A method and system for structural monitoring and anomaly identification based on refined inverse finite element during full-scale testing of a wind turbine blade

Abstract

The application discloses a kind of structure monitoring and abnormal identification method and system based on fine inverse finite element in full-size test process of wind turbine blade, method includes: using finite element simulation and load mapping method, obtain the three-dimensional strain field and displacement field of wind turbine blade under torsional load;Each blade component is pre-classified, the corresponding monitoring scheme is determined, and full-size test experiment is carried out;For different monitoring schemes, different inverse finite element calculation modules are used to process the monitoring data of wind turbine blade, and the three-dimensional strain field and displacement field of wind turbine blade are solved;The spatial correlation of finite element analysis result and inverse finite element monitoring result is combined, and the specific damage position of wind turbine blade is accurately identified and positioned using numerical optimization method;In addition, the simulation data obtained by finite element analysis and the measured data obtained by inverse finite element monitoring can be directly solved to obtain the torsional stiffness of wind turbine blade.

Description

TECHNICAL FIELD

[0001] The present application relates to a structure monitoring and anomaly identification method in a full-size test process of a wind turbine blade, in particular to a structure monitoring and anomaly identification method and system based on refined inverse finite element in a full-size test process of a wind turbine blade. BACKGROUND

[0002] Wind energy, as a renewable energy, plays an increasingly important role in the development and utilization of the world's power supply from fossil fuels to clean energy to combat climate change. Wind turbine blades are developing towards super-long flexibility.

[0003] The wind turbine blade structure is complex, including main beam, web, small web, blade root bolt, C-shaped connecting piece, skin, leading edge, trailing edge, UD, structural adhesive, core material and other key components. The main materials used are glass fiber, carbon fiber, multi-axial cloth, biaxial cloth, uniaxial cloth, balsa wood, etc. The process includes pultruded plate and vacuum infusion, etc. During manufacturing, transportation and working process, cracks, interlaminar shear, interlaminar delamination, fiber rupture, separation of resin and fiber, degradation and dissolution and other damages are often formed. The damage often causes significant loss of blade cost and construction cost.

[0004] IEC 61400 and DNV(GL) and other industry series standards stipulate that wind turbine blades must undergo full-size test before passing type approval. Currently, the industry commonly conducts forward flapping, reverse flapping, forward edgewise, reverse edgewise and corresponding fatigue tests. The current full-size test generally uses strain gauge-based structure detection method, and a small amount of fiber Bragg grating monitoring method is used. However, these methods have many measuring points, high cost, long cycle, and relatively few available data. Specifically, the measuring points are as high as hundreds, the labor cost of laying strain gauges is high; the cost of a single type blade full-size test is hundreds of thousands of yuan; the duration is as long as several months; relatively speaking, the main test data are bearing capacity and stiffness, and there is a lack of effective testing means for important parameters such as torsional stiffness.

[0005] The inverse finite element method has many advantages in structural monitoring, including: (1) no prior knowledge is required, other shape sensing techniques such as Ko displacement theory method and modal method require prior acquisition of modal characteristics, load data and material properties of wind turbine blades and other parameters, while the inverse finite element method does not require prior acquisition of relevant information; (2) real-time monitoring, the inverse finite element method combines the strain data collected by the sensor in real time, which can monitor the state and performance changes of the wind turbine blade in time, and help to find potential problems and carry out preventive maintenance; (3) global information acquisition, through the inverse finite element method, the global information of the structure can be obtained from the discrete strain data of the wind turbine blade, and the global reconstruction of the displacement field and strain field of the blade can be realized, which is helpful to the comprehensive evaluation of the health condition of the wind turbine blade; high accuracy, the inverse finite element method can accurately back-propagate the structural characteristics of the wind turbine blade through mathematical model and optimization algorithm, which improves the accuracy and reliability of the monitoring results; (4) non-destructive testing, the inverse finite element method does not need to carry out destructive testing on the wind turbine blade, only needs to collect the strain data of the blade through the sensor, so it is a non-destructive testing method, which will not cause additional damage to the structure of the blade; (5) data-driven: the inverse finite element method is a data-driven technology, which can adjust the model parameters according to the measured data, better reflect the actual state of the wind turbine blade, and adapt to different working conditions and environmental changes.

