Intelligent method for monitoring health state of underwater structure of offshore wind turbine

Abstract

The invention provides an intelligent method for monitoring the health state of an underwater structure of an offshore wind turbine, and belongs to the technical field of ocean engineering structure monitoring based on computer data processing. Measuring an acceleration signal at a typical node position on water of the fan; obtaining low-order actual measurement frequency and modal shape data of the offshore wind turbine structure; establishing a fan structure finite element model; setting a damage degree vector for the underwater structure unit, and obtaining simulation frequencies and modal shape data under different damage degrees; according to the actually measured frequency and modal shape data and the simulation frequency and modal shape data of the fan structure, establishing a fitness objective function; the fitness values are classified based on clustering analysis, a multi-population updating strategy is introduced into a longcattle herd optimization algorithm, a target function is solved, and an optimal damage degree vector is obtained; and monitoring the health state of the underwater structure through the damage degree. According to the method, the health state of the underwater structure can be monitored only by arranging limited monitoring sensors on the overwater structure of the offshore wind turbine for sparse measurement.

Description

Technical Field

[0001] The present invention belongs to the technical field of marine engineering structure monitoring based on computer data processing, and in particular relates to an intelligent method for monitoring the health status of underwater structures of offshore wind turbines. Background Art

[0002] Offshore wind power is a significant new source of clean energy worldwide, and the safe operation of its supporting structures—offshore wind turbines—is crucial. Typically, offshore wind turbines consist of a top-mounted turbine unit and blade structure, a central tower structure, and an underwater foundation structure. These structures are exposed to a complex and harsh marine environment, subject to numerous natural factors, including wave impacts, current erosion, seawater corrosion, and seabed geological changes. Furthermore, these structures are susceptible to structural defects, damage, and foundation scour, potentially threatening the overall safety of the offshore wind turbine structure.

[0003] Offshore wind turbine underwater structure monitoring primarily relies on shipborne multibeam bathymetry systems, diver inspections, and underwater sensor networks. While shipborne multibeam bathymetry systems can obtain relatively detailed seabed topography, they require stringent operating conditions, are expensive to operate at sea, and are subject to weather and other constraints, making them incapable of real-time and continuous monitoring of underwater structures. Diver inspections are inefficient, dangerous, and expensive, and prolonged underwater operations pose significant risks to divers. Furthermore, comprehensive coverage of all underwater structures is difficult. While underwater sensor networks enable real-time monitoring, sensor installation, deployment, maintenance, and replacement are difficult and costly. Furthermore, in complex marine environments, sensors are susceptible to interference, resulting in significant packet loss and data reliability issues. Offshore wind turbine structures typically have a limited number of monitoring sensors installed, but this limited measurement information is not fully utilized, resulting in insufficient intelligent monitoring and early warning capabilities, making it difficult to accurately and timely assess the health of underwater structures. In summary, there is an urgent need and potential for widespread application to develop a lightweight, low-cost, and intelligent method for monitoring the health of offshore wind turbine underwater structures, leveraging the limited measurement information of offshore wind turbine tower structures. Summary of the Invention

[0004] In response to the above problems, the present invention uses the limited measurement information of the offshore wind turbine's above-water structure to perform real-time and continuous monitoring of the underwater structure's health status, and develops a lightweight, low-cost, and intelligent offshore wind turbine underwater structure health status monitoring method to solve the problems of high cost, low efficiency, poor real-time performance, and insufficient intelligence of existing monitoring methods, thereby achieving efficient, accurate, and real-time monitoring and evaluation of the offshore wind turbine's underwater structure health status, thereby ensuring the safe operation of the offshore wind turbine structure.

