine operating parameters.

[0081] Specifically, if the stress balance does not reach the safe range, a deep learning model (such as CNN (Convolutional Neural Network) + LSTM (Long Short-Term Memory)) is used to perform secondary analysis on the feedback data (e.g., "stress concentration in a certain area of ​​the blade has not been relieved") to obtain the analysis results of not meeting the safe range, and to determine the optimized values ​​of each parameter in the optimization parameter set (e.g., "a larger adjustment of the blade angle is needed, or additional adjustment of the tower support parameters is required"). The deep learning model outputs the direction and range of parameter optimization. Blade angle: Current value ±5° (e.g., current value 10°, optimization range 5°-15°); Rotational speed: Current value ±3 rpm (e.g., current 15 rpm, optimization range 12 rpm-18 rpm); Tower damping coefficient: 0.6-0.9; The genetic algorithm uses this range as the initial search space and iteratively calculates the optimal combination of parameters.

[0082] Specifically, the deep learning model designed for the special scenarios of offshore wind power adopts a CNN-LSTM hybrid architecture with an anti-interference module. The CNN layer has 3 convolutional layers (3×3 kernels, with 16, 32, and 64 kernels respectively, stride 1), ReLU activation function, and 2×2 max pooling; the LSTM layer has 128 hidden units. Training uses the Adam optimizer (learning rate 0.01) and cross-entropy loss function, and is trained for 50 rounds (batch size 32). A Dropout layer (rate=0.2) is introduced to prevent overfitting. The CNN layer is specifically designed to extract high-dimensional spatial features, capturing the stress distribution differences in different regions of the blade (such as the blade tip and blade root) and the radial / axial vibration coupling features of the tower under the impact of sea waves through multi-scale convolutional kernels, especially strengthening the recognition of local stress concentration patterns caused by salt spray corrosion. The LSTM layer focuses on time series analysis, focusing on learning the dynamic stress law under the alternating action of wind and waves.

[0083] The anti-interference module is implemented through a combination of deep learning model architecture design and maritime-specific dataset enhancement strategies. These two aspects together constitute the core of the anti-interference module, specifically including the following two aspects: On the one hand, the anti-interference module is reflected in the architecture design of the CNN-LSTM hybrid model: the CNN layer adopts 3 convolutional layers (3×3 convolutional kernels, number of kernels 16, 32, and 64, stride 1), which accurately captures the stress distribution of different regions of the blade (blade tip and blade root) and the radial / axial vibration coupling characteristics of the tower through multi-scale convolutional kernels, while filtering out local interference signals generated by wave impact; the LSTM layer has 128 hidden units, and uses gating mechanisms (input gate, forget gate, output gate) to screen the key force evolution trend under the alternating action of wind and waves, and eliminate electromagnetic interference and high-frequency noise generated by sensor fluctuations in the salt spray environment; in addition, the model also introduces a Dropout layer (rate=0.2) to avoid model overfitting caused by complex interference at sea, and further improve the anti-interference robustness.

[0084] On the other hand, the marine-specific dataset augmentation strategy is an important supplement to the anti-interference module: It expands sparse samples under extreme sea conditions (gusts of wind exceeding force 12, storm surges) by using DCGAN (Deep Convolutional Generative Adversarial Network), increasing the number of samples from 500 to 5000. Simultaneously, it simulates sensor drift characteristics under wave heights of 2-5m and salt spray concentrations of 50-100mg / m³, allowing the model to fully learn the data flow characteristics of typical marine interference scenarios during training. This enables the model to effectively identify and counteract the impact of such interference on force analysis in practical applications. In short, the marine-specific dataset augmentation strategy is a means for the anti-interference module to improve its anti-interference capabilities from the "data source," while the targeted design of the CNN-LSTM architecture is the core of the anti-interference module's interference filtering at the "model processing" level. Together, they effectively suppress complex marine interference.

