Modular wind turbine tower design monitoring method and system

Abstract

This invention discloses a modular wind turbine tower design monitoring method and system. The system includes a modular, segmented, multi-faceted three-dimensional wind turbine tower structure, an intelligent design optimization module, a real-time monitoring module, and a cloud-based data-driven platform. The segmented connection interfaces utilize high-strength bolt friction-type connections. The real-time monitoring module collects preload change data via wireless stress sensors and acquires connection surface image data via an image recognition module, generating anomaly identification results. The cloud-based data-driven platform integrates the modular segmented design results and connection status data to generate stress relaxation trends and corrosion trends, and based on this, outputs preventative maintenance reminders and dynamic design correction instructions to adjust connection parameters and / or material configuration parameters for subsequent similar wind farms or operating conditions. This solution improves the accuracy of status identification at the modular wind turbine tower connection interfaces and enhances the targeted nature of subsequent design corrections.

Description

Technical Field

[0001] This invention relates to the field of digital design and structural monitoring technology for wind turbine towers, and in particular to a modular wind turbine tower design monitoring method and system. Background Technology

[0002] As a key load-bearing structure for wind turbine generators, wind turbine towers typically need to maintain sufficient strength, stiffness, and stability under complex wind loads, vibration loads, and long-term alternating stress conditions. Current technologies usually employ integral or segmented tower structures, combined with methods such as 3D modeling and finite element analysis to design, verify, and optimize the tower structure parameters.

[0003] For example, international patent document WO2017131953A1 discloses an optimized design scheme for wind turbine towers, aiming to optimize the design of wind turbine tower structures under cost constraints. This technical solution establishes a tower structure model and, combined with physical constraints, load conditions, and optimization algorithms, analyzes and optimizes the geometric and structural parameters of the tower, thereby outputting tower design results that meet strength and stability requirements, which can improve tower design efficiency and structural rationality to a certain extent. Another example is Chinese patent CN120140136A, which discloses a ring-shaped prestressed device and assembly, a concrete tower, and a concrete wind turbine tower, aiming to improve the overall stress performance of the assembled segmented tower. This technical solution adopts a segmented tower structure and enhances the connection stability between segments through a ring-shaped prestressed device and an anchoring system, thereby improving the tower's torsional, shear, and overall stability performance.

[0004] However, existing technologies still have the following technical shortcomings: First, existing intelligent optimization design schemes mainly target the overall parameter optimization of conventional towers, lacking a collaborative optimization mechanism for the segment size, connection form, and material gradient arrangement of modular segmented multi-ribbed three-dimensional steel towers; second, existing segmented tower connection technologies are mostly concentrated in prestressed concrete systems, failing to be applicable to steel modular towers using high-strength bolt friction connections; third, existing technologies generally lack real-time monitoring, degradation prediction, and dynamic correction mechanisms for the service status of connection parts, making it difficult to achieve integrated closed-loop processing of wind power tower design and structural monitoring. Summary of the Invention

[0005] The purpose of this invention is to provide a modular wind turbine tower design monitoring method and system to solve the problem in the prior art that, for the segmented connection interface of modular segmented multi-ribbed three-dimensional wind turbine towers, there is a lack of technical solutions to correlate and process the design parameters of high-strength bolt friction type connections, preload change data during operation, abnormal information of connection surface images, and degradation trend prediction results, and further feed them back for subsequent design corrections. This results in the difficulty in timely identification of stress relaxation, corrosion expansion, and fatigue risks at the connection interface, and the difficulty in making targeted adjustments to connection parameters and material configuration parameters for similar wind farms or similar operating conditions based on historical service results.

[0006] To achieve the above objectives, the first technical solution provided by this invention is as follows: A modular wind turbine tower design monitoring system Including modular, segmented, multi-ribbed, three-dimensional wind turbine towers, The modular segmented multi-ribbed three-dimensional structure wind turbine tower is formed by splicing multiple segments, and a segment connection interface is provided between adjacent segments. The segment connection interface adopts a high-strength bolt friction type connection. The system also includes an intelligent design optimization module, a real-time monitoring module, and a cloud-based data-driven platform. The intelligent design optimization module is used to acquire wind farm environmental data and, based on the wind farm environmental data, jointly optimize the segmented structural parameters, material configuration parameters, and connection parameters of the modular segmented multi-ribbed three-dimensional wind turbine tower to generate modular segmented design results. The real-time monitoring module is located at the segmented connection interface and includes a wireless stress sensor and an image recognition module. The wireless stress sensor is used to collect data on the preload change at the segmented connection interface, and the image recognition module is used to collect image data of the connection surface and generate anomaly identification results. The cloud-based data-driven platform is communicatively connected to the intelligent design optimization module and the real-time monitoring module. It is used to integrate the modular segment design results and connection status data, including the preload change data and the anomaly identification results, to generate the stress relaxation trend and corrosion trend of the segment connection interface. Based on the stress relaxation trend and the corrosion trend, it outputs preventive maintenance reminders and dynamic design correction instructions.

[0007] Furthermore, the dynamic design correction command is used to adjust the connection parameters and / or material configuration parameters for subsequent similar wind farms or similar operating conditions. The connection parameters and / or material configuration parameters include at least one of the following: bolt specifications, bolt spacing, target preload, steel grade of local connection area, and friction surface treatment grade.