[0005] The present invention provides an intelligent method for monitoring the health status of underwater structures of offshore wind turbines, which is characterized by comprising the following processes: S1, real-time measurement of acceleration signals at typical node positions of the offshore wind turbine tower structure; S2, using online modal identification methods to obtain real-time low-order measured frequencies and modal vibration shape data of offshore wind turbine structures; S3, establish the offshore wind turbine structure finite element model based on the overall structural size parameters, material parameters, constraints and design data of the offshore wind turbine; S4, setting the underwater unit damage degree vector according to the number and properties of the finite element model units, and obtaining simulation frequency and mode vibration shape data under different damage degree vectors based on numerical simulation; S5, based on the measured frequency and modal vibration data of the offshore wind turbine structure and the simulated frequency and modal vibration data of the finite element model, establish an objective function that comprehensively considers the frequency and modal vibration data, and transform the condition monitoring problem into a problem of minimizing the difference between the measured modal parameters and the simulated modal parameters; S6, based on cluster analysis, classifies fitness values ​​and introduces a multi-population update strategy into the beetle swarm optimization algorithm to obtain the optimal (with the minimum fitness value) damage degree vector by minimizing the objective function; S7, the optimal damage severity vector represents the damage severity of the underwater unit of the offshore wind turbine structure. The damage severity is used to monitor the health status of the underwater structure and assess the potential damage severity.

[0006] Furthermore, the typical positions in S1 include the tower top position, the connection position between towers, and the connection position between the tower and the foundation.

[0007] Furthermore, the number of measured nodes in S2 is ≥3, and the modal order identified by the measured response is ≥2.

[0008] Furthermore, the offshore wind turbine structure finite element model in S3 is established in MATLAB, ANSYS or ABAQUS software, and includes at least a tower, three pile foundations, a nacelle, and blades, wherein the nacelle and blades can be equivalently replaced by applying a concentrated mass method at the top of the tower; the three-leg foundation at the bottom of the model adopts a fixed constraint.

[0009] Furthermore, the S4 includes the following settings according to the number and attributes of the finite element units: Damage vector for underwater units: Selected For the underwater units of offshore wind turbines, a damage degree value is randomly set for each unit, where the value range is (-1, 0), forming a damage degree vector, as follows: ; in: Represents initialization A damage severity vector; For the Set damage level for each underwater unit; Create a package containing Initialize the damage severity matrix with the damage severity vector , as follows: ; Based on numerical simulation, the simulation frequency and modal vibration shape data under different damage degree vectors are obtained; Damage severity vector In terms of the above, the benchmark stiffness matrix of the offshore wind turbine structure finite element model is updated , thereby forming a potential damage stiffness matrix including damage The specific process is: ; in: For the The element stiffness matrix corresponding to the underwater element; By solving the characteristic equation , obtain the frequency of potential damage structures and mode shapes ;in is the benchmark mass matrix of the offshore wind turbine structure finite element model.

[0010] Furthermore, the objective function in S5 is an objective function that integrates measured and potential damage modal parameters, specifically: ; in: ; ; Where: and The first Mode shape and Middle Modal displacements of degrees of freedom; and is the measured modal order and degree of freedom; is the weight coefficient between the frequency and coordinate modal confidence criteria; when the modal parameters (frequency, mode shape) of the potential damaged structure are the same as the measured structural parameters, = 0; otherwise, >0; Through The damage degree vector is numerically simulated and the characteristics are solved to obtain A set of potential damaged structure frequency and mode shape data; by this calculation, we can obtain A fitness value.

[0011] Furthermore, the fitness data is classified based on cluster analysis in S6, and the specific process is as follows: Step 1, Fitness value As input data, for fitness values; Set the number of cluster populations to , randomly selected The cluster centers are Where For the cluster centers; Step 2: Calculate the fitness value and cluster centers The minimum distance to divide the group to which it should belong : ; Step 3: Recalculate each category based on the re-divided clusters The cluster center : ; Step 4: Repeat steps 1 to 3 until convergence, and output the final clustering results and cluster centers. The distance between the obtained cluster center and the optimal fitness value is used as the basis for distinguishing the quality of the population, and the population is divided into the best population, the medium population, and the poor population.

[0012] Furthermore, in S6, the multi-population update strategy is introduced into the beetle swarm optimization algorithm to obtain the optimal damage degree vector by minimizing the objective function. The specific process is as follows: Step 1: Solve In the iteration Damage severity vector ; Step 2: Generate a random direction vector ; Step 3: The left and right tentacles of the longhorn beetle search for the left tentacles in the variable space. ;in For the The distance between the two antennae of the longicorn in the iteration or the antenna sensing diameter; Step 4: Judge The population classification level; Step 5: Establish different population update strategies for different population classification levels, specifically: like is the best population, then the damage degree vector of the potential damaged structure is updated as follows: ;in For the The iteration step size, and is the cognitive coefficient, represents the symbolic function, is the optimal damage severity vector; like If it is a medium population, the damage degree vector of the potential damaged structure is updated as follows: ;in is the Levy flight disturbance variable; like If it is a poor population, the damage degree vector of the potential damaged structure is updated as follows: ;in is a random disturbance variable; Step 6: Update the stiffness matrix ; Step 7: Solve the characteristic equation Obtain predicted structural parameters and ; Step 8: Calculate the fitness value through the constructed fitness function ; Step 9: If , update the best fitness value and the optimal damage severity vector ; Repeat steps 1 to 9 above until the iteration termination condition is met and the optimal damage degree vector is output. .

