Diffusion model framework construction method based on wind simulation and building performance evaluation

Abstract

The invention is suitable for the technical field of wind dynamics simulation, and provides a diffusion model framework construction method based on wind simulation and building performance evaluation, and the method comprises the steps: firstly obtaining multiple pieces of multi-angle wind field data information, then rapidly generating wind simulation data set information according to the multiple pieces of multi-angle wind field data information, and carrying out the construction of a diffusion model framework. And finally, based on the wind simulation data set information, training a preset multi-angle diffusion model, and effectively generating target wind simulation model information. According to the method, a simplified solution can be provided for rapid wind field simulation and building performance evaluation, multi-angle wind dynamic prediction in a high-density city environment is realized, a designer and a planning person can seamlessly integrate wind field analysis to an early stage of design exploration, a more dynamic evaluation process is realized, and the evaluation efficiency is improved. The time and the calculation workload required by performance driving decision are greatly reduced, and the reliability is remarkably improved.

Description

Technical Field

[0001] The present application relates to the technical field of wind dynamics simulation. Specifically, it relates to a method for constructing a diffusion model framework based on wind simulation and building performance evaluation. Background Art

[0002] With the advancement of the global urbanization process, more and more people choose to live in cities and their suburbs. High population density will greatly increase the urban space density, reduce the ventilation volume, and exacerbate the urban heat island effect. Moreover, the increase in temperature and the reduction in ventilation will directly affect human health, comfort, and overall quality of life. Therefore, understanding the urban wind environment is crucial for urban and building research, which helps to improve the microclimate and enhance the resilience and sustainability of cities and human settlements.

[0003] Currently, traditional wind simulation methods are usually limited to two-dimensional planes, lacking the ability to model three-dimensional environments and unable to simulate time changes, which limits their applicability in dynamic scenarios and has the problem of relatively low reliability, thus requiring further improvement. Summary of the Invention

[0004] Based on this, the embodiments of the present application provide a method for constructing a diffusion model framework based on wind simulation and building performance evaluation to solve the problem of relatively low reliability in the prior art.

[0005] In a first aspect, the embodiments of the present application provide a method for constructing a diffusion model framework based on wind simulation and building performance evaluation, the method comprising:

[0006] Obtaining a plurality of multi-angle wind field data information;

[0007] Generating wind simulation dataset information according to the plurality of multi-angle wind field data information;

[0008] Training a preset multi-angle diffusion model based on the wind simulation dataset information to generate target wind simulation model information.

[0009] The beneficial effects compared with the prior art are as follows: For the method for constructing a diffusion model framework based on wind simulation and building performance evaluation provided by the embodiments of the present application, the terminal device can first obtain a plurality of multi-angle wind field data information, then quickly generate wind simulation dataset information according to the plurality of multi-angle wind field data information, and finally train a preset multi-angle diffusion model based on the wind simulation dataset information to effectively generate target wind simulation model information, thereby providing a simplified solution for rapid wind field simulation and building performance evaluation, realizing the prediction of multi-angle wind dynamics in a high-density urban environment, greatly reducing the time and computational workload required for performance-driven decision-making, enabling wind simulation to not only be limited to two-dimensional planes but also improving its applicability in dynamic scenarios, and solving the current problem of relatively low reliability to a certain extent.

[0010] In a second aspect, an embodiment of the present application provides a diffusion model framework construction system based on wind simulation and building performance evaluation. The system includes:

[0011] A multi-angle wind field data information acquisition module: used to acquire multiple multi-angle wind field data information;

[0012] A wind simulation data set information generation module: used to generate wind simulation data set information according to the multiple multi-angle wind field data information;

[0013] A target wind simulation model information generation module: used to train a preset multi-angle diffusion model based on the wind simulation data set information to generate target wind simulation model information.

[0014] In a third aspect, an embodiment of the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method in the first aspect as described above are implemented.

[0015] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method in the first aspect as described above are implemented.

