A method and system for real-time monitoring of pretension force of steel cables in mixed towers

Abstract

This invention discloses a method and system for real-time monitoring of pretension force of mixed-tower steel cables, relating to the field of wind power generation technology. The method includes the following steps: acquiring known dynamic characteristic data of mixed-tower steel cables under multiple operating conditions and structures; using global sensitivity analysis to screen key parameters from the dynamic characteristic data that effectively characterize the dynamic response of the mixed-tower steel cable system and the system structure and operating loads, and constructing a dynamic characteristic database based on the key parameters; based on the dynamic characteristic database, constructing a structure-operating condition-response relationship neural network surrogate model using neural networks or regression prediction methods; and using the structure-operating condition-response relationship neural network surrogate model to inversely calculate the pretension force of the mixed-tower steel cables from the real-time operating condition data and motion response data sensed by sensors. This invention enables real-time and accurate monitoring of the pretension force of mixed-tower steel cables, realizing the prediction of pretension force under extreme and fault conditions.

Description

Technical Field

[0001] This invention relates to the field of wind power generation technology, and in particular to a method and system for real-time monitoring of the pretension force of steel cables in mixed towers. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Currently, in the stress research of mixed-tower cable systems, due to the large number of cables, the tight arrangement of each bundle, and limited space, traditional wire rope tension measurement methods and equipment cannot measure the preload force within the cable bundles. Therefore, developing a real-time online preload monitoring system is of significant engineering importance. Firstly, a real-time online preload monitoring system can promptly detect cable failures, allowing for early maintenance measures to ensure effective preload control and reduce the probability of accidents. Secondly, real-time preload monitoring data strongly supports the calculation of cable fatigue life, enabling monitoring-based maintenance for faults such as insufficient preload and cable fatigue failure. Finally, the real-time cable preload monitoring system can collect preload data under multiple operating conditions over extended periods, clarifying the relationship between the preload force of mixed-tower cables and operating conditions and the motion response of the mixed-tower system, providing a basis for subsequent mixed-tower cable system design.

[0004] Current technologies often calculate the pretension of hybrid wind turbine cables using the hydraulic pressure of the tensioning cables. However, after the hydraulic tensioning system is tensioned, the hydraulic station no longer provides hydraulic pressure during the natural contraction phase of the cables, making it impossible to monitor the tension after contraction. Furthermore, because wind loads fluctuate significantly and have large average values ​​during normal wind turbine operation, real-time monitoring of cable pretension is necessary. Although methods such as fiber optics and magnetic flux leakage can achieve real-time monitoring, these require sensors to be deployed on each cable, resulting in high monitoring costs and poor applicability. Therefore, the current lack of effective equipment and methods for measuring cable pretension hinders wind farm operation and maintenance.

[0005] In conclusion, it is necessary to develop an economical and reliable real-time monitoring method for the pretension force of mixed-tower steel cables to reduce the cost of real-time monitoring and prevent the risk of pretension force failure. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for real-time monitoring of the pretension force of mixed tower cables, which can monitor the pretension force of mixed tower cables in real time and accurately, and realize the prediction of pretension force of mixed towers under extreme and fault conditions.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0008] The first aspect of this invention provides a method for real-time monitoring of the preload of steel cables in mixed-tower construction, comprising the following steps:

[0009] Acquire known dynamic characteristic data of mixed tower steel cables under multiple working conditions and multiple structures. The dynamic characteristic data includes structural data, working condition data and corresponding motion response data.

[0010] Global sensitivity analysis was used to screen key parameters from dynamic characteristic data that effectively characterize the dynamic response of the hybrid tower cable system and the system structure and working load. A dynamic characteristic database was then constructed based on the key parameters.

[0011] Based on a dynamic characteristic database, a neural network surrogate model of structure-operating condition-response relationship is constructed using neural network or regression prediction methods.

[0012] By using a structure-condition-response relationship neural network surrogate model, the preload of the mixed tower steel cable is obtained by reverse calculation of the real-time condition data and motion response data sensed by the sensor.

[0013] Furthermore, the specific steps for obtaining known dynamic characteristic data of mixed-tower steel cables under multiple working conditions and multiple structures are as follows:

[0014] A dynamic model of the hybrid tower cable system was constructed. Based on the dynamic model, simulations were performed on hybrid towers of different structural dimensions under different working conditions to determine the motion response of the hybrid tower under wind load.

