CN-122021387-A - Urban design parameter optimization method and system based on high-resolution thermal environment rapid prediction

CN122021387ACN 122021387 ACN122021387 ACN 122021387ACN-122021387-A

Abstract

The invention discloses a city design parameter optimization method and system based on high-resolution thermal environment fast prediction, which comprises the steps of designing a plurality of simulation scenes aiming at the form of an infrastructure of a target area, carrying out CFD simulation, constructing an original high-dimensional dataset matrix, projecting the original high-dimensional dataset matrix into a low-dimensional space, obtaining a low-dimensional feature dataset matrix with reduced dimensions, using the low-dimensional feature dataset matrix for predictive training of a radial basis function neural network model, predicting a new design scene of the target area by a trained model, obtaining a predicted value of a new principal component score with low dimensions, restoring a complete high-resolution temperature field and wind speed field space distribution diagram under the new design scene by ascending dimension reconstruction, and finally constructing a comprehensive evaluation index system by combining average temperature and wind speed and construction and maintenance cost, thereby recommending an optimal city design parameter scheme. The method improves the efficiency and the precision of urban thermal environment prediction, and can optimize the urban design scheme according to the prediction result.

Inventors

CAO SHIJIE
XI CHANG
REN CHEN

Assignees

东南大学

Dates

Publication Date: 20260512
Application Date: 20251218

Claims (10)