[0008] Furthermore, the modular segmented multi-ribbed three-dimensional wind turbine tower uses high-strength steel arranged in a gradient. The high-strength steel includes at least two of Q355, Q460, Q550, Q690, and Q890, and is configured differently along the tower height or in local connection areas according to the structural stress level and environmental conditions.

[0009] Furthermore, the wireless stress sensor is placed near the high-strength bolt washer or on the surface of the connecting plate to invert the preload change data through strain changes.

[0010] Furthermore, the image recognition module performs edge recognition on the on-site side to preprocess the connection surface image data, extract edges, and segment abnormal regions to generate the anomaly recognition result. The anomaly identification results include at least one of the following: rust boundary, coating damage area, slippage mark, or suspected crack edge.

[0011] Furthermore, the cloud-based data-driven platform uses the preload retention rate to quantify the stress relaxation trend and the abnormal area ratio of the connection surface to quantify the corrosion trend.

[0012] Furthermore, when the corrosion trend exceeds a preset threshold, the dynamic design correction instruction includes increasing the steel grade in the local connection area and / or increasing the treatment level of the friction surface; When the stress relaxation trend exceeds a preset threshold, the dynamic design correction instruction includes increasing the bolt size, adjusting the bolt spacing, and / or adjusting the target preload.

[0013] Based on the first technical solution, the second technical solution proposed by this invention is as follows: A modular wind turbine tower design monitoring method is applied to the above system. Includes the following steps: Acquire wind farm environmental data and establish an initial design model for a modular, segmented, multi-ribbed, three-dimensional wind turbine tower. Based on the wind field environmental data, the segmented structural parameters, material configuration parameters, and connection parameters of the initial design model are jointly optimized to generate modular segmented design results. At the segmented connection interface, a wireless stress sensor collects preload change data, and an image recognition module collects connection surface image data and generates anomaly recognition results. The modular segmentation design results and connection status data, including the preload change data and the anomaly identification results, are uploaded to the cloud data-driven platform. The cloud-based data-driven platform integrates the modular segmentation design results and the connection status data to generate the stress relaxation trend and corrosion trend of the segmentation connection interface; Based on the stress relaxation trend and the corrosion trend, preventive maintenance reminders and dynamic design correction instructions are output to adjust the connection parameters and / or material configuration parameters for subsequent similar wind farms or similar operating conditions.

[0014] Furthermore, when jointly optimizing the segmented structure parameters, material configuration parameters, and connection parameters, finite element analysis and genetic algorithms are used to perform multi-objective collaborative optimization. The optimization objectives include structural strength, material cost, and fatigue resistance.

[0015] Furthermore, the image recognition module performs edge recognition on the connection surface image data on the field side, generates the anomaly recognition result, and then uploads the anomaly recognition result to the cloud data-driven platform.

[0016] The above-described one or more technical solutions in the embodiments of the present invention have at least one of the following technical effects: 1. This invention does not simply set up tower design, condition monitoring, and maintenance analysis in parallel. Instead, it focuses on the specific object of the segmented connection interface, and correlates the design parameters of the high-strength bolt friction connection, the preload variation data during operation, and the abnormal information of the connection surface image. Since the preload variation data can characterize the attenuation process of the connection clamping state, and the abnormal information of the connection surface image can characterize surface anomalies such as corrosion, slip marks, coating damage, or suspected crack edges, the combined use of these two types of data on the same segmented connection interface can more accurately distinguish stress relaxation-dominated degradation, corrosion-dominated degradation, and the coupled degradation of the two than single sensor data or single image recognition results, thereby improving the accuracy of identifying the true service state of the connection interface.

[0017] 2. This invention inputs modular segmented design results and connection status data into a cloud-based data-driven platform to generate stress relaxation trends and corrosion trends, rather than simply outputting general health assessment results. This allows degradation prediction results to directly correspond to specific parameter correction directions in subsequent designs. When corrosion trends are prominent, corrections can be made to the steel grade and friction surface treatment level of local connection areas; when stress relaxation trends are prominent, corrections can be made to bolt specifications, bolt spacing, and target preload; when both are prominent, combined corrections can be made to connection parameters and material configuration parameters. Thus, a clear technical correspondence is established between degradation prediction results and subsequent design corrections.

[0018] 3. This invention applies dynamic design correction commands to subsequent new projects in similar wind farms or under similar operating conditions, ensuring that service monitoring results from historical projects are not merely recorded in maintenance logs, but are instead incorporated into specific engineering parameters and fed back into the next design process. Compared to existing technologies that separate the design and operation phases, this invention improves the operational reliability, maintenance focus, and adaptability to subsequent designs of modular wind turbine tower connection interfaces.

[0019] 4. The gradient arrangement of multiple grades of high-strength steel, friction-type connection of high-strength bolts, monitoring of preload changes, identification of connection surface anomalies, and degradation trend-driven design corrections in this invention are not isolated from each other. The gradient arrangement of multiple grades of high-strength steel and friction-type connection of high-strength bolts provide an adjustable structural foundation for the connection interface; monitoring of preload changes and identification of connection surface anomalies provide complementary information for degradation judgment; and dynamic design corrections transform the aforementioned monitoring and prediction results into targeted adjustments to subsequent design parameters. The resulting technical effect is not a simple summation of the effects of individual measures, but rather an improvement in the accuracy of connection interface state identification and the effectiveness of subsequent design corrections through multi-stage linkage.