[0085] "High-dimensional spatial features" are features specifically extracted by the CNN layer (convolutional neural network layer), focusing on "complex correlation features in the spatial dimension": capturing the differences in stress distribution in different spatial regions of the same component through multi-scale convolutional kernels (such as the stress distribution gradient between the blade tip and root, and the difference in vibration intensity between the bottom and top of the tower), as well as the coupling features of different force directions of the same component (such as the coordinated change pattern of radial and axial vibration of the tower under wave impact), and also including the special local stress state of the component (such as the stress concentration pattern in a specific area of ​​the blade caused by salt spray corrosion). These features are multi-dimensional and correlated features formed by "spatial location, regional distribution, and multi-directional coupling", focusing on reflecting the complex distribution and correlation of forces in the spatial dimension.

[0086] For example, the frequency changes of blade flutter caused by sudden changes in wind speed during typhoons, or the cumulative effect of alternating loads on the tower foundation during tidal cycles, are filtered by a gating mechanism to remove high-frequency noise generated by wave impacts, while preserving key stress evolution trends. During the model training phase, a marine-specific dataset augmentation strategy is introduced: sparse samples under extreme sea conditions (such as gusts of wind exceeding level 12 and storm surges) are augmented using a generative adversarial network (DCGAN) to simulate sensor drift characteristics under different wave heights and salt spray concentrations, ensuring robustness to complex marine environments. Specifically, DCGAN can be used to augment the samples. Generator: 4 layers of deconvolution (output channels are 64, 32, 16, and 3 respectively), ReLU activation, batch normalization; Discriminator: 4 convolutional layers (3 input channels, 16, 32, 64, and 1 output channels respectively), LeakyReLU activation; Training: Adam optimizer (learning rate 0.0002), batch size 64, training epochs 200, expanding 500 samples under level 12 gusts to 5000 samples, simulating sensor drift characteristics under wave height of 2 to 5 m and salt spray concentration of 50 to 100 mg / m³.

[0087] The output layer adopts a multi-task learning design. In addition to identifying specific anomalies such as "stress concentration at the leading edge of the blade", it can also directly associate with marine equipment parameters such as pitch system response delay and tower damping characteristics, providing directional guidance for subsequent genetic algorithm optimization (such as outputting decision suggestions for intertidal units to "prioritize adjusting the blade angle to counteract the additional torque of waves"), achieving end-to-end adaptation from anomaly diagnosis to control optimization.

[0088] For the optimized values, a genetic algorithm (adept at global optimization) is used to iteratively calculate the optimal set of operating parameters (e.g., 15 rpm rotation speed + 12° blade angle + 0.8 tower damping coefficient). The population size is 100, the crossover probability is 0.8, the mutation probability is 0.01, the number of iterations is 50 generations, and the fitness function is 0.7 × (1 - stress overshoot rate) + 0.3 × (1 - power loss rate). This yields an adaptive control scheme for complex sea conditions (e.g., alternating wind and waves, sudden gusts), resulting in an optimal set of parameters that balances safety (stress not exceeding the limit) and efficiency (output power loss < 5%).

[0089] Among them, "stress exceedance rate" refers to the proportion of the number of times the measured stress value of the core components of the wind turbine (blades, towers, gearboxes, i.e. components with strain gauges deployed to collect stress data) exceeds the "safety threshold" of the corresponding stress parameter during the monitoring period of the genetic algorithm iterative optimization, to the total number of stress monitoring times in the same period.

[0090] The determination of the "stress safety threshold" is based on the "inherent value" specified in the document (the design rated stress value of each component of the wind turbine model, such as the inherent value of blade axial stress of 100MPa, and the inherent stress value of tower and gearbox determined according to GB / T19073-2008 "Wind Turbine Generator Gearbox" and the manufacturer's design manual). It is usually a stress safety upper limit set based on the inherent value (to avoid exceeding the design rated value, which may lead to structural fatigue or failure). The "number of monitoring" is matched with the high-frequency sampling rate (1000Hz, i.e., 1000 times per second) and monitoring cycle mentioned above. For example, if the monitoring is performed for 10 seconds in a certain iteration cycle, the total number of monitoring is 10,000. If 500 of the measured stress values ​​exceed the safety threshold, the stress exceedance rate is 500÷10000=5%.

