Urban layout optimization simulation method, system and device for alleviating urban heat island effect

Abstract

The present invention discloses an urban layout optimization simulation method, system and device for alleviating the urban heat island effect. The urban layout optimization simulation method includes: S100: dividing the target urban area into grids, adding urban surface temperature data and urban data to the corresponding grid points to obtain a grid-based urban model; S200: extracting the average urban surface temperature data and various influencing factor data of each grid point in the grid-based urban model, and respectively constructing a surface temperature prediction model for each grid point by using a geographically weighted regression model; S300: using the average urban surface temperature data and influencing factor data of each grid point as the state space, taking each influencing factor and action type as the action space, and performing reinforcement learning to simulate the optimization scheme of the urban layout. This application can automatically learn the optimal urban layout simulation scheme that can maintain urban functions, control optimization costs and alleviate the heat island effect, so as to provide scientific guidance for urban layout optimization.

Description

Technical Field

[0001] This application belongs to the technical field of urban layout optimization simulation, and particularly relates to an urban layout optimization simulation method, system and device for alleviating the urban heat island effect. Background Art

[0002] The process of urbanization is advancing at an unprecedented speed, changing the urban landscape and ecological environment. This rapid urbanization has led to the formation and increasing severity of the urban heat island effect. The urban heat island effect is a phenomenon in which heat accumulates within the urban spatial range due to factors such as urban buildings and human activities. Currently, most methods for alleviating the urban heat island effect include increasing urban greening, improving urban infrastructure, enhancing urban ventilation, and controlling artificial heat sources. However, it is impossible to quantify the impact of different planning measures on the urban thermal environment, and it is difficult to judge the advantages and disadvantages of various planning schemes. Summary of the Invention

[0003] The purpose of this application is to provide an urban layout optimization simulation method, system and device for alleviating the urban heat island effect. This application can quantify the impact of different urban planning schemes on the heat island effect and automatically learn the optimal layout scheme.

[0004] On the one hand, this application provides an urban layout optimization simulation method for alleviating the urban heat island effect, including:

[0005] S100: Conduct grid division on the target urban area, and add the urban surface temperature data and urban data of the target city to the corresponding grid points to obtain a grid-based urban model; the urban data refers to the urban data related to the causes of the urban heat island effect;

[0006] S200: Extract the average urban surface temperature data and various influencing factor data of each grid point in the grid-based urban model, and use the geographically weighted regression model to construct the surface temperature prediction model of each grid point. The influencing factors are preset and are quantifiable influencing factors that affect the urban heat island effect and are obtained from the urban data;

[0007] S300: Use the average urban surface temperature data and influencing factor data of each grid point as the state space, use each influencing factor and action type as the action space, and adopt ε -greedy strategy and perform reinforcement learning based on the constructed reward function to simulate the optimization scheme of the urban layout;

[0008] The reward function is , where:

[0009] , , are weights, + + = 1; , 、 respectively represent the surface temperature matrices of grid points before and after the execution of the action, which are the predicted values using the surface temperature prediction model; represents the sum of all elements in the matrix;

[0010] is the urban function change value of the urban functional area p , which is the sum of the urban function change values of the grid points included in the urban functional area p ; is the sum of the urban function change values of P urban functional areas of the target city; the calculation of the urban function change value of a grid point is as follows: First, calculate the change values of each influencing factor related to the urban functional area p at the grid point after the execution of the action respectively; then, multiply the change value of each influencing factor by the regression coefficient of each influencing factor at the grid point, and the regression coefficient is obtained from the surface temperature prediction model of the grid point; finally, sum up the products corresponding to all influencing factors, which is the urban function change value of the grid point;

[0011] is the sum of the values of the cost of changing the land use layout of the grid point, and the value of the cost of changing the land use layout of the grid point is: the product of the sum of the change values of each influencing factor at the grid point after the execution of the action, the weight of the grid point, and the current housing price; the weight of the grid point is the comprehensive weight plus 1, and the comprehensive weight is pre-allocated and is positively correlated with the centralization degree and economic level of the grid point location.

[0012] In some specific embodiments, step S100 is specifically: using the fishnet tool in ArcGIS to divide the target urban area into grids, and then using the extract values to points tool in ArcGIS to add the urban surface temperature data and urban data of the target city to the corresponding grid points.