[0015] Furthermore, the multi-structure parameters include the height of the mixed tower, the number of concrete segments, the radius of the mixed tower, the wall thickness of the mixed tower, the radius of the steel cable, the length of the steel cable, the arrangement of the steel cable, and the pretension force. The multi-condition parameters include the steel cable pretension force under normal condition, insufficient pretension force condition, excessive pretension force condition, unbalanced pretension force condition, and steel cable breakage condition. The structural data includes the height of the mixed tower, the number of concrete segments, the radius of the mixed tower, the wall thickness of the mixed tower, and the arrangement, quantity, diameter, axial stiffness, damping, anchor point position, and pretension force of the steel cable. The condition data includes wind speed, wind direction, wind turbine power, wind turbine speed, blade airfoil, blade chord length, and blade pitch angle. The motion response parameters include the dynamic tilt angle of the tower top, angular velocity, and angular acceleration.

[0016] Furthermore, the specific steps for determining the motion response of the hybrid tower under wind load are as follows:

[0017] Based on Markov chain analysis of local wind speed data, wind speed data is combined with time data, i.e., a one-to-one correspondence is established to establish a speed-time state.

[0018] The local wind speed data is traversed, and all states of the next second are integrated with a set size of 1 m / s. The frequency of different states in the next second is counted.

[0019] Calculate the state transition probability from the current second to the next second;

[0020] Calculate the velocity-time state transition probability matrix from the current second to the next second at different times;

[0021] Monte Carlo simulation of the wind speed time state transition probability matrix was performed using the acceptance-rejection sampling method to obtain the equivalent wind speed curve;

[0022] Multiple equivalent wind speed curves are generated through repeated calculations to ensure that their mean and standard deviation are the same as the local wind speed data, thus obtaining multi-condition simulated wind speed curves.

[0023] Furthermore, by utilizing global sensitivity analysis to screen key parameters from the dynamic characteristic data that effectively characterize the dynamic response of the hybrid tower cable system and the system structure and operating loads, and constructing a dynamic characteristic database based on these key parameters, the specific steps are as follows:

[0024] The contribution rates of structural data and working condition data to the dynamic response of the hybrid tower cable system are evaluated, including the contribution rates of each structural data and working condition data to the dynamic response of the hybrid tower cable system, as well as the contribution rates of the interaction between each structural data and working condition data to the dynamic response of the hybrid tower cable system. Representative structures, working conditions, and corresponding motion response data are then selected.

[0025] The method of fusion based on variance weights fuses the response experimental data, and the optimal weights for data fusion are obtained by using the variance of the data signals.

[0026] A database of dynamic characteristics of hybrid tower cable systems was established based on the fused response data.

[0027] Furthermore, based on the dynamic characteristic database of hybrid tower cable systems, using structural data and operating condition data as input, high-dimensional features are extracted through neural networks, and motion response data is output to clarify the structure-operating condition-response relationship. With the help of neural networks and regression prediction methods, a neural network surrogate model of the structure-operating condition-response relationship is obtained.

[0028] Furthermore, the reverse calculation capability of the structure-operating condition-response relationship neural network proxy model is developed. Using motion response data and operating condition data as input, high-dimensional features are extracted through the neural network, and structural data is output to realize the monitoring of steel cable pretension.

[0029] A second aspect of the present invention provides a real-time monitoring system for the pretension force of mixed-tower steel cables, comprising:

[0030] The data acquisition module is configured to acquire known dynamic characteristic data of mixed tower steel cables under multiple working conditions and multiple structures. The dynamic characteristic data includes structural data, working condition data and corresponding motion response data.

[0031] The data filtering module is configured to use global sensitivity analysis to filter key parameters in the dynamic characteristic data that effectively characterize the dynamic response of the hybrid tower cable system and the system structure and working load, and to build a dynamic characteristic database based on the key parameters;

[0032] The model building module is configured to construct a structure-condition-response relationship neural network surrogate model based on a dynamic characteristic database and using neural network or regression prediction methods.

[0033] The preload monitoring module is configured to use a structure-condition-response relationship neural network proxy model to reverse calculate the preload of the mixed tower steel cable by using the real-time condition data and motion response data sensed by the sensor.

[0034] A third aspect of the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed the steps of the real-time monitoring method for pretensioning of mixed-tower steel cables as described in the first aspect of the present invention.

[0035] A fourth aspect of the present invention provides a computer device comprising:

[0036] A processor, adapted to execute computer programs;

[0037] A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the real-time monitoring method for pretensioning of mixed-tower steel cables as described in the first aspect of the present invention.

