Urban and rural space planning dynamic simulation teaching system

Abstract

The invention relates to the technical field of dynamic simulation teaching, in particular to a dynamic simulation teaching system for urban and rural space planning, which comprises a data acquisition unit, an enhanced analysis unit and a multi-modal feedback processing unit. The method comprises the following steps: calculating the load deformation quantity of a component through a finite element algorithm, generating a ventilation efficiency cloud picture and a shadow coverage rate curve in combination with a computational fluid mechanics model and a ray tracing algorithm, analyzing a user operation instruction to trigger stress over-limit early warning, and realizing force feedback and multi-physics field data visualization through VR equipment. The problems of insufficient multi-source data fusion and lack of physical authenticity in traditional teaching are solved, dynamic physical verification and interactive teaching of a planning scheme are realized, and the planning decision ability and engineering practice cognition of students are improved.

Description

Technical Field

[0001] This invention relates to the field of dynamic simulation teaching technology, and more specifically, to a dynamic simulation teaching system for urban and rural spatial planning. Background Technology

[0002] Dynamic simulation teaching is an important technology. In urban and rural planning teaching practice, dynamic simulation teaching systems, by constructing interactive virtual planning scenarios, are a core tool to help students understand the principles of spatial layout and verify the feasibility of planning schemes.

[0003] As urban and rural planning education moves towards digitalization and practical application, constructing a teaching system that integrates multi-source data, physical simulation, and interactive feedback is crucial for enhancing students' planning decision-making abilities and bridging the gap between theory and practice. Traditional simulation models based solely on geometric modeling are no longer sufficient to meet the teaching needs of complex planning scenarios. However, traditional urban and rural spatial planning simulation teaching technologies suffer from core problems such as insufficient multi-source data fusion and a lack of physical realism. Existing solutions rely solely on building geometric models, failing to integrate material physical properties and dynamic environmental data. When students adjust building layouts, the impact of climate factors on planning schemes cannot be simulated. Furthermore, when changing building orientations, traditional systems cannot calculate in real-time the impact of changes in light intensity on shadow coverage. The system suffers from several problems: excessive light and shadow analysis errors, making it difficult for students to understand the impact of the lighting environment on spatial experience; the use of a simplified mechanical model increases the deviation in structural stress calculations; when students delete load-bearing components, the system cannot accurately warn of stress exceeding limits, leading to incorrect structural safety perceptions; furthermore, traditional systems lack fluid dynamics simulations such as ventilation efficiency, and cannot output ventilation efficiency cloud maps when the planning layout affects airflow, causing students to ignore the impact of the wind environment on building functions. Ultimately, this results in problems such as poor light and heat environment and structural safety hazards in the actual application of the planning scheme, increasing the disconnect between teaching content and engineering practice. To solve this technical problem, we provide a dynamic simulation teaching system for urban and rural spatial planning. Summary of the Invention

[0004] The purpose of this invention is to provide a dynamic simulation teaching system for urban and rural spatial planning to solve the problems mentioned in the background art.

[0005] 1. Due to the insufficient fusion of multi-source data in traditional technologies and the lack of integration of material physical properties and environmental data, shadow analysis has large errors. Therefore, this case study uses a data acquisition unit to extract component-level parameters and environmental dynamic data, and enhances the analysis unit to generate shadow coverage curves, which can accurately simulate the effects of illumination and help students understand the light environment.

[0006] 2. Due to the lack of physical realism in traditional systems, structural stress calculations have large deviations and cannot accurately warn of stress exceeding limits. Therefore, this case study uses finite element algorithms and adaptive mesh refinement to trigger stress exceeding limit warning coordinates, enabling accurate calculation of structural stress and improving students' understanding of structural safety.

[0007] To achieve the above objectives, a dynamic simulation teaching system for urban and rural spatial planning is provided, comprising:

[0008] The data acquisition unit extracts component-level geometric topology parameters and dynamic environmental data of the teaching scenario through building information modeling and climate sensors, respectively.

[0009] The enhanced analysis unit generates a 3D solid model based on component-level geometric topology parameters, inputs the physical properties of component materials into the 3D solid model, calculates the deformation of the component under load through the finite element algorithm, extracts the real-time airflow velocity value of the environmental dynamic data and inputs it into the computational fluid dynamics model, outputs a ventilation efficiency cloud map in the teaching scene, then drives the ray tracing algorithm based on the light intensity value of the environmental dynamic data to generate a curve of the building shadow coverage rate changing over time, finally detects user operation commands, analyzes and calculates the structural stability loss rate based on the user operation commands, and triggers the stress over-limit warning coordinates of the load-bearing components.

[0010] The multimodal feedback processing unit converts user operation commands into force feedback vibration intensity values ​​of VR devices, and overlays ventilation efficiency cloud maps and building shadow coverage curves over time onto the 3D scene, marks the coordinates of stress over-limit warnings for load-bearing components, and generates violation operation reports with physical parameters.

[0011] As a further improvement to this technical solution, the generation of a three-dimensional solid model based on component-level geometric topology parameters employs a non-uniform rational B-spline surface reconstruction technique with control point reverse calibration, as detailed below:

[0012] The topological connection type of beam-column nodes in the building information model is analyzed to generate an initial control point mesh. The initial control point mesh is iteratively optimized based on the measured wall thickness and the tolerance zone of the beam-column cross-section to generate control point coordinates. The semantic label level of the load-bearing components is extracted from the component-level geometric topology parameters. A three-dimensional solid model with mechanical transfer path is established based on the semantic label level of the load-bearing components and the control point coordinates.

[0013] As a further improvement to this technical solution, an inversion compensation mechanism based on material spectral characteristics is adopted to input the physical properties of the component material into the three-dimensional solid model:

[0014] For components lacking physical properties, a near-infrared spectroscopy scanner is activated to obtain the molecular vibration spectrum of the component surface, and then input into a convolutional neural network. The convolutional neural network outputs the compensated density distribution matrix and elastic modulus gradient tensor by matching the spectral-mechanical property mapping relationship of the molecular vibration spectrum of the component surface in the material database.

