Fan blade bending form monitoring method based on node-segment chain model

Abstract

The invention relates to the technical field of structural health monitoring and data processing, in particular to a fan blade bending form monitoring method based on a node-segment chain model, which comprises the following steps: arranging a plurality of sensing nodes on a fan blade; generating a plurality of space line segments by taking the three-dimensional coordinates of the sensing nodes as midpoints; sequentially connecting all spatial line segments end to end, and constructing a node-line segment chain model; performing fitting optimization on the direction of each space line segment in the node-line segment chain model by taking minimization of geometric deviation and constraint of an included angle between adjacent line segments as targets; and according to the optimized node-segment chain model, the three-dimensional coordinates of the sensing nodes and a pre-stored design model of the fan blade, calculating a geometric error, a smoothness index and structural consistency of the node-segment chain model, and generating a comprehensive score. By adopting the scheme, the local deformation feature and the overall smoothness can be considered, the calculation efficiency is relatively high, and the quantifiable evaluation of the fitting quality is realized.

Description

Technical Field

[0001] This invention relates to the field of structural health monitoring and data processing technology, and in particular to a method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model. Background Technology

[0002] As the capacity of wind turbine generators continues to increase, the size of wind turbine blades has significantly increased, with their length generally exceeding 100 meters. Under the long-term action of complex aerodynamic loads, gravity, and inertial forces, the blades are prone to bending and torsional deformation. Real-time and accurate acquisition of the actual bending morphology of the blades is crucial for structural health monitoring, fatigue life assessment, operational status diagnosis, and preventive maintenance decisions.

[0003] Currently, the main technical means for monitoring the morphology of wind turbine blades include the following categories: First, vision-based 3D reconstruction methods, which acquire point cloud data of the blade surface through photogrammetry or laser scanning and then perform 3D reconstruction. However, this method is significantly affected by ambient light, weather conditions, blade surface cleanliness, and shading in practical engineering applications. It also has high requirements for equipment installation location, calibration accuracy, and maintenance, making it difficult to achieve long-term stable, all-weather real-time monitoring in harsh outdoor environments. Second, curve fitting methods based on mathematical models, which typically use a single high-order polynomial or spline curve to globally fit discrete measurement points. While this method is relatively simple to calculate, for complex bending morphologies with characteristics such as local abrupt changes and drastic curvature variations, the globally continuous curve often fails to accurately capture local details, easily leading to overfitting or underfitting, resulting in significant deviations between the fitted curve and the actual morphology in the overall or local areas. Third, finite element inversion methods based on mechanical principles, which acquire strain information through strain sensors deployed on the blade and combine this with a finite element model and material constitutive relations to invert the overall deformation. This method is theoretically rigorous, but it is computationally intensive and time-consuming. It relies heavily on accurate material property parameters and boundary condition assumptions, has poor real-time performance, and is difficult to deploy on edge embedded devices with limited computing resources.

[0004] Furthermore, existing technologies generally focus on the morphology fitting algorithm itself, lacking a systematic, multi-dimensional, and quantifiable evaluation system for fitting quality. Evaluations often rely on a single, coarse indicator such as the overall root mean square error. This lack of an evaluation system makes it difficult for engineers to objectively judge the merits of different fitting methods or the same method under different parameters. It fails to provide clear guidance for iterative optimization of the fitting process and weakens the reliability and persuasiveness of morphology monitoring results for subsequent safety assessments and decision support.

[0005] Therefore, there is an urgent need to provide a method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model, which can take into account both local deformation characteristics and overall smoothness, has high computational efficiency to meet real-time or near-real-time processing requirements, and enables quantifiable evaluation of fitting quality. Summary of the Invention

[0006] This invention provides a method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model. It can take into account both local deformation characteristics and overall smoothness, has high computational efficiency to meet real-time or near-real-time processing requirements, and enables quantifiable evaluation of fitting quality.

[0007] To achieve the above objectives, this application provides the following technical solution:

[0008] A method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model includes the following steps:

[0009] Several sensing nodes are arranged on the wind turbine blades, and the three-dimensional coordinates of each sensing node are obtained.

[0010] Using the three-dimensional coordinates of each sensing node as the midpoint, several spatial line segments are generated, and all spatial line segments are connected end to end in sequence to construct a node-line segment chain model for characterizing the bending morphology of wind turbine blades.

[0011] With the goal of minimizing geometric deviation and constraining the included angle between adjacent line segments, the orientation of each spatial line segment in the node-line segment chain model is fitted and optimized to generate an optimized node-line segment chain model.

