Heat island cooling method and system based on hot knowledge guided multi-type ant colony algorithm

Abstract

The application discloses a heat island cooling method and system based on a heat knowledge guided multi-type ant colony algorithm, and the heat island cooling method based on the heat knowledge guided multi-type ant colony algorithm is used for cooling the urban heat island, so that the spatial position of an optimized pixel is obtained, and an optimized land use layout scheme is output. The application provides strong technical support for urban sustainable heat environment planning and relieving of the urban heat island effect.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of geographic information science, and particularly relates to a heat island cooling method and system based on hot knowledge guided multi-type ant colony algorithm. BACKGROUND

[0002] Under the background of rapid urbanization, the conversion speed of natural green land in urban areas to impervious surface has greatly accelerated, leading to a series of ecological and environmental problems related to human activities in major cities around the world. Among them, the heat island effect is the most widespread climate change caused by urbanization. The heat island effect is characterized by higher ground surface temperature in urban areas than in rural areas, causing significant negative impacts, including extreme heat injury events, reduced biodiversity, increased energy consumption, and endangering the health of urban residents and the sustainable development of cities. In 2022, extreme high-temperature weather occurred frequently in the summer in many countries, with record-breaking maximum temperatures, and heads of state began to actively seek solutions to curb urban warming. Among them, urban green land can produce significant cooling effects, and optimizing the spatial layout of green land can effectively curb high temperatures and alleviate the heat island effect.

[0003] As an important means of urban heat island cooling, developing a reasonable and effective green space optimization plan needs to solve two problems: ① accurately quantifying the range and intensity of the urban heat island; ② designing a model to identify the location of impervious surface conversion to green space at the city level. First, in quantifying the range and intensity of the urban heat island, the key is to determine the background temperature value of the rural area. Existing methods are mostly divided into two categories: boundary-based and temperature jump threshold-based. Although the boundary-based method is simple and effective, the accuracy of the urban-rural boundary greatly affects the results of the urban-rural temperature difference calculation; the temperature jump threshold-based method does not require the definition of urban-rural boundaries, but the sudden point (or breakpoint) of the surface temperature decay curve along the urban-rural direction is often determined subjectively, and the parameters in the fitting function have no clear physical meaning in the context of the urban heat island. Second, in terms of urban heat island cooling, determining the number and spatial location of increased green land to improve the cooling effect of the heat island is still a challenge. Some methods based on urban system dynamics or physical heat models simulate the heat exchange process in the building environment, although they have high precision, but the calculation time cost is large, making it difficult to guide urban greening planning; some bionic intelligent algorithms such as ant colony algorithm, genetic algorithm, and particle swarm algorithm, after spatialization, show good performance in solving the spatial optimization of land resources, but the optimization algorithm often pays too much attention to land suitability and surface features, ignoring the multi-dimensional comprehensive features of the city, and the spatial distribution of optimized land patches is scattered, making it difficult to focus on areas in urgent need of cooling in the city.

[0004] Therefore, it is urgent to establish a method for defining the intensity of urban heat island that can avoid subjective interference, integrate the identified urban heat environment information into an optimization model, minimize the current land use layout changes, and automatically adjust and optimize the distribution of land use types at a high spatial resolution at the city level, so as to reduce the urban land surface temperature and alleviate the urban heat island effect. SUMMARY

[0005] In view of the deficiencies in the prior art, the present application provides a heat island cooling method and system based on a multi-type ant colony algorithm guided by thermal knowledge, which automatically identifies the range and intensity of the urban heat island based on remote sensing image data, and uses the urban heat island cooling method based on the multi-type ant colony algorithm guided by thermal knowledge to cool the identified urban heat island by optimizing the land use layout.

[0006] To achieve the above purpose, the technical scheme provided by the present application is a heat island cooling method based on a multi-type ant colony algorithm guided by thermal knowledge, comprising the following steps:

[0007] Step 1, collect remote sensing image data and population thermal data of the study area, and preprocess them, calculate the biophysical component index, and extract the impervious surface of the study area according to the set threshold value;

[0008] Step 2, calculate the density of the impervious surface of the study area;

[0009] Step 3, based on the different densities of the impervious surface calculated in step 2, use the urban clustering algorithm to extract the main built-up area boundary of the study area;

[0010] Step 4, divide the land use types of the study area according to the NDVI index and K-means clustering algorithm, use the split window method to retrieve the land surface temperature of the study area, based on the city gravity center to the periphery to make an equal area multi-ring buffer zone, after removing the water area and high DEM area in the buffer zone, use the inverse S-shaped equation to fit the attenuation process of the average land surface temperature in each buffer zone;