[0015] As a further improvement to this technical solution, the calculation of the component's deformation under load using the finite element algorithm employs an adaptive mesh refinement strategy based on the stress concentration factor:

[0016] The von Mises stress gradient change rate in the beam-column joint region is identified. For beam-column joint regions where the von Mises stress gradient change rate exceeds the threshold, an octree mesh is recursively subdivided to refine the local mesh. The subdivision level is positively correlated with the von Mises stress gradient change rate. An explicit dynamic algorithm is used to solve the transient deformation displacement field of the refined local mesh. The load deformation of the component is calculated based on the transient deformation displacement field.

[0017] As a further improvement to this technical solution, the operation of the computational fluid dynamics model adopts the building surface boundary layer vortex stripping simulation technology:

[0018] A body-fitted boundary layer mesh is generated on the outer surface of the building. Velocity slip boundary conditions are set, and the real-time airflow velocity value is decomposed into the mean flow field and turbulent fluctuation components. These components are then input into the large eddy simulation algorithm to capture the evolution path of the separated eddies. Based on the evolution path of the separated eddies, a ventilation efficiency cloud map of the eddy intensity distribution is output.

[0019] As a further improvement to this technical solution, the curve of the generated building shadow coverage over time is obtained using the photon beam scattering cumulative statistical method:

[0020] A continuous motion model of the sun's position is constructed, and an anisotropic photon emission source in the model is driven to emit photon beams. The reflection and scattering paths of each photon beam on the building facade are tracked, and the photon flux density of the building facade blocked per unit time is calculated based on the reflection and scattering paths of each photon beam on the building facade. A time-series heat map of shadow coverage is generated based on the shadow coverage flux density, and the curve of the building shadow coverage over time is extracted from the time-series heat map of shadow coverage.

[0021] As a further improvement to this technical solution, the analysis and calculation of the structural stability loss rate adopts the eigenvalue buckling mode displacement superposition method:

[0022] The semantic parsing engine converts user operation commands into a set of computer-processable component deletion operations, generates a list of topological relationships of the components to be deleted, reconstructs the stiffness matrix of the remaining components, calculates the first three buckling mode eigenvectors of the remaining components based on the stiffness matrix of the remaining components, and quantifies the overall stiffness attenuation ratio of the components by weighted superposition of the first three buckling mode eigenvectors through modal displacement. Finally, the structural stability loss rate is calculated based on the overall stiffness attenuation ratio of the components.

[0023] As a further improvement to this technical solution, the triggering coordinate for the over-stress warning of the load-bearing component is achieved through the following steps:

[0024] The deformation of the component under load output by the finite element algorithm is converted into Mises stress values ​​and mapped to the surface mesh nodes of the three-dimensional solid model.

[0025] A stress transmission path tree diagram is constructed based on the component connection relationship in the building information model. The stress transmission path tree diagram takes the high-stress component as the root node and traces the stress distribution of adjacent components level by level along the topological connection relationship of the building information model. This is used to locate the source of stress concentration propagation. The coordinates of grid nodes whose stress values ​​exceed the material yield strength threshold are marked in the stress transmission path tree diagram, and a pulse warning is triggered in the multimodal feedback processing unit.

[0026] As a further improvement to this technical solution, when calculating the deformation of the component under load using the finite element algorithm, the following steps are performed simultaneously:

[0027] The transient deformation displacement field is decomposed into intrinsic orthogonal decomposition to separate the main deformation modes. The deformation amplitude data of the main deformation modes are converted into vertex animation sequences of the three-dimensional solid model and then superimposed onto the corresponding component surfaces in the VR device scene.

[0028] As a further improvement to this technical solution, the ventilation efficiency cloud map output by the computational fluid dynamics model is linked to the structural analysis in the following way:

[0029] Wind pressure distribution data on the building surface is extracted from the ventilation efficiency cloud map and used as an additional load input to the finite element algorithm to recalculate the deformation of the component under load. When the wind-induced vibration displacement deviates from the original design allowable value by more than the preset deviation threshold, the coordinates of the high-risk wind pressure zone and the stress over-limit warning are simultaneously marked in the multimodal feedback processing unit.

[0030] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0031] In a dynamic simulation teaching system for urban and rural spatial planning, a data acquisition unit synchronously extracts component-level geometric topology parameters and environmental dynamic data. An enhanced analysis unit, based on finite element method, computational fluid dynamics model, and ray tracing algorithm, achieves multi-physics coupled simulation of building structure stress deformation, ventilation efficiency, and shadow coverage. This addresses the problems of insufficient multi-source data fusion and lack of physical realism in traditional technologies. The system utilizes control point reverse calibration technology to reconstruct a three-dimensional solid model, combined with a material spectral inversion compensation mechanism to ensure the accuracy of the model's mechanical transmission path and material properties. Through adaptive mesh refinement and stress over-limit early warning, it accurately locates structurally weak areas. With the help of VR force feedback and multimodal data overlay, physical parameters are visualized, improving the accuracy of physical verification of planning schemes, helping students establish correct structural safety awareness, bridging the gap between teaching and engineering practice, and enhancing planning decision-making capabilities in complex scenarios. Attached Figure Description

[0032] Figure 1 This is an overall block diagram of the present invention.

[0033] The meanings of the labels in the diagram are as follows:

[0034] 1. Data acquisition unit; 2. Enhanced analysis unit; 3. Multimodal feedback processing unit. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] This invention provides a dynamic simulation teaching system for urban and rural spatial planning. Please refer to [link / reference]. Figure 1 As shown, it includes the following units:

[0037] Data acquisition unit 1 extracts component-level geometric topology parameters and environmental dynamic data of the teaching scenario through building information modeling and climate sensors, respectively.