[0091] The stress exceedance rate is directly related to the stress safety of wind turbine units. It has a weight of 0.7 in the fitness function (higher than the 0.3 weight of the power loss rate), reflecting the core objective of the genetic algorithm to prioritize structural stress safety during optimization.

[0092] By updating the operating status of wind turbines with the optimal parameter set and continuously collecting stress data (e.g., over several months), the trend of stress balance under long-term operation is obtained to determine whether the system's adaptability to dynamic sea conditions (e.g., seasonal changes, typhoon seasons) is stable (e.g., "under different sea conditions, the fusion index can be stable at 70-90"). The long-term verification stability standards are as follows: Time dimension: within a continuous 3-month monitoring period, the number of times the fusion index exceeds the standard ≤ 3 times (the duration of each exceedance < 10 minutes); Numerical dimension: the index fluctuation range ≤ ±5% (e.g., average 80, fluctuation range 76-84); Sea state coverage: covering normal wind (3-8 m / s), strong wind (8-15 m / s), typhoon (>15 m / s), and storm surge (wave height > 3 m) scenarios, with the indicators for each scenario meeting the above requirements.

[0093] The core of the "optimal parameter set" includes three key parameters: rotational speed, blade angle, and tower damping coefficient. In step S405, when the primary control is not within the safe range, the deep learning model outputs the direction and range of parameter optimization (e.g., current blade angle ±5°, current rotational speed ±3rpm, tower damping coefficient 0.6-0.9). Subsequently, the genetic algorithm iteratively optimizes the parameters within this range as the initial search space. A specific example of the final optimal parameter set is "rotational speed 15rpm + blade angle 12° + tower damping coefficient 0.8". These three parameters correspond to: rotational speed controlling the rotational force of the unit, blade angle adjusting the windward area to affect the magnitude of the force, and tower damping coefficient suppressing tower vibration. All three work together to optimize the force balance of the wind turbine unit. Therefore, the "optimal parameter set" is the optimal combination determined around these three parameters.

[0094] The offshore wind turbine stress imbalance monitoring method provided in this embodiment constructs a closed-loop offshore wind turbine stress balance management system of "precise perception - dynamic control - deep optimization". At the perception level, it innovatively adopts a multi-source sensor collaborative acquisition and weighted fusion strategy: by deploying stress, vibration and displacement sensors to cover core components, and combining signal denoising and feature extraction technology to separate the independent stress components of each component, a fusion index is generated according to the safety priority weights of 50% for blades, 30% for towers and 20% for gearboxes. This breaks through the limitations of traditional single-parameter monitoring and achieves precise quantification of the overall stress balance state.

[0095] At the level of regulation and optimization, a pioneering hierarchical intelligent decision-making mechanism is implemented: Primary regulation employs model predictive control algorithms to optimize rotational speed and blade angle in real time for rapid response to force imbalances; when primary regulation fails to achieve its target, a deep learning model is introduced for secondary diagnosis, precisely locating the direction of parameter optimization, and then generating the optimal parameter set through iterative calculation, forming a closed loop of "real-time adjustment - deep optimization - continuous monitoring." This system ensures the timeliness of force balance regulation while improving optimization accuracy under complex sea conditions through intelligent algorithms, balancing safety and operational efficiency.

[0096] As one or more specific application embodiments of the present invention, combined with Figure 5 The method for monitoring the force imbalance of offshore wind turbines provided by this invention will be further described in detail, such as... Figure 5 As shown, the specific process is as follows: Step S501: Deploy multi-source sensors on core components such as blades, towers, and gearboxes to collect stress, vibration, and displacement data to form multi-dimensional signals; after denoising and feature extraction to determine the independent force components of each component, integrate the feature components according to the weights of 50% for blades, 30% for towers, and 20% for gearboxes through a multi-source data fusion model to generate fusion indexes. If the fusion index value exceeds the threshold...