[0011] Step 5, calculate the intensity of the urban heat island based on the land surface temperature of the city and the countryside, and divide it into different grades;

[0012] Step 6, establish the mapping relationship between the multi-type ant colony algorithm and the heat island cooling land use optimization problem, and determine the objective function;

[0013] Step 7, couple the pheromone intensity updating process based on thermal environment perception into the multi-type ant colony algorithm, and construct the multi-type ant colony algorithm guided by thermal knowledge;

[0014] Step 8, based on the multi-type ant colony algorithm guided by thermal knowledge, cool the identified urban heat island by optimizing the land use layout, and output the heat island cooling result and the land use layout optimization result.

[0015] Moreover, the calculation method of the biophysical component index in step 1 is as follows:

[0016]

[0017] In the formula, BCI is the biophysical component index, H is the normalized TC1 or "high reflectivity", L is the normalized TC3 or "low reflectivity", and V is the normalized TC2 or "vegetation". The three indexes are represented by the following formula:

[0018]

[0019]

[0020]

[0021] In the formula, TCi (i = 1, 2, 3) represents the i-th component of the fur coat, and TCi max and TCi min are the maximum and minimum values of the i-th component of the fur coat, respectively.

[0022] The threshold value of BCI is set when extracting the impervious surface, and the impervious surface of the study area is extracted.

[0023] Moreover, the density calculation formula of the impervious surface in step 2 is as follows:

[0024]

[0025] In the formula, Density s (r) represents the density of the impervious surface within a radius of r from the center pixel s, D i represents the distance between pixel i and center pixel s within a radius of r from the center pixel s, if pixel i is an impervious surface pixel, B si = 1, otherwise, B si = 0, and n is the number of pixels within a radius of r from the center pixel s.

[0026] Moreover, in step 3 of the city clustering algorithm, the impervious surface pixel is randomly selected as the center pixel, and the connected region of the impervious surface within the Moore neighborhood is calculated. Then, a distance threshold L is defined, and the connected regions with a distance less than L are merged. An aggregation threshold S is determined, and the connected regions with a pixel number less than S are removed. Finally, according to different input values of the impervious surface density, the boundaries of the built-up areas with different density levels in the study area are obtained.

[0027] Moreover, in step 4, the NDVI index calculation formula is as follows:

[0028]

[0029] In the formula, R NIR , R RED are the reflectivity of near-infrared band and red band respectively.

[0030] The study area is divided into four types of land cover, i.e., water area, low NDVI area, medium NDVI area and high NDVI area by using K-means clustering algorithm, the land surface temperature of the study area is retrieved by using the split window method, the average value of land surface temperature of the concentric ring except water and high DEM area is calculated, and the inverse S-shaped equation is used to fit the land surface temperature points, and the inverse S-shaped equation is expressed as follows:

[0031]

[0032] The second derivative of the equation f(r) with respect to the variable r is obtained as follows:

[0033]

[0034]

[0035] In the formula, f(r) is the land surface temperature of the rth multi-ring buffer zone, f''(r) is the second derivative of f(r), t, c, alpha and D are constants to be fitted, e is a natural constant, and h is an intermediate variable in the derivation process of the equation.

[0036] The rural background temperature is the land surface temperature corresponding to the extreme point of the second derivative f''(r), i.e.,

[0037]

[0038] At this time, there are two solutions r1 and r2, and f(r1) and f(r2) represent the urban core area and the rural background temperature respectively, i.e.,

[0039]

[0040]

[0041] Based on the average temperature of each equal-area buffer zone, the four parameters of the inverse S-shaped equation are solved by using the nonlinear least squares method, wherein the parameters t and c are two horizontal asymptotes, when r approaches infinity, f(r) approaches c, i.e., the average temperature of the urban fringe area; when r=0, f(r) approaches t, i.e., the average temperature of the urban core area, since r1+r2=D, the parameter D represents the approximate city range.

[0042] Moreover, the city heat island intensity ΔT in step 5 is calculated as follows:

[0043] ΔT=T Urban -T Rural (13)

[0044] where T Urban represents the urban background temperature, which is the average land surface temperature in the main built-up area of the city, T Rural represents the rural background temperature.

[0045] The obtained urban heat island intensity is divided into different grades.