[0038] To construct a high-precision 3D solid model with mechanical transmission paths to support subsequent structural analysis and physical simulation, Enhanced Analysis Unit 2 generates a 3D solid model based on component-level geometric topology parameters using a non-uniform rational B-spline surface reconstruction technique with control point reverse calibration. The physical properties of the component materials are then input into the 3D solid model, as detailed below:

[0039] First, since the topological connection type of beam-column nodes in the Building Information Model (BIM) directly determines the mechanical transmission path of the structure, it is necessary to first analyze the topological connection type of beam-column nodes in the BIM and generate an initial control point mesh based on the analysis results. This mesh adopts a control point architecture of a non-uniform rational B-spline surface. The distribution density of the initial control points is dynamically adjusted according to the node complexity to ensure that the model can initially represent the geometric shape of the components. The geometric representation error of the initial mesh is controlled within ±2 mm. Subsequently, to improve the geometric consistency between the model and the actual components, the initial control point mesh is iteratively optimized based on the measured wall thickness and the tolerance zone of the beam-column cross-section dimensions to generate control point coordinates. The measured wall thickness and the tolerance zone of the beam-column cross-section dimensions are both component-level parameters obtained through data acquisition unit 1. The specific optimization process is as follows: calculate the deviation between the measured dimensions and the dimensions represented by the initial mesh. If the deviation exceeds the upper limit of the tolerance zone, the control point coordinates are adjusted using the gradient descent algorithm until more than 90% of the component dimensions deviate from the tolerance zone. If the difference falls within the tolerance zone, iterative optimization is performed 3 to 5 times to improve the geometric accuracy of the model to ±1 mm, laying a geometric foundation for subsequent mechanical calculations. Next, the semantic label level of the load-bearing components is extracted from the component-level geometric topology parameters. This label is defined by the component attributes in the building information model, such as "main beam", "secondary beam", and "column". This level reflects the stress importance of the component in the structural system. Based on the semantic label level of the load-bearing components and the optimized control point coordinates, a three-dimensional solid model with a mechanical transfer path is established. The construction logic of the mechanical transfer path is as follows: with the semantic label level of the load-bearing components as the priority, the force flow transfer relationship is established step by step from the basic components to the upper components. Through the control point weight allocation of the non-uniform rational B-spline surface, the control points of high-priority components have a 30% higher weight in mechanical calculations, ensuring that the model can accurately reflect the mechanical transfer characteristics of the structure. This provides a more realistic structural foundation for finite element analysis and effectively solves the problem of the disconnect between the model and engineering practice in traditional teaching.

[0040] To ensure the accurate material physical properties of the 3D solid model and support the physical authenticity of subsequent finite element analysis, the enhanced analysis unit 2 employs an inversion compensation mechanism based on material spectral characteristics to input the component's material physical properties into the 3D solid model, as detailed below:

[0041] First, for components lacking physical properties in the Building Information Model (BIM), the system automatically triggers a near-infrared spectroscopy scanner to scan the component surface point-by-point and acquire the molecular vibrational spectrum. During scanning, the scanner moves along the component surface at a speed of 1 mm / s, acquiring a set of spectral data every 50 micrometers. Each set of data contains absorbance values ​​at 4096 wavelengths, forming a molecular vibrational spectrum image of the component surface. This scanning precision can capture differences in the material's microstructure, ensuring the integrity of the spectral data. Subsequently, the acquired molecular vibrational spectrum image is input into a pre-trained convolutional neural network. The network comprises 5 convolutional layers, 3 pooling layers, and 2 fully connected layers, employing the ReLU activation function. The training data originates from a database containing the spectral-mechanical properties of over 2000 building materials. This database experimentally calibrates the mapping relationship between the near-infrared spectrum of each material and mechanical parameters such as density and elastic modulus. The specific processing flow is as follows: the convolutional layers extract characteristic bands from the spectral images using 3×3 convolutional kernels, such as the hydroxyl absorption peak at 1450 nm and the carbon-carbon double bond characteristic peak at 2350 nm; the pooling layers reduce the dimensionality of these features, preserving key spectral features while reducing computational cost; and the fully connected layers... The layer matches the extracted spectral features with the mapping relationships in the database, adjusts the weights through the backpropagation algorithm, and the convolutional neural network outputs the compensated density distribution matrix and elastic modulus gradient tensor by matching the spectral-mechanical property mapping relationship of the molecular vibration spectrum of the component surface in the material database. The density distribution matrix is ​​distributed according to the grid nodes of the component surface, with each node corresponding to a density value. The matrix resolution is consistent with the grid accuracy of the component surface. The elastic modulus gradient tensor characterizes the variation trend of the material's elastic modulus in three-dimensional space. The anisotropic characteristics are described by the nine components of the tensor. The gradient change rate of adjacent nodes is considered when calculating the tensor to make the spatial transition of the elastic modulus smooth. This inversion compensation mechanism forms a technical linkage with the three-dimensional solid model construction mentioned above. The material properties obtained by inverting the spectral features are directly assigned to the grid nodes of the model, so that the model can truly reflect the mechanical behavior of the material when performing finite element analysis. Compared with the traditional method of manually inputting properties, it improves the accuracy of material parameters, especially for non-standard components or new materials. It has a significant effect on property compensation and effectively solves the analysis error problem caused by the lack of material parameters in traditional simulation, providing a reliable material physics basis for subsequent stress over-limit early warning.