[0006] The inverse finite element structure monitoring method has not been applied in the fan blade. In a small number of related literatures and patents, the image reconstruction identification and the convolutional neural network image recognition algorithm model damage detection training are mainly used to carry out the wind turbine blade damage detection. However, (1) the length of the current super-long flexible wind turbine blade is close to 130 meters, and the single layer thickness is in the order of millimeters, and the damage size is in the order of sub-millimeters. The existing finite element model node number in the industry is in the order of millions. If the neural network method is used to carry out a large number of finite element forward calculation and calibrate the inverse finite element calculation results, the calculation amount will be astronomical, and it is difficult to realize real-time monitoring and abnormal identification. (2) The existing inverse finite element method for shape sensing of the wind turbine blade is mainly based on the plate shell theory. This method needs to monitor the strain on both sides of the wind turbine blade to realize shape sensing. However, it is difficult to install test devices such as strain gauges or optical fibers on the outside of the super-long flexible wind turbine blade, and it will also affect the aerodynamic characteristics of the wind turbine blade. (3) The wind turbine blade itself is a composite material structure, which has multiple layers and is composed of a skin, a core material, a main beam and other composite structures. The single image recognition method can only reconstruct the strain field and displacement field on the inner and outer surfaces of the wind turbine blade, and cannot consider the damage and complex stress mode of the leading edge, trailing edge and structural adhesive. (4) The image recognition method cannot solve the key problems such as inaccurate torsional stiffness measurement, difficulty in accurately determining the cause of strain abnormalities, and cannot consider the pre-bending, pre-torsion, detailed structure damage, material interface property mutation, clamp tightness, pre-tightening force not reaching the design index and other problems. (5) The existing inverse finite element method for wind turbine blade damage monitoring needs to arrange sensors regularly, which is difficult to realize in places where personnel cannot enter such as the blade tip. SUMMARY

[0007] In order to solve the problems in the background art, the present application provides a wind turbine blade full-size test process based on a refined inverse finite element structure monitoring and abnormal identification method and system, which can greatly reduce the amount of forward finite element calculation for calibration, reduce the number of strain gauges or optical fiber strain sensors needed on both sides, reduce the number of three-way strain gauges needed, and can identify detailed structure and material damage under complex stress modes such as bending and torsion coupling.

[0008] The technical scheme adopted by the present application is:

[0009] The wind turbine blade full-size test process based on the refined inverse finite element structure monitoring and abnormal identification method of the present application comprises the following steps:

[0010] Step 1, various loads transmitted to the wind turbine blade during the full-size test process of the wind turbine blade are simulated by using finite element, and a load mapping method is used to perform multi-scale refined analysis on the wind turbine blade, and a three-dimensional strain field and displacement field of the wind turbine blade under torsional load are obtained;

[0011] Step 2, for multi-scale three-dimensional refined analysis of each loading level, pre-classification of stress modes of each blade component is realized, and corresponding monitoring schemes are determined for different stress modes of different components, and full-size test experiments are performed;

[0012] Step 3, for different monitoring schemes, different inverse finite element calculation modules are used, the stiffness matrix is extracted and coupled with the inverse finite element calculation module, the monitoring data of the wind turbine blade are processed, and the three-dimensional strain field and displacement field of the wind turbine blade are solved;

[0013] Step 4, combining the spatial correlation of the finite element analysis result and the inverse finite element monitoring result, the numerical optimization method is used to accurately identify and locate the specific damage position of the wind turbine blade; in addition, the simulation data obtained by the finite element analysis and the measured data obtained by the inverse finite element monitoring can be directly used to solve the torsional stiffness of the wind turbine blade.

[0014] In the above technical solution, further, the load mapping method in step 1 specifically includes:

[0015] In the blade root coordinate system, according to the principle that the sum of the internal force and the bending moment caused by the distributed force is consistent with the aerodynamic elastic response calculation result, a balance condition is established, the blade beam model design load in the loading direction is mapped on all nodes on the surface of the shell finite element model, and it is assumed that the node force is linearly distributed in the pitch and flap directions, and the axial force is uniformly distributed. Through the full-size test loading load mapping method, multi-scale refined analysis during the full-size test process of the wind turbine blade is realized. Through the parameterized analysis program, the displacement, strain and stress of the key components such as the main beam, the web, the small web, the blade root bolt, the C-shaped connecting piece, the skin, the leading edge, the trailing edge, the UD, the structural adhesive, the core material and the like during the loading process can be refinedly predicted, especially the interlaminar shear stress inside the composite material, the shear force caused by torsion and the like. The pre-bending, pre-torsion and three-dimensional refined finite element model of the wind turbine blade are considered, the stiffness matrix is extracted, and stored in the memory of the computer for subsequent rapid real-time calculation of the forward result.