[0013] In some specific embodiments, the urban data includes urban building and building physical property data, urban ecology and landscape data, urban infrastructure and transportation data, and urban social and economic data.

[0014] In some specific embodiments, the surface temperature prediction model is ; where, represents the position of the grid point i ; represents the urban surface temperature of the grid point i ; represents the intercept of the grid point i ; represents the k ​​The regression coefficient of an influencing factor at a grid point i ; denotes the k th influencing factor at the grid point i value, m denotes the number of influencing factors; denotes the error term of the grid point i .

[0015] In some specific embodiments, the influencing factors include:

[0016] ① One or more of the floor area, average building height, building heat capacity, and thermal conductivity of the grid points extracted from urban building and building physical property data;

[0017] ② One or more of the urban landscape index and the floor area of each type of urban blue-green space of the grid points extracted from urban ecological and landscape data;

[0018] ③ The shortest distance from the grid point extracted from urban infrastructure and traffic data to the urban infrastructure;

[0019] ④ One or more of the population quantity, GDP, and electricity consumption of the grid points extracted from urban social and economic data.

[0020] In some specific embodiments, the action types include three types of action operations: increasing the influencing factor value of the grid point by a preset amplitude, decreasing the influencing factor value of the grid point by a preset amplitude, and remaining unchanged.

[0021] In some specific embodiments, the comprehensive weight of the grid points is allocated by the following method:

[0022] The target city is divided into several regions with different centrality degrees according to the geographical location, and centrality weights are respectively allocated to each region; the centrality weight is positively correlated with the centrality degree of the region where it is located;

[0023] The target city is divided into several regions with different economic levels according to the economic situation, and economic level weights are respectively allocated to each region; the economic level weight is positively correlated with the economic level of the region;

[0024] The product of the centrality weight and the economic level weight of the region where the grid point is located is the comprehensive weight of the grid point.

[0025] In some specific embodiments, step S300 further includes:

[0026] S310: The agent observes the current state space;

[0027] S320: The agent uses the ε -greedy strategy, based on the current probability value εSelect an action in the action space a ;

[0028] S330: The agent executes the action a ;

[0029] S340: The agent interacts with the state space, updates the state space, and feeds back the reward function value;

[0030] S350: The agent observes the updated state space and updates the Q value using the reward function value:

[0031] S360: The agent determines whether the stop condition is met. If it is met, the iteration ends, and the final state space is the optimized urban layout simulation scheme; otherwise, continue to execute steps S310 - S350.

[0032] On the other hand, the present application also provides an urban layout optimization simulation system for alleviating the urban heat island effect, including:

[0033] The first module is used to divide the target urban area into grids, add the urban surface temperature data and urban data of the target city to the corresponding grid points to obtain a grid-based urban model; the urban data refers to the urban data related to the causes of the urban heat island effect;

[0034] The second module is used to extract the average urban surface temperature data and the data of each influencing factor of each grid point in the grid-based urban model, and respectively construct a surface temperature prediction model for each grid point using a geographically weighted regression model. The influencing factors are preset and are quantifiable influencing factors affecting the urban heat island effect, obtained from the urban data;

[0035] The third module is used to use the average urban surface temperature data and the data of the influencing factors of each grid point as the state space, use the influencing factors and the action types as the action space, and adopt ε -greedy strategy and perform reinforcement learning based on the constructed reward function to simulate the optimization scheme of the urban layout;

[0036] The reward function is , where:

[0037] , , are weights, + + = 1; , , respectively represent the surface temperature matrices of the grid points before and after executing the action, the predicted value using the surface temperature prediction model; Denotes the sum of all elements in the matrix;

[0038] Is the urban functional area p Of the urban function change value, which is the sum of the urban function change values of the grid points included in the urban functional area p ; Is the target city P The sum of the urban function change values of the urban functional areas; The calculation of the urban function change value of the grid point is as follows: First, calculate the change values of each influencing factor related to the urban functional area p After the execution of the action at the grid point respectively; Then, multiply the change value of each influencing factor by the regression coefficient of each influencing factor at the grid point, and the regression coefficient is obtained from the surface temperature prediction model of the grid point; Finally, sum the products corresponding to all influencing factors, that is, the urban function change value of the grid point;

[0039] Is the sum of the land use layout change cost values of the grid points, and the land use layout change cost value of the grid point is: the product of the sum of the change values of each influencing factor of the grid point after the execution of the action, the weight of the grid point, and the current housing price; The weight of the grid point is the comprehensive weight plus 1, and the comprehensive weight is pre-allocated and is positively correlated with the centralization degree and economic level of the grid point location.

