Automatic optimization method and system for spatial form of urban high-temperature risk area

Abstract

The invention discloses an automatic optimization method and system for the spatial form of an urban high-temperature risk area, and the method comprises the steps: firstly building an urban scale basic space sand table according to vector building form data with a street as a spatial analysis unit; acquiring temperature and humidity data in each spatial analysis unit, and calculating a thermal risk index of the urban layout; then, the urban high-temperature risk area is partitioned according to the thermal risk index, and importance indexes of spatial form elements are calculated through a clustering algorithm; and finally, establishing a multi-agent system by taking each monomer in the city as an agent, exciting the agents to adopt actions capable of reducing the high-temperature risk through interaction iteration of the multi-agents until optimization conditions are met, and outputting a spatial form optimization result of the city high-temperature risk area. The method can meet the requirements of large-scale high-temperature risk identification and form automatic optimization in a range of more than 100km < 2 >, and improves the accuracy and scientificity of subsequent form element adjustment.

Description

Technical Field

[0001] The present invention relates to the technical field of urban planning, and in particular to an automatic optimization method and system for the spatial form of a high temperature risk area in an urban area. Background Art

[0002] High temperatures in cities have significant impacts on people's physical and mental health and quality of life, such as causing a series of potential risks such as heat stroke and respiratory diseases; for the city itself, it will also lead to an intensified heat island effect and affect the sustainable development of the urban ecological environment. Spatial form is closely related to urban high temperatures. Factors such as the greening rate and building layout of urban neighborhoods will directly affect the urban thermal environment. By optimizing spatial form, the urban heat island effect can be effectively alleviated, the adaptability of the city can be improved, and thus livability and sustainable development can be improved. Therefore, in research and practice, the optimization of spatial form in urban high temperature risk areas is considered to be a regulatory measure that can effectively cope with the heat island effect in urban planning and construction.

[0003] However, in the traditional methods of dealing with urban high temperature risk spaces, on the one hand, professionals often pay too much attention to the early stage of high temperature identification, and judge the current situation through actual measurement or simulation, but pay insufficient attention to the optimization and adjustment of spatial form in the later stage, and fail to effectively capture the mechanism and intrinsic relationship between urban spatial form and high temperature risk. On the other hand, professionals often propose universal climate adaptation strategies for spatial form based on subjective experience, lacking scientific data support and systematic analysis methods, and it is difficult to propose practical optimization plans for large-scale high temperature risk areas in a short period of time. In this context, it is urgent to study the automatic optimization method of spatial form in high temperature risk areas to provide more scientific, systematic and operational urban renewal plans to meet the challenges of urban high temperature risks. Summary of the invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a method and system for automatic optimization of the spatial morphology of urban high-temperature risk areas, so as to realize automatic adjustment of the spatial layout of green spaces, industrial buildings, residential buildings and commercial buildings in urban high-temperature risk areas, and promote high-temperature risk investigation, hidden danger elimination and sustainable and healthy development of urban space.

[0005] Technical solution: The automatic optimization method of the spatial morphology of urban high temperature risk areas described in the present invention comprises the following steps:

[0006] Based on the vector building form data, the neighborhood unit formed by the urban road network is used as the spatial analysis unit to build a basic space sand table at the urban scale;

[0007] Acquiring temperature and humidity data in each of the spatial analysis units and calculating a thermal risk index for the urban layout;

[0008] The urban high temperature risk areas are divided according to the heat risk index and displayed in the urban scale basic space sand table;

[0009] According to the spatial morphological elements and thermal risk index of each of the spatial analysis units, an importance index of the spatial morphological elements is calculated by a clustering algorithm;

[0010] Each single entity in the city is taken as an agent to establish a multi-agent system. The reward function is fitted according to the importance index of the spatial morphological elements to evaluate the change of the heat risk index. Through the interactive iteration of multiple agents, the agents are encouraged to take actions that can reduce the high temperature risk until the optimization conditions are met. The spatial morphological optimization results of the high temperature risk area in the city are output. The actions of the agents include removal, addition, displacement and scaling, and the state of the agents is the spatial morphological elements.

[0011] The spatial morphology optimization results of the urban high temperature risk area are displayed in the urban scale basic space sand table.

[0012] Further, partitioning the urban high temperature risk area according to the heat risk index includes: calculating the average value of the heat risk index of each spatial analysis unit and standard deviation σ, The spatial analysis unit is the general risk area. The spatial analysis unit is the key monitoring area. The spatial analysis unit is the area that needs to be optimized; i is the thermal risk index of the ith spatial analysis unit.