[0016] Further, the multi-scale three-dimensional refinement analysis in step 2 for each load level realizes the pre-classification of the stress mode of each wind turbine blade component, and on this basis, the refinement design of the strain monitoring scheme is carried out. For example, the main beam is mainly in a pure bending stress mode, and a one-way strain monitoring scheme is used on the outside of the wind turbine blade; if there is strong torsional shear force on the leading edge and trailing edge, three-way strain monitoring points are used on the inside and outside of the wind turbine blade. Strain gauges or optical fibers are used for monitoring equipment; among them, optical fibers are used to collect continuous strain data along the spanwise of the wind turbine blade; strain gauges are used to collect discrete strain data, and interpolation algorithms are used to generate strain along the spanwise path; strain measuring devices should be arranged on the inside and outside of the same position of the wind turbine blade, and for positions on the wind turbine blade where sensors cannot be arranged on the inside of the blade, the same assumption of the inside surface strain data and the outside surface strain data is adopted. Further, on the basis of three-dimensional refinement analysis, composite material damage criteria can also be integrated, a composite material damage judgment factor is proposed, and the monitoring scheme is refined according to the damage judgment factor. For example, at the maximum chord length, the damage factor is usually large, so the monitoring scheme at this position is refined. The failure criteria used include Hashin failure criterion, Tsai-Wu failure criterion, Puck failure criterion, Hoffman failure criterion, Christensen failure criterion, and Azzi-Tsai-Hill failure criterion, which can be fine-tuned for parameters, and at the same time, it is not limited to a certain type of failure criterion, and has strong scalability and flexibility.

[0017] Further, different inverse finite element calculation modules are used for different monitoring schemes, such as a pure bending inverse finite element calculation module for the main beam and a bending-shear-torsion inverse finite element calculation module for the leading edge and trailing edge, and a coupling calculation method of different inverse finite element calculation modules is proposed. For the pure bending module, a one-way strain measuring point is used, and a weight coefficient of a fraction is taken for directions other than the main strain direction; for the bending-shear-torsion coupling module, a three-way strain measuring point is used, and a weight coefficient of 1 is taken. According to the monitoring points in the monitoring scheme, a pre-storage method of the inverse finite element stiffness matrix is established and stored in the memory of the computer, realizing the fast real-time calculation of the three-dimensional strain field and displacement field of the wind turbine blade by the inverse finite element method.

[0018] Further, the method can be directly used to solve the torsional stiffness of an ultra-long flexible wind turbine blade. Due to the difficulty in accurately controlling the loading angle, the current test of the torsional stiffness of an ultra-long flexible wind turbine blade is a major problem faced by the industry. There is a large error between the measured data and the fine finite element data simulation, and even different tests may have different torsional stiffness. At the same time, combined with the spatial correlation algorithm of high-precision finite element calculation and inverse finite element analysis, the specific damage position of the wind turbine blade can be accurately identified and located.

[0019] By further deepening the damage location analysis, using numerical optimization methods such as least squares method and regularization theory, and spatial correlation techniques to accurately assess the type, location and size of damage, to provide scientific basis for the maintenance and strengthening of wind turbine blades.

[0020] Firstly, the strain field ε sim and displacement field u sim of the wind turbine blade under torsional load are obtained by finite element analysis.

[0021] Then full-size test is carried out, and three-dimensional strain field ε meas or displacement field u meas in the actual test process is realized by inverse finite element.

[0022] Comparing the simulated strain field and the measured strain field, the difference δε is calculated:

[0023] δε i = ε meas,i - ε sim,i (K torsion ), where K torsion is the torsional stiffness.