[0042] To accurately calculate the deformation of components under stress and address the insufficient accuracy of traditional uniform meshes in stress concentration regions, Enhanced Analysis Unit 2 calculates the deformation of components under load using the finite element method, employing an adaptive mesh refinement strategy based on stress concentration factors, as detailed below:

[0043] To identify the rate of change of the von Mises stress gradient in the beam-column joint region, firstly, for the beam-column joint region in the 3D solid model, which is prone to stress concentration due to geometrical abrupt changes, calculate the von Mises stress value of each mesh node to characterize the overall stress state of the material. Then, calculate the stress gradient between adjacent nodes using the finite difference method, i.e., the rate of change of stress along the three coordinate axes. The direction with the largest absolute value is taken as the principal direction of the stress gradient for that node. For example, the rate of change of stress at a node might be 5 MPa / mm² along the X-axis, 3 MPa / mm² along the Y-axis, and 4 MPa / mm² along the Z-axis. If the stress gradient rate is set to 5 MPa / mm, then a threshold for the stress gradient rate is set, initially at 10 MPa / mm, which can be dynamically adjusted by the material yield strength. For beam-column joint regions where the von Mises stress gradient rate exceeds the threshold, an octree mesh recursively subdivides to refine the local mesh. The octree subdivision logic is as follows: the mesh element to be refined is taken as the root node, recursively divided into 8 sub-elements, corresponding to the 8 octaves of 3D space. After each subdivision, the stress gradient of the new element is calculated. If it still exceeds the threshold, subdivision continues until the stress gradient of the element is reached. If the force gradient is less than a threshold or reaches the maximum subdivision level (default 6 levels), the subdivision level is linearly positively correlated with the stress gradient change rate, ensuring that the mesh density in high-stress areas is sufficient to capture stress abrupt changes. Next, an explicit dynamic algorithm is used to solve the transient deformation displacement field of the refined local mesh. The algorithm iterates with time steps, calculating node acceleration based on the stress state of the current mesh node using Newton's second law within each time step. The acceleration is then integrated over time to obtain velocity and displacement, updating node positions and checking for element distortion. If the distortion rate exceeds 15%, mesh re-division is triggered. This algorithm does not require solving large systems of equations and is suitable for handling transient responses in high-stress areas. Finally, the load-bearing deformation of the component is calculated based on the transient deformation displacement field. The deformation calculation is divided into global and local parts: the global deformation is the displacement difference between the two endpoints of the component, and the local deformation is the maximum displacement value of nodes in the high-stress area. For example, if the global displacement of a beam component is 2 mm and the maximum displacement in the local high-stress area is 0.5 mm, the load-bearing deformation is expressed as 2 mm globally and 0.5 mm locally, providing a quantitative basis for subsequent stress over-limit warnings.

[0044] Further explanation is needed regarding the optimization of stress gradient calculation. Cubic spline interpolation is used to smooth the nodal stress, avoiding abrupt gradient changes caused by mesh discretization. Specifically, for each node, a cubic spline function is constructed using the stress values ​​of its 26 neighboring nodes. When calculating the gradient, the interpolated continuous stress field is used to reduce the gradient calculation error. The octree subdivision termination condition is as follows: in addition to the stress gradient threshold, a lower limit for the element size is added to prevent the computational workload from exploding due to excessive subdivision. When the element size is less than the lower limit, subdivision stops even if the stress gradient exceeds the limit, and a higher-order shape function is used to improve the calculation accuracy, ensuring a balance between accuracy and efficiency in the algorithm. Explicit dynamic time step control: based on the minimum mesh size h and the longitudinal wave propagation velocity c in the material, the time step Δt is set to h / (2c) to ensure that the Courant stability condition is met.

[0045] Furthermore, in order to intuitively present the component deformation process obtained from finite element calculations in the VR teaching scenario and enhance students' dynamic understanding of structural stress and deformation, the following steps are performed simultaneously when calculating the component's deformation under load using the finite element algorithm:

[0046] First, the transient deformation displacement field output by the finite element method is subjected to eigenorthogonal decomposition to separate the main deformation modes. Eigenorthogonal decomposition is a data dimensionality reduction technique that characterizes complex deformations by extracting the main feature modes. In practice, the displacement field is first organized into a two-dimensional matrix according to the time series, with each row of the matrix corresponding to the three-dimensional displacement vector of all grid nodes at a certain time. Then, the covariance matrix of this matrix is ​​calculated, and its eigenvalues ​​and eigenvectors are solved by singular value decomposition. The eigenvectors correspond to different deformation modes. They are sorted by the size of the eigenvalues, and the top N eigenvectors with a cumulative energy ratio exceeding 90% are selected as the main deformation modes. N is 5-10. The complexity is dynamically adjusted while preserving the main deformation features. The deformation amplitude data of the main deformation modes is converted into a vertex animation sequence of the 3D solid model. Then, the deformation amplitude data of the main deformation modes is separated. For the displacement field at each time point, its projection coefficient on each main mode is calculated. This coefficient reflects the contribution of the corresponding mode at the current moment. For example, if the projection coefficient of the displacement field on the first main mode is 0.8 at a certain moment, it indicates that this mode accounts for 80% of the current deformation. By tracking the change curve of the projection coefficient over time, the dynamic amplitude characteristics of each mode are obtained. Then, the deformation amplitude data of the main deformation modes is converted into a vertex animation sequence of the 3D solid model. The specific conversion process is as follows: Establish a mapping relationship between the main pattern feature vectors and model vertices. Each feature vector corresponds to a set of vertex displacement offsets. Linearly combine the projection coefficients with the feature vectors according to the time sequence to generate vertex displacement increments at each moment. Perform smooth interpolation on the displacement increments (using cubic spline interpolation) to achieve an animation frame rate of 30 frames per second, avoiding screen stuttering. Encode the interpolated vertex displacement sequence into an animation format supported by the VR engine. Each vertex contains position, normal, and tangent information to ensure rendering accuracy. Finally, overlay the vertex animation sequence onto the corresponding component surface in the VR device scene, using V... The R engine's material system binds animation sequences to the geometric models of components, enabling real-time rendering of the deformation process. The specific binding logic is as follows: for static components, vertex blending technology is used to fuse vertex positions at different times according to the animation frame weights; for dynamic interactive components, the latest displacement data from finite element calculations is received in real time, vertex positions are updated, and force feedback from VR devices is triggered. For example, vibration intensity is positively correlated with deformation, and stress cloud diagrams are displayed synchronously when superimposed. Visual cues are enhanced through color mapping, enabling students to intuitively associate deformation with stress distribution, effectively bridging the teaching gap between abstract mechanical calculations and intuitive engineering cognition.