[0013] Furthermore, the heat risk index of each spatial analysis unit is the sum of the heat risk indexes of all land use type plots in the spatial analysis unit. The heat risk index H of the nth land use type plot is n for:

[0014]

[0015] Among them, τ is the temperature coefficient, γ is the humidity coefficient, t n is the duration of the highest temperature of the nth land use type, T nmax is the maximum surface temperature, T′ nmax is the maximum air temperature, RH is t n The average air humidity during the time period, V is the wind speed of the nth land use type plot.

[0016] Furthermore, through the interactive iteration of multiple agents, the agents are encouraged to take actions that can reduce the risk of high temperature until the optimization conditions are met, and the optimization results of the spatial morphology of the urban high temperature risk area are output, including:

[0017] During the interactive iteration process, it is judged whether the loss function of the multi-agent system converges to the optimal solution. When it converges to the optimal solution, the urban high temperature risk area is re-divided according to the heat risk index. When the number of areas to be optimized and key monitoring areas are both zero, the optimization conditions are met and the spatial morphological optimization results of the urban high temperature risk area are output.

[0018] Furthermore, the reward function is:

[0019]

[0020] Among them, R(s, a) represents the total reward obtained by taking action a in state s, γ is the discount factor, and r t represents the immediate reward obtained at time step t.

[0021] Furthermore, according to the spatial morphological elements and thermal risk index of each of the spatial analysis units, the importance index of the spatial morphological elements is calculated by a clustering algorithm, including: the spatial morphological elements are used as nodes of a decision tree, and the average impurity value of each node is calculated as the importance index of the node.

[0022] Furthermore, the vector building form data includes a city road network SHP file, an existing closed building and number of floors or height SHP file, and a green space outline or area SHP file.

[0023] Furthermore, the spatial morphological elements include building height, building density, building orientation, building form, building function mixture, building materials and building color indicators at the building level; street aspect ratio, street direction, road area ratio and line-contacting rate indicators at the street level; green space rate, water body area, building vegetation coverage rate and hardness rate indicators at the green space and water body level, as well as the sky visibility coefficient.

[0024] The automatic optimization system of the spatial form of the urban high temperature risk area described in the present invention comprises:

[0025] The city-scale basic space sand table establishment unit is used to build a city-scale basic space sand table based on vector building form data and the neighborhood unit formed by the urban road network as the spatial analysis unit;

[0026] A heat risk index calculation unit, used to obtain temperature and humidity data in each of the spatial analysis units and calculate a heat risk index for the urban layout;

[0027] An urban high temperature risk zone zoning unit is used to zone the urban high temperature risk zone according to the heat risk index and display it in the urban scale basic space sand table;

[0028] A spatial morphological element importance calculation unit, used to calculate the importance index of the spatial morphological element through a clustering algorithm according to the spatial morphological element and the thermal risk index of each of the spatial analysis units;

[0029] The urban high temperature risk area spatial morphology optimization unit is used to establish a multi-agent system by treating each monomer in the city as an agent, fitting a reward function according to the importance index of the spatial morphological elements to evaluate the change of the heat risk index, and motivating the agents to take actions that can reduce the high temperature risk through the interactive iteration of multiple agents until the optimization conditions are met, and outputting the optimization results of the urban high temperature risk area spatial morphology. The actions of the agents include removal, addition, displacement and scaling, and the state of the agents is the spatial morphological elements;

[0030] The spatial morphology optimization result display unit is used to display the spatial morphology optimization result of the urban high temperature risk area in the basic spatial sand table of the urban scale.

[0031] The electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, an automatic optimization method for the spatial morphology of the urban high temperature risk area is implemented.

[0032] Beneficial effects: Compared with the prior art, the advantages of the present invention are that the measurement and classification of high temperature risk areas are realized based on the urban space sand table, the correlation between the urban space morphological element indicators and the heat risk index under the spatial analysis unit is established, and the automatic adjustment and holographic display of the spatial morphological scheme that meets the requirements of urban high temperature risk control are realized through the training and interactive iteration of multiple intelligent agents. Specifically, the advantages include the following:

[0033] 1. This invention establishes a high temperature index measurement system and realizes the measurement and classification of high temperature risk areas based on the urban space sand table. In addition to the optimization of small and medium-scale spatial morphology, it also expands the judgment results of the zoning status of urban high temperature risk areas to the urban scale, which can cope with 100km 2 Large-scale high temperature risk identification and automatic morphological optimization within the above range are required.

[0034] 2. The present invention constructs a spatial morphological prediction model of the thermal risk index and establishes the correlation between the urban spatial morphological element indicators and the thermal risk index under the spatial analysis unit, breaking through the deficiency of only expert experience judgment in the traditional urban planning and design layout, and improving the accuracy and scientificity of subsequent morphological element adjustments.