[0024] Define the objective function J(K torsion ) to quantify the error between the measured data and the simulated data:

[0025] J(K torsion ) = ∑ i (ε meas,i - ε sim,i (K torsion )) 2 .

[0026] The goal is to minimize J(K torsion ) to estimate K torsion .

[0027] Use optimization algorithm to minimize J(K torsion ) and estimate K torsion . Assuming the use of gradient descent algorithm, the update rule can be expressed as:

[0028]

[0029] Where, α is the step size, n is the iteration step, is the gradient of J

[0030] Due to the complexity of wind turbine blade structure and the nonlinearity of the structure, the complexity of the system free vibration increases, and the superposition of multiple vibration modes appears. The above inverse finite element monitoring method obtains the three-dimensional displacement field and strain field of the wind turbine blade, which can be used to monitor the high-order mode and damping ratio of the super-long flexible wind turbine blade in free vibration test.

[0031] First, the displacement field from the high-precision finite element calculation is analyzed by least square method, and the inverse finite element calculation is used to locate the damage. A three-dimensional model is used to illustrate the method. The displacement field from the high-precision finite element calculation and the displacement field from the inverse finite element calculation are u sim and u meas respectively, and x, y, z are the spatial coordinates.

[0032] The high-precision finite element calculation result u sim (x, y, z) and the inverse finite element calculation result u meas (x, y, z) are formatted into numerical matrix form.

[0033] The spatial cross-correlation matrix R(m, n, t) is calculated using the following formula.

[0034] R(m, n, t) = ∑ x ∑ y ∑ z u sim (x, y, z) · u meas (x + m, y + n, z + t).

[0035] Where m, n, t represent the relative displacement in different directions of the measurement points, which is used to translate the u meas data in space to test the correlation with u sim .

[0036] In order to make the results independent of the size of the displacement, the normalized spatial cross-correlation formula is used:

[0037]

[0038] Where the cross-correlation value p(m, n, t) has a value range of [-1, 1], a value of 1 indicates a complete positive correlation, -1 indicates a complete negative correlation, and 0 indicates no correlation.

[0039] Check the normalized cross-correlation matrix. In the area with small cross-correlation values, especially near zero or negative values, these areas may indicate that there is a large difference between u sim and u meas , which may be due to the damage to the structure of the wind turbine blade in this area, changing the displacement response of this part.

[0040] By comparing the distribution of cross-correlation values with the geometry of the structure, the damage can be located. By observing the positions of low correlation values in the cross-correlation distribution map, and mapping these positions to the geometry of the wind turbine blade. If the displacement distribution of a certain area has indeed changed significantly compared to the undamaged state, then this area is likely to be the damaged area.

[0041] The beneficial effects of the present application are: the present application combines fine inverse finite element analysis and efficient strain monitoring scheme, and provides an innovative, efficient and accurate structure monitoring and abnormal identification method for full-size test of wind turbine blades. Through optimized monitoring design and high-precision damage identification algorithm, the efficiency and accuracy of wind turbine blade test are greatly improved. Among them, since each component is fine modeled in the fine modeling of the wind turbine blade, spatial cross-correlation analysis can be performed on individual components to locate damage, thereby greatly reducing the amount of calculation. The overall structure can also be analyzed for correlation and damage location. In addition, the spatial correlation method used in the present application does not require a large number of non-physical assumptions, and has strong applicability. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 is a flow chart of the structure monitoring and abnormal identification method of the wind turbine blade based on fine inverse finite element of the present application;

[0043] Figure 2 is a flow chart of the wind turbine blade damage accurate identification and positioning based on spatial correlation algorithm of high-precision finite element calculation and inverse finite element analysis;

[0044] Figure 3 is an example of a strain monitoring scheme for a wind turbine blade;

[0045] Figure 4 is a schematic diagram of a strain sensor. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application. It should be noted that, unless otherwise defined, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art of the present application.