[0013] Furthermore, the optimal damage degree vector obtained in S7 includes A solved damage level value: If the solved damage level value is ≥1%, the corresponding underwater unit may be damaged, and the severity of the damage can be determined according to the damage level value; if the solved damage level value is <1%, the structure is assumed to be in a healthy state, and the damage level change trend needs to be continuously tracked.

[0014] Compared with the prior art, the present invention has the following beneficial effects: Significant cost-effectiveness: Compared with comprehensive monitoring methods such as underwater sensor networks, this method only requires a limited number of monitoring sensors to be deployed on the offshore wind turbine's surface structure for sparse measurement. It can monitor the health of the underwater structure without installing sensors underwater, significantly reducing the cost of purchasing, installing, and maintaining underwater equipment. Improved real-time performance and accuracy: By collecting dynamic responses such as acceleration of above-water structures in real time and dynamically updating the health status assessment model, it can promptly reflect changes in the health status of underwater structures caused by environmental changes and damage to the structures themselves, thereby improving the real-time performance and accuracy of monitoring. High monitoring and early warning efficiency: Combined with the newly proposed clustering beetle swarm optimization algorithm for data processing and analysis, it can complete large amounts of data processing and health status assessment in a short period of time, improving monitoring and early warning efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the overall process flow of the present invention.

[0016] Figure 2 This is a schematic diagram of the offshore wind turbine structure model of the present invention.

[0017] Figure 3 This is a diagram of damage identification results for working condition A based on the KSBSOA method in an embodiment of the present invention.

[0018] Figure 4 This is a diagram of damage identification results for working condition B based on the KSBSOA method in an embodiment of the present invention.

[0019] Figure 5 This is a diagram of damage identification results for working condition C based on the KSBSOA method in an embodiment of the present invention.

[0020] Figure 6 This is a diagram of damage identification results for working condition D based on the KSBSOA method in an embodiment of the present invention.

[0021] Figure 7 This is a diagram of damage identification results for working condition E based on the KSBSOA method in an embodiment of the present invention.

[0022] Figure 8 This is a diagram of damage identification results for working condition E based on the PSO method in an embodiment of the present invention.

[0023] Figure 9 This is a diagram of damage identification results for working condition E based on the BSO method in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The present invention will be further described below with reference to the following embodiments. It should be understood that the embodiments described are only a portion of the present invention, not all of the embodiments. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are intended to fall within the scope of protection of the present invention.

[0025] In this embodiment, an intelligent method for monitoring the health status of underwater structures of offshore wind turbines is provided. The overall process is as follows: Figure 1 shown.

[0026] 1. Real-time measurement of acceleration signals at typical node locations of the offshore wind turbine tower structure.

[0027] First, set the typical nodes of the offshore wind turbine water structure - tower, such as Figure 2 As shown, typical node positions include the tower top position, the connection position between towers, and the connection position between towers and foundation; specifically, nodes 15, 16, 17, and 18; by deploying acceleration sensors at the node positions above water, acceleration signals at the node positions above water under the action of wind, waves and currents are obtained.

[0028] 2. Obtain low-order measured frequencies and modal vibration shape data of offshore wind turbine structures in real time through online modal identification methods.

[0029] Based on the acquired acceleration signal, the low-order measured frequencies of the offshore wind turbine structure are obtained in real time through the modal identification method. and mode shape data ; In this embodiment, the measured number of nodes is 4 and the measured modal order is 5.

[0030] Among them, online modal identification is to construct a covariance matrix or Hankel matrix through the vibration acceleration output response data, and use linear algebraic decomposition (such as QR decomposition, SVD) to extract the system subspace, and then identify modal parameters such as frequency, damping ratio, and vibration shape.