[0035] 3. The present invention realizes automatic adjustment and holographic display of spatial morphology solutions that meet the requirements of urban high temperature risk management through multi-agent training and interactive iteration of four types of green spaces, industrial buildings, residential buildings and commercial buildings, shortening the work cycle that originally required more than one month to less than one day, and improving the standardization and standardized output of the solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 is a flow chart of the automatic optimization method of the present invention;

[0037] Figure 2 This is a diagram showing the high temperature risk zone classification results according to an embodiment of the present invention;

[0038] Figure 3 A schematic diagram of a multi-agent system according to an embodiment of the present invention;

[0039] Figure 4 This is a diagram showing the spatial morphology optimization results of an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0041] like Figure 1 As shown, the automatic optimization method for the spatial morphology of urban high temperature risk areas includes the following steps.

[0042] S01. Build a basic urban-scale space sand table. Based on Landsat8 remote sensing images and site observations, obtain high-precision land use remote sensing monitoring maps to obtain land use type data. Based on the GIS geographic information platform and the mapping drone equipped with the Beidou satellite navigation system, obtain vector building form data to clarify the spatial distribution of existing green spaces, industrial buildings, residential buildings and commercial buildings in the urban space. Take the neighborhood unit formed by the urban road network as the spatial analysis unit, label each spatial analysis unit, and match it with the land use data and vector building form data in coordinates and elevations. The data format is unified and input into the geographic information system platform to generate a basic urban-scale space sand table.

[0043] The obtained vector building form data includes a road network SHP file, an existing closed building and number of floors (height) SHP file, and a green space outline (area) SHP file.

[0044] The city-scale basic spatial sand table uses a projection operation to convert data coordinates and adjust the city's three-dimensional vector data to the 2000 National Geodetic Coordinate System; uses a Layer3DToFeatureClass operation in combination with building height information to stretch the building to form a three-dimensional model, and matches the spatial analysis unit attribute information with the land use data and vector building form data in coordinates and elevations to ultimately generate a city-scale basic spatial sand table.

[0045] S02. Measurement and classification of high temperature risk areas. Set up and establish a high temperature index measurement system within the spatial analysis unit, use temperature and humidity sensors to obtain spatial temperature data and air humidity data in the measurement area; combine the infrared channel inversion of the satellite-borne sensor to obtain surface temperature data and form a high temperature index database. Call the corresponding formula to calculate the heat risk index of the existing layout and construct a heat risk index data set.

[0046] The heat risk index of urban spatial areas is standardized and imported into the basic spatial sand table at the urban scale. The average and standard deviation of the heat risk index are calculated, and three numerical ranges are constructed. The heat risk index is divided into three levels based on the standard deviation: low risk (index value < average), medium risk (average value ≤ index value ≤ average value + standard deviation) and high risk (index value > average value + standard deviation). According to this standard, the urban high temperature disaster risk areas are divided into general risk areas, key monitoring areas, and areas to be optimized. The zoning status of urban high temperature risk areas is displayed based on the spatial sand table.

[0047] The high temperature index measurement system includes measurement point layout, measurement height setting, time resolution setting, spatial resolution setting, wind environment measurement setting and high temperature index calibration. The measurement point layout is to evenly set temperature and humidity sensors and drones in the measurement area with a scale of 50M×50M as the minimum measurement unit to meet the full coverage of the target area; the measurement height is set as the temperature and humidity measurement height and the wind environment measurement height, and the temperature and humidity sensors are set at 2m as the temperature and humidity measurement height, and the drone flight route is arranged at 5m as the wind environment measurement height. The time resolution is set to 30 minutes as the time resolution; the wind environment measurement facility is to use a drone equipped with an intelligent wind measuring instrument to measure the wind speed and wind direction values ​​in the target area with a height of 5m as the reference. The surface temperature data is obtained by combining the infrared channel inversion of the satellite-borne sensor.

[0048] The heat risk index uses the spatial analysis unit as the statistical spatial scale of the heat risk index, and the single land use type plot as the calculation spatial scale of the heat risk index. The heat risk index of the spatial analysis unit is the sum of the heat risk index of the single land use type plot. The heat risk index calculation integrates air temperature, surface temperature, high temperature duration, air humidity and wind speed factors, constructs a calculation formula that is positively correlated with the heat risk index, and constructs a heat risk index data set. The formula used is:

[0049]

[0050] Where G is the thermal risk index of the i-th spatial analysis unit, H n is the heat risk index of the nth land use type plot within the ith spatial analysis unit;

[0051]

[0052] Among them, t n is the duration of the highest temperature of the nth land use type, T nmax is the maximum surface temperature, T′ nmax is the maximum air temperature, RH is t n The average air humidity during the time period, V is the wind speed value of the nth land use type plot.