[0097] Step S502: Determine whether the wind turbine has an unbalanced state that deviates from the normal range: If the fusion index is within the preset threshold range, it is determined that the wind turbine does not have a force imbalance; if it is not within the preset threshold range, it is determined that the wind turbine has a force imbalance.

[0098] In step S503, model predictive control is adopted to optimize and adjust the rotational speed and blade angle in real time, generate an optimized parameter set, and drive actuators such as pitch and speed control to adjust the operating state so that the force balance is restored to the preset safe range.

[0099] Step S504: If the force balance does not reach the safe range, the optimized values ​​of each parameter in the optimized parameter set are determined by the deep learning model, and the optimal parameter set is calculated iteratively to update the wind turbine's operating status and continuously monitor it.

[0100] Example: To address the structural stress monitoring needs of offshore wind turbines under complex sea conditions, the implementation process revolves around the entire workflow of "data acquisition - signal processing - balance assessment - control optimization - long-term verification," as detailed below: First, multi-source data acquisition and preprocessing are performed. Adaptive sensors are deployed on core components of the wind turbine, such as blades, tower, and gearbox: strain gauges are installed on the blade surface to collect stress data; accelerometers are placed inside the tower to monitor vibration; and displacement sensors are installed on the gearbox casing to capture displacement changes. All sensors continuously collect data at a sampling frequency of 1000Hz (1000 times per second) and transmit it in real time to the central processing unit via industrial Ethernet (TCP / IP protocol), with a transmission delay of ≤100ms, forming a multi-dimensional raw dataset containing stress, vibration, and displacement. Sensor deployment scheme: Blade: 2 strain gauges (axial + radial) at the blade root (1m from the hub), 1 accelerometer at the blade middle (1 / 2 length of the blade), and 1 displacement sensor at the blade tip (0.5m from the blade tip); Tower: 2 accelerometers (horizontal + vertical) at the bottom (2m above sea level), 1 strain gauge at the middle (1 / 2 height of the tower), and 1 displacement sensor at the top (1m from the nacelle); Gearbox: 1 strain gauge at the input shaft end, 1 vibration sensor at the output shaft end, and 1 displacement sensor at the top of the gearbox.

[0101] First, the multidimensional original dataset is denoised. Signal decomposition algorithm is used to filter out marine-specific noise such as wave impact and electromagnetic interference. Then, feature extraction technology is used to separate key feature vectors such as vibration frequency, peak stress, and axial / radial force from the denoised signal. At the same time, the ratio of the measured value of each feature vector to the inherent value of the corresponding component is calculated to standardize data of different magnitudes to a unified numerical range, laying the foundation for subsequent analysis.

[0102] Next, multi-source data fusion and imbalance judgment are performed. The preset multi-source data fusion model is invoked, and according to the safety priority of offshore wind turbine components, the stress characteristic vectors of blades, towers, and gearboxes are assigned weights and weighted and integrated to generate a fusion index characterizing the stress balance of the whole machine.

[0103] The fusion index is compared with the preset safety threshold range. If the index is within the threshold, the unit is determined to be in stress balance and continuous monitoring is sufficient. If the index exceeds the threshold, the imbalance analysis process is initiated. Combining historical stress data, the degree and duration of the current imbalance are quantified using time series analysis methods. For example, an ARIMA (p=2, d=1, q=1) time series model (with parameters determined by the AIC criterion) is used. Based on historical data collected at 1000Hz over the past 30 days, the degree of anomaly is quantified as (current stress value - historical average value for the same period) / historical average value for the same period × 100% (e.g., 20%). The duration of the anomaly is statistically analyzed using a sliding window algorithm (window size 1 hour) (e.g., 3 hours) to further refine the dynamic evolution law of the stress anomaly, such as the stress fluctuation trend with changes in wind speed and wave height.