[0016] It can be understood that the beneficial effects of the above second aspect to the fourth aspect can refer to the relevant descriptions in the first aspect above, and will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art.

[0018] Figure 1 is a schematic flowchart of a diffusion model framework construction method provided by an embodiment of the present application;

[0019] Figure 2 is a schematic flowchart of step S210 in the diffusion model framework construction method provided by an embodiment of the present application;

[0020] Figure 3 is a first schematic diagram of impervious area information provided by an embodiment of the present application;

[0021] Figure 4 is a second schematic diagram of impervious area information provided by an embodiment of the present application;

[0022] Figure 5 is a third schematic diagram of impervious area information provided by an embodiment of the present application;

[0023] Figure 6 It is a schematic diagram of wind field image information provided by an embodiment of the present application;

[0024] Figure 7 It is a schematic flowchart of step S300 in the method for constructing a diffusion model framework provided by an embodiment of the present application;

[0025] Figure 8 It is a schematic diagram of an adjustment process provided by an embodiment of the present application;

[0026] Figure 9 It is a schematic flowchart after step S320 in the method for constructing a diffusion model framework provided by an embodiment of the present application;

[0027] Figure 10 It is a schematic diagram of wind field reconstruction with wind intensity information provided by an embodiment of the present application;

[0028] Figure 11 It is a schematic diagram of wind field reconstruction with wind direction information provided by an embodiment of the present application;

[0029] Figure 12 It is a schematic diagram of the training loss and learning rate in the training process provided by an embodiment of the present application;

[0030] Figure 13 It is a schematic diagram of a prediction result provided by an embodiment of the present application;

[0031] Figure 14 It is a block diagram of modules of a diffusion model framework construction system provided by an embodiment of the present application;

[0032] Figure 15 It is a schematic diagram of a terminal device provided by an embodiment of the present application. Detailed implementation manners

[0033] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0034] In the description of the specification and claims of the present application, the terms "first", "second", "third", etc. are only used for descriptive distinction and cannot be understood as indicating or implying relative importance.

[0035] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that specific features, structures, or characteristics described in connection with that embodiment are included in one or more embodiments of this application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized. The terms "comprising", "including", "having" and their variants all mean "including but not limited to", unless otherwise specifically emphasized.

[0036] In order to illustrate the technical solutions described in this application, the following will be described through specific embodiments.

[0037] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of a method for constructing a diffusion model framework based on wind simulation and building performance evaluation provided by an embodiment of this application. In this embodiment, the execution subject of the method for constructing the diffusion model framework is a terminal device. It can be understood that the types of terminal devices include but are not limited to mobile phones, tablet computers, laptop computers, Ultra-Mobile Personal Computers (UMPCs), netbooks, Personal Digital Assistants (PDAs), etc., and this application embodiment does not impose any restrictions on the specific types of terminal devices.

[0038] Please refer to Figure 1 , the method for constructing the diffusion model framework provided by an embodiment of this application includes but is not limited to the following steps:

[0039] In S100, obtain a plurality of multi-angle wind field data information.

[0040] Specifically, the terminal device can first obtain a plurality of multi-angle wind field data information, where the multi-angle wind field data information is used to describe the data of the multi-angle wind field.

[0041] In some possible implementation manners, in order to facilitate improving subsequent data quality, before step S100, the method further includes but is not limited to the following steps:

[0042] In S101, generate a three-dimensional vector data grid based on a preset visual programming language.

[0043] Specifically, the terminal device can first generate a three-dimensional vector data grid based on a preset visual programming language. Each three-dimensional point in the three-dimensional vector data grid represents the behavior of local wind, thereby providing detailed information on the interaction between the airflow and the building environment. Among them, the visual programming language can be Grasshopper; this three-dimensional vector data grid is used to capture wind intensity information and wind direction information.

[0044] Correspondingly, the above step S100 includes but is not limited to the following steps:

[0045] In S110, based on a preset simulation tool, generate multiple multi-angle wind field data information.