[0053] τ is the temperature coefficient (when the maximum temperature T<30℃, τ is 0; when the maximum temperature 30℃≤T≤35℃T, τ is 1; when the maximum temperature T≥35℃, τ is 2); γ is the humidity coefficient (when the maximum humidity and the maximum temperature overlap for more than 70%, γ is 2; when the maximum humidity and the maximum temperature overlap for more than 40%, γ is 1; when the maximum humidity and the maximum temperature overlap for 0%, γ is 0).

[0054] The standard deviation classification is used to extract the heat risk index data x i , where the data with a heat risk index of zero are not included in the calculation of the mean and standard deviation, and the heat risk index mean is calculated n is the number of spatial analysis units, i is the label of the spatial analysis unit, and the standard deviation of the heat risk index is calculated. The heat risk index was classified into low risk, Medium risk and high risk There are three levels. Corresponding to this standard, urban high temperature disaster risk areas are divided into general risk areas, key monitoring areas, and areas that need urgent optimization.

[0055] The urban high temperature risk zone zoning status imports the thermal risk index value of the spatial analysis unit into the geographic information platform for spatial analysis, and sets different grayscale colors to express different degrees of urban high temperature risk conditions.

[0056] S03. Construction of the spatial morphological prediction model of the heat risk index. The spatial morphological element indicator set of different spatial analysis units and the heat risk index calculation results are input into the heat risk index spatial morphological element prediction relationship library. The extracted feature data are cleaned and standardized, and divided into 70% training set and 30% test set data samples. The random forest algorithm is used to train the target city data set, and the root mean square error (RMSE) and mean absolute error (MAE) are used to evaluate the prediction performance of the model. Based on the average reduction impurity, the impurity reduction used to divide the nodes in all decision trees is calculated, and its average value is further calculated to measure the importance of spatial morphological element characteristics.

[0057] The spatial morphological element index set includes the building level, street level, green space and water level, and sky viewing angle coefficient. The building level includes building height, building density, building orientation, building form, building function mixture, building materials, and building color indicators; the street level includes street aspect ratio, street direction, road area ratio, and line-contacting rate indicators; the green space and water level includes green space rate, water area, building vegetation coverage rate, and hardness rate indicators; the sky viewing angle coefficient is calculated using fisheye photography method with the help of image processing technology. The spatial morphological element index set is constructed by integrating the indicators of the four levels. Some of the indicators are obtained through urban planning indicators.

[0058] The average reduction of impurity refers to the feature data extracted from the prediction relationship library of the spatial morphological elements of the heat risk index, splitting them into nodes in the construction of the decision tree, and calculating the probability of randomly selecting two samples belonging to different categories in the node, that is, the impurity. The higher the probability, the lower the purity of the node, and the lower the importance ranking for urban high temperature. At the same time, the impurity of the node in other decision trees is calculated, and the average value is taken to obtain the importance of the feature data corresponding to the node for urban high temperature, and the importance ranking of each feature data can be obtained based on this.

[0059] S04. Definition of multi-agent system for automatic optimization of spatial morphology. In the DGX computer workstation, a multi-agent system is defined. Each unit of green space, industrial building, residential building and commercial building (including green space or building unit) is regarded as an agent. Multi-agent deep reinforcement learning (RIAL) is used to solve the spatial morphology optimization problem of high temperature risk areas in different neighborhoods of the target city.

[0060] Specifically, it includes defining the state space of the environmental system of the four types of intelligent agents at different times as the spatial morphological element index parameters of the neighborhood space unit where they are located, defining the action space of the four types of intelligent agents as the "removal", "increase", "displacement" and "scaling" of the monomers under the requirements of the urban design management regulations of the target city, using a neural network to output the action probability distribution of each intelligent agent, fitting a reward function according to the contribution value of the influencing factor of the spatial morphological prediction model of the thermal risk index of the target city in step S3 to evaluate the change of the thermal risk index, training the neural network of the intelligent agent, setting the loss function of the neural network, and updating the weight and bias of the neural network in the opposite direction of the gradient by calculating the gradient of the loss function.

[0061] In the action space of the intelligent agent, "moving out" refers to removing the green space or building unit from the current position; "adding" refers to copying and generating new green space and building units at the current position, and the newly added green space and building units inherit the neural network parameters of the original intelligent agent; "displacement" refers to moving the plane position coordinates of the green space or building unit; "scaling" refers to changing the size of the green space or building unit, and after scaling, it cannot overlap with other intelligent agent spaces.