[0104] The core of extracting the dynamic evolution law of stress anomalies is achieved through the logic of "sliding window anomaly period locking + multi-source data correlation analysis + trend quantification". This needs to be carried out in conjunction with the settings of "high-frequency data collection, environmental parameter linkage, and time series analysis" specified in the document. The specific process is as follows: First, a sliding window algorithm is used to lock in the complete time period of the anomaly: with a fixed window size of 1 hour, the real-time collected stress data (1000Hz sampling rate) and fusion index are judged window by window: if the fusion index in a certain window exceeds the safe range of 70-90 (or the stress deviation percentage reaches the abnormal standard), then the window is marked as an "abnormal window"; when 3 consecutive abnormal windows appear, the duration of the anomaly is determined to be 3 hours, and the complete data period corresponding to these 3 hours is locked (including the stress, vibration, and displacement standardized values ​​in each window, as well as the environmental data of wind speed and wave height at the same time), avoiding isolated analysis of single point-in-time data and ensuring coverage of the entire process of anomaly evolution.

[0105] Secondly, by correlating the "stress data - environmental data" during the duration of the anomaly, a correlation is established: Stress parameters for each minute (or finer time granularity, based on a 1000Hz sampling rate, minute-level statistical values ​​can be extracted, such as peak stress per minute and average vibration frequency) within the 3-hour anomaly duration are mapped one-to-one with concurrent wind speed and wave height data, forming a three-dimensional data matrix of time-environmental parameters-stress parameters. For example, during the 3-hour period of wind speed increasing from 8 m / s to 12 m / s and wave height increasing from 1.5 m to 2.8 m, the peak axial stress of the blades increases from 50 MPa to 65 MPa, and the radial vibration of the tower increases from 2.5 mm / s... 2 Increased to 4mm / s 2 The corresponding changes were used to preliminarily identify the "linkage relationship between changes in environmental parameters and abnormal stress".

[0106] Finally, dynamic evolution patterns are extracted through trend quantitative analysis: Based on the above-mentioned related data, the time series analysis methods (such as the ARIMA model) or trend fitting algorithms mentioned in the document are used to quantitatively model the changes in "environmental parameters - stress parameters" during the period of abnormality: For example, the average increase in peak blade stress is calculated when the wind speed increases by 1 m / s during the 3-hour period of abnormality (as shown in the document example "as the wind speed increases, the stress increases at a rate of 5% per hour"); or the period of change in tower vibration frequency after the wave height exceeds 3 m (such as 1 fluctuation every 10 minutes, with an amplitude increase of 0.3 mm / s²). Meanwhile, by comparing the stress data under the same environmental conditions in the same period in history (such as when the wind speed was 8-12 m / s in the past 3 hours, the stress increase was only 3% per hour), we can further verify whether the current abnormal evolution conforms to the normal trend, or whether there are new evolutionary characteristics such as "the increase in salt spray concentration leading to accelerated local stress concentration". In the end, we can form a quantifiable and reusable dynamic evolution law (such as "when the wind speed is ≥10 m / s and the wave height is ≥2.5 m, the gearbox displacement parameter exceeds the standard at a rate of 8% per hour"), and provide accurate trend basis for subsequent graded regulation.

[0107] Subsequently, tiered control and optimization were implemented. Based on the abnormal evolution trend, model predictive control algorithms were prioritized for primary control, dynamically adjusting the wind turbine's rotational speed and blade angle. Adjusting the rotational speed controlled the rotational forces on the turbine, while changing the blade angle adjusted the windward area to control the magnitude of the forces. This generated a preliminary optimized parameter set and drove the pitch and speed control systems to perform adjustments. After control, new force data was collected in real time and fed back to determine if the fused indicators had recovered to a safe range. If recovered, the current operating parameters were maintained and continuous monitoring continued; if not recovered, secondary deep optimization was initiated. A CNN-LSTM hybrid deep learning model, specifically designed for offshore scenarios, was used to perform secondary analysis of the feedback data. This model captures the spatial force characteristics of different regions of the blades and the tower through the CNN layer, and learns the time-series force patterns under alternating wind and waves through the LSTM layer, accurately locating potential anomalies such as unrelieved stress concentration and clarifying the direction of parameter optimization. Based on this direction, a genetic algorithm was used for global optimization iteration to calculate the optimal combination of parameters such as rotational speed, blade angle, and tower damping coefficient, balancing stress safety and power generation efficiency, forming the final optimized parameter set.

