Fan tower footing structure optimization method and equipment based on AI load model and medium
By using a pre-trained load prediction model and an automated iterative loop mechanism, the problem of low iteration efficiency in wind turbine tower foundation design was solved, achieving efficient structural parameter optimization and finding the global optimal solution.
Patent Information
- Application Number
- CN202511612745.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-05
- Publication Date
- 2026-03-03
AI Technical Summary
The reliance on high-precision physical simulation in wind turbine tower design leads to low design iteration efficiency and makes it difficult to efficiently achieve global optimization.
A pre-trained load prediction model is used, and the structural parameters are automatically adjusted through an iterative loop mechanism of "judgment-adjustment-repetition" until the convergence condition is met, thereby achieving efficient optimization of the structural parameters.
It improves the efficiency of design iteration, eliminates the reliance on human experience and repeated trial and error, and enables efficient exploration of parameter combinations to find the optimal solution that meets the convergence conditions.
Smart Images
Figure CN121598745A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present invention relate to the field of wind turbines, and particularly to a method, equipment and medium for optimizing wind turbine tower foundation structure based on AI load model. Background Technology
[0002] Wind energy, as a clean and renewable energy source, plays an increasingly important role in the global energy structure. Wind turbines are key equipment for capturing wind energy, and their towers and foundations (collectively referred to as the tower base) are the core structures supporting the entire turbine and withstanding complex environmental loads. The safety, reliability, and economic efficiency of their design are crucial to the overall return on investment of wind farms. During the early bidding and detailed design phases of wind turbine projects, the tower base structure needs to be repeatedly designed and verified to ensure that its structural strength and fatigue life meet requirements throughout its entire life cycle.
[0003] In current technological practices, the design of wind turbine tower foundations typically relies on high-precision physical simulation models. Designers first propose an initial set of tower foundation structural parameters based on experience or reference designs, and then use specialized simulation software to perform load calculations on the structure under given environmental conditions (such as wind conditions and sea conditions).
[0004] However, this design method based on physical simulation has significant technical bottlenecks. On the one hand, the high-precision simulation process is extremely time-consuming and consumes a large amount of computational resources, resulting in a long verification cycle for a single design scheme and severely restricting the efficiency of design iteration. On the other hand, this method is essentially a "forward verification" process, that is, inputting certain structural parameters to output corresponding load results, making the search for the optimal design largely dependent on the designer's experience and repeated trial and error, making it difficult to efficiently achieve the global optimum. Summary of the Invention
[0005] The purpose of this invention is to provide a method, device, and medium for optimizing wind turbine tower foundation structure based on an AI load model, thereby solving the problem of low design iteration efficiency of tower foundation structure parameters and difficulty in efficiently achieving global optimization.
[0006] To address the aforementioned technical problems, embodiments of the present invention provide a wind turbine tower foundation structure optimization method based on an AI load model, comprising: acquiring environmental parameters under a target operating condition and structural parameters of the tower foundation to be optimized; inputting the environmental parameters and the structural parameters into a pre-trained load prediction model to obtain a predicted load on the tower foundation under the target operating condition; wherein, the load prediction model is used to characterize the mapping relationship between the environmental parameters, structural parameters, and tower foundation load; based on the predicted load, determining whether the structural parameters of the tower foundation meet a convergence condition; if the convergence condition is not met, adjusting the structural parameters, and repeatedly executing the steps of inputting the load prediction model, obtaining the predicted load, and determining whether the convergence condition is met using the adjusted structural parameters, until the convergence condition is met; when the convergence condition is met, outputting the structural parameters as the optimization result.
[0007] Embodiments of the present invention also provide an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the wind turbine tower foundation structure optimization method based on the AI load model as described above.
[0008] Embodiments of the present invention also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the wind turbine tower foundation structure optimization method based on the AI load model as described above.