[0040] On the other hand, the present application also provides an urban layout optimization simulation device for alleviating the urban heat island effect, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above method can be implemented.

[0041] Compared with the prior art, the present application has the following advantages and beneficial effects:

[0042] The present application combines the geographically weighted regression model with reinforcement learning, uses the geographically weighted regression model to construct a surface temperature prediction model, so as to obtain the influence coefficient of each influencing factor on the heat island effect at the grid point (i.e., the regression coefficient of the surface temperature prediction model), and introduces the influence coefficient into the reward function of reinforcement learning, so as to quantify the influence of the influencing factor on the heat island effect. Combining the reward function, using reinforcement learning to quantitatively simulate the influence of different urban planning schemes on the heat island effect, so as to automatically learn the optimal urban layout simulation scheme that can alleviate the heat island effect while satisfying the maintenance of urban functions and controlling the optimization cost. The obtained optimal urban layout simulation scheme can provide technical support and scientific guidance for urban layout optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 Is the flowchart of the urban layout optimization simulation method in the embodiment of the present application;

[0044] Figure 2 In the embodiments of the present application ε - Flowchart of reinforcement learning using the greedy strategy. Detailed implementation manners

[0045] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be clearly and completely described below with reference to the accompanying drawings. Apparently, the described embodiments are some but not all of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without making creative efforts shall fall within the protection scope of the present application.

[0046] Please refer to Figure 1 , which shows a schematic flowchart of a method for optimizing the urban layout to mitigate the urban heat island effect provided by the embodiments of the present application, including the following specific steps:

[0047] S100: Divide the target urban area into grids, and add the urban surface temperature data and urban data of the target city to the corresponding grid points to obtain a grid-based urban model; wherein, the urban data refers to the urban data related to the causes of the urban heat island effect;

[0048] Specifically, use the fishnet tool in ArcGIS to divide the target urban area into evenly distributed grid points of equal size, for example, divide them into grid points of 50m×50m; then use the Extract Values to Points tool in ArcGIS to add the urban surface temperature data and urban data of the target city to the corresponding grid points.

[0049] In this embodiment, the urban surface temperature data is obtained by inverting the remote sensing image of the target city, and the specific process is as follows:

[0050] Use the ENVI software tool to obtain the near-infrared band and infrared band data from the remote sensing image, and crop and merge the target urban area; in this embodiment, the remote sensing image is the Landsat8 OLI_TIRS remote sensing image;

[0051] Perform radiometric calibration and atmospheric correction on the remote sensing image of the target city, and then use the single-window algorithm to invert the remote sensing image to obtain the urban surface temperature.

[0052] Furthermore, using the single-window algorithm to invert the remote sensing image includes:

[0053] Calculate the normalized difference vegetation index (NDVI) for the remote sensing image, estimate the land cover type based on the NDVI, and calculate the land surface emissivity in combination with the NDVI;

[0054] According to the land surface emissivity, the radiative transfer equation and Planck's law are used to invert the urban land surface temperature.

[0055] In this application, the urban data can be selected as long as it is related to the causes of the urban heat island effect, without specific limitations. In this embodiment, the urban data includes urban building and building physical property data, urban ecology and landscape data, urban infrastructure and transportation data, and urban social and economic data. Further, the urban building and building physical property data includes, but is not limited to, building layout, building area and height, heat capacity and thermal conductivity of the building; the urban ecology and landscape data includes, but is not limited to, urban landscape index, urban blue-green space area; the urban infrastructure and transportation data includes, but is not limited to, urban road network, shortest distance from grid points to infrastructure such as parks and bus stops; the urban social and economic data includes, but is not limited to, population quantity, GDP, electricity consumption. It should be noted that the urban blue-green space is a general term for urban green spaces and urban water bodies, including, but not limited to, water areas, green spaces, parks.