[0012] Based on the optimized node-segment chain model, the three-dimensional coordinates of each sensing node, and the pre-stored design model of the wind turbine blade, the geometric error, smoothness index, and structural consistency of the node-segment chain model are calculated.

[0013] A comprehensive score is generated based on the geometric error, smoothness index, and structural consistency.

[0014] Furthermore, a bi-objective optimization function is employed to fit and optimize the orientation of each spatial line segment in the node-segment chain model; the bi-objective optimization function is:

[0015]

[0016] In the formula, The objective function value, Let be the coordinates of the j-th fitted point on the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. Let be the angle between the i-th spatial line segment and the (i+1)-th spatial line segment. and These are the geometric error weighting coefficient and the smoothness weighting coefficient, respectively. This represents the total number of spatial line segments. The number of points within the preset fitting range is set for the i-th spatial line segment.

[0017] Furthermore, the formula for calculating the geometric error is as follows:

[0018]

[0019] In the formula, Let be the geometric error of the i-th spatial line segment. Preset the number of points within the fitting range for the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. Let be the coordinates of the j-th fitted point on the i-th spatial line segment.

[0020] Furthermore, the formula for calculating the smoothness index is as follows:

[0021]

[0022] In the formula, This represents the smoothness index at the connection point between the i-th spatial line segment and the (i+1)-th spatial line segment. Let be the slope of the i-th spatial line segment. Let be the slope of the (i+1)th spatial line segment.

[0023] Furthermore, the formula for calculating the structural consistency is as follows:

[0024]

[0025] In the formula, To ensure structural consistency between the node-segment chain model and the design model of the wind turbine blades. To determine the coordinates of the farthest point of the blade tip in the design model, Let be the length of the i-th spatial line segment. Let be the unit direction vector of the i-th spatial line segment. Let be the geometric error of the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. In the design model and The theoretical coordinates of the corresponding location This represents the total number of spatial line segments. The number of points within the preset fitting range is set for the i-th spatial line segment.

[0026] Furthermore, the formula for calculating the comprehensive score is as follows:

[0027]

[0028] In the formula, For comprehensive scoring, This represents the total number of spatial line segments. Let be the length of the i-th spatial line segment. Let be the geometric error of the i-th spatial line segment. This represents the smoothness index at the connection point between the i-th spatial line segment and the (i+1)-th spatial line segment. To ensure structural consistency between the node-segment chain model and the design model of the wind turbine blades. , , These are the weighting coefficients for geometric error, smoothness, and structural consistency, respectively.

[0029] Furthermore, each sensing node is arranged at equal intervals along the span of the wind turbine blades.

[0030] Furthermore, it also includes:

[0031] Based on the comprehensive score, a rating level is generated; the rating level ranges from low to high as unqualified, qualified, and excellent; if the rating level is unqualified, the spacing between each sensor node is adjusted.

[0032] The principles and advantages of this invention are as follows:

[0033] First, in this scheme, the complex continuous shape is transformed into a piecewise linear structure through the node-segment chain model. This not only accurately depicts the local abrupt changes and non-smooth features of blade bending through local dense point placement, avoiding the oscillation and distortion of the overall fitting of high-order polynomials, but also naturally maintains the continuity of the overall shape through the connection between line segments, thus achieving accurate adaptation and expression of complex deformation patterns.

[0034] Secondly, this scheme adopts a dual-objective optimization function that integrates the minimization of geometric deviation and the constraint of the angle between adjacent line segments. This achieves a close fit between the fitted line segments and the measured data points, while introducing an angle penalty term to effectively suppress the slope abruptness problem that may occur in the piecewise linear model. Thus, while ensuring the accuracy of local fitting, it also ensures the smoothness and naturalness of the overall fitting shape, achieving the optimal balance between fidelity and smoothness.

[0035] Furthermore, by employing an evaluation system that incorporates geometric error, smoothness index, and structural consistency, the quality of the fit is transformed from a traditional, vague, and singular judgment into a clear, multidimensional, and quantifiable precise assessment. This not only ensures that the fitting results have clear engineering interpretability and comparability but also provides precise guidance for the iterative optimization of the fitting parameters.

[0036] Finally, this solution does not rely on precise material parameters or harsh ambient lighting conditions and is compatible with various types of sensors. Its model and algorithm complexity are suitable for operation on edge computing terminals, meeting the engineering requirements for long-term, online, and real-time monitoring of wind turbine blades, and greatly improving the environmental robustness, deployment flexibility, and operation and maintenance economy of the morphology monitoring system.