[0020] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating an embodiment of the present invention. Detailed Implementation

[0023] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0024] Example 1 This embodiment provides a basic example of a modular wind turbine tower intelligent design and monitoring system. The system includes a modular, segmented, multi-faceted, three-dimensional wind turbine tower structure, an intelligent design optimization module, a real-time monitoring module, and a cloud-based data-driven platform. The modular, segmented, multi-faceted, three-dimensional wind turbine tower structure is formed by splicing multiple segments, with connecting interfaces between adjacent segments. These connecting interfaces employ a friction-type connection structure. The intelligent design optimization module, the real-time monitoring module, and the cloud-based data-driven platform are communicatively connected, forming an integrated closed-loop architecture for design, monitoring, prediction, and feedback.

[0025] In this embodiment, the modular, segmented, multi-ribbed three-dimensional wind turbine tower is a steel tower with a multi-ribbed three-dimensional outer contour formed circumferentially. Each tower section is formed by multiple segments. Each segment can be an arc-shaped plate or a zigzag plate, as long as it can form a multi-ribbed three-dimensional structure after splicing. Adjacent segments are assembled and connected through the segment connection interface. The friction-type connection structure preferably uses a high-strength bolt friction-type connection so that the load is mainly transmitted through the friction force of the connection surface, thereby reducing the risk of pressure concentration on the hole wall and connection slippage. The segment connection interface includes a connecting plate, a friction surface, and a bolt hole area. The friction surface is treated with sandblasting or shot blasting to improve the interface friction coefficient and enhance connection stability.

[0026] The intelligent design optimization module is used to acquire wind farm environmental data and establish an initial design model for a modular, segmented, multi-ribbed, three-dimensional wind turbine tower based on this data. The wind farm environmental data includes at least annual average wind speed, extreme wind speed, turbulence intensity, wind direction frequency distribution, ambient temperature, relative humidity, salt spray level, and altitude information. The initial design model includes at least the tower height, number of tower sections, number of segments, segment thickness, number of ribs, variable cross-section parameters of the tower sections, material configuration parameters, and connection parameters. The segmented structural parameters include segment thickness, segment width, rib angle, tower section diameter, and tower section height; the material configuration parameters include steel grade and corresponding distribution location; and the connection parameters include high-strength bolt specifications, spacing, edge distance, target preload value, and friction surface treatment grade.

[0027] The intelligent design optimization module employs finite element analysis and genetic algorithms to perform multi-objective collaborative optimization. Specifically, the finite element analysis submodule first establishes a finite element model based on the initial design model, including segments, edges, connecting plates, and high-strength bolt connection interfaces. It then loads rated wind conditions, extreme wind conditions, start-up and shutdown conditions, and fatigue load spectra using the wind field environmental data to solve for the static, buckling, and fatigue responses of the tower as a whole and the connection interfaces. This yields the tower top displacement, stress distribution of segments and connecting plates, overall buckling margin, shear force transmission distribution at the connection interfaces, and fatigue hotspot locations corresponding to high stress amplitudes. Subsequently, the genetic algorithm submodule encodes the segmented structure parameters, material configuration parameters, and connection parameters into chromosomes. The encoded variables include at least the segment thickness, segment width, edge angle, tower segment diameter, steel grade distribution, bolt specifications, bolt spacing, edge distance, target preload, and friction surface treatment grade. Then, a fitness is constructed based on the finite element analysis results. Selection, crossover, and mutation are performed on the parent schemes that meet the strength, stiffness, buckling stability, and connection anti-slip constraints to generate the next generation of candidate schemes. The next generation of candidate schemes is then substituted back into the finite element analysis submodule for repeated solving until the comprehensive objective function meets the convergence condition, at which point the optimal scheme or Pareto optimal scheme is output.

[0028] To ensure the feasibility of the optimization process, this embodiment adopts the following comprehensive objective function: in, To comprehensively optimize the target value; The total mass of the tower is expressed in tons. This is the sum of material costs and connection manufacturing costs, expressed in yuan. This represents the peak displacement at the top of the tower, in millimeters. This is the fatigue damage evaluation value, which is a dimensionless quantity. , , and These are the normalized bases corresponding to the baseline scheme; , , and Let be the weight coefficient, and satisfy... In this embodiment, the weighting coefficients are obtained through adaptive optimization using a genetic algorithm on a training sample composed of historical wind field samples and existing tower service samples. When the rate of change of the comprehensive optimization target value is less than [a certain value] for 20 consecutive generations, the weighting coefficients are determined. When convergence is reached, it is determined that the convergence has occurred.

[0029] For example, in a sample of medium-to-high wind speed wind fields on land, the optimized weighting coefficients could be: , , , In the coastal high-salt-fog wind field sample, the optimized weighting coefficients can be: , , , The former point value focuses more on the balance between quality and cost, while the latter point value focuses more on fatigue damage control and long-term reliability.

[0030] The real-time monitoring module is located at the segmented connection interface and is used to perform multimodal acquisition of connection status data of the segmented connection interface. In this embodiment, the connection status data includes at least preload change data and connection surface image data. The preload change data is acquired by a wireless stress sensor located near the high-strength bolt washer or on the surface of the connection plate, and the connection surface image data is acquired by an industrial camera unit located near the segmented connection interface. The real-time monitoring module sends the acquired connection status data to the cloud data-driven platform after timestamping, marking, and marking the tower segment. The timestamp is used to reflect the acquisition time, the location mark is used to reflect the location of the tower segment and connection point, and the tower segment mark is used to establish a mapping relationship between the monitoring data and the corresponding design model parameters.