[0031] 3. Establish a finite element model of the offshore wind turbine structure based on the overall structural size parameters, material parameters, constraints and design data of the offshore wind turbine.

[0032] The finite element model of the offshore wind turbine structure is established based on the overall structural size parameters, material parameters, constraints and design data of the offshore wind turbine. The finite element model is established in MATLAB or ANSYS or ABAQUS software. The mechanical properties of the material include density, Poisson's ratio, elastic modulus and yield strength.

[0033] Establishment of offshore wind power finite element model: Offshore wind power mainly consists of tower, three-pile foundation, nacelle, blades and other components. The nacelle and blades can be replaced by the concentrated mass method at the top of the tower. The finite element model in this embodiment is shown in the figure below. Figure 2As shown. The structure contains 18 nodes and 20 units, and each unit is a homogeneous beam unit. Among them: units 1-9 are pile leg units; units 10-12 are cross bracing units; units 13-15 are diagonal bracing units; units 16-17 are column units; units 18-20 are tower units. The height of the structure is 2.5m. The outer diameters of the pile legs, cross bracing and diagonal bracing units are all 20mm, the outer diameters of the columns and tower units are all 30mm, and the wall thickness of each unit is 2mm. The bottom of the model adopts fixed constraints, and the top unit of the wind turbine adopts equivalent simulation, specifically, a 3.3kg concentrated mass is added to the top node. The basic material properties of the structure include a density of 7.85×10 - 6 kg / mm 3 , Poisson's ratio is 0.3, and the elastic modulus is 2.1×10 5 MPa.

[0034] 4. Set the underwater unit damage degree vector based on the number and properties of the finite element model units, and obtain the simulation frequency and modal vibration shape data under different damage degree vectors based on numerical simulation.

[0035] In this example, a damage severity vector consisting of 17 underwater elements is set based on the number and properties of finite element elements. 17 underwater elements of an offshore wind turbine are selected, and a damage severity value is randomly assigned to each element, with the value range being (-1, 0). This forms a damage severity vector as follows: ; in: Representative A damage severity vector; For the Set damage level for each underwater unit; The damage degree is specifically the ratio of the value obtained by subtracting the initial elastic modulus of the unit from the equivalent elastic modulus of the unit to the initial elastic modulus of the unit. The damage degree is a dimensionless modulus, and the damage degree value range is (-1,0). When the damage degree is 0, it indicates that the unit is in a healthy state, and when the damage degree is -1, it indicates that the unit is completely damaged.

[0036] Create a damage severity matrix containing 60 damage severity vectors , as follows: .

[0037] Based on numerical simulation, the simulation frequency and modal vibration shape data under different damage degree vectors are obtained; Damage severity vector In terms of the above, the benchmark stiffness matrix of the offshore wind turbine structure finite element model is updated , thereby forming a potential damage stiffness matrix including damage The specific process is: ; in: For the The element stiffness matrix corresponding to the underwater element; By solving the characteristic equation , obtain the frequency of potential damage structures and mode shapes ;in is the benchmark mass matrix of the offshore wind turbine structure finite element model.

[0038] 5. Based on the measured frequency and modal vibration data of the offshore wind turbine structure and the simulated frequency and modal vibration data of the finite element model, an objective function that comprehensively considers the frequency and modal vibration is established, and the condition monitoring problem is converted into a problem of minimizing the difference between the measured modal parameters and the simulated modal parameters.

[0039] The objective function is a fusion function of measured and potential damage modal parameters, specifically: ; in: ; ; Where: and The first Mode shape and Middle Modal displacements of degrees of freedom; is the measured number of degrees of freedom; when the modal parameters of the potential damage structure are the same as the measured structural parameters, = 0; otherwise, >0; In this embodiment, the total number of degrees of freedom is 24.

[0040] By numerical simulation and characteristic solution of 60 damage degree vectors, we can obtain A set of potential damaged structure frequency and mode shape data is collected; based on this calculation, 60 fitness values ​​can be obtained.

[0041] 6. Based on cluster analysis, the fitness values ​​are classified, and the multi-population update strategy is introduced into the beetle swarm optimization algorithm to obtain the optimal damage degree vector by minimizing the objective function.