[0062] The reward function is fitted based on the contribution value of the influencing factors of the spatial morphology prediction model of the heat risk index of different spatial analysis units in the target city. The greater the contribution value of the spatial morphology factor index to the negative feedback of the heat risk index, the greater the characteristic weight of the set reward function is, so as to encourage the intelligent agent to take actions that can reduce the risk of high temperature. The specific formula is:

[0063]

[0064] Among them, R(s,a) represents the total reward obtained by taking action a in state s, γ is the discount factor, and r t Represents the immediate reward obtained at time step t. By calculating the cumulative reward, the overall benefit obtained by the agent after taking a specific action can be evaluated.

[0065] S05, multi-agent training and interactive iteration. During the interactive iteration process, determine whether the loss function converges to the optimal solution. When the convergence condition is met, check the number of areas to be optimized and key monitoring areas after each step according to the numerical interval standard determined by the average value and standard deviation of the current thermal risk index in step (3). When the number of the above two types of risk areas is 0 at the same time, stop iterating and output the spatial optimization plan for urban high temperature disaster risk. Otherwise, re-execute the multi-agent interactive iteration process until the number of risk areas is met.

[0066] The convergence condition refers to the loss function becoming stable after several consecutive iterations and no longer changing significantly. When the loss function converges to below 0.1 and the reward function converges to the optimal solution, it is considered that the optimal solution or a state close to the optimal solution has been reached, and the iteration is stopped for the next step.

[0067] S06. Holographic display and output of spatial optimization plans. The spatial optimization plan for the city's high-temperature risk areas is input into the urban space holographic sandbox. Users wear MR mixed reality glasses to compare and observe the current plan and the spatial optimization plan. Planning, design and management personnel can use handheld control handles to make detailed adjustments and verifications. Finally, the thermal risk indexes and high-temperature risk areas of different spatial analysis units in the city before and after optimization are printed out in a table for output.

[0068] The technical solution of the present invention will be described in detail below by taking Nanjing Jiangbei New District as an example.

[0069] (1) Taking Nanjing Jiangbei New District as the target urban area, we obtained land use data and urban vector building form data, and used neighborhood units as spatial analysis units for labeling, elevation and coordinate matching, and generated the basic spatial sand table of Nanjing Jiangbei New District, including:

[0070] (1.1) Through Landsat8 remote sensing images and site observations, the land use remote sensing monitoring map of Nanjing Jiangbei New District was obtained and translated into land use type data.

[0071] (1.2) Through the GIS geographic information platform and the mapping drone with a resolution of 1920*1080 or above equipped with the Beidou satellite navigation system, the vector building form data and road data of Nanjing Jiangbei New District are obtained, including the road network SHP file, the current closed building and number of floors (height) SHP file, and the green space outline (area) SHP file, and the existing green space, industrial buildings, residential buildings, commercial buildings and roads in Nanjing Jiangbei New District are clearly identified.

[0072] (1.3) Import the existing road network SHP file of Nanjing Jiangbei New District, the existing closed building and number of floors (height) SHP file, and the green space outline (area) SHP file into the geographic information system software, spatially associate the building closed surface with the building number of floors, and attach the number of floors (height) information to each building; and use the neighborhood units divided by the road network as spatial analysis units, perform attribute labeling in the geographic information system software, and divide them into 46 small-scale spatial units.

[0073] (1.4) Use the projection operation to transform the data coordinates and adjust the urban three-dimensional vector data to the 2000 National Geodetic Coordinate System; use the Layer3DToFeatureClass operation, combined with the building height information, to stretch the buildings to form a three-dimensional model; and coordinate and match the spatial analysis unit attribute information with the land use data and vector building form data to finally generate the three-dimensional base model of Nanjing Jiangbei New District.

[0074] (2) Select each spatial analysis unit within Nanjing Jiangbei New Area, set up a high temperature index measurement system within it, obtain high temperature index and calculate heat risk index, and realize the measurement and classification of high temperature risk areas in Nanjing Jiangbei New Area, including:

[0075] (2.1) A high temperature index measurement system is set up inside each spatial analysis unit. The temperature and humidity sensors and drones are evenly set up with a time resolution of 30 minutes and a minimum measurement unit of 50M×50M. The temperature and humidity sensors are set up with a temperature and humidity measurement height of 2m, and the drone flight route is arranged with a wind environment measurement height of 5m. A drone equipped with an intelligent wind measuring instrument is used to measure the wind speed and wind direction values ​​in the target area with a height of 5m as the benchmark to meet the full coverage of Nanjing Jiangbei New District; thereby obtaining the spatial temperature data and air humidity data inside the spatial analysis unit,

[0076] (2.2) Using the thermal infrared sensor of the Landsat satellite, obtain the infrared channel data of the satellite-borne sensor, and perform atmospheric correction and inversion algorithm on it to obtain accurate surface temperature. Combined with the air temperature and humidity in step (1), a high temperature index database for Nanjing Jiangbei New District is constructed.