[0047] Real-time airflow velocity values ​​from dynamic environmental data are extracted and input into a computational fluid dynamics (CFD) model. To accurately simulate the airflow state within the teaching scenario and generate a ventilation efficiency cloud map, Enhanced Analysis Unit 2 employs a building surface boundary layer vortex stripping simulation technique to create a CFD model, outputting a ventilation efficiency cloud map within the teaching scenario, as detailed below:

[0048] First, a body-fitted boundary layer mesh is generated on the building's outer surface to capture near-wall flow characteristics. A triangular mesh is generated using the advancing wavefront method. The thickness of the first mesh layer is determined by the wall Reynolds number, calculated using the formula δ = 0.03 × L × Re. -0.2 Where L is the building feature length and Re is the Reynolds number, ensuring the dimensionless wall distance is controlled within the range of 30-300, the mesh expands layer by layer along the wall normal at a growth rate of 1.2 times, with a total of 15 layers, so that the mesh within the boundary layer can accurately resolve the velocity gradient. Compared with a uniform mesh, this improves the ability to capture near-wall flow characteristics. Subsequently, velocity slip boundary conditions are set to conform to the actual flow state. At the building wall, the tangential velocity component adopts a partial slip model, with the slip length set to 0.01×δ, where δ is the boundary layer thickness. Setting the velocity component to zero, this condition is achieved by modifying the wall boundary term of the momentum equation, making the calculation closer to the flow separation phenomenon on the real building surface. Compared with the no-slip condition, this reduces the prediction error of the separation vortex position. Next, the real-time airflow velocity values ​​acquired by data acquisition unit 1 are decomposed into the mean flow field and turbulent fluctuation components. The Reynolds time-averaged method is used for decomposition, with a time averaging window of 10 flow cycles (if the wind speed is 5 m / s, the cycle is 2 seconds), to obtain the velocity vector field U(x,y,z) of the mean flow field and the turbulent fluctuation component u. After decomposition (x,y,z,t), the mean flow field reflects the overall wind direction, and the fluctuating component characterizes the random fluctuations of turbulence, providing initial conditions for large eddy simulation (Leddy Simulation). These initial conditions are then input into the Leddy Simulation algorithm to capture the evolution path of the separated eddies. A dynamic Smagorinsky subgrid model is used to close the control equations. Spatial discretization uses a second-order central difference scheme, and time propagation employs a third-order Runge-Kutta method. The computational domain size is set to 10 times × 8 times × 6 times the length of the building feature (length × width × height). The inlet boundary uses velocity ingress... Under the following conditions, the outlet is set as a pressure outlet (gauge pressure is zero), and the top and side boundaries are set as free-slip boundaries. The evolution of vortices with a scale larger than the mesh size is analyzed by solving the filtered Navier-Stokes equations. The time step is set to Δt = 0.001 × L / U∞ (U∞ is the incoming flow velocity), ensuring the stability condition of a Coulomb number less than 0.8 is met. The computational efficiency is improved compared to direct numerical simulation. Finally, the ventilation efficiency contour map of the vortex intensity distribution is output based on the separated vortex evolution path, using the Q criterion (Q = 0.5 × (Ω...). 2 -S 2The vortex structure is identified by a method (where Ω is the vortex tensor and S is the strain rate tensor). When Q>0, it is determined to be a vortex region. The vortex intensity is standardized to a dimensionless value between 0 and 1 and mapped to the computational domain mesh through trilinear interpolation. The volume is rendered using Paraview software. Red indicates a strong vortex region (low ventilation efficiency) and blue indicates a weak vortex region (high ventilation efficiency). The cloud map resolution is set to 200×200×150 voxels, which can clearly show the vortex distribution around the building and provide an intuitive basis for the ventilation assessment of the planning scheme.

[0049] Then, based on the light intensity values ​​from the dynamic environmental data, the ray tracing algorithm is driven. To accurately simulate the building's shadow coverage at different times and generate a shadow coverage curve that reflects the dynamic changes in the light environment, the enhanced analysis unit 2 uses the photon beam scattering cumulative statistical method to generate a curve showing the change in building shadow coverage over time, as follows:

[0050] First, a continuous solar position motion model is constructed to simulate the real solar trajectory. This model, based on astronomical algorithms, calculates the solar altitude and azimuth angles according to the geographical location (latitude and longitude), date, and time of the teaching scenario. Specifically, a simplified Saros period model combined with a real-time clock is used to update the solar position every minute, ensuring the continuity of solar motion. For example, for a region at 30 degrees North latitude, on the summer solstice of June 21st, the solar altitude angle gradually rises from 0 degrees at sunrise to 83 degrees at noon, and then drops to 0 degrees at sunset. The model can accurately reproduce this process, with the calculation error of the solar position controlled within ±0.5 degrees, providing an accurate light source reference for subsequent shadow calculations. Next, the anisotropic photon emission source in the continuous solar position motion model is driven to emit photon beams, taking into account the direction of solar radiation. The photon emission source emits photons according to the cosine law, meaning the emission intensity is highest when perpendicular to the sun and decreases as the angle increases. The emission parameters are set as follows: 100,000 photons are emitted every 1 minute, the photon energy is distributed according to the solar spectrum, the initial direction of the photon beam is consistent with the direction of sunlight, and the beam diameter is set to 0.1 meters to balance calculation accuracy and efficiency. This anisotropic emission mechanism makes the photon distribution conform to the characteristics of real solar radiation, improving the simulation accuracy of shadow edges compared to uniform emission. Subsequently, the reflection and scattering paths of each photon beam on the building facade are tracked using a recursive ray tracing algorithm. When each photon beam intersects the building surface, the reflection and scattering directions are calculated based on the bidirectional reflection distribution function of the material. For example, for a concrete wall, the reflectivity is set to 0.3 and the scattering rate to 0.7. After a photon beam collides, it scatters with a 70% probability and reflects specularly with a 30% probability. The tracking depth is set to 3 times, meaning each photon reflects a maximum of 3 times; tracking terminates if the depth is exceeded. This setting strikes a balance between computational efficiency and physical realism, ensuring that over 90% of the effective lighting paths are captured, reducing computation time. Then, based on the reflection and scattering paths of each photon beam on the building facade, the photon flux density obstructed by the building facade per unit time is calculated. The building facade is divided into 1m × 1m grid cells. For each photon beam, if it is obstructed by other building components during the tracking process (i.e., it does not reach the target facade), then... An occlusion count is incremented in the grid cell corresponding to the time of occlusion. For example, if a photon beam is blocked by the building on the east side at 10:00 and does not reach the west facade, the occlusion count of the corresponding grid cell on the west facade at 10:00 is incremented by 1. The occlusion count per unit time is converted into photon flux density (unit: beam / m²·min). This statistical method can quantify the shadow intensity of different facade areas, reducing statistical errors. Finally, a time-series heatmap of shadow coverage is generated based on the blocked photon flux density. The curve of building shadow coverage over time is extracted from the time-series heatmap. The heatmap is plotted with time as the horizontal axis (0-). Using a 24-hour timeframe and building facade grid as the vertical axis, the occlusion density is represented by color depth (black represents 100% occlusion, white represents no occlusion). A Gaussian filter (kernel size 3×3 time-space window) is used to smooth the heatmap, eliminating random noise in photon statistics. The shadow coverage rate of each facade grid over time is extracted from the heatmap. Coverage rate is defined as the ratio of the occluded photon flux density to the theoretical photon flux density without occlusion, expressed as a percentage. For example, if a grid has an occlusion density of 800 beams / m²·min at 10:00 and 1000 beams / min without occlusion, then the shadow coverage rate is... With an 80% coverage rate, a time resolution of 1 minute, and a spatial resolution of 1 meter, the curve clearly demonstrates the dynamic evolution of shadows over time, providing a quantitative basis for the light environment assessment of planning schemes. This photon beam scattering cumulative statistical method is linked with the light intensity monitoring technology of data acquisition unit 1. Real-time light data is used to calibrate photon emission energy, and the generated shadow coverage curve can inversely represent the spatiotemporal distribution of light intensity. Compared with traditional solar radiation analysis software, this improves the accuracy of shadow simulation for complex building shapes, enabling students to intuitively understand the impact of building orientation and shape on the light environment, and assisting in optimizing the daylighting design of planning schemes.

[0051] Finally, user operation commands are detected, and the structural stability loss rate is calculated based on these commands. To quantify the impact of user operation commands on structural stability, enhanced analysis unit 2 uses the eigenvalue buckling modal displacement superposition method to analyze and calculate the structural stability loss rate, as detailed below:

[0052] First, a semantic parsing engine converts user commands into a set of computer-processable component deletion operations. This engine, based on a hybrid architecture of rule-based engine and machine learning, uses a pre-trained model containing over 2000 planning-related commands. Keyword matching and syntactic analysis are used to generate the operation set, ensuring accurate command conversion. Next, a list of topological relationships for the components to be deleted is generated. All adjacency relationships of the components are extracted from the topological connection database of the building information model. An adjacency list data structure records the parent node (e.g., the column supporting it) and child nodes (e.g., the floor slab it supports) of each component, providing complete connection information for subsequent stiffness matrix reconstruction. Finally, the stiffness matrix of the remaining components is reconstructed based on the 3D solid model. The geometric parameters and material properties of the remaining components are used. The element stiffness matrix formula in the finite element method is employed to calculate the local stiffness matrix for each remaining component. This matrix is ​​then transformed into a global stiffness matrix using coordinate transformation. The components are assembled into a global stiffness matrix according to the connection method in the topology list. For n remaining components, the global stiffness matrix has a size of 6n×6n, which is stored using a sparse matrix to improve computational efficiency. Then, the eigenvectors of the first three buckling modes of the remaining components are calculated based on their stiffness matrices. The eigenvalue problem (K-λM)φ=0 is solved using the subspace iteration method, where K is the stiffness matrix, M is the mass matrix (simplified to an identity matrix), λ is the eigenvalue, and φ is the eigenvector. A convergence tolerance of 10 is set during the iteration process. -6 After 10-15 iterations, the first three eigenvalues ​​(λ1≤λ2≤λ3) and their corresponding eigenvectors φ1, φ2, and φ3 are extracted. These three modes typically contribute more than 85% of the buckling deformation energy of the structure and can effectively characterize the buckling tendency of the structure. Next, the eigenvectors of the first three buckling modes are superimposed using modal displacement weighting. The weighting coefficients are determined according to the reciprocal of the eigenvalues ​​(w1=1 / λ1, w2=1 / λ2, w3=1 / λ3) to ensure that lower-order modes (more likely buckling modes) receive higher weights. The superimposed total displacement vector φ_total=w1φ1+w2φ2+w3φ3 is then used to quantify the buckling tendency of the structure by calculating the ratio of the Euclidean norm of this vector to the modal displacement norm of the original structure (without deleting components). The overall stiffness attenuation ratio of the components is calculated, and finally, the structural stability loss rate is calculated based on the overall stiffness attenuation ratio of the components. The stability loss rate is defined as (1 - current stiffness matrix determinant / original stiffness matrix determinant) × 100%. This formula, combined with the sensitivity of the determinant to matrix singularity, can intuitively reflect the degree of decline in the structure's ability to resist buckling. When the loss rate exceeds a preset threshold, a stress over-limit warning for the load-bearing components is triggered. In the VR scene, potential buckling areas are marked with red highlights. This method and the construction of the three-dimensional solid model form a technical closed loop: the mechanical transmission path of the model provides the basis for stiffness matrix calculation, and the stability loss rate calculation after user operation can provide real-time feedback on the structural safety of the planned scheme, improving the accuracy of stability assessment for complex structural modifications.