[0047] According to a specific embodiment of the present application, a structure monitoring and abnormal identification method based on fine inverse finite element in a full-size test process of a wind turbine blade is as follows:

[0048] I. Load mapping and multi-scale fine analysis

[0049] In the initial simulation stage of full-scale testing of wind turbine blades, the load mapping method is used to achieve multi-scale refined analysis of various loads transferred to the wind turbine blades during the testing process. This step relies on a parameterized analysis program, which can predict the displacement, strain and stress of each key part of the wind turbine blade (such as the main beam, web, etc.) during the loading process, especially the interlaminar shear stress and shear force caused by twisting in the composite material. The load mapping method involved is as follows:

[0050] In the blade root coordinate system, the blade beam model design load in a certain loading direction is mapped on all nodes on the surface of the shell finite element model, and the balance condition is established according to the following equation. Where X i , Y i and r are the coordinates of the i-th node in the blade beam model. F X,i , F Y,i , F z,i and M Z,i represent the flapwise force, edgewise force, axial force and pitch moment acting on the i-th node. In the beam model, the next (i+1) node from the i-th node to the end corresponds to a total of N nodes in the blade shell finite element model for this segment. For a given finite element model node j with coordinates (X j , Y j , Z j ) in this segment, it is assumed that the node forces (f X,j and f Y,j ) associated with the coordinates are linearly distributed in the edgewise and flapwise directions, while the axial force f Z,j is uniformly distributed. The internal force and bending moment caused by the distributed force at the intersection of the pitch axis and the cross section i should be consistent with the aeroelastic response calculation results. Where k1-k7 are parameters to be solved. In order to obtain a unique solution, a set of assumptions is added, i.e. only the flapwise force contributes to the torque, and the edgewise force does not generate torque.

[0051]

[0052] II. Refined design of strain monitoring scheme

[0053] Based on the above analysis, the main components of the wind turbine blade are pre-classified according to different stress modes and strain states, and the refined design of the strain monitoring scheme is further carried out. For different stress modes of the main components, appropriate monitoring schemes are selected, unidirectional strain measuring points are used for components mainly under bending stress mode, and three-directional strain measuring points are used for components mainly under bending-shear-torsion coupled stress mode, to ensure efficient and accurate monitoring of the strain state of the key components. Based on the strain state of different parts of the wind turbine blade, a monitoring scheme is used, such as Figure 3The monitoring points are arranged along the spanwise direction at the positions of the main beam of the suction surface and the main beam of the pressure surface of the wind turbine blade mainly in bending, and three-direction strain monitoring points are arranged along the spanwise path at the pressure side of the leading edge beam and the trailing edge beam of the wind turbine blade in bending-shear-torsion coupling and the suction side of the trailing edge beam.

[0054] Different strain measurement devices such as strain gauges or optical fibers can be used. Continuous strain data along the spanwise direction of the wind turbine blade can be collected by using optical fibers, and discrete strain data can be collected by using strain gauges, and an interpolation algorithm can be used to generate strain along the spanwise path. The strain measurement devices should be arranged on the inside and outside of the same position of the wind turbine blade. In addition, for the problem that sensors cannot be arranged on the inside of the blade due to difficulty in entering for some positions in the wind turbine blade, based on the mechanical properties of thin shells, the same assumption of the strain data on the inner surface and the outer surface is made, thereby greatly reducing the number of required sensors.

[0055] In order to reasonably arrange and distribute the positions of the strain monitoring points, on the basis of three-dimensional refinement analysis, a composite material damage criterion is integrated, a composite material damage judgment factor is proposed, and a monitoring scheme is refined according to the damage judgment factor. For example, at the position of the maximum chord length of the blade, the damage factor is usually large, and therefore the monitoring scheme at this position is refined. The failure criteria used include Hashin failure criterion, Tsai-Wu failure criterion, Puck failure criterion, Hoffman failure criterion, Christensen failure criterion, and Azzi-Tsai-Hill failure criterion, which can be fine-tuned for parameters, and are not limited to a certain type of failure criterion, and have strong scalability and flexibility.

[0056] III. Coupling and optimization of inverse finite element calculation module

[0057] Considering factors such as pre-bending and pre-torsion of the wind turbine blade, a stable stiffness matrix is extracted and coupled with the inverse finite element calculation module. The monitoring data of the wind turbine blade is processed by different inverse finite element calculation modules (such as pure bending, bending-shear-torsion, etc.). In order to ensure the real-time and accuracy of the calculation, a pre-storage method of the inverse finite element stiffness matrix is also established.