[0077] (2.3) The heat risk index of the spatial analysis unit calculated from the high temperature index database in step (2.2) was used to construct a heat risk index set; the heat risk index was divided into three levels: low risk, medium risk, and high risk according to the standard deviation classification, and corresponded to the general risk area, key monitoring area, and area to be optimized in the space of Nanjing Jiangbei New District, respectively. The index was imported into the basic space sandbox of Nanjing Jiangbei New District for hot spot analysis, and the degree of urban high temperature risk was expressed by different depths of red, forming a status map of high temperature risk areas in Nanjing Jiangbei New District. The general risk area H n (2.8≤H n ≤6.5), key monitoring area H n (6.5≤H n ≤10.5), urgently need to optimize area H n (H n ≥10.5). Figure 2 Shown is a diagram showing the high temperature risk zone classification results of this embodiment.

[0078] (3) Construct a spatial morphological prediction model of the heat risk index in Nanjing Jiangbei New Area, including:

[0079] (3.1) Obtain the spatial morphological elements of different spatial analysis units in Nanjing Jiangbei New Area, including the building level, street level, green space and water level, and sky viewing angle coefficient. Obtain building height, building density, building orientation, building morphological index, green space ratio, water area, building vegetation coverage rate, and hardness ratio index through remote sensing images; obtain building material and building color index through street view photo collection and analysis technology; obtain street height-to-width ratio, street direction, road area ratio, and line-fitting rate index through vector building data and road network data; use fisheye photography method with the help of image processing technology to calculate and obtain sky viewing angle coefficient. The spatial morphological element index set of each spatial analysis unit in Nanjing Jiangbei New Area is constructed by integrating the indicators of the four levels.

[0080] (3.2) The spatial morphological element indicator sets of different spatial analysis units and the heat risk index calculation results are input into the heat risk index spatial morphological element prediction relationship library, and the extracted feature data are cleaned and standardized, and divided into 70% training set and 30% test set data samples.

[0081] (3.3) The training set in step 2 is trained on the target city dataset through the random forest algorithm, and the root mean square error (RMSE) and mean absolute error (MAE) are used to evaluate the prediction performance of the spatial morphological prediction model of the heat risk index of Nanjing Jiangbei New District.

[0082] (3.4) The feature data extracted from the prediction relationship library of the spatial morphological elements of the heat risk index are used to determine the importance of the feature data for urban high temperatures. Based on this, the importance ranking of each feature data can be obtained to measure the importance of the spatial morphological elements.

[0083] (4) Figure 3 As shown in the figure, a multi-agent system for automatic optimization of the spatial form of Nanjing Jiangbei New District is constructed. In the DGX computer workstation, a multi-agent system is defined, and each monomer of the green space, industrial building, residential building and commercial building in Nanjing Jiangbei New District is defined as each agent. Multi-agent deep reinforcement learning (RIAL) is used to solve the spatial form optimization problem of the urban high temperature risk area in different spatial analysis units of Nanjing Jiangbei New District, including:

[0084] (4.1) Define the state space of the environmental system of four types of intelligent entities, namely green space, industrial building, commercial building and residential building, at different times as the spatial morphological element index parameters of the spatial analysis unit;

[0085] (4.2) Defined that under the requirements of the urban design management regulations of Nanjing Jiangbei New District, the action spaces of the four types of agents are "removal", "increase", "displacement" and "scaling" of green space or building units. The neural network is used to output the action probability distribution of each agent.

[0086] (4.3) The reward function is fitted based on the contribution value of the influencing factors of the spatial morphological prediction model of the heat risk index of different spatial analysis units in Nanjing Jiangbei New Area. The greater the contribution value of the spatial morphological element index to the negative feedback of the heat risk index, the greater the characteristic weight of the reward function is set, so as to encourage the intelligent agent to take actions that can reduce the risk of high temperature.