[0038] The above one or more technical solutions have the following beneficial effects:

[0039] This invention discloses a method and system for real-time monitoring of pretension force of mixed-tower steel cables. The dynamic response characteristics of a mixed-tower steel cable system are closely related to its structural characteristics and operating loads, while the pretension force affects the system's stiffness characteristics. Therefore, when the pretension force is insufficient, excessive, or uneven, the dynamic response characteristics of the mixed-tower steel cable system differ under the same operating conditions. Structural characteristics such as the pretension force can be monitored based on the system's operating loads and dynamic response characteristics. Given that current wind turbine nacelles already have wind measurement systems installed to capture system operating information, and installed sensors for tilt angles and acceleration to capture the system's macroscopic dynamic response characteristics, this invention aims to calculate the pretension force of steel cables based on existing wind turbine operating data and macroscopic dynamic response characteristics. A structure-operating condition-response relationship for the mixed-tower steel cable system is constructed based on a neural network surrogate model. Based on the reverse calculation capability of this model, the operating condition and response information acquired by the sensors are input into the structure-operating condition-response relationship neural network surrogate model to achieve the monitoring of the pretension force of the steel cables.

[0040] This invention fulfills the requirements for real-time or near-real-time monitoring. After the neural network model is trained, its monitoring process is computationally fast, which is crucial for the safety monitoring of mixed-tower cable systems. Furthermore, existing methods for directly and accurately measuring cable preload (such as the hydraulic jack method or frequency method) are typically difficult, time-consuming, and costly. This method provides a new approach for real-time, continuous monitoring based on easily obtainable indirect signals.

[0041] This invention also reduces reliance on high-precision, high-cost dedicated preload measurement equipment, instead utilizing more economical and easier-to-install sensors (such as angle and acceleration sensors) to acquire signals. The installation of indirect measurement sensors is generally simpler than that of direct measurement equipment, with less structural interference and lower maintenance costs. Automated, continuous preload monitoring can be achieved, significantly reducing the frequency of high-cost, high-risk manual inspections.

[0042] This invention can promptly detect abnormal preload, providing early warning of potential tower failures and preventing safety accidents. Based on accurate preload status information, it enables monitoring-based maintenance, allowing for precise tension adjustments when necessary, avoiding over- or under-maintenance, thereby optimizing maintenance costs and extending the structure's service life.

[0043] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 This is a flowchart of the real-time monitoring method for pretension force of steel cables in mixed towers according to Embodiment 1 of the present invention;

[0046] Figure 2 This is a schematic diagram of the neural network proxy model of the structure-operating condition-response relationship of the hybrid tower cable system in Embodiment 1 of the present invention;

[0047] Figure 3 This is a schematic diagram of the process of monitoring the preload of steel cables in Embodiment 1 of the present invention. Detailed Implementation

[0048] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0049] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0050] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0051] Example 1:

[0052] To address the problem that traditional equipment struggles to effectively monitor the pretension force of mixed-tower steel cables due to their numerous, densely packed arrangement and the risk of fatigue failure, Embodiment 1 of this invention provides a real-time monitoring method for the pretension force of mixed-tower steel cables. This method enables pretension force monitoring during the tensioning process and under extreme and fault conditions. Figure 1 As shown, firstly, dynamic simulations of multiple working conditions and structures are carried out based on the parametric model of the hybrid tower cable system. Then, sensitivity analysis is conducted, response experimental data are integrated, a dynamic characteristic database is established, and a neural network surrogate model of the structure-working condition-response relationship is established and trained. The working conditions and motion response data sensed by the sensors are transmitted into the surrogate model, and the reverse calculation capability of the surrogate model is used to monitor the cable preload.

[0053] Specifically, the following steps are included:

[0054] Step 1: Obtain known dynamic characteristic data of mixed tower steel cables under multiple working conditions and multiple structures. The dynamic characteristic data includes structural data, working condition data and corresponding motion response data.

[0055] This embodiment constructs a dynamic model of a hybrid tower cable system. Based on this model, simulations are performed on hybrid towers of different structural dimensions under various operating conditions to determine their motion response under wind load. The specific steps are as follows:

[0056] Step 1.1: Construct a dynamic model of the hybrid tower cable system and perform dynamic characteristic analysis to determine the motion response of the hybrid tower under wind load.

[0057] In one specific implementation, dynamic characteristic analysis is performed based on the dynamic model of the hybrid tower cable system to clarify the motion response of the hybrid tower under wind load. During operation, the hybrid tower is mainly subjected to wind load, which in this embodiment is considered as the effects of tower wind pressure load and blade aerodynamic load.

[0058] The formula for calculating the wind pressure load on the tower is as follows:

[0059] ,

[0060] .