[0053] The coordinates for triggering an over-stress warning for load-bearing components are established through the following steps:

[0054] Since the deformation of the component under load output by the finite element algorithm needs to be converted into stress index to assess safety, the deformation of the component under load output by the finite element algorithm is first converted into Mises stress value by Hooke's law. In specific implementation, for each mesh node, the normal stress σ=E·ε is calculated according to the elastic modulus E and deformation ε of the component to which it is located. Then, the Mises stress is synthesized by combining the shear stress component. Subsequently, the stress value is mapped to the surface mesh node of the three-dimensional solid model. Trilinear interpolation is used to ensure that the stress distribution is accurately matched with the geometric model, so that the stress cloud map can intuitively reflect the stress state of the component.

[0055] To locate the source of stress concentration propagation, a stress transmission path tree diagram is constructed based on the component connection relationships in the Building Information Model (BIM). This tree diagram uses high-stress components as root nodes and components with Mises stress values ​​exceeding 80% of the material's yield strength as root nodes. The yield strength threshold is obtained from a material database. A breadth-first search algorithm is used to trace the stress distribution of adjacent components along the topological connections to locate the source of stress concentration propagation. The coordinates of grid nodes with stress values ​​exceeding the material's yield strength threshold are marked in the stress transmission path tree diagram, and a pulse warning is triggered in the multimodal feedback processing unit 3. When the node stress value exceeds the material's yield strength, the corresponding three-dimensional grid coordinates (X, Y, Z) are extracted and associated with the component surface in the VR scene through a coordinate mapping table. Simultaneously, a pulse signal is sent to the multimodal feedback processing unit 3, triggering a triple warning: ① the corresponding coordinate point in the VR scene flashes a red highlight; ② the VR controller generates a 50Hz pulse vibration; ③ the system interface generates a text warning. The warning response time is controlled within 100 milliseconds to ensure students promptly notice structural safety hazards.

[0056] Since the impact of wind load on the structure needs to be included in stress analysis, wind pressure distribution data on the building surface is extracted from the ventilation efficiency cloud map output by the computational fluid dynamics model and input as an additional load into the finite element algorithm. Specifically, the cloud map is sampled at a grid density of 0.5m × 0.5m to obtain the static pressure value at the center of each grid, with negative values ​​representing suction and positive values ​​representing pressure. The cloud map grid is mapped to the outer surface nodes of the 3D solid model through coordinate transformation. The wind pressure is converted into nodal loads using an area-weighted method (formula: F = p·A, where p is the wind pressure and A is the surface area associated with the node). The load direction is perpendicular to the building surface. The wind pressure load is input as an additional load into the finite element algorithm, and the deformation of the components under load is recalculated using an implicit dynamics algorithm. After the calculation, the wind-induced vibration displacement of key parts of the building is extracted and compared with the original design allowable value, which is determined by building codes. When the deviation between the measured displacement and the allowable value exceeds 15%, it is judged as a high-risk state of wind load. The difference calculation uses the relative error formula: δ=(|D_measured - D_design| / D_design)×100%. When the wind-induced vibration displacement deviates from the original design allowable value by more than the preset deviation threshold, the system automatically performs double annotation: ① In the ventilation efficiency cloud map, areas with an absolute wind pressure value exceeding 200 Pa are marked as red high-risk areas. ② In the stress over-limit early warning system, the stress over-limit coordinates caused by wind load are simultaneously highlighted. GPU-accelerated volume rendering technology is used during annotation to keep the frame rate of the superimposed display of the cloud map and the structural model above 60 FPS, ensuring real-time visualization of multi-physics data. In addition, a report with wind pressure-stress coupling analysis is generated, including parameters such as high-risk area coordinates, wind pressure values, and stress over-limit amplitude, to help students understand the impact mechanism of wind environment on structural safety. In the multimodal feedback processing unit 3, the wind pressure high-risk area and stress over-limit early warning coordinates are simultaneously annotated to help students establish interdisciplinary planning safety cognition and effectively shorten the gap between classroom teaching and engineering practice.

[0057] The multimodal feedback processing unit 3 converts user operation commands into force feedback vibration intensity values ​​of VR devices, and overlays ventilation efficiency cloud maps and building shadow coverage curves over time onto the three-dimensional scene, marks the coordinates of stress over-limit warnings for load-bearing components, and generates a violation operation report with physical parameters.

[0058] This invention extracts component-level geometric topology parameters and environmental dynamic data through building information modeling and climate sensors, calculates component deformation under load using finite element method, and generates ventilation efficiency cloud maps and shadow coverage curves by combining computational fluid dynamics model and ray tracing algorithm. It analyzes user operation commands to trigger stress over-limit warnings and realizes force feedback and multi-physics field data visualization through VR devices. This solves the problems of insufficient multi-source data fusion and lack of physical authenticity in traditional teaching, realizes dynamic physical verification of planning schemes and interactive teaching, and improves students' planning decision-making ability and engineering practice cognition.

[0059] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A dynamic simulation teaching system for urban and rural spatial planning, characterized in that, Includes the following units: The data acquisition unit (1) extracts component-level geometric topology parameters and environmental dynamic data of the teaching scene through building information model and climate sensor, respectively; The enhanced analysis unit (2) generates a three-dimensional solid model based on the component-level geometric topology parameters, inputs the physical properties of the component material into the three-dimensional solid model, calculates the deformation of the component under load through the finite element algorithm, extracts the real-time airflow velocity value of the environmental dynamic data and inputs it into the computational fluid dynamics model, outputs the ventilation efficiency cloud map in the teaching scene, and then drives the ray tracing algorithm according to the light intensity value of the environmental dynamic data to generate the curve of the building shadow coverage rate changing with time. Finally, it detects the user operation command, analyzes and calculates the structural stability loss rate based on the user operation command, and triggers the stress over-limit warning coordinate of the load-bearing component. The multimodal feedback processing unit (3) converts the user's operation instructions into the force feedback vibration intensity value of the VR device, and overlays the ventilation efficiency cloud map and the curve of the building shadow coverage over time onto the three-dimensional scene, marks the stress over-limit warning coordinates of the load-bearing components, and generates a violation operation report with physical parameters.