[0009] In this embodiment of the invention, the predicted load is first obtained by employing a pre-trained load prediction model. Since this model is pre-trained, its prediction calculation process has a significant speed advantage compared to physical simulation, thus fundamentally solving the problem of low efficiency in design iteration. Secondly, this solution does not simply perform a single load calculation, but establishes an iterative loop mechanism of "judgment-adjustment-repetition." Based on the rapidly obtained predicted load, it judges whether the current structural parameters meet the "convergence condition." If not, it automatically "adjusts the structural parameters" and "repeates the process." This automated closed-loop adjustment mechanism frees the optimization process of structural parameters from dependence on human experience and repeated trial and error, enabling efficient and systematic exploration of different parameter combinations until the optimal solution satisfying the convergence condition is found, thereby solving the problem that existing technologies struggle to efficiently achieve global optima. Attached Figure Description
[0010] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.
[0011] Figure 1 This is a flowchart of a wind turbine tower foundation structure optimization method based on an AI load model provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the construction of the load prediction model in the wind turbine tower foundation structure optimization method based on the AI load model provided in an embodiment of the present invention. Figure 3 This is a flowchart illustrating the design of the number of tower base structures in the wind turbine tower base structure optimization method based on an AI load model provided in an embodiment of the present invention. Figure 4 This is a comparison chart of the prediction accuracy of the load prediction model in the wind turbine tower foundation structure optimization method based on the AI load model provided in the embodiments of the present invention; Figure 5 This is a diagram showing the relationship between damage equivalent load and structural parameters in the wind turbine tower foundation structure optimization method based on AI load model provided in an embodiment of the present invention. Figure 6 This is a structural diagram of the electronic equipment used in the wind turbine tower foundation structure optimization method based on the AI load model provided in the embodiments of the present invention. Detailed Implementation
[0012] Wind energy, as a clean and renewable energy source, plays an increasingly important role in the global energy structure. Wind turbines are key equipment for capturing wind energy, and their towers and foundations (collectively referred to as the tower base) are the core structures supporting the entire turbine and withstanding complex environmental loads. The safety, reliability, and economic efficiency of their design are crucial to the overall return on investment of wind farms. During the early bidding and detailed design phases of wind turbine projects, the tower base structure needs to be repeatedly designed and verified to ensure that its structural strength and fatigue life meet requirements throughout its entire life cycle.
[0013] In current technological practices, the design of wind turbine tower foundations typically relies on high-precision physical simulation models. Designers first propose an initial set of tower foundation structural parameters based on experience or reference designs, and then use specialized simulation software to perform load calculations on the structure under given environmental conditions (such as wind conditions and sea conditions).
[0014] However, this design method based on physical simulation has obvious technical bottlenecks. On the one hand, the high-precision simulation process is extremely time-consuming and consumes a large amount of computing resources, resulting in a long verification cycle for a single design scheme, which seriously restricts the efficiency of design iteration. On the other hand, this method is essentially a "forward verification" process, that is, inputting determined structural parameters to output corresponding load results, making the search for the optimal design largely dependent on the designer's experience and repeated trial and error, making it difficult to efficiently achieve the global optimum. To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details are presented in the various embodiments of the present invention to enable readers to better understand this application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in this application can be implemented. The division of the following embodiments is for the convenience of description and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined and referenced with each other without contradiction.
[0015] One embodiment of the present invention relates to a wind turbine tower foundation structure optimization method based on an AI load model, which can be applied to electronic devices or computing systems capable of numerical calculation and data processing, such as high-performance computers, servers, and cloud computing platforms. The method includes: acquiring environmental parameters under a target operating condition and structural parameters of the tower foundation to be optimized; inputting the environmental parameters and structural parameters into a pre-trained load prediction model to obtain the predicted load of the tower foundation under the target operating condition; wherein, the load prediction model is used to characterize the mapping relationship between the environmental parameters, structural parameters, and tower foundation load; based on the predicted load, determining whether the structural parameters of the tower foundation meet the convergence condition; if the convergence condition is not met, adjusting the structural parameters, and repeatedly executing the steps of inputting the load prediction model, obtaining the predicted load, and determining whether the convergence condition is met using the adjusted structural parameters until the convergence condition is met; when the convergence condition is met, outputting the structural parameters as the optimization result. In this embodiment of the invention, the predicted load is first obtained by using a pre-trained load prediction model. Since the model is pre-trained, its prediction calculation process has a significant speed advantage compared to physical simulation, thus fundamentally solving the problem of low design iteration efficiency. Secondly, this scheme does not simply perform a single load calculation, but establishes an iterative loop mechanism of "judgment-adjustment-repetition". Based on the rapidly obtained predicted load, it judges whether the current structural parameters meet the "convergence condition". If not, it automatically "adjusts the structural parameters" and "repeates the execution" of the process. This automated closed-loop adjustment mechanism frees the optimization process of structural parameters from dependence on human experience and repeated trial and error, enabling efficient and systematic exploration of different parameter combinations until the optimal solution that meets the convergence condition is found. This solves the problem that existing technologies struggle to efficiently achieve global optima.