[0046] Specifically, after the terminal device generates a three-dimensional vector data grid, the terminal device can generate multiple multi-angle wind field data information based on a preset simulation tool, thereby realizing the simulation of a high-resolution wind field, generating an accurate airflow pattern around urban buildings under different wind conditions, and providing key input for subsequent processing. The simulation tool can be Eddy CFD.

[0047] Exemplarily, the high-fidelity simulation can be carried out on a specified 1-kilometer by 1-kilometer slice of the urban core area, aiming to capture complex urban wind dynamics, ensuring that the results accurately reflect local changes in wind intensity and wind direction changes, and the output data can be used as a basic data set for coding and further analysis, while effectively retaining the intricate geometric and aerodynamic characteristics in these dense urban environments.

[0048] In S200, generate wind simulation data set information according to multiple multi-angle wind field data information.

[0049] In some possible implementation manners, in order to create a comprehensive data set for training, step S200 includes but is not limited to the following steps:

[0050] In S210, for each multi-angle wind field data information: perform encoding processing on the multi-angle wind field data information to generate wind simulation data set information.

[0051] Specifically, after the terminal device generates multiple multi-angle wind field data information, the terminal device can perform processing for each multi-angle wind field data information: perform encoding processing on the multi-angle wind field data information, encode the generated wind field as an RGB image, and generate wind simulation data set information. Among them, the wind simulation data set information includes multiple wind field image information, and the wind field image information is an RGB image.

[0052] In some possible implementation manners, in order to generate wind simulation data set information, please refer to Figure 2 , step S210 includes but is not limited to the following steps:

[0053] In S211, for each multi-angle wind field data information: Based on the logarithmic transformation algorithm, map the wind intensity information of the multi-angle wind field data information to the green channel to generate green channel information.

[0054] Specifically, the terminal device can perform this processing for each multi-angle wind field data information: Based on the logarithmic transformation algorithm, map the wind intensity information of the multi-angle wind field data information to the green channel to generate green channel information, where the green channel information can be expressed as:

[0055] G = round[255×log(w i + 1)÷log(W mmm + 1)],

[0056] In the formula, G represents the green channel information, round[] represents the mathematical function for rounding the parameter, w i represents the wind intensity information, and W mmm represents the preset maximum wind intensity information; it should be noted that the base number in the green channel information can be any positive number, such as e, 2, or 10.

[0057] In S212, obtain the relative angle information.

[0058] Specifically, after the terminal device generates the green channel information, the terminal device can obtain the relative angle information, where the relative angle information is used to describe the relative angle between the wind direction vector and the prevailing wind direction.

[0059] In S213, based on the relative angle information, map the wind direction information of the multi-angle wind field data information to the blue channel to generate blue channel information.

[0060] Specifically, after the terminal device obtains the relative angle information, the terminal device can, based on the relative angle information, map the wind direction information of the multi-angle wind field data information to the blue channel to generate blue channel information, where the blue channel information can be expressed as:

[0061] B = round[255×w d ·θ÷360],

[0062] where B represents the blue channel information, w d represents the wind direction information, and θ represents the relative angle information.

[0063] In S214, obtain the information of the impermeable terrain area and the building area.

[0064] Exemplarily, please refer to Figure 3 、 Figure 4 and Figure 5, after the terminal device generates the blue channel information, the terminal device can obtain the information of the impermeable terrain area and the building area, where the information of the impermeable terrain area is used to describe the area corresponding to the impermeable terrain, and the information of the building area is used to describe the area corresponding to the impermeable building.

[0065] In S215, according to the information of the impermeable terrain area and the building area, generate the information of the impermeable area.

[0066] Exemplarily, please refer to Figure 4 and Figure 5 , after the terminal device obtains the information of the impermeable terrain area and the building area, the terminal device can generate the information of the impermeable area according to the information of the impermeable terrain area and the building area, so as to determine the impermeable areas such as buildings and terrain.