1. The city design parameter optimization method based on the rapid prediction of the high-resolution thermal environment is characterized by comprising the following steps of: S100, different city design parameter items and variables thereof are proposed according to the form of the infrastructure in a target area, a plurality of simulation scenes are designed according to the city design parameter items and the variables thereof, and each simulation scene has scene parameters comprising respective corresponding city design parameter items and variable values thereof; S200, carrying out CFD simulation on all the simulation scenes in sequence to obtain CFD simulation data of each grid node in all the simulation scenes, and constructing an original high-dimensional dataset matrix according to the CFD simulation data to obtain a high-resolution thermal environment database; S300, performing principal component analysis on the high-resolution thermal environment database, and projecting the original high-dimensional dataset matrix into a low-dimensional space to obtain a low-dimensional feature dataset matrix with reduced dimensions; S400, training the radial basis function neural network model by utilizing the low-dimensional characteristic data set matrix, so that the model learns a nonlinear mapping relation from the urban design parameter vector to the principal component score vector until the model can accurately output a predicted value of the principal component score with low dimensionality according to the urban design parameter; S500, predicting a new design scene with the design parameter items in the target area kept unchanged and only variable changes by the trained radial basis function neural network model to obtain a predicted value of a low-dimensional new principal component score of the target area under the design parameters of newcastle disease; S600, performing dimension-lifting reconstruction on the predicted value of the low-dimension new principal component score vector by using a PCA reconstruction algorithm, and restoring a complete high-resolution temperature field and wind speed field spatial distribution diagram of a target area in a new design scene; s700, constructing a comprehensive evaluation index system based on the high-resolution temperature field and wind speed field spatial distribution diagram and combining the average temperature and wind speed and the construction and maintenance cost, and recommending an optimal urban design parameter scheme according to the comprehensive evaluation index system.
2. The city design parameter optimization method based on high-resolution thermal environment fast prediction according to claim 1, wherein in step S100, the design method of the simulation scene specifically comprises: Firstly, providing a plurality of optimized city design parameter items according to the infrastructure form in the target area, wherein the infrastructure form comprises buildings, roads, water areas and greening; then designing variables according to the proposed urban design parameter items, and respectively selecting a plurality of variable values for the variables of each urban design parameter item; Finally, m simulation scenes are formed according to a full factor design method, wherein the number m of the simulation scenes is equal to the product of the variable numbers of each city design parameter item, and m is a positive integer; the variable values of all the urban design parameter items in each of the simulated scenes together constitute scene parameters of the simulated scene.
3. The city design parameter optimizing method based on fast prediction of high resolution heat environment of claim 2, wherein the setting method of the variables of the city design parameter item specifically comprises: when building in the target area is selected as one of the city design parameter items, building height and/or building density are/is set as variables, and then a plurality of different building height values and/or different building density values are selected as variable values of the city design parameter item; When the road in the target area is selected as one of the city design parameter items, firstly setting the road width and/or the road area as variables, and then selecting a plurality of different road width values and/or different road density values as variable values of the city design parameter item; When the water area in the target area is selected as the urban design parameter item, firstly setting the water area and/or the water area density as variables, and then selecting a plurality of different water area values and/or different water area density values as the variable values of the urban design parameter item; When the greening in the target area is selected as the urban design parameter item, firstly, the greening position and/or the greening coverage rate are set as variables, and then a plurality of different greening position points and/or different greening coverage rate values are selected as the variable values of the urban design parameter item.
4. The city design parameter optimization method based on high-resolution thermal environment fast prediction according to claim 1, wherein in step S200, the method for constructing the high-resolution thermal environment database specifically comprises: Firstly, sequentially carrying out geometric modeling, grid division and thermal environment simulation on all m simulation scenes by using a computational fluid dynamics method, and obtaining CFD simulation data of all N grid nodes in each simulation scene by generating a plurality of high-resolution thermal environment scenes covering different scene parameters, wherein m and N are positive integers; Then, sorting all the grid node data, combining the grid node data belonging to the same simulation scene to form m data matrixes with 5 multiplied by N dimensions respectively, and recording the data matrixes as high-resolution thermal environment data matrixes, wherein each simulation scene corresponds to one high-resolution thermal environment data matrix respectively; and finally, combining all m high-resolution thermal environment data matrixes to finally form an original high-dimensional data set matrix with the dimension of m multiplied by 5 multiplied by N, namely the high-resolution thermal environment database.
5. The urban design parameter optimization method based on high-resolution thermal environment fast prediction according to claim 4, wherein each grid node data contains 5 characteristic parameters, in particular including spatial coordinate data (x, y, z) describing the geometrical position of the grid node at 1.5m above ground, and air temperature data T and wind speed data U describing the thermal environment characteristics of the grid node at 1.5m above ground, so the grid node data are denoted as [ x, y, z, T, U ].
6. The city design parameter optimization method based on high-resolution thermal environment rapid prediction according to claim 1, wherein in step S300, the specific steps of the principal component analysis are as follows: s301, preprocessing the original high-dimensional data set matrix by adopting a Z-score method, so as to obtain a standardized data matrix; s302, performing covariance calculation based on the obtained standardized data matrix to obtain a covariance matrix of the standardized data matrix; S303, carrying out feature decomposition based on the obtained covariance matrix to obtain a plurality of feature values and feature vectors, and then arranging the feature values in a descending order, and defining the corresponding feature vectors as main components; s304, firstly calculating the variance contribution rate of each principal component, then sequentially selecting the calculated accumulated interpretation variance contribution rates of the first k principal components each time by using the obtained variance contribution rate of each principal component, wherein k is a positive integer, and finally stopping calculation when the calculated accumulated interpretation variance contribution rate is more than or equal to 95%, and selecting k principal components corresponding to the current accumulated interpretation variance contribution rate; S305, constructing a projection matrix belonging to each simulation scene by using the first k selected principal components, and combining the projection matrices of each simulation scene to finally obtain the low-dimensional characteristic data set matrix of all the simulation scenes.
7. The method for optimizing urban design parameters based on rapid prediction of high-resolution thermal environment according to claim 6, wherein in step S400, the architecture of the radial basis function neural network model comprises: The input layer is responsible for receiving new variable values of set design parameter items during prediction so as to form a newcastle disease design parameter vector, wherein the new variable values are used as input of the radial basis function neural network model, and a plurality of sample data are selected from the standardized data matrix through a K-means clustering algorithm during training so as to form a city design parameter vector for training and serve as input of the radial basis function neural network model; The hidden layer comprises a plurality of nodes and is responsible for mapping the urban design parameter vector input by the input layer to a low-dimensional space through nonlinear transformation to realize data dimension reduction and obtain a predicted value of a principal component score vector; and the output layer is responsible for outputting the predicted value of the principal component score vector obtained by the hiding layer to form the principal component score vector which is used as the output of the radial basis function neural network model.
8. The city design parameter optimization method based on high-resolution thermal environment rapid prediction of claim 7, wherein in step S400, the training method of the radial basis function neural network model is as follows: Firstly, selecting new grid node data at K different positions in the target area and scene parameters thereof from the standardized data matrix by the input layer through a K-means clustering algorithm as new sample data, thereby forming city design parameter vectors for training as input; then training the radial basis function neural network model by utilizing the low-dimensional characteristic data set matrix to enable the hidden layer to learn a nonlinear mapping relation from the urban design parameter vector to the principal component score vector; Finally, the radial basis function neural network model can accurately output the predicted value of the principal component score with low dimensionality according to the city design parameters, and k principal component score vectors are obtained through the output layer.
9. The method for optimizing urban design parameters based on rapid prediction of high-resolution thermal environment according to claim 1, wherein in step S700, the recommended method for the optimal urban design parameter scheme is specifically as follows: Firstly, calculating the average temperature and average wind speed of a target area in a new design scene according to a complete high-resolution temperature field and wind speed field spatial distribution diagram of the target area in the new design scene obtained by reconstruction; then, combining the calculated average temperature and average wind speed of the target area under the new design scene with the construction and maintenance cost to construct a comprehensive evaluation index system; then, determining objective weights of all indexes in the comprehensive evaluation index system by adopting an entropy weight method; And finally, calculating the scores of the comprehensive evaluation index systems of all different design scenes, and recommending an optimal urban design parameter scheme according to the score conditions.
10. A system for implementing the urban design parameter optimization method based on high-resolution thermal environment rapid prediction according to any one of claims 1-9, characterized by comprising a high-resolution thermal environment database module, a high-dimension reduction and feature extraction module, a low-dimension feature prediction module, a high-resolution thermal environment reconstruction module and a multi-objective optimization decision module, The high-resolution thermal environment database module is responsible for providing different city design parameter items and variables thereof aiming at the forms of infrastructures in a target area, designing a plurality of simulation scenes according to the city design parameter items and the variables, sequentially performing CFD simulation on all the simulation scenes to obtain CFD simulation data of each grid node in all the simulation scenes, and constructing an original high-dimension data set matrix according to the CFD simulation data to obtain a high-resolution thermal environment database; The high-dimensional data dimension reduction and feature extraction module is responsible for carrying out principal component analysis on the high-resolution thermal environment database, projecting the original high-dimensional data set matrix into a low-dimensional space, and obtaining a dimension-reduced low-dimensional feature data set matrix; The low-dimensional feature prediction module is responsible for training a radial basis function neural network model by utilizing the low-dimensional feature data set matrix on one hand, so that the model learns a nonlinear mapping relation from a city design parameter vector to a principal component score vector until the model can accurately output a predicted value of a low-dimensional principal component score according to the city design parameter, and on the other hand, is responsible for predicting a new design scene of the target area by utilizing the trained radial basis function neural network model to obtain the predicted value of the low-dimensional new principal component score of the target area under the design parameter of the newcastle; The high-resolution thermal environment reconstruction module is responsible for carrying out ascending dimension reconstruction on the predicted value of the new principal component score vector with low dimension by utilizing a PCA reconstruction algorithm, and restoring a complete high-resolution temperature field and wind speed field spatial distribution diagram of a target area in a new design scene; The multi-objective optimization decision module is responsible for constructing a comprehensive evaluation index system based on the high-resolution temperature field and wind speed field spatial distribution diagram and combining the average temperature and wind speed and the construction and maintenance cost, and accordingly recommending an optimal urban design parameter scheme.