[0058] In the inverse finite element method, a least squares function Φ representing the difference between the measured strain value and the numerical strain value is first defined, and Φ is minimized with respect to the entire discrete node freedom to reconstruct the deformation shape of the discrete structure:

[0059] Φ e (u e )=w e ||e(u e )-e ε || 2 +w k ||k(u e )-k ε|| 2 +w g ||g(u e )-g ε || 2

[0060] where e(u e ), k(u e ) and g(u e ) represent the theoretical values of membrane strain, bending curvature and transverse shear strain, respectively, e ε , k ε and g ε represent the corresponding measured values, the membrane strain e ε and the bending curvature k ε are calculated by the following equations, and the transverse shear strain g ε can be ignored. Figure 4 The strain sensors arranged on the surface of the shell element are shown together with the direct strain measurement values, h is the half thickness of the shell element, w e , w k and w g are weight coefficients.

[0061]

[0062]

[0063] The monitoring data of the wind turbine blade is processed by different inverse finite element calculation modules (such as pure bending, bending-shear-torsion, etc.). For the pure bending module, a one-way strain measuring point is used, and a weight coefficient of a small number, such as 10 -4 , is taken for the direction other than the main strain direction. For the bending-shear-torsion coupling module, a three-way strain measuring point is used, and a weight coefficient of 1 is taken. Further, the following can be obtained:

[0064]

[0065]

[0066]

[0067] where A e is the neutral surface of the shell element, and n is the number of strain measuring points in the shell element. These conditions satisfy the strain compatibility relationship, and the equation is minimized with respect to the node displacement freedom, that is, the matrix equation is solved:

[0068]

[0069] Where k is the stiffness matrix, which is only related to the inverse finite element model's geometric information and boundary conditions, and can be extracted and stored in advance. The global stiffness matrix of the inverse finite element system is a symmetric matrix, and a one-dimensional variable bandwidth storage technology is adopted for the pre-storage of the global stiffness matrix to reduce the storage space of the matrix. The storage technology only stores the lower triangular part of the stiffness matrix, and all rows are placed in a one-dimensional floating point array in order, and the diagonal elements are stored in another integer array. e is the displacement of the node in the element.

[0070] Then the strain-displacement relationship of the shell element is applied, the structure is discretized by isoparametric transformation and shape function, and the elements are assembled. Further, the global matrix equation is solved, and finally the inverted global displacement field u of the structure is obtained. On the basis of the global displacement field, the global strain field ε can be obtained.

[0071] Four, solution of torsional stiffness and damage positioning

[0072] The present application can use the obtained monitoring data to directly solve the torsional stiffness of the super-long flexible wind turbine blade through inverse finite element analysis, effectively solving the problem of torsional stiffness testing in the industry. At the same time, combined with high-precision finite element calculation and spatial correlation algorithm of inverse finite element analysis, the specific damage position of the wind turbine blade is accurately identified and positioned.

[0073] First, the strain field ε of the wind turbine blade under torsional load is obtained by finite element analysis sim and displacement field u sim .

[0074] Then full-size testing is carried out, and three-dimensional strain field ε meas or displacement field u meas is realized by inverse finite element during actual testing.

[0075] Compare the simulated strain field and the measured strain field, and calculate the difference δε:

[0076] δε i = ε meas,i - ε sim,i (K torsion ), where K torsion is the torsional stiffness.

[0077] Define the objective function J(K torsion ) to quantify the error between the measured data and the simulated data:

[0078] J(K torsion ) = ∑ i (ε meas,i - ε sim,i (Ktorsion )) 2 .

[0079] Objective is to minimize J(K torsion ) to estimate K torsion .

[0080] Optimization algorithm is used to minimize J(K torsion ) and estimate K torsion . Assuming we use gradient descent algorithm, the update rule can be expressed as:

[0081]

[0082] Where, α is the step size, n is the number of iteration steps, is the gradient with respect to J;

[0083] Due to the complexity of wind turbine blade structure and the nonlinearity of the structure, the complexity of the system free vibration increases, and the superposition of multiple vibration modes appears. The above inverse finite element monitoring method obtains the three-dimensional displacement field and strain field of the wind turbine blade, which can be used to monitor the high-order mode and damping ratio of the super-long flexible wind turbine blade in free vibration test.