[0056] The above urban data can be obtained from multi-source data such as remote sensing images, GIS data, urban planning documents, field survey sampling data, and meteorological satellite data of the target city. In this application, the geographically weighted regression model of grid points is constructed using urban land surface temperature data and urban data. Therefore, the urban land surface temperature data and urban data added to the grid points should be data of the same period.

[0057] S200: Extract the average urban land surface temperature data and various influencing factor data of each grid point in the gridded urban model, and use the geographically weighted regression model to construct the surface temperature prediction model of each grid point, where the influencing factors are quantifiable influencing factors that affect the urban heat island effect;

[0058] When constructing the surface temperature prediction model, the urban land surface temperature is used as the dependent variable, and the influencing factors are used as the independent variables. The constructed surface temperature prediction model is expressed as follows:

[0059] (1)

[0060] In formula (1): represents the position of grid point i ; represents the urban land surface temperature of grid point i ; represents the intercept of grid point i ; represents the regression coefficient of the k th influencing factor at grid point i ; represents the k th influencing factor at grid point i value,m Represents the number of influencing factors; Represents the grid point i of the error term.

[0061] In this embodiment, the regional statistical method is used to calculate and extract the average urban surface temperature data of each grid point.

[0062] In this application, the influencing factors of the grid point are preset, and some quantifiable influencing factors affecting the urban heat island effect are selected for the quantitative performance of the urban data of the grid point.

[0063] In this embodiment, the influencing factors include:

[0064] ① One or more of the building area, average building height, heat capacity and thermal conductivity of the grid point extracted from the data of urban buildings and building physical characteristics;

[0065] ② One or more of the urban landscape index of the grid point extracted from the data of urban ecology and landscape, and the floor area of each type of urban blue-green space;

[0066] ③ The shortest distance from the grid point to urban infrastructure extracted from the data of urban infrastructure and transportation, such as the shortest distance from the grid point to a park, a bus stop, etc.;

[0067] ④ One or more of the population quantity, GDP, and electricity consumption of the grid point extracted from the data of urban social economy, and the population quantity, GDP, and electricity consumption of the grid point can be obtained by the downscaling method.

[0068] In this embodiment, the values of the influencing factors of the grid point are extracted using ENVI and ArcGIS software tools.

[0069] The building layout, such as the floor area and height of the building, will change the air circulation in the city, affect the heat dissipation in the city, and thus affect the surface temperature. The heat capacity and thermal conductivity of the building will change the reflectivity and absorptivity of sunlight, affect the heating rate of the surface, and thus affect the surface temperature. Vegetation and water bodies can alleviate the urban heat island effect. Therefore, the floor area of urban blue-green spaces such as parks, water areas, and green spaces, as well as the urban landscape index, are also influencing factors of the urban heat island effect. The heat release during factory production and transportation processes, as well as the heat emissions from electrical equipment such as air conditioners and appliances in buildings, can exacerbate the urban heat island effect, and the heat emissions are positively correlated with the electricity consumption, so the electricity consumption is also an influencing factor of the urban heat island effect.

[0070] Please refer to Table 1 below, which shows the average urban surface temperature data of some grid points and some influencing factor data in this embodiment. Among them, uid represents the unique identifier of the grid point, which is used to locate and distinguish grid points; some influencing factors include water area, building area, vegetation index, number of people, and GDP.

[0071] Table 1 Average urban surface temperature data of some grid points and some influencing factor data in the embodiment

[0072]

[0073] S300: Using the average urban surface temperature data and influencing factor data of each grid point as the state space, and using each influencing factor and action type as the action space, establish a reward function with the goals of reducing surface temperature, maintaining urban functions, and controlling optimization costs, and adopt ε -greedy strategy for reinforcement learning to simulate the optimization scheme of the urban layout.

[0074] Reinforcement learning belongs to machine learning, which mainly learns the optimal behavior by the interaction between the agent and the environment according to the reward value feedback by the environment. ε -greedy strategy is a strategy used to balance exploration and exploitation in reinforcement learning; in the exploration stage, the agent randomly selects actions for exploration with a probability ε , and all possible actions have the same probability of being selected; in the exploitation stage, the agent selects actions with a probability of 1 - ε , and selects the currently considered optimal action based on the current Q value, and the Q value contains the reward value and is used to reflect future rewards.