[0031] The cloud-based data-driven platform integrates the modular segmented design results with the connection status data to generate connection performance degradation prediction results. Based on these predictions, it outputs preventative maintenance reminders and dynamic design correction instructions. Specifically, the platform first establishes design data representation vectors and monitoring data representation vectors. The design data representation vectors include at least tower segment location, segment thickness, local steel grade, bolt specifications, bolt spacing, target preload, and friction surface treatment level. The monitoring data representation vectors include at least a preload retention rate sequence, abnormal area percentage, abnormality type code, and abnormal location code. Subsequently, the two types of vectors are aligned according to tower segment location, connection interface number, and acquisition time window, and then input into the prediction model for inference. The connection performance degradation prediction results include at least stress relaxation trend and corrosion trend. The stress relaxation trend characterizes the decrease or rate of decrease in the preload retention rate within a predetermined time window, while the corrosion trend characterizes the increase or rate of increase in the percentage of abnormal area on the connection surface within a predetermined time window.

[0032] The stress relaxation trend can be quantified using the following degradation indices: in, Preload retention rate; For a moment The measured preload force is expressed in kilonewtons. The initial preload is measured in kilonewtons. The smaller the value, the more pronounced the stress relaxation. Corrosion trend can be quantified by the percentage of abnormal area at the interface: in, The ratio represents the corrosion characterization ratio; The area of ​​the abnormal region obtained from image recognition is expressed in square millimeters. The effective observation area of ​​the connecting surface is expressed in square millimeters. The cloud-based data-driven platform is based on… and The time-varying curves or predicted values ​​generate degradation prediction results, and preventative maintenance reminders are generated when the predicted values ​​exceed a preset threshold. The dynamic design correction instructions are then fed back to the intelligent design optimization module, which is used to fine-tune the segmented structure parameters, material configuration parameters, and connection parameters under similar wind farms or similar operating conditions, thereby forming a complete closed loop.

[0033] In this invention, the preload variation data and the anomaly identification results are not used in isolation, but rather jointly to characterize the connection status of the same segmented connection interface. The cloud-based data-driven platform aligns and analyzes the preload retention rate sequence acquired by the wireless stress sensor with the anomaly area, anomaly type, and anomaly location output by the image recognition module to distinguish between stress relaxation-dominated degradation caused by high-frequency alternating loads, corrosion-dominated degradation caused by high-salt spray and humid heat environments, and coupled degradation where stress relaxation and corrosion coexist. This reduces the risk of misjudgment due to relying on only a single monitoring quantity and improves the targeting of subsequent dynamic design corrections.

[0034] To verify the feasibility of this embodiment, two specific designs are selected for illustration. The first group is an onshore wind farm with a tower height of 120 meters. The thickness of the sections is compared between 18 mm and 24 mm, with a target preload of 220 kN for the high-strength bolts. The second group is a coastal wind farm with a tower height of 140 meters. The thickness of the sections is compared between 22 mm and 28 mm, with a target preload of 260 kN for the high-strength bolts. After optimization, in the first case, using a combination of a 24 mm thick bottom plate and an 18 mm thin top plate reduces the tower top displacement by approximately 8.6% compared to a uniform thickness design. In the second case, using a combination of a 28 mm thick bottom plate and a 22 mm thin top plate, and improving the friction surface treatment level at the connection interface, reduces the predicted stress relaxation risk by approximately 11.2%. Therefore, this embodiment can completely realize a closed-loop technology from design to monitoring, prediction, and feedback.

[0035] Example 2 This embodiment provides a preferred embodiment, focusing on the specific structure of the gradient arrangement of multiple grades of high-strength steel, the friction-type connection of high-strength bolts, and the real-time monitoring module. In this embodiment, the modular segmented multi-ribbed three-dimensional wind turbine tower uses multiple grades of high-strength steel arranged in a gradient. The multiple grades of high-strength steel include at least two of Q355, Q460, Q550, Q690, and Q890, and are configured differently along the tower height direction or in local connection areas according to the structural stress level and environmental conditions.

[0036] Specifically, Q690 or Q890 is preferred in the high bending moment region at the bottom of the tower; Q460 or Q550 is preferred in the main load-bearing region in the middle; and Q355 or Q460 is preferred in the upper low-stress region. Near the segmented connection interfaces, due to factors such as localized stress concentration, frictional force transmission, and fatigue accumulation, the material grade can be locally increased. For example, in one specific configuration, the bottom two tower sections use Q690, the middle three tower sections use Q550, the upper two tower sections use Q460, and the locally reinforced area of ​​the connecting plate uses Q890; in another specific configuration, the bottom tower section uses Q890, the lower two tower sections use Q690, the upper three tower sections use Q460, and the upper one tower section uses Q355. The above two point value schemes are applicable to high-load nearshore marine environments and conventional onshore wind farm environments, respectively, and both fall within the scope of protection of the claims.