[0084] Five, accurate assessment of damage type, location, size

[0085] Combined with high-precision finite element calculation and spatial correlation algorithm of inverse finite element analysis, the specific damage position of wind turbine blade can be accurately identified and located. Specifically, numerical optimization methods such as least squares method and regularization theory, and spatial correlation technology are used to accurately assess the type, location and size of damage, and provide scientific basis for the maintenance and strengthening of wind turbine blades. The technical route is as shown in Figure 2 .

[0086] High-precision finite element calculation displacement field u sim and inverse finite element calculation displacement field u meas Spatial correlation algorithm is used to identify the specific damage position of wind turbine blade. A three-dimensional model is used to illustrate, x, y, and z are spatial coordinates.

[0087] High-precision finite element calculation results u sim (x,y,z) and inverse finite element calculation results u meas (x,y,z) are formatted into numerical matrix form, and the spatial cross-correlation matrix R(m,n,t) is calculated.

[0088] R(m,n,t)=∑ x ∑ y ∑ z u sim (x,y,z)·u meas(x + m, y + n, z + t).

[0089] where m, n, t represent the relative displacement of the measuring point in different directions, respectively, for the correlation between u meas and u sim in space.

[0090] In order to make the result independent of the size of the displacement, the normalized cross-correlation formula is used:

[0091]

[0092] where the cross-correlation value p(m, n, t) has a value range of [-1, 1], and a value of 1 indicates a complete positive correlation, -1 indicates a complete negative correlation, and 0 indicates no correlation.

[0093] The normalized cross-correlation matrix is checked, and in the area with a small cross-correlation value, especially in the area close to zero or negative, it may mean that there is a large difference between u sim and u meas , which may be due to the damage of the structure of the wind turbine blade in this area, changing the displacement response of this part.

[0094] By comparing the distribution map of the cross-correlation value with the geometric map of the structure, the damage can be located. By observing the position of the low correlation value in the cross-correlation distribution map, and mapping these positions to the geometric map of the wind turbine blade. If the displacement distribution of a certain area indeed also changes significantly compared with the undamaged state, then this area may be the damage area.

[0095] The present application combines fine inverse finite element analysis and efficient strain monitoring scheme, and provides an innovative, efficient and accurate structure monitoring and abnormal identification method for full-size testing of wind turbine blades (such as forward flapping, reverse flapping, forward edgewise, reverse edgewise and fatigue test). Through optimized monitoring design and high-precision damage identification algorithm, it provides strong support for long-term stable operation and maintenance of wind turbine blades.

[0096] Those skilled in the art will appreciate that embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0097] The above described embodiments only express the more specific and detailed embodiments of the present application, but are not construed as limiting the scope of the patent application. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades, characterized in that: The following steps are involved: Step 1: Using finite element simulation to simulate the various loads transmitted to the wind turbine blades during the full-scale test of the wind turbine blades, and using the load mapping method to perform multi-scale refined analysis of the wind turbine blades, and obtain the three-dimensional strain field and displacement field of the wind turbine blades under the torsional load; Step 2: Perform a multi-scale, three-dimensional, refined analysis of each loading level to pre-classify the stress patterns of each blade component. Based on the different stress patterns of different components, determine the corresponding monitoring scheme and conduct full-scale testing experiments. Step 3: For different monitoring schemes, different inverse finite element calculation modules are used to extract the stiffness matrix and couple it with the inverse finite element calculation module to process the monitoring data of the wind turbine blades and obtain the three-dimensional strain field and displacement field of the wind turbine blades; Step 4: Combining the spatial correlation between the finite element analysis results and the inverse finite element monitoring results, a numerical optimization method is used to accurately identify and locate the specific damage location of the wind turbine blade. In addition, the torsional stiffness of the wind turbine blade can be directly solved by using the simulation data obtained by finite element analysis and the measured data obtained by inverse finite element monitoring.

2. The method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades according to claim 1, characterized in that: The load mapping method described in step 1 specifically includes: In the blade root coordinate system, the equilibrium condition is established based on the principle that the sum of the internal forces and bending moments caused by the distributed forces is consistent with the calculated results of the aeroelastic response. The design load of the blade beam model in the loading direction is mapped to all nodes on the surface of the shell finite element model. It is assumed that the nodal forces are linearly distributed in the shimmy and flapping directions, while the axial forces are uniformly distributed.