[0067] In S216, map the information of the impermeable area to the red channel to generate the red channel information.

[0068] Specifically, after the terminal device generates the information of the impermeable area, the terminal device can map the information of the impermeable area to the red channel to generate the red channel information, where the red channel information can be represented as R, and the specific value of the red channel information can be 255.

[0069] In S217, generate the wind field image information according to the green channel information, the blue channel information and the red channel information.

[0070] Exemplarily, please refer to Figure 6 , after the terminal device generates the information of the impermeable area, the terminal device can generate the wind field image information according to the green channel information, the blue channel information and the red channel information, so as to complete the encoding process of the multi-angle wind field data information.

[0071] Without loss of generality, the terminal device can form a data set containing 200 samples, and each image is accompanied by a text prompt to describe the angle of view and the wind condition before the wind.

[0072] In S300, based on the wind simulation data set information, train the preset multi-angle diffusion model to generate the target wind simulation model information.

[0073] Specifically, after the terminal device generates the wind simulation data set information, the terminal device can train the preset multi-angle diffusion model based on the wind simulation data set information to generate the target wind simulation model information, so as to not only be limited to the two-dimensional plane, but also fully consider the time change, improve the applicability in the dynamic scene, provide an extensible and accurate solution for urban wind field analysis, and improve the reliability.

[0074] In some possible implementation manners, to generate effective target wind simulation model information, please refer to Figure 7 , step S300 includes but is not limited to the following steps:

[0075] In S310, integrate a preset LoRA fine-tuning mechanism into a preset multi-angle diffusion model to generate a diffusion model to be trained.

[0076] Specifically, after the terminal device generates wind simulation dataset information, the terminal device can integrate a preset LoRA fine-tuning mechanism into a preset multi-angle diffusion model to generate a diffusion model to be trained, where the LoRA fine-tuning mechanism includes a text embedding prompt adjustment mechanism and a control network adjustment mechanism.

[0077] In S320, based on the wind simulation dataset information, train the diffusion model to be trained to generate target wind simulation model information.

[0078] Specifically, after the terminal device generates the diffusion model to be trained, the terminal device can train the diffusion model to be trained based on the wind simulation dataset information, complete the LoRA training of the diffusion model, and generate target wind simulation model information.

[0079] Exemplarily, please refer to Figure 8 , the terminal device can use two core adjustment mechanisms, namely the text embedding prompt adjustment mechanism and the control network adjustment mechanism, to efficiently and accurately simulate the wind field; among them, the adjustment process can start from the encoder, and the encoder extracts latent representations (Z t ) from the input wind field data (W), and these data are encoded in RGB and are used to capture wind intensity, wind direction, and impervious areas. Moreover, these latent representations are continuously improved through the diffusion process and gradually learn to approximate the multi-dimensional complexity of wind dynamics.

[0080] Without loss of generality, the generated text embeddings can encode two key aspects: the wind direction and cross-section angle of the simulated wind field. These embedding information can be used as prompts for specific angles to guide the model to generate outputs consistent with the specific direction and cross-section input. At the same time, the control network (ControlNet) adjustment mechanism can also be applied to the geometric and structural information of the building environment; through this dual adjustment mechanism, it can be ensured that the model adapts to the high-level context features and local structural changes in the dataset at the same time.

[0081] Specifically, LoRA fine-tuning can introduce an efficient parameter adjustment layer within this framework. By integrating a low-order adaptation layer, the model optimizes computing resources while maintaining high accuracy. Exemplarily, the key training parameters can include 50 learning steps, 20 epochs, and a learning rate of 1×10 for the text prompt embedding - - and a learning rate of 1×10 for the UNet backbone network-- Learning rate. These parameters can be adjusted using a cosine annealing strategy to avoid overshooting the optimal value or getting stuck in local minima, thus ensuring robust convergence.