[0037] The friction connection structure is a high-strength bolt friction connection, employing a torque-shear type high-strength bolt connection pair conforming to EN 14399-10-2018 standard. Preferably, the high-strength bolt friction connection uses M24 or M30 torque-shear type high-strength bolt connection pairs. M24 is suitable for small and medium-sized tower section connections, with a target preload set to 225 kN; M30 is suitable for large-diameter tower section connections, with a target preload set to 355 kN. After installation, the real-time monitoring module tracks preload changes over a long period. When the preload retention rate... When the value is below 0.90, a Level 1 maintenance reminder can be generated; when the preload retention rate... When the threshold is below 0.85, a level-two maintenance reminder can be generated, and early inspection can be recommended. The threshold is derived from the joint calibration of existing wind turbine steel structure connection service data and simulation results of this embodiment, which can provide a clear basis for on-site implementation.

[0038] The real-time monitoring module in this embodiment includes a wireless stress sensor and an image recognition module. The wireless stress sensor is used to collect preload change data, and can be implemented in two ways. In the first implementation, the wireless stress sensor is placed near the high-strength bolt washer and forms a stable coupling with the washer or bolt end. It collects the local strain signal generated in the washer area under the axial clamping force of the bolt. After temperature compensation, zero-point correction, and filtering, the preload at the corresponding moment is calculated based on the pre-established calibration relationship between the washer's local strain and the bolt's axial tensile force. In the second implementation, the wireless stress sensor is placed at predetermined measuring points on the surface of the connecting plate. By monitoring the changes in the local strain field in the bolt hole area and the lap area, strain characteristic parameters related to the connection clamping state are extracted, and the preload change data is obtained by combining the strain-preload inversion model of the connecting plate calibrated before installation. Both implementations use the initial preload as a reference value to output a preload retention rate sequence, and both can achieve continuous acquisition of preload change data without changing the main connection structure and the main force path.

[0039] The preload inversion can be performed using the following formula: in, For a moment The preload force, measured in kilonewtons; For a moment The strain measurement value is expressed in microstrain. The slope coefficient is used for calibration, and the unit is kilonewtons per microstrain. The intercept is calibrated in kilonewtons. and Obtained through pre-installation calibration tests. For example, for an M24 specification connection pair, a set of calibration results might be... , Another set of calibration results may be: , Those skilled in the art can obtain the corresponding coefficients through standard calibration procedures based on the differences in connection configurations.

[0040] The image recognition module is used to acquire image data of the connection surface and generate anomaly recognition results. The anomaly recognition results include at least one or more of the following: rust boundaries, coating damage areas, slip marks, or suspected crack edges. To meet different on-site deployment conditions, this embodiment provides two specific implementation schemes. In the first implementation scheme, the image recognition module performs edge recognition on the on-site side, preprocessing the connection surface image data, performing grayscale enhancement, edge extraction, and anomaly region segmentation, generating the anomaly recognition results, and then uploading them to the cloud-based data-driven platform, thereby reducing upload bandwidth pressure. In the second implementation scheme, the image recognition module only performs image compression and sharpening processing on the on-site side before uploading the connection surface image data to the cloud-based data-driven platform, where the anomaly recognition is performed. Both schemes can achieve the function of "the image recognition module acquiring image data of the connection surface and generating anomaly recognition results" as described in the claims, and are suitable for deployment scenarios with strong edge computing resources and strong cloud computing resources, respectively.

[0041] In this embodiment, the cloud-based data-driven platform can provide design correction suggestions for material configuration parameters and connection parameters based on environmental differences. For example, in wind fields with high salt spray levels and frequent alternation of heat and humidity, if the corrosion characterization ratio of the segmented connection interface is monitored... If the preload increases from 0.8% to 2.6% within six months, the dynamic design correction instruction may require higher steel grades, increased anti-corrosion coating levels, or improved friction surface treatment requirements in subsequent similar wind farm projects; if the preload retention rate is monitored... If the value decreases from 0.98 to 0.88 within three months, the dynamic design correction instruction can require increasing bolt specifications, optimizing bolt spacing, or adjusting the target preload. This enables the gradient arrangement of multiple grades of high-strength steel, friction-type connections of high-strength bolts, and multi-modal monitoring to truly form a collaborative working relationship.

[0042] Preferably, the dynamic design correction command does not merely provide general maintenance prompts, but directly corresponds to specific parameters in the subsequent design process. When the corrosion trend reaches a preset threshold, design correction suggestions can be output to increase the steel grade in the local connection area and / or increase the friction surface treatment level; when the stress relaxation trend reaches a preset threshold, design correction suggestions can be output to increase the bolt size, adjust the bolt spacing, and / or increase the target preload; when the stress relaxation trend and corrosion trend increase simultaneously, combined corrections can be implemented for connection parameters and material configuration parameters. This allows the degradation prediction results to be directly converted into executable engineering parameter correction actions in the subsequent design stage.

[0043] Example 3 This embodiment provides a modular wind turbine tower intelligent design monitoring method applied to the modular wind turbine tower intelligent design monitoring system described in any of the foregoing embodiments, including the following steps: Wind farm environmental data is acquired, and an initial design model for a modular, segmented, multi-faceted, three-dimensional wind turbine tower structure is established. The wind farm environmental data can be jointly provided by anemometers, turbine operation platforms, meteorological databases, and geographic environment databases. To ensure modeling accuracy, the wind farm environmental data is statistically analyzed at daily and seasonal scales to generate extreme value indices, mean indices, and fluctuation indices. The initial design model is established based on the target turbine capacity, hub height, site topographic roughness, and transportation and installation constraints.

[0044] Based on the wind field environmental data, the segmented structural parameters, material configuration parameters, and connection parameters of the initial design model are jointly optimized to generate modular segmented design results. In this embodiment, the joint optimization still employs finite element analysis and genetic algorithms to perform multi-objective collaborative optimization.