[0037] In summary, this scheme can balance local deformation characteristics with overall smoothness, has high computational efficiency to meet real-time or near-real-time processing requirements, and enables quantifiable evaluation of fitting quality. Attached Figure Description

[0038] Figure 1 This is a flowchart of an embodiment of a wind turbine blade bending morphology monitoring method based on a node-segment chain model according to the present invention.

[0039] Figure 2 This is a schematic diagram of the node-segment chain model in an embodiment of the wind turbine blade bending morphology monitoring method based on the node-segment chain model of the present invention. Detailed Implementation

[0040] The following detailed description illustrates the specific implementation method:

[0041] It should be noted that, in order to enable weighted and comprehensive calculations of parameters with different physical meanings, the input parameters of all calculation formulas involved in this embodiment and the foregoing invention have been standardized or normalized, converted into dimensionless values. Specifically: all length and coordinate parameters are divided by a characteristic length; in this scheme, they are normalized by dividing by the blade length. All angle parameters are normalized by dividing by 1 degree. Therefore, all variables appearing in the embodiments and formulas below are dimensionless values, thus ensuring the consistency of dimensions in the various operations within the formulas.

[0042] Example 1:

[0043] A method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model is presented in this embodiment, using a wind turbine generator with a blade length of 80 meters as the monitoring object. Figure 1 As shown, it includes the following steps:

[0044] S100: Several sensor nodes are arranged on the wind turbine blades, and the three-dimensional coordinates of each sensor node are acquired. The sensor nodes are arranged equidistantly along the spanwise direction of the wind turbine blades according to a preset initial spacing. Specifically, on the surface or internal main beam of the wind turbine blades, sensor nodes are arranged equidistantly along the spanwise direction, i.e., from the blade root to the blade tip, according to a preset initial spacing. In this embodiment, considering both monitoring accuracy and cost, the preset initial spacing is set to 5% of the blade length, i.e., 4 meters, with a total of 21 nodes (including the blade root and blade tip positions). In this embodiment, a high-precision inertial measurement unit (IMU) is used as the sensor node. Each IMU node can output its three-dimensional coordinates and three-axis attitude angles in real time. A unified world coordinate system is established for all IMU nodes using the RTK-GNSS system or total station of the wind turbine tower base station, and initial calibration is completed to acquire the three-dimensional coordinates of each sensor node.

[0045] S200, such as Figure 2 As shown, several spatial line segments are generated with the three-dimensional coordinates of each sensor node as the midpoint, and all spatial line segments are connected end to end in sequence to construct a node-segment chain model to represent the bending morphology of the wind turbine blade. Specifically, firstly, the length of the spatial line segments is determined: the length of the spatial line segments is 1.8 times the preset initial spacing, that is, 7.2m, to ensure that adjacent spatial line segments have sufficient overlap area; using the attitude angle provided by the IMU, the direction vector of the blade tangent at the sensor node is used as the initial direction for generating the spatial line segments. If there is no reliable direction data, the average direction of the line connecting the sensor node with the adjacent sensor nodes before and after it is used as the initial direction; the 21 generated spatial line segments are connected end to end in sequence from the sensor node to the blade tip to form an initial node-segment chain model representing the current blade morphology.

[0046] S300, with the goal of minimizing geometric deviations and constraining the angles between adjacent line segments, the orientations of each spatial line segment in the node-line segment chain model are fitted and optimized to generate an optimized node-line segment chain model. Specifically, firstly, a preset fitting range is determined. In this embodiment, for the i-th spatial line segment, the preset fitting range is set to a range centered on the midpoint of the spatial line segment with a length of... The leaf segment, and the coordinates of all sensor nodes within that range. This refers to the actual points on the spatial line segment that need to be fitted. Then, using the direction vector of each line segment as the optimization variable, an optimization algorithm library (such as using SciPy's minimize function) is called in the edge computing terminal. The algorithm iteratively solves the problem with a bi-objective optimization function as the objective, minimizing the objective function value by adjusting the direction of each spatial line segment. The bi-objective optimization function is:

[0047]

[0048] In the formula, The objective function value, Let be the coordinates of the j-th fitted point on the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. Let be the angle between the i-th spatial line segment and the (i+1)-th spatial line segment. and These are the geometric error weighting coefficient and the smoothness weighting coefficient, respectively. This represents the total number of spatial line segments. The number of points within the preset fitting range is set for the i-th spatial line segment. In this embodiment, , The included angle threshold is set to 40°, meaning that when the included angle between adjacent spatial line segments approaches 40°, a penalty term is applied. It will significantly increase and suppress unreasonable mutations.