[0108] Finally, long-term operation verification and adaptability assessment are conducted. The final optimized parameter set is applied to the wind turbine, the operating status is updated, and stress data is continuously collected for several months. Time series analysis and clustering algorithms are used to analyze the stress balance trend under long-term operation, focusing on verifying the turbine's adaptability to dynamic sea conditions such as seasonal changes and frequent typhoons. If the integrated indicators remain stable within a safe range under different sea conditions, the monitoring and control scheme is deemed suitable for the current marine environment and can be put into long-term use. If fluctuations exist, the model parameters are further fine-tuned based on the monitoring data to ensure continuous and stable system operation.

[0109] The offshore wind turbine unbalance monitoring method provided in this embodiment continuously monitors and optimizes the force balance of the wind turbine, effectively improving the stability and operating efficiency of the system under dynamic sea conditions, and providing technical support for the safe and efficient operation of offshore wind turbines.

[0110] This embodiment also provides a device for monitoring the force imbalance of offshore wind turbines. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0111] This embodiment provides a device for monitoring the force imbalance of offshore wind turbines, such as... Figure 6 As shown, it includes: The multidimensional signal acquisition module 601 is used to acquire multidimensional signals that reflect the real-time stress state of the wind turbine.

[0112] The fusion index generation module 602 is used to extract features from multi-dimensional signals to obtain multi-dimensional force feature vectors, and to fuse the multi-dimensional force feature vectors to generate fusion indexes that characterize the force balance of wind turbine units.

[0113] The force balance judgment module 603 is used to judge the force balance of the wind turbine based on the relationship between the fusion index and the preset threshold range.

[0114] The predictive control module 604 is used to adjust the operating parameters of wind turbines with force imbalance in real time using model predictive control until the force balance of the wind turbine is restored to the preset safe range.

[0115] In some optional implementations, the multidimensional signal acquisition module 601 includes: The multidimensional signal acquisition unit is used to collect stress signals, vibration signals and displacement signals of each target component through multi-source sensors deployed on multiple target components of the wind turbine, so as to obtain multidimensional signals reflecting the real-time stress state of the wind turbine.

[0116] In some alternative implementations, the fusion index generation module 602 includes: The independent force feature vector determination unit is used to perform noise reduction and feature extraction on multidimensional signals using a signal decomposition algorithm to obtain multiple independent force feature vectors. After parameter standardization of the multiple independent force feature vectors, multiple parameter-standardized independent force feature vectors are obtained.

[0117] The fusion index generation unit is used to generate a fusion index that characterizes the force balance of a wind turbine by weighting and fusing multiple independent force feature vectors after parameter standardization using a preset multi-source data fusion model.

[0118] In some optional implementations, the fusion index generation unit includes: The component weight allocation subunit is used to assign component weights to each target component based on the importance of each target component.

[0119] The parameter weight allocation subunit is used to assign parameter weights to each independent force feature vector based on the importance of each independent force feature vector within each target.

[0120] The fusion calculation subunit is used to perform weighted fusion calculation on all force feature vectors based on component weights and parameter weights, using a preset multi-source data fusion model, to generate a fusion index characterizing the force balance of the wind turbine.

[0121] In some optional implementations, the fusion computing subunit is also used for: Based on parameter weights, a pre-set multi-source data fusion model is used to perform weighted fusion calculations on each force feature vector within each target component to obtain the comprehensive state value of all feature vectors of each component; based on component weights, a pre-set multi-source data fusion model is used to perform weighted fusion calculations on the comprehensive state value of all independent force feature vectors of each component to obtain a fusion index characterizing the force balance of the wind turbine.