[0075] The action types include three types of action operations: increase, decrease, and remain unchanged. Specifically, they are operations to increase, decrease, and remain unchanged the values of the influencing factors of the grid points. The increase and decrease amplitudes are preset, and in this embodiment, the amplitude is set to 5% of the original value. It should be noted that when performing actions, it is necessary to ensure that the influencing factor values have practical significance after the actions are executed.

[0076] The reward function constructed in this application is expressed as follows:

[0077] (2)

[0078] Among them, , , respectively represent the weights of the three goals of reducing surface temperature, maintaining urban functions, and controlling optimization costs, + + = 1; , , It is preset according to the degree of emphasis on the three goals of reducing the surface temperature, maintaining urban functions, and controlling and optimizing costs.

[0079] In this embodiment, it is highly desired to alleviate the urban heat island effect. Therefore, the weights of the three goals of reducing the surface temperature, maintaining urban functions, and controlling and optimizing costs are set to 0.5, 0.3, and 0.2 respectively.

[0080] In the reward function, is used to describe the change in the surface temperature of the target city after performing an action, where represents the sum of all elements in the matrix ; is the surface temperature change matrix of grid points, which is used to reflect whether the surface temperature of each grid point decreases after performing an action; the rows and columns of the surface temperature change matrix correspond to longitude and dimension , where each element is the surface temperature change value of the grid point corresponding to each position after performing the action; specifically, i , , is the surface temperature matrix of grid points before performing the action, is the surface temperature matrix of grid points after performing the action; the surface temperature of each grid point after performing the action is the predicted value obtained by the surface temperature prediction model; when the surface temperature decreases after performing the action, a positive value is fed back.

[0081] In the reward function, is used to describe the change in the urban functions of the target city after performing an action, represents the number of urban functional areas in the target city, p represents the p th urban functional area; an urban functional area refers to the area where various functional activities in the city gather spatially, including but not limited to commercial areas, residential areas, industrial areas, public service facilities, and green areas; represents the p th number of influencing factors related to this type of urban functional area in the q th urban functional area, q represents the

[0082] The influencing factors related to various urban functional areas come from mIndependent variable influencing factors. For commercial areas, the related influencing factors include building area, average building height, heat capacity and thermal conductivity of buildings, shortest distance from grid points to bus stops, electricity consumption, etc.; for residential areas, the related influencing factors include building area, average building height, heat capacity and thermal conductivity of buildings, shortest distance from grid points to bus stops, shortest distance from grid points to parks, population quantity, electricity consumption, etc.; for industrial areas, there are multiple ones among building area, shortest distance from grid points to urban infrastructure, GDP, electricity consumption, etc.; for public service facilities, there are multiple ones among building area, average building height, urban landscape index, floor areas of various urban blue-green spaces respectively, shortest distance from grid points to urban infrastructure, population quantity, etc.; for green areas, the related influencing factors include urban landscape index, floor areas of various urban blue-green spaces respectively, etc.

[0083] represents the urban functional area p of the urban function change value, then is the sum of the urban function change values of all urban functional areas in the target city, P is the number of urban functional areas in the target city. The urban functional area p of the urban function change value is for the urban functional area p which is the sum of the urban function change values of the grid points included.

[0084] The calculation of the urban function change value of a grid point is as follows: First, calculate the change values of each influencing factor related to the urban functional area p at the grid point after the execution of the action. Here, the change value refers to the value of the influencing factor at the grid point after the execution of the action minus that before the execution of the action; then, multiply the change value of each influencing factor by the regression coefficient of each influencing factor at the grid point respectively. The regression coefficient is obtained from the surface temperature prediction model of the grid point, and the regression coefficient is the parameter in formula (1) which represents the regression coefficient of the influencing factor k at the grid point i ; finally, sum up the products corresponding to all influencing factors, which is the urban function change value of the grid point.

[0085] For this embodiment, the number of influencing factors related to the urban functional area p is Q , then first calculate the change values of Q influencing factors at the grid point after the execution of the action respectively, and then multiply the change values of Q influencing factors by the regression coefficients of the corresponding influencing factors at the grid point respectively to obtain the products corresponding to each influencing factor; finally, sum up Q the products.