[0061] In the formula, For tower wind pressure load, Basic wind pressure, air density, As the reference wind speed, For wind vibration coefficient, This is the wind load shape coefficient. This is the wind pressure height variation coefficient.

[0062] The aerodynamic loads of the fan are as follows:

[0063] .

[0064] In the formula, T represents the aerodynamic load of the fan. air density, The wind speed at the height of the wind turbine hub. It is an axial inducing factor. The area swept by the wind turbine.

[0065] During impeller startup, the impeller speed gradually increases, and the aerodynamic load also gradually increases from zero to its maximum. Furthermore, during actual operation, wind speed varies randomly, and the wind pressure load on the hybrid tower changes constantly. To ensure that the constructed wind speed equation more accurately reflects the actual wind load on the hybrid tower, this embodiment uses local actual wind speed data to construct transient wind speed conditions based on Markov chains. In a Markov chain, the current state depends only on the state at the previous moment and is independent of past states. Specifically, given the current state, the probability distribution of future states depends only on the current state and is unaffected by past states; this property is called the Markov property. In this embodiment, the wind load conditions during the hybrid tower operation are considered as a stochastic process with the Markov property, and the wind speed constituting the wind load conditions is used as the basic unit for construction. The specific steps are as follows:

[0066] Step 1.1.1: Analyze local wind speed data based on Markov chain analysis, combine wind speed data with time data to establish a one-to-one correspondence and establish a speed-time state.

[0067] Specifically, the velocity-time state at the current moment is represented by... Indicate, then The possible velocity-time state in the next second can be represented by the set Q:

[0068] .

[0069] In the formula, for The speed-time state of the next second, where This represents the velocity-time state index in set Q. This represents the total number of velocity-time states in set Q.

[0070] Step 1.1.2: Iterate through the local wind speed data, integrate all states of the next second of the current moment with a set size of 1 m / s, and count the frequency of different states in the next second.

[0071] Specifically, the set of frequencies of different states occurring in the next second is denoted by B:

[0072] .

[0073] In the formula, represent The frequency of.

[0074] Step 1.1.3: Calculate the state transition probability from the current second to the next second.

[0075] Specifically, the calculation starts from... arrive The state transition probability is:

[0076] .

[0077] Step 1.1.4: Calculate the velocity-time state transition probability matrix from the current second to the next second at different times.

[0078] Specifically, the velocity-time state transition probability matrix at different times is described by the following formula:

[0079] .

[0080] Where P is the velocity-time state transition probability matrix, and when the state transitions from the current time b to the next time j, it can be described by the following equation:

[0081] .

[0082] In the formula, S is the state space composed of local wind speed data. This indicates the previous state of the wind speed condition. This indicates the next state of the wind speed condition.

[0083] Step 1.1.5: Use the accept-reject sampling method to perform Monte Carlo simulation on the wind speed time state transition probability matrix to obtain the equivalent wind speed curve.

[0084] Because the probabilities in the wind speed time-state transition probability matrix are not uniformly distributed, using uniform sampling in Monte Carlo simulations will produce very large errors, thus affecting the accuracy and precision of the model. Therefore, an accept-reject sampling method is used for Monte Carlo simulations. The accept-reject sampling method in Monte Carlo is used programmatically to construct the load case results, which are stored in a candidate chain. Finally, an error evaluation function is used to calculate the error between each load case curve and the actual load case, and the optimal load case with the smallest error is output.

[0085] The specific steps are as follows:

[0086] 1. Sample from distribution G to obtain a sample Y.

[0087] 2. Sample from the uniform distribution of [0,1] to obtain a sample U.

[0088] 3. Determine if If the value is positive, then accept this Y as the recorded sample value; otherwise, reject the sample value, discard it, and resample.

[0089] Where G represents a uniform distribution; c is a constant value. Let be the probability density function of the target distribution. Let x be the probability density function of the proposed distribution. For any x, we have: In order to improve the efficiency of sampling, c should be taken as a smaller value under the above conditions.

[0090] Step 1.1.6: Generate multiple equivalent wind speed curves through multiple calculations to ensure that their mean and standard deviation are the same as the local wind speed data, and obtain multi-condition simulated wind speed curves.

[0091] Specifically, the time frame is set to 100 seconds, and the wind speed results for 100 seconds are sampled according to the above steps to achieve equivalence for local wind speed operating conditions data within one year.

[0092] Step 1.2: Based on the dynamic model of the hybrid tower cable system, simulations are performed on hybrid towers with different structural dimensions under different working conditions.