[0046] Moreover, the mapping relationship between the multi-type ant colony algorithm and the heat island cooling land use optimization problem in step 6 is as follows: the number of types of ants is the same as the number of land use types, the number of initial ants in each type of ant is determined by the number of pixels of each land use type on the remote sensing image, each pixel in the initial state remote sensing image is randomly occupied by an ant, the final ant type of each pixel is derived as the best land use layout scheme, the current search space of each pixel is K, K is the number of types of ants, which is also the number of land use types, in the iteration process, if the objective function of the pixel in this iteration is greater than that in the last iteration, the pixel selects the k types of ants with the maximum selection probability in this iteration, and the land use type of the pixel becomes k; otherwise, the pixel will keep the current land use type; the selection probability P ijk (t) of the pixel (i, j) by the ant of the kth type in the tth iteration is defined as follows:

[0047]

[0048] where Suit ijk (t) represents the heuristic value of the ant of the kth type at the pixel (i, j), τ ijk (t) represents the pheromone intensity of the ant of the kth type at the pixel (i, j), η ijtype (t) represents the heuristic value of the ant of the kth type at the pixel (i, j), τ ijtype (t) represents the pheromone intensity of the ant of the kth type at the pixel (i, j), 0 < k < K, and the constants α and β are used to adjust the relative importance of the two factors.

[0049] The heuristic value η ijk is defined as follows:

[0050]

[0051] where Suit ijk represents the land use suitability of the ant of the kth type at the pixel (i, j), ∑ (i,j)∈x Suit ijk represents the total land use suitability in the entire study area, and x represents the entire study area.

[0052] The pheromone intensity τ ijk(t) is the only factor guiding the multi-type ant colony algorithm to achieve land use optimization, which is initialized as:

[0053]

[0054] where G is the total number of pixels in the entire study area.

[0055] The pheromone intensity is updated at each iteration as follows: if the land use type at pixel (i, j) changes in this iteration, the pheromone intensity of the land use type after the change at pixel (i, j) is increased at the end of this iteration, and the pheromone intensity of other land use types at pixel (i, j) is reduced, i.e.:

[0056] τ ijk (t+1) = τ ijk (t) (1-ρ) + Δτ ijk (t) (17)

[0057]

[0058] where τ ijk (t), τ ijk (t+1) are the pheromone intensities of the tth and t+1th iterations, respectively, ρ is the coefficient controlling the evaporation rate, r is the coefficient adjusting the pheromone additional value, and Δτ ijk (t) is the pheromone left by the ant of population k at pixel (i, j), which is determined by the initialized τ ijk (0).

[0059] After multiple iterations, the pheromone of a specific ant colony at different pixels will dominate according to the pheromone update process, so that a stable and optimal land use allocation is output in the final stage.

[0060] The goals of the multi-type ant colony algorithm are to maximize land suitability and to minimize the difference between land surface temperature and human comfort temperature. The objective function F ij is defined as follows:

[0061] F ij = max k (μ×Suit ijk -v×abs (LST ijhtt -LST ijk )) (19)

[0062] where μ and v are the weights of land suitability and land surface temperature, respectively, Suit ijk represents the land suitability of the ant of population k at pixel (i, j), LST ijhtt represents the human comfort temperature, and LSTijk is the land surface temperature of the land use type k after the change of the pixel in each iteration.

[0063] Since the thermal infrared band data and the land surface emissivity are two important factors for the inversion of the land surface temperature LST, and considering that the land surface emissivity of different land types is only related to the NDVI index, after the change of the land use type, the following two strategies are used to update the values of the thermal infrared band and NDVI after the modification of the land use type in different stages: 1) after the end of each iteration, the minimum value of the thermal infrared band and the maximum value of NDVI corresponding to the changed land use type k in the original image are used, and this method is used in the process of iteration of the ant colony algorithm; 2) a buffer area is made with the current pixel as the center, and the average value of the thermal infrared band and the average value of NDVI corresponding to the changed land type k in the buffer area are used, and this method is used for the output of the final result after the end of all iterations.

[0064] Moreover, the pheromone intensity updating process based on thermal environment perception in step 7 includes two aspects of self-perception and environment assessment, wherein the self-perception ability enhancement process of the ant is expressed as:

[0065]

[0066] wherein τ ijk (t) is the pheromone intensity of the tthiteration, pop ij is the population density of the pixel (i, j), pop mean is the average value of the population density of the study area, pop max is the maximum value of the population density.

[0067] If the heat island intensity level is very high, the pheromone intensity of the land type of the pixel (i, j) will be reduced, that is:

[0068]

[0069] wherein τ ijk (t) is the pheromone intensity of the tthiteration, θ l is the pheromone concentration decay coefficient based on the heat island intensity, and m is the level of the high-level heat island intensity.