[0082] Specifically, the denoising UNet plays a crucial role in the decoding stage, which can convert the latent representation (Z t ) back to the spatial resolution wind field prediction . These predictions can be conditioned on the fused output of text embeddings and structural features processed by a conditional projector, which combines information from both sources to enable the UNet to accurately reconstruct wind dynamics in multi-angle and different urban scenarios. The final output is a dynamic 3D wind field simulation; reconstructed from multi-angle inputs, it can capture the complexity of the interaction between wind flow and urban geometry. By combining the advantages of LoRA fine-tuning, text-embedded prompts, and structural adjustment via a control network, the proposed framework addresses the limitations of traditional simulation methods and provides a scalable and accurate solution for urban wind field analysis.

[0083] In some possible implementations, to achieve comprehensive multi-angle prediction and reconstruct a cohesive wind field for new on-site data, refer to Figure 9 , after step S320, the method further includes but is not limited to the following steps:

[0084] In S321, obtain on-site wind field data information.

[0085] Specifically, the terminal device can obtain on-site wind field data information, where the on-site wind field data information is used to describe the on-site wind field data.

[0086] In S322, input the on-site wind field data information into the target wind simulation model information to generate output result information.

[0087] Specifically, after the terminal device obtains the on-site wind field data information, the terminal device can input the on-site wind field data information into the target wind simulation model information to generate output result information, where the output result information is used to describe the output result of the target wind simulation model information.

[0088] In S323, based on inverse transformation processing, generate wind intensity information and wind direction information according to the output result information.

[0089] Specifically, after the terminal device generates the output result information, the terminal device can, based on inverse transformation processing, generate wind intensity information and wind direction information according to the output result information, thereby realizing the decoding of the RGB-encoded output generated by the model. Among them, the inverse transformation processing can refer to the relevant descriptions in the above steps S211 to S216.

[0090] Exemplarily, refer to Figure 10 and Figure 11 , the terminal device maps the green channel back to wind intensity information through inverse conversion. At the same time, the blue channel reconstructs the angular relationship of the wind direction vector relative to the prevailing wind direction and maps it back to wind direction information. Through a dual decoding process, it ensures the accurate acquisition of wind intensity and wind direction from the model output.

[0091] Exemplarily, refer to Figure 10 and Figure 11 , multi-angle prediction fusion can generate a comprehensive three-dimensional representation of wind dynamics, merging individual two-dimensional cross-section outputs into a single spatially consistent field; this synthesis takes into account the overlapping and interdependent wind behaviors between adjacent cross-sections, ensuring the preservation of local variations and interactions. In one possible implementation, the fusion process can employ an interpolation technique to integrate the outputs at all angles under the guidance of the geometric and structural constraints of the urban environment, thereby effectively reconstructing complex airflow patterns and capturing the intricate interactions between wind, buildings, and impervious surfaces.

[0092] Without loss of generality, by combining data from different angles, this method improves the accuracy and robustness of wind field simulations, especially in high-density urban environments. The resulting three-dimensional wind field can provide actionable insights into wind intensity gradients, flow directions, and local turbulence, enabling practitioners to more confidently assess the impact of wind on urban planning and building performance. In addition, this method is flexible and can dynamically adapt to different spatial layouts and boundary conditions, so it can be applied to urban configurations that have never been seen before.

[0093] Specifically, this multi-angle fusion and reconstruction process can address the limitations of traditional two-dimensional simulation techniques by providing a dynamic three-dimensional representation of wind interactions. The inverse conversion from RGB coding to physical parameters ensures the interpretability of the prediction results and consistency with real-world wind dynamics, providing a scalable and accurate tool for urban wind analysis and great application potential for complex building and urban planning scenarios.