[0016] The following is a detailed description of the implementation details of the wind turbine tower foundation structure optimization method based on the AI load model according to the embodiments of the present invention. The following content is only for the convenience of understanding the implementation details and is not necessary for implementing this solution.
[0017] like Figure 1 As shown, the embodiments of the present invention mainly include steps 110 to 140.
[0018] In step 110, the environmental parameters under the target operating conditions and the structural parameters of the tower base to be optimized are obtained.
[0019] In a specific example, the aforementioned environmental parameters include at least wind and sea condition information for the region of the tower base, and / or the structural parameters include at least the geometric dimensions and structural response characteristics of the tower base. For example, environmental parameters may include, but are not limited to, data such as wind speed, and structural parameters may include, but are not limited to, data such as the mass of the tower base. Those skilled in the art will understand that the aforementioned environmental parameters include parameters used to collectively define the external excitation borne by the wind turbine tower base, while the aforementioned structural parameters include parameters used to define the physical and dynamic characteristics of the tower base itself. Together, they constitute the key factors affecting the load on the tower base. In traditional high-precision physical simulations, it is precisely the complex combination of these parameters and their nonlinear relationship with the load that leads to the time-consuming, labor-intensive, and resource-intensive calculation process. This invention aims to comprehensively capture the core factors affecting the load on the tower base by using these key environmental and structural parameters as input features of the model, providing a sufficient and necessary information foundation for establishing accurate mapping relationships in subsequent load prediction models.
[0020] In step 120, environmental parameters and structural parameters are input into a pre-trained load prediction model to obtain the predicted load on the tower base under the target working condition; wherein, the load prediction model is used to characterize the mapping relationship between environmental parameters and structural parameters and the tower base load.
[0021] As those skilled in the art will understand, the load prediction model serves to replace the extremely time-consuming and computationally resource-intensive high-precision physical simulation calculations in the traditional design process. Through pre-training on a large amount of simulation data, the model has learned and solidified the complex nonlinear mapping relationships between environmental and structural parameters and tower foundation loads. Therefore, during the optimization iteration process, the model can be directly invoked for rapid inference, obtaining load calculation results within seconds that would otherwise take hours or even longer, significantly improving design efficiency.
[0022] In step 130, based on the predicted load, it is determined whether the structural parameters of the tower base meet the convergence condition.
[0023] It is important to note that the above convergence condition is that the design cost of the tower base is less than the preset cost. Specifically, the design cost can be a comprehensive economic indicator, which may include, but is not limited to, at least one of the following: material cost, tower cost, manufacturing cost, construction cost, total life cycle cost, operation and maintenance cost, and decommissioning cost.
[0024] In a specific example, the above-mentioned determination of whether the structural parameters of the tower base meet the convergence condition based on the predicted load includes: designing the tower based on the predicted load to obtain the tower design parameters; designing the tower base based on the tower design parameters and the predicted load to obtain the tower base design parameters; and determining whether the evaluation of the tower base design parameters meets the convergence condition based on the tower base design parameters.
[0025] In a specific example, the above-mentioned tower foundation design based on tower design parameters and predicted loads, to obtain tower foundation design parameters, includes: obtaining geological survey data of the target installation location of the tower foundation; and designing the tower foundation based on tower design parameters, predicted loads, geological survey data, and environmental parameters to obtain tower foundation design parameters.