[0042] Step 1: In this embodiment, the fitness value is 60. As input data, for fitness values; Set the number of cluster populations to 3 and randomly select 3 cluster centers, which are .

[0043] Step 2: Calculate the fitness value and cluster centers The minimum distance to divide the group to which it should belong : .

[0044] Step 3: Recalculate each category based on the re-divided clusters The cluster center : .

[0045] Step 4: Repeat steps 1 to 3 until convergence, and output the final clustering results and cluster centers. The distance between the obtained cluster center and the optimal fitness value is used as the basis for distinguishing the quality of the population, and the population is divided into the best population, the medium population, and the poor population.

[0046] The multi-population update strategy is introduced into the beetle swarm optimization algorithm to obtain the optimal damage degree vector by minimizing the objective function. The specific process is as follows: Step 1: Solve In the iteration Damage severity vector ; Step 2: Generate a random direction vector ; Step 3: The left and right tentacles of the longhorn beetle search for the left tentacles in the variable space. ;in For the The distance between the two antennae of the longicorn in the iteration or the antenna sensing diameter; Step 4: Judge The population classification level; Step 5: Establish different population update strategies for different population classification levels, specifically: like is the best population, then the damage degree vector of the potential damaged structure is updated as follows: ;in For the The iteration step size, and is the cognitive coefficient, represents the symbolic function, is the optimal damage severity vector; like If it is a medium population, the damage degree vector of the potential damaged structure is updated as follows: ;in is the Levy flight disturbance variable; like If it is a poor population, the damage degree vector of the potential damaged structure is updated as follows: ;in is a random disturbance variable; Step 6: Update the stiffness matrix ; Step 7: Solve the characteristic equation Obtain predicted structural parameters and ; Step 8: Calculate the fitness value through the constructed fitness function ; Step 9: If , update the best fitness value and the optimal damage severity vector ; Repeat steps 1 to 9 above until the iteration termination condition is met and the optimal damage degree vector is output. .

[0047] By calculating the fitness values ​​under all damage degree vectors in sequence, a vector consisting of 60 fitness values ​​can be obtained, which can be expressed as: .

[0048] The fitness values ​​were subjected to population cluster analysis and were divided into three categories, with the first category being good, the second category being medium, and the third category being poor.

[0049] After multi-strategy movement update, a new beetle position is obtained, and a clustering beetle swarm optimization algorithm is established and executed to output the optimal beetle position, thereby obtaining a new damage degree vector of the underwater structural unit of the offshore wind turbine and realizing real-time monitoring of the health status of the underwater structure of the offshore wind turbine.

[0050] Table 1 Damage conditions in numerical simulation

[0051] Taking into account different damage locations, damage degrees, random noise, modal orders, and the number of measuring points, four groups of single damage conditions and one group of double damage conditions were set. Condition A considered 15% damage to the diagonal brace element E14, 1% random noise, and the first five modes, with 4 measuring points; Condition B considered 15% damage to the diagonal brace element E14, 3% random noise, and the first five modes, with 4 measuring points; Condition C considered 15% damage to the diagonal brace element E14, 3% random noise, and the first three modes, with 4 measuring points; Condition D considered 15% damage to the diagonal brace element E14, 3% random noise, and the first three modes, with 3 measuring points; Double damage condition E considered 10% damage to the diagonal brace element E10, 15% damage to the diagonal brace E14, 3% random noise, and the first three modes, with 3 measuring points. The damage conditions are shown in Table 1. The KSBSO of the present invention is used to identify damage to offshore wind turbine structures under five working conditions (ABCDE). The identification results are as follows: Figure 3-Figure 7 shown.

[0052] For damage conditions A, B, and D, the damage position can be accurately located, and the damage degree is identified as -0.1499, which has an error of 0.067% compared to the set damage degree of -0.15.

[0053] For working condition C, the damage location can be accurately located and the damage extent can be assessed without error.

[0054] In working condition E, the damage locations of units E10 and E14 can be accurately located. The damage degree of unit E10 is -0.0995 with an identification error of 0.5%, and the damage degree of unit E14 is -0.1491 with an identification error of 0.6%. The KSBSO of the present invention can accurately locate the damage location and identify the damage degree, with an error of less than 1%.

[0055] The PSO algorithm and BSO algorithm are used to identify damage for working condition E. The results are as follows: Figure 8 and Figure 9 .