[0093] In one specific implementation, the multi-structure includes the following parameters: tower height, number of concrete segments, tower radius, tower wall thickness, cable radius, cable length, cable arrangement, and preload variable parameters. The multi-condition includes the following parameters: normal cable preload condition, insufficient preload condition, excessive preload condition, unbalanced preload condition, and cable breakage condition. The structural data includes the following parameters: tower height, number of concrete segments, tower radius, tower wall thickness, cable arrangement position, quantity, diameter, axial stiffness, damping, anchor point position, and preload. The condition data includes wind speed, wind direction, turbine power, rotor speed, blade airfoil, blade chord length, and blade pitch angle. The motion response parameters include the tower top dynamic tilt angle, angular velocity, and angular acceleration.

[0094] After simulation, the dynamic tilt angle, angular velocity, and angular acceleration of the tower top are recorded as simulation results.

[0095] Step 2: Use global sensitivity analysis to screen key parameters in the dynamic characteristic data that effectively characterize the dynamic response of the hybrid tower cable system and the system structure and working load, and construct a dynamic characteristic database based on the key parameters.

[0096] Step 2.1: Evaluate the contribution rate of structural data and operating condition data to the dynamic response of the hybrid tower cable system, including the contribution rate of each structural data and operating condition data to the dynamic response of the hybrid tower cable system, as well as the contribution rate of the interaction between each structural and operating condition data to the dynamic response of the hybrid tower cable system, and screen out representative structural, operating condition and corresponding motion response data.

[0097] This embodiment selects key variables that significantly affect the dynamic response of the hybrid tower cable system (such as the dynamic tilt angle at the top of the tower, angular velocity, angular acceleration, etc.) from numerous input parameters (hybrid tower geometry, cable size and arrangement, external load, cable preload, etc.), and evaluates the contribution rate of each parameter individually and in interaction to the output response.

[0098] Specifically, first, one parameter from each of the numerous input parameters is selected as a single variable, and the dynamic response is observed. The more obvious the result, the greater the contribution rate. Then, the numerous input parameters are combined sequentially as combined variables, and the influence of each combined variable on the dynamic response is observed. Again, the more obvious the result, the greater the contribution rate. The most representative parameters with the largest contribution rates are then selected as the input parameters for the subsequent surrogate model.

[0099] Based on hybrid tower cable system models with different geometric dimensions, motion response parameters such as dynamic tilt angle, angular velocity, and angular acceleration at the top of the hybrid tower were recorded in 5-second increments.

[0100] Step 2.2: The response experimental data is fused using a variance-weighted fusion method, and the optimal weights for data fusion are determined by the variance of the data signals. This step increases the amount of sample data, improves the accuracy of the trained monitoring model, and ensures that the fused data retains, to the greatest extent possible, the obvious fault characteristics of the simulation data and the real environmental impact characteristics of the experimental data. The response experimental data is obtained through response experiments, which are existing techniques in this field and will not be elaborated upon here.

[0101] The variance weight fusion method is as follows:

[0102] These are the experimental data in state i; This represents the variance corresponding to the experimental data; This refers to the simulation data in state i; This represents the variance corresponding to the simulated pressure data.

[0103] The formula for weighted fusion of experimental and simulation data is:

[0104] .

[0105] In the formula: The fused data for state i; The assigned weight is obtained in state i.

[0106] To alleviate the optimal state i To enable data fusion The fused data contains the most effective features from both types of data, making the fused data... variance Find the optimal weight by using the minimum value in reverse. Since the two sets of signals are independent, taking the variance of both sides of the above equation and simplifying it yields:

[0107] .

[0108] right Differentiate and let =0, solving for the equation yields:

[0109] .

[0110] Substituting the above formula into the formula for weighted fusion of experimental and simulation data yields the optimal estimate, at which point the desired fused signal is obtained. The variance is minimized, so that the fused data contains the most effective features from both types of data, thus achieving data fusion.

[0111] Step 2.3: Establish a dynamic characteristic database of the hybrid tower cable system based on the fused response data to provide data support for establishing the structure-operating condition-response relationship of the hybrid tower cable system.

[0112] Step 3: Based on the dynamic characteristic database, construct a neural network surrogate model of the structure-operating condition-response relationship using neural network or regression prediction methods.

[0113] In one specific implementation, this embodiment is based on the dynamic characteristic database of the hybrid tower cable system. With structural data and working condition data as input, high-dimensional features are extracted through neural networks, and motion response data is output to clarify the structure-working condition-response relationship. By using neural networks and regression prediction methods, a surrogate model based on the structure-working condition-response relationship is obtained.