3. The method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades according to claim 1, characterized in that: The multi-scale refined analysis described in step 1 includes: displacement, strain and stress of key components of the wind turbine blade, as well as interlaminar shear stress and torsional shear force within the composite material.

4. The method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades according to claim 1, characterized in that: In step 2, the stress modes of the blade components are pre-classified, and corresponding monitoring schemes are determined for the different stress modes and strain states of the main components. The monitoring scheme includes the arrangement of measurement points and the selection of monitoring equipment; specifically, it includes: Unidirectional strain measurement points are arranged along the span direction on the main beams of the suction and pressure sides of the wind turbine blades, which are mainly subjected to bending. Three-dimensional strain monitoring points are arranged along the span direction on the pressure and suction sides of the leading and trailing edge beams of the wind turbine, which are subjected to bending, shear and torsion coupling. The monitoring equipment is strain gauges or optical fibers. Fiber optics are used to collect continuous strain data along the span direction of the wind turbine blades; strain gauges are used to collect discrete strain data, and an interpolation algorithm is used to generate strain along the span direction. Strain measurement equipment should be arranged on the inner and outer sides of the same position on the wind turbine blade. For locations on the wind turbine blade where it is not possible to arrange sensors on the inner side of the blade, it is assumed that the strain data on the inner surface is the same as that on the outer surface.

5. The method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades according to claim 1, characterized in that: In step 2, based on the three-dimensional refined analysis, the composite material damage criteria are incorporated and the monitoring plan is refined according to the damage judgment factors.

6. The method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades according to claim 1, characterized in that: In step 3, different inverse finite element calculation modules are used for different monitoring positions. Specifically, the monitoring data of the wind turbine blades are processed by different inverse finite element calculation modules, wherein the main beam adopts the pure bending inverse finite element calculation module, and the leading edge and trailing edge adopt the bending, shear and torsion inverse finite element calculation module. For the pure bending module, a unidirectional strain measurement point is used, and the weight coefficient for the direction other than the main strain direction is taken as a decimal. For the bending, shear and torsion coupling module, a three-directional strain measurement point is used, and the weight coefficient is taken as 1.

7. The method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades according to claim 1, characterized in that: Considering the pre-bending, pre-torsion and three-dimensional refined finite element model of the blade, the stiffness matrix is ​​extracted and pre-stored in the computer memory using one-dimensional variable bandwidth storage technology to achieve fast real-time inverse finite element calculation of the blade's three-dimensional strain field and displacement field.

8. The method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades according to claim 1, characterized in that: In step 4, the spatial correlation between the finite element analysis results and the inverse finite element monitoring results is combined to use a numerical optimization method to accurately identify and locate the specific damage location of the wind turbine blade. Specifically, the following steps are performed: the displacement field calculated by the finite element and the displacement field obtained by the inverse finite element analysis are formatted into a numerical matrix form, the normalized spatial cross-correlation formula is used, the normalized cross-correlation matrix is ​​checked, and potential damage areas are identified in areas where the cross-correlation value is zero or negative; the distribution map of the cross-correlation value is compared with the geometric map of the structure, and the damage area is determined by observing the positions of low correlation values ​​and mapping these positions onto the geometric map of the wind turbine blade.

9. The method for structural monitoring and anomaly identification based on refined inverse finite element analysis during full-scale testing of wind turbine blades according to claim 1, characterized in that: In step 4, the torsional stiffness of the wind turbine blade can be directly solved by using the simulation data obtained by finite element analysis and the measured data obtained by inverse finite element monitoring, including: Finite element analysis is used to obtain the strain field of the wind turbine blade under torsional load, namely the simulated strain field, which is related to the torsional stiffness. The strain field during the full-scale test process, namely the measured strain field, is detected by inverse finite element analysis. The simulated strain field and the measured strain field are compared, and the difference between the two is calculated. The objective function is used as the objective function and is minimized through an optimization algorithm to estimate the torsional stiffness.

10. A structural monitoring and anomaly identification system based on refined inverse finite element analysis during full-scale testing of wind turbine blades, characterized in that: Used to implement the method according to any one of claims 1 to 9.

Images