[0122] In some optional implementations, the force balance determination module 603 includes: The force balance judgment is intended to determine that if the value of the fusion index is within the preset threshold range, the wind turbine is judged to have no force imbalance; if the value of the fusion index exceeds the preset threshold range, the wind turbine is judged to have force imbalance.

[0123] In some alternative implementations, the predictive control module 604 includes: The anomaly degree and duration determination unit is used to obtain the current independent force feature vector in the fusion index, and based on the current independent force feature vector and the corresponding historical data independent force feature vector, to determine the anomaly degree and duration of the force imbalance.

[0124] The trend analysis unit is used to analyze the trend of force imbalance in wind turbines based on the degree and duration of anomalies.

[0125] The primary control unit is used to dynamically optimize the operating parameters of the wind turbine under preset constraints based on the changing trend and using model predictive control. It then adjusts the operating status of the wind turbine based on the dynamically optimized operating parameters until the force balance of the wind turbine is restored to the preset safe range.

[0126] In some alternative embodiments, the device further includes: The secondary control module is used to analyze the dynamically optimized wind turbine operating parameters based on a preset deep learning model when the dynamically optimized wind turbine parameter set fails to restore the force balance of the wind turbine to a preset safe range, determine the optimized values ​​of the wind turbine operating parameters, iteratively calculate the optimized values ​​of the wind turbine operating parameters using a preset optimization algorithm, obtain the optimal wind turbine operating parameters, and update the operating status of the wind turbine based on the optimal wind turbine operating parameters.

[0127] The offshore wind turbine unbalance monitoring device provided in this embodiment of the invention can execute the offshore wind turbine unbalance monitoring method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method. Further functional descriptions of the above modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0128] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0129] The following is a detailed reference. Figure 7 This diagram illustrates a suitable structural schematic for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 701, which can perform various appropriate actions and processes based on a program stored in read-only memory (ROM) 702 or a program loaded from memory 708 into random access memory (RAM) 703. The RAM 703 also stores various programs and data required for the operation of the electronic device. The processor 701, ROM 702, and RAM 703 are interconnected via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0130] Typically, the following devices can be connected to I / O interface 705: input devices 706 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 707 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 708 including, for example, magnetic tapes, hard disks, etc.; and communication devices 709. Communication device 709 allows electronic devices to exchange data via wireless or wired communication with other devices. Although Figure 7 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0131] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 709, or installed from a memory 708, or installed from a ROM 702. When the computer program is executed by the processor 701, it performs the functions defined in the offshore wind turbine force imbalance monitoring method of the embodiments of the present invention.

[0132] Figure 7 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0133] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the method for monitoring the force imbalance of offshore wind turbines shown in the above embodiments is implemented.

[0134] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0135] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for monitoring force imbalance in offshore wind turbine units, characterized in that, The method includes: Acquire multi-dimensional signals that reflect the real-time stress state of wind turbine units; Feature extraction is performed on the multidimensional signal to obtain a multidimensional force feature vector, and the multidimensional force feature vector is fused to generate a fusion index characterizing the force balance of the wind turbine. Based on the relationship between the fusion index and the preset threshold range, the force balance of the wind turbine is determined. Model predictive control is used to adjust the operating parameters of wind turbines with stress imbalances in real time until the stress balance of the wind turbines is restored to the preset safe range.

2. The method according to claim 1, characterized in that, The acquisition of multi-dimensional signals reflecting the real-time stress state of the wind turbine includes: By deploying multi-source sensors on multiple target components of the wind turbine, stress signals, vibration signals, and displacement signals of each target component are collected to obtain multi-dimensional signals reflecting the real-time stress state of the wind turbine.