[0114] Specifically, the input and output variables of the hybrid tower structure-operating condition-response neural network surrogate model exhibit a highly nonlinear relationship. Thanks to the nonlinear activation function within the neural network units, the neural network algorithm performs exceptionally well in handling these nonlinear relationships. This embodiment selects a neural network algorithm to establish the mapping relationship between the input and output variables of the hybrid tower structure-operating condition-response relationship. Using structural data such as cable preload and time-varying external load data as input, the neural network extracts high-dimensional features and outputs hybrid tower dynamic characteristic data. This model achieves accurate monitoring of the system response under complex operating conditions at a speed a thousand times faster than traditional simulations, providing real-time early warnings for the cable safety status and significantly improving the monitoring efficiency of hybrid tower structures.

[0115] Specifically, based on the established database of dynamic characteristics of hybrid tower steel cable systems, 200,000 sets of data were randomly generated for training the neural network, and 20,000 sets of data were generated for verification. Both the training set and the verification set data were formed by fusing simulation and experimental data.

[0116] The structure-condition-response relationship neural network surrogate model consists of a linear weighting function and a nonlinear activation function. The linear weighting function can linearly transform input quantities from different sources, while the nonlinear activation function performs nonlinear processing on the values ​​after the linear transformation. The choice of activation function has a significant impact on the network performance. In this embodiment, the ReLU function, which is commonly used in regression tasks, is selected.

[0117] Neural network units are interconnected according to certain rules to form a deep neural network. For example... Figure 2 As shown, the forward neural network has two hidden layers, with hidden layer 1 containing 20 computational nodes and hidden layer 2 containing 20 computational nodes. (Wind speed...) V Blade pitch angle β Mixed tower radius r Number of steel cables a Preload F The operating conditions and structural data are input into the input layer as input variables to determine the dynamic tilt angle of the tower top. θ angular velocity ω and angular acceleration αThe motion response parameters are output by the output layer as output variables, and the nonlinear relationship between the input and output variables can be gradually approximated by the deep neural network.

[0118] Normalized mean square error (MSE) is an indicator used to quantify the accuracy of model monitoring. The calculation method is as follows:

[0119] .

[0120] In the formula, and These are the monitored value and the actual response value of the positive relational neural network, respectively. m is the total number of iterations, and o is the current step number.

[0121] To avoid model overfitting, set the learning rate to [value]. The Adam optimizer was selected, and the number of training epochs was set to 1000. The convergence of the positive relation neural network training process was determined by the mean squared error value after 1000 iterations.

[0122] like Figure 2 As shown, similar to the establishment process of the forward relational neural network, the reverse monitoring neural network also adopts a cascaded dual hidden layer structure for filtering and processing input information. Hidden layer 1 contains 20 computational nodes, and hidden layer 2 contains 40 computational nodes. (Wind speed...) V Blade pitch angle β Dynamic tilt angle of the tower top θ angular velocity ω and angular acceleration α The operating conditions and motion response parameters are input into the input layer as input variables, including the preload. F Mixed tower radius r Number of steel cables a Structured data is output by the output layer as output variables. Reverse monitoring also randomly generates 220,000 initial data sets for learning, with 200,000 sets used as the training set and the remaining 20,000 sets used as the validation set.

[0123] Step 4: Use the structure-condition-response relationship neural network surrogate model to reverse calculate the preload of the mixed tower steel cable by using the real-time condition data and motion response data sensed by the sensor.

[0124] In one specific implementation, this embodiment develops the reverse calculation capability of a structure-operating condition-response relationship neural network proxy model. Using motion response data and operating condition data as input, the neural network extracts high-dimensional features and outputs structural data to realize the monitoring of cable pretension.