[0094] Exemplarily, refer to Figure 12 , in order to facilitate more sensitive and detailed analysis in the follow-up, overcome the limitations of previous two-dimensional methods, and more realistically reproduce the wind dynamics in high-density urban areas, the terminal device can evaluate the performance of the target wind simulation model information. As Figure 12 can be seen, the target wind simulation model information shows stable convergence after a total of 40,000 training steps, indicating that it can accurately simulate multi-angle wind dynamics while maintaining computational efficiency at the same time.

[0095] Exemplarily, refer to Figure 13, the terminal device can also generate prediction results for the test site and then perform multi-view cross-sectional wind field reconstruction. Exemplarily, please refer to Table 1 below. It can be seen from Table 1 that the prediction results are in good agreement with the expected wind dynamics, verifying the general ability of the target wind simulation model information for unknown environments. Quantitatively, the structural similarity index (SSIM) of the image-level evaluation is 0.78, indicating a high similarity between the predicted wind field output and the reference wind field output. Moreover, the overall deviation of the wind intensity parameter matching within the entire range is less than 24.7%, which also confirms the accuracy of the target wind simulation model information in capturing wind intensity changes. At the same time, in terms of computational efficiency, compared with traditional CFD simulations, this diffusion model framework significantly shortens the processing time. On an NVIDIA RTX 4090 GPU, the process of generating multi-angle cross-sectional outputs and reconstructing the wind field only takes 15 minutes, saving 57% of the time compared to the more than 35 minutes usually required for directly setting up and running complex CFD simulations on-site. These results highlight the potential of this diffusion model framework in simplifying the wind field analysis workflow without significantly reducing the accuracy.

[0096] Table 1 Verification Results

[0097] Evaluation matrix Accuracy ratio Computing time savings SSIM(average) 0.78 20min(57%time saving) Wind intensity error 24.7％ Wind direction error 43.1％

[0098] The implementation principle of the method for constructing a diffusion model framework based on wind simulation and building performance evaluation in the embodiments of this application is as follows: The terminal device can first obtain multiple multi-angle wind field data information, and then quickly generate wind simulation dataset information based on the multiple multi-angle wind field data information. Finally, based on the wind simulation dataset information, train a preset multi-angle diffusion model to effectively generate target wind simulation model information, thereby providing a simplified solution for rapid wind field simulation and building performance evaluation. The ability to predict multi-angle wind dynamics in a high-density urban environment enables designers and planners to seamlessly integrate wind field analysis into the early stage of design exploration, enabling urban planners and architects to achieve a more dynamic evaluation process, that is, iteratively adjust the design according to real-time wind performance feedback. For example, establish a dynamic optimization path, continuously update the wind field prediction and related performance indicators (such as energy load and ventilation efficiency) according to design changes, thereby significantly reducing the time and computational workload required for performance-driven decisions, making wind simulation not only limited to two-dimensional planes, but also improving its applicability in dynamic scenarios and enhancing reliability.

[0099] It should be noted that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of this application.

[0100] Embodiments of the present application also provide a diffusion model framework construction system based on wind simulation and building performance evaluation. For ease of description, only parts related to the present application are shown, such as Figure 14 As shown, the system 140 includes:

[0101] Multi-angle wind field data information acquisition module 141: used to acquire multiple multi-angle wind field data information;

[0102] Wind simulation dataset information generation module 145: used to generate wind simulation dataset information according to multiple multi-angle wind field data information;

[0103] Target wind simulation model information generation module 143: used to train a preset multi-angle diffusion model based on the wind simulation dataset information to generate target wind simulation model information.

[0104] Optionally, the system 140 further includes:

[0105] Three-dimensional vector data grid generation module: used to generate a three-dimensional vector data grid based on a preset visual programming language, where the visual programming language is Grasshopper, and the three-dimensional vector data grid is used to capture wind intensity information and wind direction information;

[0106] Correspondingly, the above-mentioned multi-angle wind field data information acquisition module 141 includes:

[0107] Multi-angle wind field data information generation sub-module: used to generate multiple multi-angle wind field data information based on a preset simulation tool, where the simulation tool is Eddy CFD.