[0087] (4.4) According to the contribution value of the influencing factors of the spatial morphological prediction model of the urban heat risk index of Nanjing Jiangbei New Area in step (3), the reward function is fitted to evaluate the change of the heat risk index, the neural network of the intelligent agent is trained, and the loss function of the neural network is set. By calculating the gradient of the loss function, the weights and bias of the neural network are updated in the opposite direction of the gradient, so as to realize the training and learning of the multi-agent system of Nanjing Jiangbei New Area.

[0088] (5) Train and interactively iterate the multi-agent system of Nanjing Jiangbei New District. When the convergence conditions are met and the number of areas to be optimized and key detection areas is zero, output the spatial optimization plan for the high temperature disaster risk of Nanjing Jiangbei New District, including:

[0089] (5.1) During the interactive iteration process, determine whether the loss function converges to the optimal solution. When the loss function converges to below 0.1 and the reward function converges to the optimal solution, it is determined that the optimal solution or a state close to the optimal solution is reached, and the convergence condition is met.

[0090] (5.2) Combined with the numerical range standard determined by the average value and standard deviation of the current thermal risk index in step (3), the number of areas to be optimized and key monitoring areas after each step is checked. When the number of the above two types of risk areas is 0 at the same time, stop the iteration and output the spatial optimization plan for the high temperature disaster risk of Jiangbei New District, Nanjing. Otherwise, re-execute the multi-agent interactive iteration process until the number of risk areas is met.

[0091] (6) Conduct a holographic display of the Nanjing Jiangbei New District spatial optimization plan and output of results, including:

[0092] like Figure 4 As shown, the spatial optimization plan for the urban high temperature risk areas of Nanjing Jiangbei New District is input into the spatial holographic sandbox of Nanjing Jiangbei New District. Users wear MR mixed reality glasses to compare and observe the current plan and the spatial optimization plan, and the planning and design and management personnel use handheld control handles to make detailed adjustments and verifications. Finally, the thermal risk indexes and high temperature risk areas of different spatial analysis units in the city before and after optimization are printed out in a table for output.

[0093] The automatic optimization system of the spatial form of the urban high temperature risk area described in the present invention comprises:

[0094] The city-scale basic space sand table establishment unit is used to build a city-scale basic space sand table based on vector building form data and the neighborhood unit formed by the urban road network as the spatial analysis unit;

[0095] A heat risk index calculation unit, used to obtain temperature and humidity data in each of the spatial analysis units and calculate a heat risk index for the urban layout;

[0096] An urban high temperature risk zone zoning unit is used to zone the urban high temperature risk zone according to the heat risk index and display it in the urban scale basic space sand table;

[0097] A spatial morphological element importance calculation unit, used to calculate the importance index of the spatial morphological element through a clustering algorithm according to the spatial morphological element and the thermal risk index of each of the spatial analysis units;

[0098] The urban high temperature risk area spatial morphology optimization unit is used to establish a multi-agent system by treating each monomer in the city as an agent, fitting a reward function according to the importance index of the spatial morphological elements to evaluate the change of the heat risk index, and motivating the agents to take actions that can reduce the high temperature risk through the interactive iteration of multiple agents until the optimization conditions are met, and outputting the optimization results of the urban high temperature risk area spatial morphology. The actions of the agents include removal, addition, displacement and scaling;

[0099] The spatial morphology optimization result display unit is used to display the spatial morphology optimization result of the urban high temperature risk area in the basic spatial sand table of the urban scale.

[0100] The electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, an automatic optimization method for the spatial morphology of the urban high temperature risk area is implemented.

[0101] The computer-readable storage media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, flash memory, or any other medium that can be used to store the desired program code in the form of instructions or data structures and that can be accessed by a computer.

[0102] The processor is used to execute the computer program stored in the memory to implement each step of the method involved in the above embodiment.

Claims

1. A method for automatically optimizing the spatial morphology of urban high temperature risk areas, characterized in that: The steps include: Based on the vector building form data, the neighborhood unit formed by the urban road network is used as the spatial analysis unit to build a basic space sand table at the urban scale; Acquiring temperature and humidity data in each of the spatial analysis units and calculating a thermal risk index for the urban layout; The urban high temperature risk areas are divided according to the heat risk index and displayed in the urban scale basic space sand table; According to the spatial morphological elements and thermal risk index of each of the spatial analysis units, an importance index of the spatial morphological elements is calculated by a clustering algorithm; Each single entity in the city is taken as an agent to establish a multi-agent system. The reward function is fitted according to the importance index of the spatial morphological elements to evaluate the change of the heat risk index. Through the interactive iteration of multiple agents, the agents are encouraged to take actions that can reduce the high temperature risk until the optimization conditions are met. The spatial morphological optimization results of the high temperature risk area in the city are output. The actions of the agents include removal, addition, displacement and scaling, and the state of the agents is the spatial morphological elements. The spatial morphology optimization results of the urban high temperature risk area are displayed in the urban scale basic space sand table.