[0026] As those skilled in the art will recognize, step 130 is the core of the integrated tower-base design. First, using the predicted loads output by the load prediction model, a preliminary tower engineering design can be performed, determining its specific dimensions and material specifications. Subsequently, in the tower base design phase, it is necessary to consider not only the loads from the upper tower and the environmental loads predicted by the load prediction model, but also the actual geological conditions of the target installation location, i.e., geological survey data. Finally, through the complete design process, a complete tower base structure scheme can be obtained, and its construction cost can be calculated accordingly. This cost will serve as a quantitative basis for judging whether the current structural parameters are "optimal" or "acceptable," and will be directly compared with the convergence criteria.
[0027] In step 140, if the convergence condition is not met, the structural parameters are adjusted, and the steps of input load prediction model, obtaining predicted load and determining whether the convergence condition is met are repeated using the adjusted structural parameters until the convergence condition is met; when the convergence condition is met, the structural parameters are output as the optimization result.
[0028] Those skilled in the art will recognize that step 140 above constitutes a closed-loop optimization process for automated optimization. When the design cost corresponding to the current structural parameters does not meet the convergence criteria (e.g., the cost is too high or there is still room for reduction), the optimization process does not terminate. Instead, the system automatically, or guided by the algorithm, adjusts the input structural parameters (such as foundation diameter, tower base diameter, tower base frequency, etc.) to generate a new set of candidate parameters. This new set of structural parameters serves as the input for the next iteration, re-entering step 120, where the load prediction model quickly calculates the corresponding predicted load. The entire "prediction-design-judgment-adjustment" cycle continues until a set of structural parameters is found whose corresponding tower base design cost meets the preset convergence criteria (e.g., reaching a global or local minimum). At this point, the iteration process terminates, and the final structural parameters are locked and output as the optimal design scheme. It is precisely because of the extremely high computational speed of the load prediction model that this rapid and automatic iterative optimization becomes possible, enabling efficient searching and locking of the optimal cost design scheme within a vast design space.
[0029] In a specific example, before optimizing structural parameters, a load prediction model needs to be trained. The training method includes: obtaining corresponding load data samples through physical simulation based on multiple sets of environmental and structural parameters; and training the load prediction model using these load data samples to establish the mapping relationship between environmental and structural parameters and the tower base load.
[0030] Those skilled in the art will recognize that the process of obtaining load data samples described above involves multiple steps. First, high-precision physical simulation software is used to perform dynamic simulations for multiple different combinations of environmental and structural parameters, generating load time-series data. However, this raw time-series data cannot be directly used as training labels. Therefore, the time-series data needs to be further processed, for example, by using rainflow counting to convert it into ten-minute equivalent fatigue (STEL) data, which is more representative of structural fatigue damage. In some application scenarios, the full-lifecycle damage equivalent load (DEL) can be further calculated based on the STEL data. Only the final STEL or DEL data is used as load labels in the training data samples. During the model training phase, a dataset containing multiple sets of "environmental and structural parameters" (as input features) and "load labels" (as expected outputs) is further constructed and divided into training and test sets in a ratio of, but not limited to, 8:2. The training set is used to train the weights of the neural network, enabling it to learn the mapping relationship between input and output, while the test set is used to verify the prediction accuracy and generalization ability of the trained model.
[0031] In a specific example, the above load prediction model is a fully connected neural network model or a temporal convolutional-fully connected neural network model. When the load prediction model is a temporal convolutional-fully connected neural network model, the environmental parameters include at least wind and wave time series data.
[0032] Those skilled in the art will recognize that fully connected neural network models are chosen because of their simple structure, adeptness at handling vectorized static feature inputs, and ability to effectively establish nonlinear relationships between multidimensional parameters and single load values. When environmental parameters are provided in time-series form, such as continuous wind speed and wave height data, a temporal convolutional-fully connected neural network model is even more advantageous. The temporal convolution module can effectively capture time-dependent and dynamic features from time-series data, which is particularly important for predicting structural loads affected by dynamic environments.
[0033] In an optional embodiment, the load prediction model described above can include, but is not limited to, any model capable of characterizing the mapping relationship between environmental and structural parameters and tower foundation loads. It should be noted that, in addition to the fully connected neural network and temporal convolutional-fully connected neural network models mentioned above, other types of machine learning or deep learning models are also applicable. This invention is not limited to a specific model structure; any surrogate model capable of learning from data and rapidly reproducing the physical simulation process can be integrated into the optimization framework of this invention to achieve rapid iteration and optimization of tower foundation design parameters.