[0045] Specifically, the initial population is generated by first using the segment thickness, segment width, edge angle, tower section diameter, steel grade distribution, bolt specifications, bolt spacing, edge distance, target preload, and friction surface treatment grade as variables to be optimized. Then, each set of candidate variables is substituted back into the finite element model to calculate the tower top displacement, stress amplitude at key connection points, overall buckling coefficient, and shear force transfer results at the connection interface under rated wind, extreme wind, and fatigue load spectrum conditions. The number of iterations is then counted based on each stress amplitude interval to calculate the fatigue damage evaluation values ​​as follows: in, This is the fatigue damage evaluation value; For the first The actual number of cycles within a stress amplitude range; For the first The allowable number of cycles within a stress amplitude range; This represents the total number of stress amplitude ranges. It can be determined based on the material fatigue curve and the connection type. If A value less than 1 indicates that the fatigue requirements are met within the design life.

[0046] Based on this, the genetic algorithm selects, crosses, and iterates through mutations to screen candidate schemes that meet the strength, stiffness, buckling stability, and connection anti-slip constraints, according to fitness evaluation items such as total mass, material cost, and tower top displacement, until the modular segmented design result is output.

[0047] Connection status data is collected at the segmented connection interface and uploaded to the cloud-based data-driven platform. The connection status data includes preload variation data, connection surface image data, and anomaly identification results generated based on the connection surface image data. The connection surface image data undergoes edge recognition by an image recognition module on-site to generate the anomaly identification results before being uploaded to the cloud-based data-driven platform. In one specific process, the image recognition module first performs distortion correction and brightness normalization on the original image, then performs edge recognition based on the connection seam boundary and bolt hole boundary, extracting the abnormal edge length, abnormal area area, and abnormal area location. Preload variation data is collected by a wireless stress sensor on an hourly scale or triggered by a load event. Both are aggregated by a field gateway and then sent to the cloud-based data-driven platform.

[0048] The cloud-based data-driven platform integrates the modular sharding design results with the connection status data to generate a connection performance degradation prediction result. In this embodiment, the prediction model in the cloud-based data-driven platform is either a deep neural network model or a time series prediction model, with two specific implementation methods. The first implementation method is a deep neural network model, used to handle the nonlinear coupling relationship between design parameters, environmental parameters, image anomaly features, and preload variation features.

[0049] Specifically, samples are first constructed according to the connection interface number and continuous time window. The input vector is composed of design parameters, environmental statistics, and monitoring features of the same connection interface within a continuous predetermined time window. The design parameters include at least the slice thickness, steel grade, bolt specifications, and initial preload. The environmental parameters include at least the average wind speed, wind speed fluctuation, average temperature, average humidity, and salt spray level. The image anomaly features include at least the anomaly area percentage, anomaly edge length, anomaly location encoding, and anomaly type encoding. The preload change features include at least the preload retention rate sequence and its change slope for the most recent N sampling times. The deep neural network model may include a static feature extraction layer and a time series encoding layer. The former is used to extract the static influence features of the design parameters and environmental parameters, and the latter is used to extract the temporal decay features of the preload retention rate sequence. After fusion, the stress relaxation trend value and corrosion trend value are output within a future predetermined time window. The stress relaxation trend value characterizes the predicted decrease or rate of decrease in the preload retention rate within the future time window, and the corrosion trend value characterizes the predicted increase or rate of increase in the anomaly area percentage within the future time window. The true values ​​in the training samples can be composed of the changes in the actual preload retention rate and the changes in the proportion of abnormal areas in adjacent time windows in the historical monitoring data.

[0050] The training objective can be achieved using the following loss function: in, This is the total loss function; This represents the number of training samples; For the first Predicted stress relaxation trend values ​​for each sample; For the first The true value of the stress relaxation trend corresponding to each sample; For the first Corrosion trend prediction value corresponding to each sample; For the first The true value of the corrosion trend corresponding to each sample; and These are the loss weight coefficients. The loss weight coefficients are adaptively adjusted using the gradient descent method and satisfy... In the two specific training point values, when the monitored sample is dominated by preload decay, the following can be obtained: , When the monitored sample mainly shows corrosion anomalies, the following can be obtained: , The second approach is a time series prediction model, which takes the preload retention rate sequence, the abnormal area ratio sequence, and the environmental time series as inputs to predict the changing trend over several future periods. This approach is suitable for scenarios with stable structures and good data temporal continuity.

[0051] Based on the predicted connection performance degradation results, preventative maintenance reminders and dynamic design correction instructions are output, and the dynamic design correction instructions are fed back to the subsequent design optimization process to form a design-monitoring-prediction-optimization closed loop. The preventative maintenance reminders may include re-inspection time, re-inspection location, and maintenance priority; the dynamic design correction instructions may include suggestions for patch thickness correction, local material grade adjustment, bolt specification adjustment, and target preload correction. Thus, the method embodiment and the system embodiment completely correspond in input, processing, and output relationships and can support each other.