[0056] from Figure 7 As can be seen in the figure, the health status of 17 units of the offshore wind turbine underwater structure is monitored by measuring information from four nodes above water. KSBSO can accurately locate the damage positions of units E10 and E14. The identified damage level of unit E10 is -0.0995, with an identification error of 0.5%. The identified damage level of unit E14 is -0.1491, with an identification error of 0.6%.

[0057] from Figure 8As can be seen in the figure, the PSO algorithm identified the damage level of element E10 as -0.223%, with an identification error of 97.77% compared to the set damage level of -10%. The damage level of element E14 was -7.367%, with an identification error of 50.887% compared to the set damage level of -15%. The identified damage levels in the figure have multiple peaks, with element E5 having the highest damage level at -11.737%, followed by element E6 at -8.395%. If damage levels less than -8% are considered to be damaged, this will result in missed identification of structures with actual damage (E10 and E14) and misidentification of structures without actual damage (E5 and E6).

[0058] from Figure 9 As can be seen in the figure, the BSO algorithm identified the damage level of element E14 as -15.318%, with an identification error of 2.12%. The damage level of element E10 was identified as -0.02794%, with an identification error of 99.72%. The identified damage levels in the figure have a clear peak. If damage levels less than -15% are considered damaged, the actual damaged structure (E10) will be missed. If damage levels less than -0.2% are considered damaged, elements E1 and E11 will be misidentified.

[0059] In summary, the particle swarm optimization algorithm (PSO) and the beetle swarm optimization algorithm (BSO) cannot effectively identify damage, and other intact units seriously interfere with positioning, completely masking structural damage, resulting in missed damage detection and misjudgment. However, the present invention can accurately identify damage, and the interference from other intact units is negligible.

[0060] In summary, the intelligent method proposed in the present invention for monitoring the health status of underwater structures of offshore wind turbines using sparse on-water measurement information has high damage identification performance and high calculation accuracy.

[0061] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

[0062] Although the above describes the specific implementation methods of the present invention, it does not limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. An intelligent method for monitoring the health status of underwater structures of offshore wind turbines, characterized in that: The following processes are included: S1, real-time measurement of acceleration signals at typical node positions of the offshore wind turbine tower structure; S2, using online modal identification methods to obtain real-time low-order measured frequencies and modal vibration shape data of offshore wind turbine structures; S3, establish the offshore wind turbine structure finite element model based on the overall structural size parameters, material parameters, constraints and design data of the offshore wind turbine; S4, setting the underwater unit damage degree vector according to the number and properties of the finite element model units, and obtaining simulation frequency and mode vibration shape data under different damage degree vectors based on numerical simulation; S5, based on the measured frequency and modal vibration data of the offshore wind turbine structure and the simulated frequency and modal vibration data of the finite element model, establish an objective function that comprehensively considers the frequency and modal vibration data, and transform the condition monitoring problem into a problem of minimizing the difference between the simulated modal parameters and the measured modal parameters; S6, based on cluster analysis, classifies fitness values, introduces a multi-population update strategy into the beetle swarm optimization algorithm, and obtains the optimal damage degree vector by minimizing the objective function; S7, the optimal damage severity vector represents the damage severity of the underwater unit of the offshore wind turbine structure. The damage severity is used to monitor the health status of the underwater structure and assess the potential damage severity.

2. The intelligent method for monitoring the health status of underwater structures of offshore wind turbines according to claim 1, characterized in that: Typical positions in S1 include the tower top position, the connection position between towers, and the connection position between the tower and the foundation.

3. The intelligent method for monitoring the health status of underwater structures of offshore wind turbines according to claim 1, characterized in that: The number of measured nodes in S2 is ≥3, and the modal order identified by the measured response is ≥2.

4. The intelligent method for monitoring the health status of underwater structures of offshore wind turbines according to claim 1, characterized in that: The offshore wind turbine structure finite element model in S3 is established in MATLAB, ANSYS or ABAQUS software, and includes at least a tower, three pile foundations, a nacelle and blades. The nacelle and blades can be equivalently replaced by applying a concentrated mass method on the top of the tower; the three-leg foundation at the bottom of the model adopts fixed constraints.