[0034] In a specific example, based on multiple sets of environmental and structural parameters, corresponding load data samples are obtained through physical simulation, including: generating load time series data through physical simulation based on multiple sets of environmental and structural parameters; using the rainflow counting method on the load time series data to obtain ten-minute equivalent fatigue data of the load data samples; and calculating full life cycle damage equivalent load data based on the ten-minute equivalent fatigue data as load data samples.
[0035] Those skilled in the art will recognize that the raw output of the aforementioned physical simulation is a sequence of load changes over time. This time-series data is high-dimensional and contains a large amount of redundant information, making it unsuitable as a direct prediction target for the model. Therefore, this invention first processes the load time-series data using the rainflow counting method. This method effectively decomposes the irregular load history into a series of independent stress cycles, thereby quantifying its contribution to material fatigue. The processed data is standardized ten-minute equivalent fatigue data (STEL), which characterizes the degree of fatigue damage over a short period under specific working conditions. Finally, to obtain a single indicator representing the entire design life cycle, the life-cycle damage equivalent load (DEL) is further calculated based on the STEL database. This DEL value comprehensively considers the frequency of occurrence of various working conditions and their corresponding fatigue damage, ultimately forming a single, crucial value that can comprehensively evaluate the fatigue performance of the tower foundation structure. This value serves as the final "load data sample" or training label, with a clear physical meaning, providing a high-quality data foundation for training a high-precision load prediction model.
[0036] In a specific example, such as Figure 2This diagram illustrates the data generation, training, and testing of the load prediction model involved in this invention. The process is mainly divided into three parts: a middle data generation module, and two model training / testing modules at the top and bottom. The data generation module is the foundation of model training. First, environmental constants (such as wind conditions and sea conditions) are input to generate wind and wave time-series data. Combined with given structural parameters and a wind turbine model, load calculations are performed through high-precision physical simulation to obtain load time-series data. Subsequently, the load time-series data is processed using the rainflow counting method to calculate STEL data and form a STEL database. Finally, based on the STEL database, DEL data such as damage can be further calculated. The Fully Connected Neural Network (FCNN) model training / testing module demonstrates the implementation of FCNN. The module uses the environmental and structural parameters produced by the data generation module as input variables and the data in the STEL database as labels, dividing them into training and testing sets in, for example, an 8:2 ratio. The training set is used to train the FCNN surrogate model, and the testing set is used to evaluate the model's accuracy; the prediction results can ultimately be used for DEL prediction. The Temporal Convolutional Fully Connected Neural Network (TCN-FCNN) model training / testing module demonstrates the implementation of the TCN-FCNN model. Unlike FCNN, this model can directly use wind and wave time-series data as environmental parameter input, combined with structural parameters, to train and test STEL data. This approach can better capture the impact of temporal information on loads.
[0037] In a specific example, such as Figure 3 This document presents a flowchart illustrating the design method for the tower foundation structure involved in an embodiment of the present invention. The input data includes the structural parameters to be optimized (e.g., a wind turbine model), environmental parameters under the target operating conditions (e.g., wind and wave data), and geological survey data of the target installation location. The structural and environmental parameters are first input into a pre-trained AI load proxy model (i.e., a load prediction model) to quickly obtain the predicted load. This predicted load is then used for tower and foundation design, with the foundation design also incorporating the geological survey data. After completing the tower and foundation design, the system determines whether the design meets the "optimal cost convergence" condition. If it does not meet the condition (No), the process returns and adjusts the structural parameters for a new iteration; if it meets the condition (Yes), the process ends, and the current design scheme is locked as the final optimization result.