3. The method according to claim 1, characterized in that, Feature extraction is performed on the multidimensional signal to obtain a multidimensional force feature vector, and the multidimensional force feature vector is fused to generate a fusion index characterizing the force balance of the wind turbine, including: The multidimensional signal is denoised and its features are extracted using a signal decomposition algorithm to obtain multiple independent force feature vectors. After parameter standardization of the multiple independent force feature vectors, multiple parameter-standardized independent force feature vectors are obtained. A pre-defined multi-source data fusion model is used to weight and fuse multiple independent force feature vectors after parameter standardization to generate a fusion index characterizing the force balance of the wind turbine.

4. The method according to claim 3, characterized in that, The method employs a pre-defined multi-source data fusion model to weight and fuse multiple independent force feature vectors after parameter standardization, generating a fusion index characterizing the force balance of the wind turbine, including: Assign component weights to each target component based on its importance. Based on the importance of each independent force feature vector within each target, parameter weights are assigned to each independent force feature vector; Based on the component weights and parameter weights, a preset multi-source data fusion model is used to perform weighted fusion calculations on all force feature vectors to generate a fusion index characterizing the force balance of the wind turbine.

5. The method according to claim 4, characterized in that, Based on the component weights and parameter weights, a preset multi-source data fusion model is used to perform weighted fusion calculations on all force feature vectors, generating fusion indices characterizing the force balance of the wind turbine, including: Based on the parameter weights, a preset multi-source data fusion model is used to perform weighted fusion calculations on each force feature vector within each target component to obtain the comprehensive state value of all feature vectors of each component. Based on the component weights, a pre-set multi-source data fusion model is used to perform weighted fusion calculation on the comprehensive state values ​​of all independent force characteristic vectors of each component, so as to obtain a fusion index characterizing the force balance of the wind turbine.

6. The method according to claim 1, characterized in that, Based on the relationship between the fusion index and the preset threshold range, the force balance of the wind turbine is determined, including: If the value of the fusion index is within the preset threshold range, it is determined that the wind turbine does not have a force imbalance. If the value of the fusion index exceeds the preset threshold range, it is determined that the wind turbine has a force imbalance.

7. The method according to claim 1, characterized in that, Model predictive control is used to adjust the operating parameters of wind turbines with stress imbalances in real time until the stress balance of the wind turbines is restored to a preset safe range, including: Obtain the current independent force feature vector in the fusion index, and based on the current independent force feature vector and the corresponding historical independent force feature vector, determine the degree and duration of force imbalance. Analysis of the changing trend of force imbalance in wind turbines based on the degree and duration of anomalies; Based on the changing trend, model predictive control is used to dynamically optimize the operating parameters of the wind turbine under preset constraints, and the operating status of the wind turbine is adjusted based on the dynamically optimized operating parameters until the force balance of the wind turbine is restored to the preset safe range.

8. The method according to claim 7, characterized in that, The method further includes: When the dynamically optimized wind turbine parameter set fails to restore the wind turbine's force balance to the preset safe range, the operating parameters of the dynamically optimized wind turbine are analyzed based on a preset deep learning model to determine the optimized values ​​of the wind turbine's operating parameters. The optimal operating parameters of the wind turbine are iteratively calculated using a preset optimization algorithm to obtain the optimal operating parameters of the wind turbine, and the operating status of the wind turbine is updated based on the optimal operating parameters.

9. A device for monitoring force imbalance in offshore wind turbine units, characterized in that, The device includes: The multi-dimensional signal acquisition module is used to acquire multi-dimensional signals that reflect the real-time stress state of the wind turbine. The fusion index generation module is used to extract features from the multidimensional signal to obtain a multidimensional force feature vector, and to fuse the multidimensional force feature vector to generate a fusion index characterizing the force balance of the wind turbine. The force balance judgment module is used to judge the force balance of the wind turbine based on the relationship between the fusion index and the preset threshold range. The predictive control module is used to adjust the operating parameters of wind turbines with force imbalance in real time using model predictive control until the force balance of the wind turbine is restored to the preset safe range.

10. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the offshore wind turbine force imbalance monitoring method according to any one of claims 1 to 8.

Images