[0086] In the reward function, represents the cost value of changing the land use layout caused by the execution of the action, which is the sum of the cost values of changing the land use layout of grid points. The calculation of the cost value of changing the land use layout of grid points is: the product of the total change value of the influence factors of the grid point after the execution of the action, the weight of the grid point, and the current housing price. Among them, the total change value of the influence factors refers to the sum of the change values of each influence factor of the grid point. Specifically, each change value of the influence factor is the absolute value of the change amount of the influence factor after the execution of the action. The weight of the grid point is the comprehensive weight plus 1. The comprehensive weight is used to reflect the centralization degree and economic level of the positions of each grid point. The comprehensive weight is positively correlated with the centralization degree and economic level of the grid point position. When the centralization degree is higher and the economic level is higher, the grid point is assigned a high comprehensive weight; on the contrary, the grid point is assigned a low comprehensive weight.

[0087] Next, a method for allocating the comprehensive weight of grid points in this embodiment will be provided. Obviously, the method for allocating the comprehensive weight is not limited to the following method. The method for allocating the comprehensive weight of grid points in this embodiment is as follows:

[0088] (1) Divide the target city into several regions with different centralization degrees according to the geographical location, and allocate centralization degree weights to each region respectively; the centralization degree weight is positively correlated with the centralization degree of the region where it is located, that is: the higher the centralization degree (i.e., the closer to the city center) of the region, the higher the allocated centralization degree weight;

[0089] In this embodiment, the target city is divided into three regions with different centralization degrees: the city center area, the sub - center area, and the edge area according to the geographical location. The centralization degree weights allocated to the city center area, the sub - center area, and the edge area are 3:2:1, and the sum of the centralization degree weights allocated to the city center area, the sub - center area, and the edge area is 1. In specific implementation, the division can be referred to the existing urban ring roads of the target city.

[0090] (2) Divide the target city into several regions with different economic levels according to the economic situation, and allocate economic level weights to each region respectively; the economic level weight is positively correlated with the economic level of the region, that is: the higher the economic level of the region, the higher the allocated economic level weight;

[0091] In this embodiment, the target city is divided into a high - economic region, a medium - economic region, and a low - economic region according to the economic situation. The economic level weights allocated to the high - economic region, the medium - economic region, and the low - economic region are 3:2:1, and the sum of the economic level weights allocated to the high - economic region, the medium - economic region, and the low - economic region is 1.

[0092] Therefore, in this embodiment, 9 types of divided regions can be obtained: high - economic regions in the city center, medium - economic regions in the city center, low - economic regions in the city center, high - economic regions in the sub - center, medium - economic regions in the sub - center, low - economic regions in the sub - center, high - economic regions in the fringe area, medium - economic regions in the fringe area, and low - economic regions in the fringe area. The comprehensive weight of these 9 types of divided regions is the product of the corresponding centrality weight and economic level weight. For example, for the high - economic region in the city center, its comprehensive weight is 1 / 4.

[0093] (3) The product of the centrality weight and economic level weight of the region where the grid point is located is the comprehensive weight of the grid point.

[0094] Figure 2 As shown ε - The process of reinforcement learning using the greedy strategy includes:

[0095] S310: The agent observes the current state space S ;

[0096] S320: The agent uses ε - The greedy strategy to select an action A in the action space a , and the action is selected based on the current probability value ε either by selection based on current knowledge or randomly. Selection based on current knowledge means selecting Q the action with the maximum

[0097] S330: Execute the selected action a , and the agent interacts with the state space S . During this process, the state space S changes according to the action execution;

[0098] S340: Update the state space S and return the reward function value R ( s , a ). This reward function value is used to reflect the effect of the currently executed action; the reward function is established with the goals of reducing surface temperature, maintaining urban functions, and controlling optimization costs. A positive reward is given when the target temperature is reached, and a negative reward is given when deviating from the target;

[0099] S350: The agent observes the updated state space , and uses the returned reward function value R ( s , a ) to update Q value:

[0100] (3)

[0101] In formula (3), the initial value of is set to 0, is the learning rate, is the discount factor, and both have a value range of (0, 1). The discount factor is used to weigh the importance of the current reward and future rewards. A higher value means the agent attaches more importance to future rewards. On the contrary, a lower value pays more attention to the current reward; represents the maximum value among all possible actions under the updated state space Q which reflects the best future reward that the agent can obtain under the updated state space ;

[0102] S360: The agent determines whether the stop condition is met. If it is met, the iteration ends, and the final state space is the optimized urban layout simulation scheme; otherwise, steps S310 - S350 are continued. The stop condition is preset and includes but is not limited to (1) the urban surface temperature reaches the target value; (2) the preset maximum number of iterations is reached; (3) convergence; (4) the reward function values feedback in M successive iterations remain unchanged or decrease, M and the value is preset.