[0038] In a specific example, such as Figure 4 The diagram shows the result verification of load prediction based on the method of the above embodiments of the present invention, which is used to show the error between the predicted value and the true value. Figure 4In the graph, the horizontal axis represents multiple different wind turbine test indices (e.g., WT1, WT2, etc.), and the vertical axis represents the damage equivalent load (DEL) value. Each test index contains two bars: the dark bar represents the "True DEL," and the light bar represents the "Predicted DEL," which is the predicted fatigue load data quickly calculated and output by the AI proxy model in this embodiment after receiving input parameters under the same operating conditions. It is clear from the graph that in all test conditions, the height of the light bar representing the "Predicted DEL" is very close to the height of the dark bar representing the "True DEL," showing a highly consistent trend. Specifically, the percentages marked above each group of orange bars in the graph (e.g., -2.81%, -4.73%, -0.17%) represent the error of the predicted value relative to the true value. These data show that the prediction error of the AI proxy model remains at a low level, not exceeding 20%. This verification result strongly proves that the pre-trained AI proxy model in this embodiment has successfully learned and solidified the complex mapping relationship between environmental parameters, structural parameters, and tower base load. Therefore, this model can serve as an efficient alternative to high-precision physical simulation. While ensuring the required accuracy for engineering, it can shorten the originally time-consuming load calculation process to within a few seconds, thus providing a reliable technical foundation for the subsequent rapid iterative optimization design process.
[0039] In a specific example, such as Figure 5 This diagram illustrates the relationship between damage equivalent load and structural parameters based on the embodiments of the present invention described above. It demonstrates the complex and nonlinear relationship between the structural parameters to be optimized and the final fatigue load. The data in this diagram is generated under specific working conditions with a tower base modal mass fixed at 1,200,000 kg. In the diagram: the horizontal axis (X-axis) represents the tower diameter, the vertical axis (Y-axis) represents the foundation diameter, and the vertical axis (Z-axis) represents the tower base frequency. The trend diagram shows that different combinations of structural parameters correspond to drastically different DEL values, revealing the inherent physical correlation between design parameters and load response. Traditional design methods require time-consuming, high-precision simulations of every point in the space or a selected few points, making it difficult to quickly and comprehensively evaluate the entire design space. This invention, by employing an AI proxy model and automated iterative loops, can quickly calculate the load corresponding to any combination of parameters within the space, thereby efficiently performing global optimization and systematically finding the optimal combination of structural parameters that meets cost and performance requirements.
[0040] In this embodiment of the invention, the predicted load is first obtained by employing a pre-trained load prediction model. Since this model is pre-trained, its prediction calculation process has a significant speed advantage compared to physical simulation, thus fundamentally solving the problem of low efficiency in design iteration. Secondly, this solution does not simply perform a single load calculation, but establishes an iterative loop mechanism of "judgment-adjustment-repetition." Based on the rapidly obtained predicted load, it judges whether the current structural parameters meet the "convergence condition." If not, it automatically "adjusts the structural parameters" and "repeates" the process. This automated closed-loop adjustment mechanism frees the optimization process of structural parameters from dependence on human experience and repeated trial and error, enabling efficient and systematic exploration of different parameter combinations until the optimal solution satisfying the convergence condition is found, thereby solving the problem that existing technologies struggle to efficiently achieve global optima.
[0041] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.
[0042] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent three cases: A exists, A and B exist simultaneously, and B exists. In addition, the character " / " in this document generally indicates that the related objects before and after it have an "or" relationship.
[0043] The steps described above are for clarity only. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the protection scope of this invention. Adding insignificant modifications or introducing insignificant designs to the algorithm or process, without changing the core design of the algorithm and process, are also within the protection scope of this invention.
[0044] Furthermore, the examples mentioned in the above embodiments can be freely combined, and any combination can be understood as an embodiment. The terms "embodiment" or "example" appearing in various locations in the specification do not necessarily refer to the same embodiment, nor are they independent or alternative embodiments mutually exclusive with other embodiments. Those skilled in the art will understand that the embodiments described herein can be combined with other embodiments.
[0045] Another embodiment of the present invention relates to a terminal / electronic device / server, such as Figure 6As shown, it includes at least one processor 201; and a memory 202 communicatively connected to at least one processor 201; wherein the memory 202 stores instructions executable by at least one processor 201, the instructions being executed by at least one processor 201 to enable at least one processor 201 to perform the optimization method of the tower base structure parameters as described above.