[0052] In a preferred implementation, the generation of dynamic design correction instructions from the predicted connection performance degradation results can be performed according to the following steps: S1, comparing the predicted stress relaxation trend value and corrosion trend value with their corresponding thresholds respectively; S2, if only the stress relaxation trend value exceeds the threshold, it is determined that stress relaxation dominates the degradation, and a connection parameter correction instruction is output, which includes at least one of bolt specification adjustment, bolt spacing adjustment, and target preload adjustment; S3, if only the corrosion trend value exceeds the threshold, it is determined that corrosion dominates the degradation, and a material configuration parameter and / or interface treatment parameter correction instruction is output, which includes at least one of local connection area steel grade adjustment and friction surface treatment level adjustment; S4, if both the stress relaxation trend value and the corrosion trend value exceed the threshold, it is determined that coupling degradation, and a combination of the connection parameter correction instruction and the material configuration parameter correction instruction is output; S5, the dynamic design correction instructions are archived according to the tower segment location, connection interface number, and wind field category, and fed back to the subsequent design optimization process.

[0053] Example 4 This embodiment provides specific application scenarios, namely, the application methods in high salt spray wind fields near the coast and strong turbulent wind fields in mountainous areas on land, and further illustrates the mechanism by which dynamic design correction instructions affect the design of subsequent similar wind fields or similar working conditions.

[0054] In offshore nearshore high-salt-fog wind farm scenarios, the modular segmented multi-ribbed three-dimensional wind turbine tower can reach a height of 150 meters, with a large diameter at the bottom section. The segmented connection interfaces are constantly exposed to high humidity, high salinity, and significant diurnal temperature variations. In the initial project, the intelligent design optimization module established an initial design model based on wind farm environmental data, using Q890, Q690, and Q550 high-strength steels in a gradient arrangement. M30 torque-shear type high-strength bolt connections were selected, and the initial target preload was set at 355 kN. After the project was put into operation, the real-time monitoring module continuously collected preload change data and connection surface image data. The image recognition module performed edge recognition on-site, discovering that the abnormal area ratio of some bottom connection interfaces reached 1.9% after 8 months of operation and 3.4% after 14 months. Simultaneously, the wireless stress sensor detected that the preload retention rate at some connection points decreased to 0.87. Based on this, the cloud-based data-driven platform generates connection performance degradation prediction results, determining that the bottom local connection area in this type of wind field exhibits a coupling trend of accelerated corrosion accompanied by stress relaxation. Therefore, it outputs a Level 1 maintenance reminder and a dynamic design correction instruction. The dynamic design correction instruction requires that in subsequent similar offshore and nearshore projects, the material configuration parameters of the bottom local connection area be increased from Q690 to Q890, the friction surface treatment grade be upgraded, and the local target preload force be adjusted from 355 kN to 370 kN. After subsequent projects adopted the corrected design scheme, simulations showed that the predicted preload retention rate increased from 0.89 to 0.93 over one year, and the corrosion characterization rate growth rate decreased by approximately 18.5%.

[0055] In the scenario of strong turbulent wind fields in mountainous areas, wind speed fluctuations are frequent, alternating loads are significant, corrosion factors are relatively weak, but fatigue risks are high. In the initial project, the intelligent design optimization module used three types of high-strength steel—Q690, Q550, and Q460—in a gradient arrangement, and the segmented connection interfaces used M24 torque-shear type high-strength bolt assemblies. The initial target preload was set at 225 kN. After commissioning, the real-time monitoring module found that the connection surface image data did not show obvious corrosion expansion, but the preload change data fluctuated rapidly after strong gusts in winter. The preload retention rate of some connection points decreased to 0.91 after 6 months and to 0.86 after 10 months. The cloud-based data-driven platform judged, based on the time series prediction model, that the decrease in preload was mainly caused by stress relaxation and local fatigue accumulation due to high-frequency alternating loads. Therefore, it output dynamic design correction instructions to fine-tune the segmented structure parameters, material configuration parameters, and connection parameters for subsequent similar wind fields or similar working conditions. Specific fine-tuning includes: adjusting the thickness of the lower and middle high-load tower sections from 20 mm to 24 mm; changing the steel grade in some middle connection areas from Q550 to Q690; reducing the spacing of high-strength bolts by a predetermined increment; and increasing the target preload from 225 kN to 235 kN. Following the adoption of this modified scheme, fatigue damage evaluation values ​​at the connection interface were observed in similar wind field simulations and prototype verifications. The preload retention rate can be reduced from 0.93 to 0.79, and the minimum value of the preload retention rate during the first year of tower operation can be increased from 0.88 to 0.92.

[0056] The dynamic design correction instructions are not limited to single-project rework, but rather serve as parameter correction records from historical projects, influencing the design phase of subsequent new projects under similar wind farms or conditions. In other words, when a new wind farm project is launched, the intelligent design optimization module, in addition to using the new project's wind farm environmental data, also accesses historical correction records archived in the cloud-based data-driven platform by wind farm type, tower location, and connection interface type. This allows for pre-setting corrections to segmented structural parameters, material configuration parameters, and connection parameters during the initial design model establishment phase. Thus, historical monitoring results are transformed into directly applicable engineering parameter correction bases, rather than remaining as abstract data processing conclusions.

[0057] Furthermore, in subsequent new projects under similar wind farms or operating conditions, the intelligent design optimization module can directly call dynamic design correction instructions generated from historical projects when establishing the initial design model to pre-correct the connection parameters and local material configuration parameters of the segmented connection interfaces. In other words, the preload decay law, abnormal expansion law of the connection surface, and corresponding correction results obtained from historical projects are compiled into parameter correction rules or correction records and participate in the next round of design to achieve continuous correction of the actual engineering structure and connection parameters.