[0103] This application uses the reinforcement learning method to change the state space of the target city by performing actions, uses the reward function value to judge whether the change is beneficial to reducing the surface temperature, maintaining urban functions, and controlling the optimization cost, and continuously learns and optimizes using the greedy strategy, so as to quickly simulate an urban layout scheme that can effectively reduce the urban surface temperature. The simulated urban layout scheme can be used as a reference for urban layout optimization.

[0104] Note that the above is only the preferred embodiment of this application and the technical principles applied. Those skilled in the art will understand that this application is not limited to the specific embodiments described here. Various obvious changes, re - adjustments, and substitutions can be made by those skilled in the art without departing from the protection scope of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments. Without departing from the concept of this application, more other equivalent embodiments can be included, all of which belong to the protection scope of this application.

Claims

1. An urban layout optimization simulation method for alleviating the urban heat island effect, characterized in that Including: S100: Conduct grid division on the target urban area, add the urban surface temperature data and urban data of the target city to the corresponding grid points to obtain a grid-based urban model; the urban data refers to the urban data related to the causes of the urban heat island effect. S200: Extract the average urban surface temperature data and various influencing factor data of each grid point in the grid-based urban model, and use the geographically weighted regression model to construct the surface temperature prediction model for each grid point. The influencing factors are pre-set as quantifiable influencing factors that affect the urban heat island effect and are obtained from the urban data. S300: Using the average urban surface temperature data and influencing factor data of each grid point as the state space, and each influencing factor and action type as the action space, adopt the ε -greedy strategy and perform reinforcement learning based on the constructed reward function to simulate the optimization scheme of the urban layout; The reward function is , where: , , are weights, + + = 1; , , respectively represent the surface temperature matrices of the grid points before and after the execution of the action, adopts the predicted value of the surface temperature prediction model; represents the sum of all elements in the matrix; is the urban function change value of the urban functional area p , which is the sum of the urban function change values of the grid points included in the urban functional area p ; is the sum of the urban function change values of P urban functional areas of the target city; The calculation of the urban function change value of a grid point is as follows: First, calculate the change values of each influencing factor related to the urban functional area p at the grid point after the execution of the action respectively; Then, multiply the change value of each influencing factor by the regression coefficient of each influencing factor at the grid point, and the regression coefficient is obtained from the surface temperature prediction model of the grid point; Finally, sum up the products corresponding to all influencing factors, which is the urban function change value of the grid point; It is the sum of the changed land use layout cost values of grid points. The changed land use layout cost value of a grid point is the product of the sum of the change values of each influencing factor of the grid point after the action is executed, the weight of the grid point, and the current housing price. The weight of the grid point is the comprehensive weight plus 1, and the comprehensive weight is pre-allocated and is positively correlated with the centralization degree and economic level of the grid point location.

2. The method for optimizing and simulating urban layout to mitigate the urban heat island effect according to claim 1, wherein: Step S100 is specifically: Use the Fishnet tool in ArcGIS to conduct grid division on the target urban area, and then use the Extract Values to Points tool in ArcGIS to add the urban surface temperature data and urban data of the target city to the corresponding grid points.

3. The method for optimizing and simulating urban layout to mitigate the urban heat island effect according to claim 1, wherein: The urban data includes urban building and building physical property data, urban ecology and landscape data, urban infrastructure and transportation data, and urban social and economic data.

4. The urban layout optimization simulation method for alleviating the urban heat island effect according to claim 1, characterized in that: The surface temperature prediction model is ; where represents the position of grid point i . represents the urban surface temperature of grid point i . represents the intercept of grid point i . represents the regression coefficient of the k th influencing factor at grid point i . represents the value of the k th influencing factor at grid point i , m represents the number of influencing factors; represents the error term of grid point i .