[0046] The memory 202 and processor 201 are connected via a bus, which may include any number of interconnecting buses and bridges, connecting various circuits of one or more processors 201 and memory 202 together. The bus may also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides an interface between the bus and the transceiver. The transceiver may be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 201 is transmitted over a wireless medium via an antenna, which further receives data and transmits it to processor 201.
[0047] Processor 201 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory 202 can be used to store data used by processor 201 during operation.
[0048] Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the method embodiments described above.
[0049] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0050] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing the present invention, and in practical applications, various changes in form and detail may be made without departing from the spirit and scope of the present invention.
Claims
1. A method for optimizing wind turbine tower foundation structure based on an AI load model, characterized in that, include: Obtain environmental parameters under the target operating conditions and structural parameters of the tower base to be optimized; The environmental parameters and structural parameters are input into a pre-trained load prediction model to obtain the predicted load of the tower base under the target working condition; wherein, the load prediction model is used to characterize the mapping relationship between the environmental parameters and structural parameters and the tower base load; Based on the predicted load, determine whether the structural parameters of the tower base meet the convergence condition; If the convergence condition is not met, the structural parameters are adjusted, and the steps of input load prediction model, obtaining predicted load and determining whether the convergence condition is met are repeated using the adjusted structural parameters until the convergence condition is met. When the convergence condition is met, the structural parameters are output as the optimization result.
2. The wind turbine tower foundation structure optimization method based on AI load model according to claim 1, characterized in that, Before inputting the environmental parameters and structural parameters into the pre-trained load prediction model, the following steps are included: Based on multiple sets of environmental and structural parameters, corresponding load data samples are obtained through physical simulation. The load prediction model is trained using the load data samples to establish a mapping relationship between environmental parameters, structural parameters and tower base load.
3. The wind turbine tower foundation structure optimization method based on AI load model according to claim 2, characterized in that, The load data samples obtained through physical simulation based on multiple sets of environmental and structural parameters include: Based on multiple sets of environmental and structural parameters, load time-series data is generated through the physical simulation. The rainflow counting method was used on the load time series data to obtain the ten-minute equivalent fatigue data of the load data sample; The equivalent fatigue data for the entire life cycle damage is calculated based on the ten-minute equivalent fatigue data, and used as the load data sample.
4. The wind turbine tower foundation structure optimization method based on AI load model according to claim 1, characterized in that, The step of determining whether the structural parameters of the tower base satisfy the convergence condition based on the predicted load includes: Based on the predicted load, tower design is performed to obtain tower design parameters; Based on the tower design parameters and the predicted load, the tower foundation design is carried out to obtain the tower foundation design parameters; Based on the tower base design parameters, determine whether the evaluation of the tower base design parameters meets the convergence condition.
5. The wind turbine tower foundation structure optimization method based on AI load model according to claim 4, characterized in that, The process of designing the tower foundation based on the tower design parameters and the predicted load, to obtain the tower foundation design parameters, includes: Obtain geological survey data of the target installation location of the tower base; Based on the tower design parameters, the predicted load, the geological survey data, and the environmental parameters, the tower foundation design is carried out to obtain the tower foundation design parameters.
6. The wind turbine tower foundation structure optimization method based on an AI load model according to any one of claims 1 to 5, characterized in that, The convergence condition is that the design cost of the tower base is less than the preset cost.
7. The wind turbine tower foundation structure optimization method based on an AI load model according to any one of claims 1 to 5, characterized in that, The load prediction model is either a fully connected neural network model or a temporal convolutional-fully connected neural network model. Wherein, when the load prediction model is a temporal convolutional fully connected neural network model, the environmental parameters include at least wind and wave time series data.
8. The wind turbine tower foundation structure optimization method based on an AI load model according to any one of claims 1 to 5, characterized in that, The environmental parameters include at least wind and sea conditions in the area of the tower base, and / or the structural parameters include at least the geometric dimensions and structural response characteristics of the tower base.
9. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the wind turbine tower foundation structure optimization method based on the AI load model as described in any one of claims 1 to 8.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the wind turbine tower foundation structure optimization method based on the AI load model as described in any one of claims 1 to 8.