[0058] In the above embodiments, the deep neural network model and the time series prediction model are two specific implementations of the prediction model in the cloud-based data-driven platform; the generation and uploading of anomaly identification results on the field side and the generation of anomaly identification results on the cloud side are two specific implementations of the function of the image recognition module; strain detection near the gasket and strain detection on the surface of the connecting plate are two specific implementations of the function of the wireless stress sensor. Therefore, this invention provides at least two possible implementations for each functional technical feature in the claims, and there is a clear correspondence between each input data, processing flow, and output result. Those skilled in the art can implement this invention based on this specification.

[0059] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A modular wind turbine tower design monitoring system, characterized in that, Including modular, segmented, multi-ribbed, three-dimensional wind turbine towers, The modular segmented multi-ribbed three-dimensional structure wind turbine tower is formed by splicing multiple segments, and a segment connection interface is provided between adjacent segments. The segment connection interface adopts a high-strength bolt friction type connection. The system also includes an intelligent design optimization module, a real-time monitoring module, and a cloud-based data-driven platform. The intelligent design optimization module is used to acquire wind farm environmental data and, based on the wind farm environmental data, jointly optimize the segmented structural parameters, material configuration parameters, and connection parameters of the modular segmented multi-ribbed three-dimensional wind turbine tower to generate modular segmented design results. The real-time monitoring module is located at the segmented connection interface and includes a wireless stress sensor and an image recognition module. The wireless stress sensor is used to collect data on the preload change at the segmented connection interface, and the image recognition module is used to collect image data of the connection surface and generate anomaly identification results. The cloud-based data-driven platform is communicatively connected to the intelligent design optimization module and the real-time monitoring module. It is used to integrate the modular segment design results and connection status data, including the preload change data and the anomaly identification results, to generate the stress relaxation trend and corrosion trend of the segment connection interface. Based on the stress relaxation trend and the corrosion trend, it outputs preventive maintenance reminders and dynamic design correction instructions.

2. The modular wind turbine tower design monitoring system according to claim 1, characterized in that, The dynamic design correction command is used to adjust the connection parameters and / or material configuration parameters for subsequent similar wind farms or similar operating conditions. The connection parameters and / or material configuration parameters include at least one of the following: bolt specifications, bolt spacing, target preload, steel grade of local connection area, and friction surface treatment grade.

3. The modular wind turbine tower design monitoring system according to claim 1, characterized in that, The modular, segmented, multi-ribbed, three-dimensional wind turbine tower uses high-strength steel arranged in a gradient. The high-strength steel includes at least two of Q355, Q460, Q550, Q690, and Q890, and is configured differently along the tower height or in local connection areas according to the structural stress level and environmental conditions.

4. The modular wind turbine tower design monitoring system according to claim 1, characterized in that, The wireless stress sensor is placed near the high-strength bolt washer or on the surface of the connecting plate to invert the preload change data through strain changes.

5. The modular wind turbine tower design monitoring system according to claim 1, characterized in that, The image recognition module performs edge recognition on the field side to generate the anomaly recognition result after preprocessing, edge extraction, and abnormal region segmentation of the connection surface image data. The anomaly identification results include at least one of the following: rust boundary, coating damage area, slippage mark, or suspected crack edge.

6. The modular wind turbine tower design monitoring system according to claim 1, characterized in that, The cloud-based data-driven platform quantifies the stress relaxation trend using the preload retention rate and the corrosion trend using the percentage of abnormal area on the connection surface.

7. The modular wind turbine tower design monitoring system according to claim 6, characterized in that, When the corrosion trend exceeds a preset threshold, the dynamic design correction instruction includes increasing the steel grade in the local connection area and / or increasing the friction surface treatment level; When the stress relaxation trend exceeds a preset threshold, the dynamic design correction instruction includes increasing the bolt size, adjusting the bolt spacing, and / or adjusting the target preload.

8. A modular wind turbine tower design monitoring method, applied to the system described in any one of claims 1 to 7, Its features are, Includes the following steps: Acquire wind farm environmental data and establish an initial design model for a modular, segmented, multi-ribbed, three-dimensional wind turbine tower. Based on the wind field environmental data, the segmented structural parameters, material configuration parameters, and connection parameters of the initial design model are jointly optimized to generate modular segmented design results. At the segmented connection interface, a wireless stress sensor collects preload change data, and an image recognition module collects connection surface image data and generates anomaly recognition results. The modular segmentation design results and connection status data, including the preload change data and the anomaly identification results, are uploaded to the cloud data-driven platform. The cloud-based data-driven platform integrates the modular segmentation design results and the connection status data to generate the stress relaxation trend and corrosion trend of the segmentation connection interface; Based on the stress relaxation trend and the corrosion trend, preventive maintenance reminders and dynamic design correction instructions are output to adjust the connection parameters and / or material configuration parameters for subsequent similar wind farms or similar operating conditions.

9. The modular wind turbine tower design monitoring method according to claim 8, characterized in that, When jointly optimizing the segmented structure parameters, material configuration parameters, and connection parameters, finite element analysis and genetic algorithms are used to perform multi-objective collaborative optimization. The optimization objectives include structural strength, material cost, and fatigue resistance.

10. The modular wind turbine tower design monitoring method according to claim 8, characterized in that, The image recognition module performs edge recognition on the connection surface image data on the field side, generates the anomaly recognition result, and then uploads the anomaly recognition result to the cloud data-driven platform.

Images