5. The intelligent method for monitoring the health status of underwater structures of offshore wind turbines according to claim 1, characterized in that: The S4 includes the following settings according to the number and properties of the finite element units: Damage vector for underwater units: Selected For the underwater units of offshore wind turbines, a damage degree value is randomly set for each unit, where the value range is (-1, 0), forming a damage degree vector, as follows: ； in: Represents initialization A damage severity vector; For the Set damage level for each underwater unit; Create a package containing Initialize the damage severity matrix with the damage severity vector , as follows: ； Based on numerical simulation, the simulation frequency and modal vibration shape data under different damage degree vectors are obtained; Damage severity vector In terms of the above, the benchmark stiffness matrix of the offshore wind turbine structure finite element model is updated , thereby forming a potential damage stiffness matrix including damage The specific process is: ； in: For the The element stiffness matrix corresponding to the underwater element; By solving the characteristic equation , obtain the frequency of potential damage structures and mode shapes ;in is the benchmark mass matrix of the offshore wind turbine structure finite element model.

6. The intelligent method for monitoring the health status of underwater structures of offshore wind turbines according to claim 5, characterized in that: The objective function in S5 is an objective function that integrates measured and potential damage modal parameters, specifically: ； in: ； ； Where: and The first Mode shape and Middle Modal displacements of degrees of freedom; and is the measured modal order and degree of freedom; is the weight coefficient between the frequency and coordinate modal confidence criteria; when the modal parameters of the potential damage structure are the same as the measured structural parameters, = 0; otherwise, > 0; Through The damage degree vector is numerically simulated and the characteristics are solved to obtain A set of potential damaged structure frequency and mode shape data; by this calculation, we can obtain A fitness value.

7. The intelligent method for monitoring the health status of underwater structures of offshore wind turbines according to claim 6, characterized in that: In S6, the fitness data is classified based on cluster analysis. The specific process is as follows: Step 1, Fitness value As input data, For the fitness values; Set the number of cluster populations to , randomly selected The cluster centers are Where For the cluster centers; Step 2: Calculate the fitness value and cluster centers The minimum distance to divide the group to which it should belong : ； Step 3: Recalculate each category based on the re-divided clusters The cluster center : ； Step 4: Repeat steps 1 to 3 until convergence, and output the final clustering results and cluster centers. The distance between the obtained cluster center and the optimal fitness value is used as the basis for distinguishing the quality of the population, and the population is divided into the best population, the medium population, and the poor population.

8. The intelligent method for monitoring the health status of underwater structures of offshore wind turbines according to claim 7, characterized in that: In S6, the multi-population update strategy is introduced into the beetle swarm optimization algorithm to obtain the optimal damage degree vector by minimizing the objective function. The specific process is as follows: Step 1: Solve In the iteration Damage severity vector ; Step 2: Generate a random direction vector ; Step 3: The left and right tentacles of the longhorn beetle search for the left tentacles in the variable space. ;in For the The distance between the two antennae of the longicorn in the iteration or the antenna sensing diameter; Step 4: Judge The population classification level; Step 5: Establish different population update strategies for different population classification levels, specifically: like is the best population, then the damage degree vector of the potential damaged structure is updated as follows: ;in For the The iteration step size, and is the cognitive coefficient, represents the symbolic function, is the optimal damage severity vector; like If it is a medium population, the damage degree vector of the potential damaged structure is updated as follows: ;in is the Levy flight disturbance variable; like If it is a poor population, the damage degree vector of the potential damaged structure is updated as follows: ;in is a random disturbance variable; Step 6: Update the stiffness matrix ; Step 7: Solve the characteristic equation Obtain predicted structural parameters and ; Step 8: Calculate the fitness value through the constructed fitness function ; Step 9: If , update the best fitness value and the optimal damage severity vector ; Repeat steps 1 to 9 above until the iteration termination condition is met and the optimal damage degree vector is output. .

9. The intelligent method for monitoring the health status of underwater structures of offshore wind turbines according to claim 8, characterized in that: The optimal damage degree vector obtained in S7 includes The solved damage level value of each underwater unit: If the solved damage level value is ≥1%, the corresponding underwater unit may be damaged, and the severity of the damage can be determined according to the damage level value; if the solved damage level value is <1%, the structure is assumed to be in a healthy state, and the damage level change trend needs to be continuously tracked.

Images