[0108] Optionally, the above-mentioned wind simulation dataset information generation module 145 includes:

[0109] Wind simulation dataset information generation sub-module: used for each multi-angle wind field data information: to perform encoding processing on the multi-angle wind field data information to generate wind simulation dataset information, where the wind simulation dataset information includes multiple wind field image information, and the wind field image information is an RGB image;

[0110] Among them, the above-mentioned wind simulation dataset information generation sub-module includes:

[0111] Green channel information generation unit: used for each multi-angle wind field data information: based on the logarithmic transformation algorithm, map the wind intensity information of the multi-angle wind field data information to the green channel to generate green channel information;

[0112] Relative angle information acquisition unit: used to acquire relative angle information, where the relative angle information is used to describe the relative angle between the wind direction vector and the prevailing wind direction;

[0113] Blue channel information generation unit: used to map the wind direction information of the multi-angle wind field data information to the blue channel based on the relative angle information, and generate blue channel information;

[0114] Building area information acquisition unit: used to acquire impermeable terrain area information and building area information;

[0115] Impervious area information generation unit: used to generate impervious area information according to the impermeable terrain area information and the building area information;

[0116] Red channel information generation unit: used to map the impervious area information to the red channel and generate red channel information;

[0117] Wind field image information generation unit: used to generate wind field image information according to the green channel information, the blue channel information and the red channel information.

[0118] Optionally, the above-mentioned target wind simulation model information generation module 143 includes:

[0119] To-be-trained diffusion model generation sub-module: used to integrate a preset LoRA fine-tuning mechanism into a preset multi-angle diffusion model to generate a to-be-trained diffusion model, where the LoRA fine-tuning mechanism includes a text embedding prompt adjustment mechanism and a control network adjustment mechanism;

[0120] Target wind simulation model information generation sub-module: used to train the to-be-trained diffusion model based on the wind simulation dataset information and generate target wind simulation model information.

[0121] Optionally, the system 140 further includes:

[0122] On-site wind field data information acquisition module: used to acquire on-site wind field data information;

[0123] Output result information generation module: used to input the on-site wind field data information into the target wind simulation model information and generate output result information;

[0124] Wind intensity information generation module: used to generate wind intensity information and wind direction information based on reverse conversion processing according to the output result information.

[0125] It should be noted that the information interaction, execution process, etc. between the above modules, due to being based on the same concept as the method embodiment of the present application, for their specific functions and the technical effects brought, please refer to the method embodiment part for details, and will not be elaborated here.

[0126] The embodiment of the present application also provides a terminal device, such as Figure 15As shown, the terminal device 150 of this embodiment includes: a processor 151, a memory 152, and a computer program 153 stored in the memory 152 and executable on the processor 151. When the processor 151 executes the computer program 153, it implements the steps in the embodiment of the above diffusion model framework construction method, such as Figure 1 the steps S100 to S300 shown; or, when the processor 151 executes the computer program 153, it implements the functions of each module in the above device, such as Figure 14 the functions of the modules 141 to 143 shown.

[0127] The terminal device 150 may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device 150 includes but is not limited to the processor 151 and the memory 152. Those skilled in the art can understand that Figure 15 these are merely examples of the terminal device 150 and do not constitute a limitation on the terminal device 150. It may include more or fewer components than shown in the figure, or combine certain components, or different components. For example, the terminal device 150 may further include input / output devices, network access devices, a bus, etc.

[0128] Among them, the processor 151 may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.; the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0129] The memory 152 may be an internal storage unit of the terminal device 150, such as the hard disk or memory of the terminal device 150. The memory 152 may also be an external storage device of the terminal device 150, such as a plug-in hard disk equipped on the terminal device 150, a smart media card (SMC), a secure digital (SD) card, a flash card, etc.; further, the memory 152 may also include both the internal storage unit and the external storage device of the terminal device 150. The memory 152 may also store the computer program 153 and other programs and data required by the terminal device 150. The memory 152 may also be used to temporarily store data that has been output or will be output.