[0070] The environment assessment ability of the ant is realized by constructing the perception neighborhood U, and the perception process of the abnormal temperature area is expressed as follows:

[0071]

[0072] DIFLST ij = LST ij - LST Uijp (23)

[0073]

[0074] where τ ijk (t) is the pheromone intensity of the tth iteration, θ z is the pheromone intensity decay coefficient controlled by temperature difference, U ijp is the perception neighborhood of the pixel (i, j) in a p x p range centered on the pixel (i, j), is the average land surface temperature of the perception neighborhood, LST ij is the land surface temperature of the pixel (i, j), DIFLST ij is the temperature difference of the pixel (i, j), DIFLST abnormal and DIFLST max are constants, respectively representing the abnormal temperature difference and the maximum temperature difference.

[0075] The disintegration process of the heat core area is defined as follows:

[0076]

[0077] where τ ijk (t) is the pheromone intensity of the tth iteration, HI xy is the number of high-level heat islands in the perception neighborhood of the pixel (i, j), U ijp is the perception neighborhood of the pixel (i, j) in a p x p range centered on the pixel (i, j), N is the entire image, n is the total number of pixels of the three types of land use to be optimized, θ c is the pheromone intensity decay coefficient controlled by the heat core area.

[0078] In each iteration, the pheromone intensity is updated, and the pheromone intensity update formula is selected according to the actual situation, and if multiple conditions are met at the same time, the formulas meeting the conditions are multiplied.

[0079] In addition, the heat island cooling result in step 8 is displayed in a table form to show the change of the number of each level heat island, the number of people exposed to high-level heat islands, and the like, and is displayed in a spatial graph form to show the spatial distribution of each level heat island after cooling, and the land use layout optimization result is the number structure and spatial distribution of each type of land after optimization.

[0080] The application also provides a heat island cooling system based on the heat knowledge guided multi-type ant colony algorithm, which is used to realize the heat island cooling method based on the heat knowledge guided multi-type ant colony algorithm.

[0081] In addition, the application further provides a heat island cooling system based on the heat knowledge guided multi-type ant colony algorithm, which comprises a processor and a memory, the memory is used to store program instructions, and the processor is used to call the program instructions in the memory to execute the heat island cooling method based on the heat knowledge guided multi-type ant colony algorithm.

[0082] Compared with the prior art, the present application has the following advantages:

[0083] 1) The equation using parameters has a clear physical meaning to describe the urban and rural thermal feature change, avoiding the artificial interference of urban and rural surface temperature attenuation fitting, and improving the objectivity and accuracy of the definition of urban heat island intensity;

[0084] 2) In the identification and cooling of urban heat island, the anti-S equation is introduced to fit the urban and rural surface temperature attenuation trend, and the multi-type ant colony algorithm is constructed by coupling the urban heat island field knowledge and the population distribution characteristics into the multi-type ant colony algorithm, and the urban heat island identified by the cooling is guided by the thermal knowledge. Compared with the traditional heat island definition and optimization algorithm, the subjectivity in determining the rural background temperature is avoided, and the spatial distribution of the optimized pixels is more reasonable and effective through the thermal knowledge guided algorithm;

[0085] 3) The thermal environment, thermal aggregation characteristics, and population spatial distribution characteristics are fully considered, and the spatial optimization ability of the multi-type ant colony algorithm is combined to output a more reasonable and applicable land use spatial layout optimization scheme and heat island cooling scheme, providing a powerful tool for sustainable urban thermal environment planning and relieving urban heat island effect. BRIEF DESCRIPTION OF DRAWINGS

[0086] Figure 1 is the overall flowchart of the method of the present application.

[0087] Figure 2 is the spatial distribution map of the density of impervious surface in Wuhan City in the embodiment of the present application.

[0088] Figure 3 is the boundary of built-up area of different density levels in Wuhan City in the embodiment of the present application.

[0089] Figure 4 is the schematic diagram of the anti-S equation fitting the surface temperature points in the embodiment of the present application.

[0090] Figure 5 is the anti-S equation fitting result of Wuhan City in 2022 in the embodiment of the present application.

[0091] Figure 6 is the schematic diagram of the heat island intensity level of Wuhan City in 2022 in the embodiment of the present application.

[0092] Figure 7 is the schematic diagram of the multi-type ant colony algorithm guided by thermal knowledge in the present application.