Description

Urban design parameter optimization method and system based on high-resolution thermal environment rapid prediction Technical Field The invention belongs to the technical field of smart city design and computer simulation, and particularly relates to a city design parameter optimization method based on high-resolution thermal environment rapid prediction. Background With the acceleration of the urban process, the urban heat island effect is increasingly prominent, and serious threat is formed to the health of residents and the energy consumption. The acquisition of high-resolution thermal environment space-time distribution is a precondition for precise regulation and control. At present, the acquisition of the high-resolution thermal environment distribution mainly depends on a computational fluid dynamics simulation method, but the method has inherent technical bottlenecks of high computational cost, huge time consumption and the like, and the rapid scheme comparison and optimization application of the method in the urban planning and design stage is severely limited. In recent years, artificial intelligence technology has rapidly developed, and a potential solution can be provided for the efficiency bottleneck problem of a computational fluid dynamics simulation method. In the prior art, although machine learning is tried to be used for urban thermal environment prediction, two technical defects are faced, namely 1) a high-resolution space field cannot be generated based on single-point monitoring data prediction, 2) a machine learning model for directly processing a high-dimensional space field is faced with a high-dimensional data disaster, so that training cost is high, the model is complex and low in efficiency, and global space precision of a prediction result is difficult to ensure. Therefore, there is an urgent need in the art for an innovative solution that can compromise prediction efficiency, precision and spatial integrity. Disclosure of Invention Based on the above, the invention aims to provide a city design parameter optimization method and system based on high-resolution thermal environment rapid prediction, which are used for improving the efficiency and the precision of city thermal environment prediction by integrating computational fluid mechanics and artificial intelligence technology and optimizing a city design scheme according to a prediction result. In order to solve the technical problems and achieve the technical effects, the invention is realized by the following technical scheme: A city design parameter optimization method based on high-resolution thermal environment rapid prediction comprises the following steps: S100, different city design parameter items and variables thereof are proposed according to the form of the infrastructure in a target area, a plurality of simulation scenes are designed according to the city design parameter items and the variables thereof, and each simulation scene has scene parameters comprising respective corresponding city design parameter items and variable values thereof; S200, carrying out CFD simulation on all the simulation scenes in sequence to obtain CFD simulation data of each grid node in all the simulation scenes, and constructing an original high-dimensional dataset matrix according to the CFD simulation data to obtain a high-resolution thermal environment database; S300, performing Principal Component Analysis (PCA) on the high-resolution thermal environment database, and projecting the original high-dimensional dataset matrix into a low-dimensional space to obtain a low-dimensional feature dataset matrix with reduced dimensions; S400, training the radial basis function neural network model by utilizing the low-dimensional characteristic data set matrix, so that the model learns a nonlinear mapping relation from the urban design parameter vector to the principal component score vector until the model can accurately output a predicted value of the principal component score with low dimensionality according to the urban design parameter; S500, predicting a new design scene with the design parameter items in the target area kept unchanged and only variable changes by the trained radial basis function neural network model to obtain a predicted value of a low-dimensional new principal component score of the target area under the design parameters of newcastle disease; S600, performing dimension-lifting reconstruction on the predicted value of the low-dimension new principal component score vector by using a PCA reconstruction algorithm, and restoring a complete high-resolution temperature field and wind speed field spatial distribution diagram of a target area in a new design scene; s700, constructing a comprehensive evaluation index system based on the high-resolution temperature field and wind speed field spatial distribution diagram and combining the average temperature and wind speed and the construction and maintenance cost, and reco