[0125] Specifically, cabin sensors perceive operating condition information, while tilt and acceleration sensors perceive response information, which are then input into a trained surrogate model. The structure-operating condition-response relationship neural network surrogate model selects a neural network algorithm to establish the nonlinear mapping relationship between the input and output variables of the hybrid tower's structure-operating condition-response relationship. This model includes a forward relationship neural network and a reverse monitoring neural network. First, the forward relationship neural network establishes a positive relationship between structure, operating condition, and response, i.e., it determines the response of the hybrid tower under different structural and operating conditions. Then, based on the positive relationship between structure, operating condition, and response, the reverse monitoring neural network performs a reverse inversion. Specifically, the forward relationship neural network takes structural data such as cable preload and time-varying external operating condition loads as input, extracts high-dimensional features through the neural network, and outputs hybrid tower motion response parameters. The reverse monitoring neural network takes the hybrid tower motion response parameters and time-varying external operating condition loads as input, extracts high-dimensional features through the neural network, and outputs structural data such as cable preload. Based on real-time operating condition data and motion response data sensed by sensors, structural data such as preload are calculated in reverse, enabling real-time monitoring of the steel cable preload. Specifically, for example... Figure 3 As shown, after comprehensively considering the operating status of the wind turbine and external environmental conditions, the system monitors and records operating condition information such as wind speed, wind direction, turbine power, rotational speed, rotor speed, blade airfoil, blade chord length, and blade pitch angle, as well as response information such as the dynamic tilt angle, angular velocity, and angular acceleration at the tower top sensed by tilt and acceleration sensors. A neural network proxy model of the structure-operating condition-response relationship for the hybrid tower cable system is constructed, and the reverse calculation capability of the proxy model is developed to achieve real-time monitoring of structural parameters such as cable preload. Multiple sensors are used for fusion data acquisition to ensure information integrity and high data quality. Since the blade airfoil and blade chord length of a given wind turbine model are fixed, while other parameters are dynamically changing, the system uses nacelle sensors to sense operating condition information such as turbine power, rotational speed, rotor speed, and blade pitch angle; tilt sensors to sense the dynamic tilt angle at the tower top; and acceleration sensors to sense the motion response information such as the angular velocity and angular acceleration at the tower top. The operating condition and response data are transmitted via optical fiber to the proxy model of the structure-operating condition-response relationship in the computer.

[0126] The sensor-sensed response and operating condition data are input into the input layer as input variables, and the output structural data such as the pretension force of the steel cable of the hybrid tower is output. The monitoring results are compared with the results under operating conditions such as normal, insufficient, excessive, uneven, and broken pretension forces to determine the current health status of the hybrid tower. The monitoring results are displayed on the existing SCADA system user interface, thereby achieving accurate monitoring of the steel cable pretension force.

[0127] Example 2:

[0128] Embodiment 2 of the present invention provides a real-time monitoring system for the pretension force of mixed-tower steel cables, comprising:

[0129] The data acquisition module is configured to acquire known dynamic characteristic data of mixed tower steel cables under multiple working conditions and multiple structures. The dynamic characteristic data includes structural data, working condition data and corresponding motion response data.

[0130] The data filtering module is configured to use global sensitivity analysis to filter key parameters in the dynamic characteristic data that effectively characterize the dynamic response of the hybrid tower cable system and the system structure and working load, and to build a dynamic characteristic database based on the key parameters;

[0131] The model building module is configured to construct a neural network proxy model of the structure-working condition-response relationship based on dynamic characteristics;

[0132] The preload monitoring module is configured to use a structure-condition-response relationship neural network proxy model to reverse calculate the preload of the mixed tower steel cable by using the real-time condition data and motion response data sensed by the sensor.

[0133] Example 3:

[0134] Embodiment 3 of the present invention provides a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed the steps of the real-time monitoring method for pretensioning of mixed tower steel cables as described in Embodiment 1 of the present invention.

[0135] Example 4:

[0136] Embodiment 4 of the present invention provides a computer device, the device comprising:

[0137] A processor, adapted to execute computer programs;

[0138] A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the steps in the real-time monitoring method for pretensioning of mixed-tower steel cables as described in Embodiment 1 of the present invention.

[0139] The steps and methods involved in Examples 2, 3 and 4 above correspond to those in Example 1. For specific implementation details, please refer to the relevant description section of Example 1.

[0140] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0141] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data processing device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0142] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for real-time monitoring of preload of steel cables in mixed-tower construction, characterized in that, Includes the following steps: Acquire known dynamic characteristic data of mixed tower steel cables under multiple working conditions and multiple structures. The dynamic characteristic data includes structural data, working condition data and corresponding motion response data. Global sensitivity analysis was used to screen key parameters from dynamic characteristic data that effectively characterize the dynamic response, system structure, and operating loads of the hybrid tower cable system. A dynamic characteristic database was then constructed based on these key parameters. The specific steps are as follows: The contribution rates of structural data and working condition data to the dynamic response of the hybrid tower cable system are evaluated, including the contribution rates of each structural data and working condition data to the dynamic response of the hybrid tower cable system, as well as the contribution rates of the interaction between each structural data and working condition data to the dynamic response of the hybrid tower cable system. Representative structures, working conditions, and corresponding motion response data are then selected. The method of fusion based on variance weights fuses the response experimental data, and the optimal weights for data fusion are obtained by using the variance of the data signals. A database of dynamic characteristics of hybrid tower cable systems was established based on the fused response data. Based on a dynamic characteristic database, a neural network surrogate model of structure-operating condition-response relationship is constructed using neural network or regression prediction methods. Using a structure-condition-response relationship neural network proxy model, the preload of the hybrid tower steel cable is predicted inversely based on real-time condition data and motion response data sensed by sensors.