2. The urban and rural spatial planning dynamic simulation teaching system according to claim 1, characterized in that: The method for generating a 3D solid model based on component-level geometric topology parameters employs a non-uniform rational B-spline surface reconstruction technique with back-calibrated control points, as detailed below: The topological connection type of beam-column nodes in the building information model is analyzed to generate an initial control point mesh. The initial control point mesh is iteratively optimized based on the measured wall thickness and the tolerance zone of the beam-column cross-section to generate control point coordinates. The semantic label level of the load-bearing components is extracted from the component-level geometric topology parameters. A three-dimensional solid model with mechanical transfer path is established based on the semantic label level of the load-bearing components and the control point coordinates.

3. The urban and rural spatial planning dynamic simulation teaching system according to claim 1, characterized in that: The physical properties of the component materials are input into the three-dimensional solid model using an inversion compensation mechanism based on material spectral characteristics. For components lacking physical properties, a near-infrared spectroscopy scanner is activated to obtain the molecular vibration spectrum of the component surface, and then input into a convolutional neural network. The convolutional neural network outputs the compensated density distribution matrix and elastic modulus gradient tensor by matching the spectral-mechanical property mapping relationship of the molecular vibration spectrum of the component surface in the material database.

4. The urban and rural spatial planning dynamic simulation teaching system according to claim 1, characterized in that: The calculation of the component's deformation under load using the finite element method employs an adaptive mesh refinement strategy based on stress concentration factors. The von Mises stress gradient change rate in the beam-column joint region is identified. For beam-column joint regions where the von Mises stress gradient change rate exceeds the threshold, an octree mesh is recursively subdivided to refine the local mesh. The subdivision level is positively correlated with the von Mises stress gradient change rate. An explicit dynamic algorithm is used to solve the transient deformation displacement field of the refined local mesh. The load deformation of the component is calculated based on the transient deformation displacement field.

5. The urban and rural spatial planning dynamic simulation teaching system according to claim 1, characterized in that: The computational fluid dynamics model is operated using a building surface boundary layer vortex stripping simulation technique. A body-fitted boundary layer mesh is generated on the outer surface of the building. Velocity slip boundary conditions are set, and the real-time airflow velocity value is decomposed into the mean flow field and turbulent fluctuation components. These components are then input into the large eddy simulation algorithm to capture the evolution path of the separated eddies. Based on the evolution path of the separated eddies, a ventilation efficiency cloud map of the eddy intensity distribution is output.

6. The urban and rural spatial planning dynamic simulation teaching system according to claim 1, characterized in that: The generated building shadow coverage rate over time curve was obtained using the photon beam scattering cumulative statistical method. A continuous motion model of the sun's position is constructed, and an anisotropic photon emission source in the model is driven to emit photon beams. The reflection and scattering paths of each photon beam on the building facade are tracked, and the photon flux density of the building facade blocked per unit time is calculated based on the reflection and scattering paths of each photon beam on the building facade. A time-series heat map of shadow coverage is generated based on the shadow coverage flux density, and the curve of the building shadow coverage over time is extracted from the time-series heat map of shadow coverage.

7. The urban and rural spatial planning dynamic simulation teaching system according to claim 1, characterized in that: The structural stability loss rate is calculated using the eigenvalue buckling modal displacement superposition method. The semantic parsing engine converts user operation commands into a set of computer-processable component deletion operations, generates a list of topological relationships of the components to be deleted, reconstructs the stiffness matrix of the remaining components, calculates the first three buckling mode eigenvectors of the remaining components based on the stiffness matrix of the remaining components, and quantifies the overall stiffness attenuation ratio of the components by weighted superposition of the first three buckling mode eigenvectors through modal displacement. Finally, the structural stability loss rate is calculated based on the overall stiffness attenuation ratio of the components.

8. The urban and rural spatial planning dynamic simulation teaching system according to claim 1, characterized in that: The trigger coordinates for early warning of excessive stress in load-bearing components are achieved through the following steps: The deformation of the component under load output by the finite element algorithm is converted into Mises stress values ​​and mapped to the surface mesh nodes of the three-dimensional solid model. Based on the component connection relationship in the building information model, a stress transmission path tree diagram is constructed. The stress transmission path tree diagram takes the high-stress component as the root node and traces the stress distribution of adjacent components level by level along the topological connection relationship of the building information model. It is used to locate the source of stress concentration propagation. The coordinates of grid nodes whose stress values ​​exceed the material yield strength threshold are marked in the stress transmission path tree diagram, and a pulse warning is triggered in the multimodal feedback processing unit (3).

9. The urban and rural spatial planning dynamic simulation teaching system according to claim 4, characterized in that: When calculating the deformation of a component under load using the finite element algorithm, the following steps are performed simultaneously: The transient deformation displacement field is decomposed into intrinsic orthogonal decomposition to separate the main deformation modes. The deformation amplitude data of the main deformation modes are converted into vertex animation sequences of the three-dimensional solid model and then superimposed onto the corresponding component surfaces in the VR device scene.

10. The urban and rural spatial planning dynamic simulation teaching system according to claim 5, characterized in that: The ventilation efficiency cloud map output by the computational fluid dynamics model is linked to the structural analysis in the following ways: The wind pressure distribution data on the building surface is extracted from the ventilation efficiency cloud map and used as an additional load input to the finite element algorithm to recalculate the deformation of the component under load. When the wind-induced vibration displacement deviates from the original design allowable value by more than the preset deviation threshold, the coordinates of the high-risk wind pressure zone and the stress over-limit warning are simultaneously marked in the multimodal feedback processing unit (3).

Images