2. The automatic optimization method of urban high temperature risk area spatial morphology according to claim 1 is characterized in that: The zoning of urban high temperature risk areas according to the heat risk index includes: Calculate the average value of the heat risk index for each spatial analysis unit and standard deviation σ, The spatial analysis unit is the general risk area. The spatial analysis unit is the key monitoring area. The spatial analysis unit is the area that needs to be optimized; i is the thermal risk index of the ith spatial analysis unit.

3. The automatic optimization method of urban high temperature risk area spatial morphology according to claim 1 is characterized in that: The heat risk index of each spatial analysis unit is the sum of the heat risk indexes of all land use type plots in the spatial analysis unit. The heat risk index of the nth land use type plot is H n for: Among them, τ is the temperature coefficient, γ is the humidity coefficient, t n is the duration of the highest temperature of the nth land use type, T nmax is the maximum surface temperature, T″ nmax is the maximum air temperature, RH is t n The average air humidity during the time period, V is the wind speed of the nth land use type plot.

4. The automatic optimization method of urban high temperature risk area spatial morphology according to claim 2 is characterized in that: Through the interactive iteration of multiple agents, agents are encouraged to take actions that can reduce high temperature risks until the optimization conditions are met. The optimization results of the spatial morphology of the urban high temperature risk area are output, including: During the interactive iteration process, it is judged whether the loss function of the multi-agent system converges to the optimal solution. When it converges to the optimal solution, the urban high temperature risk area is re-divided according to the heat risk index. When the number of areas to be optimized and key monitoring areas are both zero, the optimization conditions are met and the spatial morphological optimization results of the urban high temperature risk area are output.

5. The automatic optimization method of urban high temperature risk area spatial morphology according to claim 1 is characterized in that: The reward function is: Among them, R(s,a) represents the total reward obtained by taking action a in state s, γ is the discount factor, and r t represents the immediate reward obtained at time step t.

6. The automatic optimization method of urban high temperature risk area spatial morphology according to claim 1 is characterized in that: According to the spatial morphological elements and thermal risk index of each of the spatial analysis units, the importance index of the spatial morphological elements is calculated by clustering algorithm, including: The spatial morphological elements are used as nodes of the decision tree, and the average impurity of each node is calculated as the importance indicator of the node.

7. The automatic optimization method of urban high temperature risk area spatial morphology according to claim 1 is characterized in that: The vector building form data includes the city's road network SHP file, the current closed building and the number of floors or height SHP file, and the green space outline or area SHP file.

8. The automatic optimization method of urban high temperature risk area spatial morphology according to claim 1 is characterized in that: The spatial morphological elements include building height, building density, building orientation, building form, building function mixture, building materials and building color indicators at the building level; street aspect ratio, street direction, road area ratio and line-contacting rate indicators at the street level; green space rate, water body area, building vegetation coverage rate and hardness rate indicators at the green space and water body level, as well as the sky visibility coefficient.

9. An automatic optimization system for the spatial morphology of urban high temperature risk areas, characterized in that: include: The city-scale basic space sand table establishment unit is used to build a city-scale basic space sand table based on vector building form data and the neighborhood unit formed by the urban road network as the spatial analysis unit; A heat risk index calculation unit, used to obtain temperature and humidity data in each of the spatial analysis units and calculate a heat risk index for the urban layout; An urban high temperature risk zone zoning unit is used to zone the urban high temperature risk zone according to the heat risk index and display it in the urban scale basic space sand table; A spatial morphological element importance calculation unit, used to calculate the importance index of the spatial morphological element through a clustering algorithm according to the spatial morphological element and the thermal risk index of each of the spatial analysis units; The urban high temperature risk area spatial morphology optimization unit is used to establish a multi-agent system by treating each monomer in the city as an agent, fitting a reward function according to the importance index of the spatial morphological elements to evaluate the change of the heat risk index, and motivating the agents to take actions that can reduce the high temperature risk through the interactive iteration of multiple agents until the optimization conditions are met, and outputting the optimization results of the urban high temperature risk area spatial morphology. The actions of the agents include removal, addition, displacement and scaling, and the state of the agents is the spatial morphological elements; The spatial morphology optimization result display unit is used to display the spatial morphology optimization result of the urban high temperature risk area in the basic spatial sand table of the urban scale.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is loaded into the processor, the method for automatically optimizing the spatial morphology of urban high temperature risk areas according to any one of claims 1 to 8 is implemented.

Images