[0049] After optimization, a new set of spatial line segment direction vectors is obtained. Based on the optimized spatial line segment directions and the three-dimensional coordinates of the sensing nodes, the coordinates of the two endpoints of each spatial line segment are recalculated, and the connection between the beginning and end is ensured again, thus obtaining the optimized node-line segment chain model, which is smoother and more closely matches the actual measurement points.

[0050] S400: Based on the optimized node-segment chain model, the three-dimensional coordinates of each sensing node, and the pre-stored design model of the wind turbine blade, calculate the geometric error, smoothness index, and structural consistency of the node-segment chain model.

[0051] Calculate the geometric error of each spatial line segment, and then calculate the length-weighted sum of the geometric errors of all spatial line segments. The formula for calculating the geometric error is as follows:

[0052]

[0053] In the formula, Let be the geometric error of the i-th spatial line segment. Preset the number of points within the fitting range for the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. Let be the coordinates of the j-th fitted point on the i-th spatial line segment.

[0054] In this embodiment, the calculated weighted geometric error is 0.12.

[0055] To evaluate the smoothness of the model morphology, the slope changes of adjacent spatial line segments need to be calculated. Specifically, the slope difference at each connection point is calculated within the two-dimensional projection plane of the wind turbine blade's flaring direction. The slope of the spatial line segment is defined as the tangent of the angle between the projection line of the spatial line segment on the wind turbine blade's flaring plane and the reference axis (the axis pointing from the blade root to the blade tip). The two-dimensional projection plane of the wind turbine blade's flaring direction is determined as follows: the direction of the line connecting the blade root sensing node and the blade tip sensing node, or the direction of the blade's main axis obtained by fitting the coordinates of all sensing nodes, is defined as the reference axis, i.e., the axis pointing from the blade root to the blade tip; the flaring direction defined in the wind turbine blade's design model is used; the two-dimensional projection plane is the plane formed by the aforementioned reference axis and the flaring direction, and the three-dimensional coordinates of the endpoints of all spatial line segments are vertically projected onto this plane for subsequent slope calculations.

[0056] Calculate the smoothness exponents of all connection points and sum their squares. The formula for calculating the smoothness exponent at the i-th connection point is as follows:

[0057]

[0058] In the formula, This represents the smoothness index at the connection point between the i-th spatial line segment and the (i+1)-th spatial line segment. Let be the slope of the i-th spatial line segment. Let be the slope of the (i+1)th spatial line segment.

[0059] In this example, the smoothness component is calculated to be 0.05.

[0060] The optimized line segment chain model was compared with the CAD design model. The tip position deviation and overall morphological similarity were calculated. The formula for calculating structural consistency is as follows:

[0061]

[0062] In the formula, To ensure structural consistency between the node-segment chain model and the design model of the wind turbine blades. To determine the coordinates of the farthest point of the blade tip in the design model, Let be the length of the i-th spatial line segment. Let be the unit direction vector of the i-th spatial line segment. Let be the geometric error of the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. In the design model and The theoretical coordinates of the corresponding position, that is, the coordinates of the point within the preset fitting range of the spatial line segment that replaces the actual coordinates, are manually input in this embodiment. This represents the total number of spatial line segments. The number of points within the preset fitting range is set for the i-th spatial line segment.

[0063] The structural consistency calculated in this example is 0.85.

[0064] S500, based on the geometric error, smoothness index, and structural consistency, a comprehensive score is generated. The formula for calculating the comprehensive score is as follows:

[0065]

[0066] In the formula, For comprehensive scoring, This represents the total number of spatial line segments. Let be the length of the i-th spatial line segment. Let be the geometric error of the i-th spatial line segment. This represents the smoothness index at the connection point between the i-th spatial line segment and the (i+1)-th spatial line segment. To ensure structural consistency between the node-segment chain model and the design model of the wind turbine blades. , , These are the weighting coefficients for geometric error, smoothness, and structural consistency, respectively. In this embodiment, , , The overall score calculated in this embodiment is 0.245.

[0067] S600, Based on the comprehensive score, a rating level is generated; the rating levels, from low to high, include unqualified, qualified, and excellent. Specifically, a threshold range for each rating level is set. In this embodiment, At that time, the rating was excellent. At that time, the rating level was qualified. At that time, the rating level is unqualified.