2. The method for real-time monitoring of pretension force of steel cables in mixed-tower construction as described in claim 1, characterized in that, The specific steps for obtaining a database of known dynamic characteristics of hybrid tower cable systems under multiple working conditions and structures are as follows: construct a dynamic model of the hybrid tower cable system; based on the dynamic model, simulate hybrid towers of different structural dimensions under different working conditions to determine the motion response of the hybrid towers under wind load; collect measured data of existing hybrid tower cable systems, integrate simulation data and measured data, and create a database of dynamic characteristics of hybrid tower cable systems under multiple working conditions and structures.

3. The method for real-time monitoring of pretension force of steel cables in mixed-tower construction as described in claim 2, characterized in that, The system includes multiple structural parameters such as tower height, number of concrete segments, tower radius, tower wall thickness, cable radius, cable length, cable arrangement, and preload. It also includes multiple operating conditions such as normal cable preload, insufficient preload, excessive preload, unbalanced preload, and cable breakage. Structural data includes tower height, number of concrete segments, tower radius, tower wall thickness, cable arrangement location, quantity, diameter, axial stiffness, damping, anchor point location, and preload. Operating condition data includes wind speed, wind direction, turbine power, rotor speed, blade airfoil, blade chord length, and blade pitch angle. Motion response parameters include tilt angle, angular velocity, and angular acceleration.

4. The method for real-time monitoring of pretension force of mixed-tower steel cables as described in claim 2, characterized in that, The specific steps for determining the kinematic response of a hybrid tower under wind load are as follows: Based on Markov chain analysis of local wind speed data, wind speed data is combined with time data, i.e., a one-to-one correspondence is established to establish a speed-time state. The local wind speed data is traversed, and all states of the next second are integrated with a set size of 1 m / s. The frequency of different states in the next second is counted. Calculate the state transition probability from the current second to the next second; Calculate the velocity-time state transition probability matrix from the current second to the next second at different times; Monte Carlo simulation of the wind speed time state transition probability matrix was performed using the acceptance-rejection sampling method to obtain the equivalent wind speed curve; Multiple equivalent wind speed curves are generated through repeated calculations to ensure that their mean and standard deviation are the same as the local wind speed data, thus obtaining multi-condition simulated wind speed curves.

5. The method for real-time monitoring of pretension force of steel cables in mixed-tower construction as described in claim 1, characterized in that, Based on the dynamic characteristic database of hybrid tower cable systems, this study uses the structural and operational data of hybrid tower cable systems as input, extracts high-dimensional features through neural networks, outputs motion response data, clarifies the structure-operating condition-response relationship of hybrid tower cable systems, and obtains a neural network surrogate model of the structure-operating condition-response relationship by using neural networks and regression prediction methods.

6. The method for real-time monitoring of pretension force of steel cables in mixed-tower construction as described in claim 1, characterized in that, The reverse calculation capability of the neural network surrogate model of structure-operating condition-response relationship is developed. Using the motion response data and operating condition data of the hybrid tower cable system as input, high-dimensional features are extracted through the neural network, structural characteristic data are output, and the cable pretension is predicted to realize the monitoring of cable pretension.

7. A real-time monitoring system for the pretension force of mixed-tower steel cables, implementing the real-time monitoring method for the pretension force of mixed-tower steel cables according to any one of claims 1-6, characterized in that, include: The data acquisition module is configured to acquire known dynamic characteristic data of mixed tower steel cables under multiple working conditions and multiple structures. The dynamic characteristic data includes structural data, working condition data and corresponding motion response data. The data filtering module is configured to use global sensitivity analysis to filter key parameters in the dynamic characteristic data that effectively characterize the dynamic response of the hybrid tower cable system and the system structure and working load, and to build a dynamic characteristic database based on the key parameters; The model building module is configured to construct a structure-condition-response relationship neural network surrogate model based on a dynamic characteristic database and using neural network or regression prediction methods. The preload monitoring module is configured to use a structure-condition-response relationship neural network proxy model to reverse calculate the preload of the mixed tower steel cable by using the real-time condition data and motion response data sensed by the sensor.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed by the real-time monitoring method for pretensioning of mixed-tower steel cables according to any one of claims 1-6.

9. A computer device, characterized in that, include: A processor, adapted to execute computer programs; A computer-readable storage medium storing a computer program, which, when executed by the processor, implements the real-time monitoring method for pretensioning of steel cables in mixed towers according to any one of claims 1-6.

Images