5. The method for optimizing and simulating urban layout to mitigate the urban heat island effect according to claim 1, wherein: The influencing factors include: ① One or more of the building area, average building height, heat capacity, and thermal conductivity of the grid points extracted from the urban building and building physical property data; ② One or more of the urban landscape index and the floor area of each type of urban blue-green space of the grid points extracted from the urban ecology and landscape data; ③ The shortest distance from the grid points extracted from the urban infrastructure and transportation data to the urban infrastructure; ④ One or more of the population quantity, GDP, and electricity consumption of the grid points extracted from the urban social and economic data.

6. The method for optimizing and simulating urban layout to mitigate the urban heat island effect according to claim 1, wherein: The action types include three types of action operations: increasing the influencing factor value of the grid point by a preset amplitude, decreasing the influencing factor value by a preset amplitude, and remaining unchanged.

7. The method for optimizing and simulating urban layout to mitigate the urban heat island effect according to claim 1, wherein: The comprehensive weight of the grid points is allocated by the following method: Divide the target city into several regions with different degrees of centrality according to the geographical location, and allocate the centrality weights to each region respectively; the centrality weight is positively correlated with the degree of centrality of the region's location; Divide the target city into several regions with different economic levels according to the economic situation, and allocate the economic level weights to each region respectively; the economic level weight is positively correlated with the economic level of the region; The product of the centrality weight and the economic level weight of the region where the grid point is located is the comprehensive weight of the grid point.

8. The method for optimizing and simulating urban layout to mitigate the urban heat island effect according to claim 1, wherein: Step S300 further includes: S310: The agent observes the current state space; S320: The agent utilizes ε - the greedy policy to select an action in the action space based on the current probability value ε ; a ​ S330: The agent performs an action a ; S340: The agent interacts with the state space, updates the state space, and feeds back the reward function value; S350: The agent observes the updated state space and updates using the reward function value Q Value: S360: The agent determines whether the stop condition is met. If it is met, the iteration ends, and the final state space is the optimized urban layout simulation scheme; otherwise, steps S310 - S350 are continued.

9. An urban layout optimization simulation system for alleviating the urban heat island effect, characterized in that, including: The first module is used to perform grid division on the target urban area, add the urban surface temperature data and urban data of the target city to the corresponding grid points to obtain a grid-based urban model; the urban data refers to urban data related to the causes of the urban heat island effect; The second module is used to extract the average urban surface temperature data and the data of each influencing factor of each grid point in the grid-based urban model, and respectively construct a surface temperature prediction model for each grid point using a geographically weighted regression model. The influencing factors are preset as quantifiable influencing factors that affect the urban heat island effect and are obtained from the urban data; The third module uses the average urban surface temperature data and influencing factor data of each grid point as the state space, takes each influencing factor and action type as the action space, and adopts the ε -greedy strategy and conducts reinforcement learning based on the constructed reward function to simulate the optimization scheme of the urban layout; The reward function is , where: , , is the weight, + + = 1; , , respectively represent the surface temperature matrices of the grid points before and after the execution of the action, adopt the predicted values of the surface temperature prediction model; represents the sum of all elements in the matrix; is the urban function change value of the urban functional area p which is the sum of the urban function change values of the grid points included in the urban functional area p ; is the sum of the urban function change values of P urban functional areas of the target city; the calculation of the urban function change value of a grid point is as follows: First, calculate the change values of the respective impact factors related to the urban functional area p after the execution of the action at the grid point; then, multiply the change value of each impact factor by the regression coefficient of each impact factor at the grid point, where the regression coefficient is obtained from the surface temperature prediction model of the grid point; finally, sum up the products corresponding to all the impact factors, which is the urban function change value of the grid point; It is the sum of the costs of changing the land use layout for grid points. The cost value of changing the land use layout for a grid point is the product of the sum of the change values of each influencing factor of the grid point after performing the action, the weight of the grid point, and the current housing price. The weight of the grid point is the comprehensive weight plus 1, and the comprehensive weight is pre-allocated and positively correlated with the centralization degree and economic level of the grid point location.

10. An urban layout optimization simulation device for alleviating the urban heat island effect, characterized in that: It includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the method described in any one of claims 1 - 8 can be implemented.

Images