[0068] The system analyzes whether the scoring level is unqualified. If the scoring level is unqualified, the spacing between each sensor node is adjusted. Specifically, the spacing between each sensor node is reduced, and the process returns to S100. In this embodiment, the spacing between sensor nodes in the entire area is reduced by 1 meter. In other embodiments of this application, the spacing between sensor nodes in a preset key area can be reduced only. If the scoring result is qualified or excellent, the optimized node-line segment chain model is output as the monitoring result, and a new round of monitoring process is re-executed starting from S100 in the next monitoring cycle. The detection result can be connected to the wind turbine SCADA system for visualization or used for subsequent load analysis, fatigue assessment, and operation and maintenance early warning.

[0069] The above are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model, characterized in that: Includes the following steps: Several sensing nodes are arranged on the wind turbine blades, and the three-dimensional coordinates of each sensing node are obtained. Using the three-dimensional coordinates of each sensing node as the midpoint, several spatial line segments are generated, and all spatial line segments are connected end to end in sequence to construct a node-line segment chain model for characterizing the bending morphology of wind turbine blades. With the goal of minimizing geometric deviation and constraining the included angle between adjacent line segments, the orientation of each spatial line segment in the node-line segment chain model is fitted and optimized to generate an optimized node-line segment chain model. Based on the optimized node-segment chain model, the three-dimensional coordinates of each sensing node, and the pre-stored design model of the wind turbine blade, the geometric error, smoothness index, and structural consistency of the node-segment chain model are calculated. A comprehensive score is generated based on the geometric error, smoothness index, and structural consistency.

2. The method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model according to claim 1, characterized in that: A bi-objective optimization function is used to fit and optimize the orientation of each spatial line segment in the node-segment chain model; the bi-objective optimization function is: In the formula, The objective function value, Let be the coordinates of the j-th fitted point on the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. Let be the angle between the i-th spatial line segment and the (i+1)-th spatial line segment. and These are the geometric error weighting coefficient and the smoothness weighting coefficient, respectively. This represents the total number of spatial line segments. The number of points within the preset fitting range is set for the i-th spatial line segment.

3. The method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model according to claim 1, characterized in that: The formula for calculating the geometric error is as follows: In the formula, Let be the geometric error of the i-th spatial line segment. Preset the number of points within the fitting range for the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. Let be the coordinates of the j-th fitted point on the i-th spatial line segment.

4. The method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model according to claim 1, characterized in that: The formula for calculating the smoothness index is as follows: In the formula, This represents the smoothness index at the connection point between the i-th spatial line segment and the (i+1)-th spatial line segment. Let be the slope of the i-th spatial line segment. Let be the slope of the (i+1)th spatial line segment.

5. The method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model according to claim 1, characterized in that: The formula for calculating the structural consistency is as follows: In the formula, To ensure structural consistency between the node-segment chain model and the design model of the wind turbine blades. To determine the coordinates of the farthest point of the blade tip in the design model, Let be the length of the i-th spatial line segment. Let be the unit direction vector of the i-th spatial line segment. Let be the geometric error of the i-th spatial line segment. The coordinates of the j-th actual point within the preset fitting range are given for the i-th spatial line segment. In the design model and The theoretical coordinates of the corresponding location This represents the total number of spatial line segments. The number of points within the preset fitting range is set for the i-th spatial line segment.

6. The method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model according to claim 1, characterized in that: The formula for calculating the overall score is as follows: In the formula, For comprehensive scoring, This represents the total number of spatial line segments. Let be the length of the i-th spatial line segment. Let be the geometric error of the i-th spatial line segment. This represents the smoothness index at the connection point between the i-th spatial line segment and the (i+1)-th spatial line segment. To ensure structural consistency between the node-segment chain model and the design model of the wind turbine blades. , , These are the weighting coefficients for geometric error, smoothness, and structural consistency, respectively.

7. The method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model according to claim 1, characterized in that: Each sensing node is arranged at equal intervals along the span of the wind turbine blades.

8. The method for monitoring the bending morphology of wind turbine blades based on a node-segment chain model according to claim 7, characterized in that: Also includes: Based on the comprehensive score, a rating level is generated; The rating levels, from lowest to highest, include unsatisfactory, satisfactory, and excellent. If the rating is unqualified, the spacing between each sensor node will be adjusted.

Images