[0130] An embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the steps of the above-mentioned method embodiments. The computer program includes computer program code, which may be in the form of source code, object code, executable file or some intermediate form, etc.; the computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0131] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereby. Therefore, all equivalent changes made according to the method, principle and structure of the present application should be covered within the protection scope of the present application.

Claims

1. A method for constructing a diffusion model framework based on wind simulation and building performance evaluation, characterized in that: The method comprises: Obtain multiple wind field data information from multiple angles; Generate wind simulation data set information according to the plurality of multi-angle wind field data information; Based on the wind simulation data set information, a preset multi-angle diffusion model is trained to generate target wind simulation model information.

2. The method according to claim 1, characterized in that Before acquiring a plurality of multi-angle wind field data information, the method further includes: Generate a three-dimensional vector data grid based on a preset visual programming language, wherein the visual programming language is Grasshopper, and the three-dimensional vector data grid is used to capture wind intensity information and wind direction information; Accordingly, the obtaining of multiple multi-angle wind field data information includes: Based on a preset simulation tool, a plurality of multi-angle wind field data information is generated, wherein the simulation tool is Eddy CFD.

3. The method according to claim 1, characterized in that Generating wind simulation data set information according to the plurality of multi-angle wind field data information comprises: For each of the multi-angle wind field data information: encoding the multi-angle wind field data information to generate wind simulation data set information, wherein the wind simulation data set information includes a plurality of wind field image information, and the wind field image information is an RGB image; Wherein, for each of the multi-angle wind field data information: encoding the multi-angle wind field data information to generate wind simulation data set information includes: For each of the multi-angle wind field data information: based on a logarithmic transformation algorithm, the wind intensity information of the multi-angle wind field data information is mapped to a green channel to generate green channel information; Acquiring relative angle information, wherein the relative angle information is used to describe the relative angle between the wind direction vector and the prevailing wind direction; Based on the relative angle information, mapping the wind direction information of the multi-angle wind field data information to a blue channel to generate blue channel information; Obtain information on impervious terrain areas and building areas; Generate impermeable area information according to the impermeable terrain area information and the building area information; Mapping the impermeable area information to a red channel to generate red channel information; The wind field image information is generated according to the green channel information, the blue channel information and the red channel information.

4. The method according to claim 1, characterized in that: The method of training a preset multi-angle diffusion model based on the wind simulation data set information to generate target wind simulation model information includes: Integrate a preset LoRA fine-tuning mechanism into a preset multi-angle diffusion model to generate a diffusion model to be trained, wherein the LoRA fine-tuning mechanism includes a text embedded prompt adjustment mechanism and a control network adjustment mechanism; Based on the wind simulation data set information, the diffusion model to be trained is trained to generate target wind simulation model information.

5. The method according to claim 3, characterized in that: After training the preset multi-angle diffusion model based on the wind simulation data set information to generate target wind simulation model information, the method further includes: Obtain on-site wind farm data information; Inputting the on-site wind field data information into the target wind simulation model information to generate output result information; Based on the reverse conversion process, the wind intensity information and the wind direction information are generated according to the output result information.

6. A diffusion model framework construction system based on wind simulation and building performance evaluation, characterized in that: The system comprises: Multi-angle wind field data information acquisition module: used to obtain multiple multi-angle wind field data information; A wind simulation data set information generation module is used to generate wind simulation data set information according to a plurality of multi-angle wind field data information; Target wind simulation model information generation module: used to train a preset multi-angle diffusion model based on the wind simulation data set information to generate target wind simulation model information.

7. The system according to claim 6, characterized in that The system further comprises: A three-dimensional vector data grid generation module is used to generate a three-dimensional vector data grid based on a preset visual programming language, wherein the visual programming language is Grasshopper, and the three-dimensional vector data grid is used to capture wind intensity information and wind direction information.

8. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images