[0093] Figure 8 is the comparison diagram of the spatial distribution of the optimized pixels of the multi-type ant colony algorithm guided by thermal knowledge in the present application and the traditional multi-type ant colony algorithm. DETAILED DESCRIPTION

[0094] The application provides a heat island cooling method and system based on heat knowledge guided multi-type ant colony algorithm, and the technical scheme of the application is further described below with Wuhan as an example in combination with the drawings.

[0095] Embodiment one

[0096] As shown in the figure, the application provides a heat island cooling method based on heat knowledge guided multi-type ant colony algorithm, which comprises the following steps: Figure 1 Step 1, collect remote sensing image data and population thermal data of Wuhan, and pretreat them, calculate the biophysical component index, and extract the impervious surface of Wuhan according to the set threshold.

[0097] The collected Landsat8 OLI remote sensing image time node is August 3, 2022, and the weather condition is good on that day, and the image cloud content is less than 1%, after atmospheric correction and image seamless splicing pretreatment operation, the data used for subsequent analysis is obtained. The biophysical component index BCI is calculated based on the pretreated data, and the specific calculation method is as follows:

[0098]

[0099]

[0100] In the formula, BCI is the biophysical component index, H is the normalized TC1 or "high reflectivity", L is the normalized TC3 or "low reflectivity", and V is the normalized TC2 or "vegetation", and the three indexes can be represented by the following formula:

[0101]

[0102]

[0103]

[0104] In the formula, TCi (i = 1, 2, 3) represents the i-th type of TM component, and TCi max and TCi min are the maximum and minimum values of the i-th TM component, respectively.

[0105] The threshold value of BCI for extracting the impervious surface is set to -0.1, and the impervious surface of Wuhan is extracted. By comparing the 100 randomly generated points on the extracted impervious surface of Wuhan with the remote sensing image of Google Maps, it is found that the consistency of the impervious surface extraction is high, reaching 91%.

[0106] Step 2, calculate the density of the impervious surface of Wuhan.

[0107] The density calculation formula of the impervious surface is:

[0108] ​

[0109] where Density s (r) represents the impervious surface density within the radius r of the central pixel s, D i represents the distance between pixel i and the central pixel s within the radius r of the central pixel s, if pixel i is an impervious surface pixel, B si = 1, otherwise, B si = 0, and n is the number of pixels within the radius r of the central pixel s.

[0110] The spatial distribution of impervious surface density was calculated based on the density formula, as shown in Fig. 2. The high-density area (density > 70%) was concentrated in the commercial center with the densest buildings in Wuhan, and the edge of the low-density area (density within 35-50%) was similar to the planning boundary of the urban development zone in Wuhan. Figure 2

[0111] Step 3, based on the impervious surface with different densities calculated in step 2, the urban main built-up area boundary of the study area was extracted using the city clustering algorithm.

[0112] The impervious surface with water surface density > 35%, > 50%, and > 70% was used as the input data of the city clustering algorithm (CCA), a random impervious surface pixel was selected as the central pixel, and the connected region of impervious surface within the Moore neighborhood (8 adjacent pixels around the central pixel) was calculated. The distance threshold L was set to 200 m, the connected regions with a distance less than 200 m were merged, the aggregation threshold S was set to 10,000, and the connected regions with a pixel number less than 10,000 were removed. Finally, according to different input values of impervious surface density, the built-up area boundaries of different impervious surface density levels in Wuhan were obtained, as shown in Fig. 3. Figure 3

[0113] Step 4, according to the NDVI index and K-means clustering algorithm, the land use types in the study area were divided, the split-window method was used to retrieve the land surface temperature in the study area, and based on the city gravity center, the equal-area multi-ring buffer zone was made to the periphery. After removing the water area and high DEM area in the buffer zone, the inverse S-shaped equation was used to fit the decay process of the average land surface temperature in each buffer zone.

[0114] The calculation formula of the NDVI index is as follows:

[0115]

[0116] where R NIR , R RED are the reflectivity of near-infrared band and red band, respectively.

[0117] ​​The study area was divided into four land cover types using the K-means clustering algorithm, including water, low NDVI areas (LNDVI, such as roads, built-up areas, bare land, and industrial areas), medium NDVI areas (MNDVI, such as grasslands, sparse vegetation cover areas, and bare land), and high NDVI areas (HNDVI, such as forests and vegetation areas). Referring to the land use and cover classification product in 2022, the average classification accuracy is 81%.

[0118] The land surface temperature of Wuhan City was inverted using the split-window method, and the results are shown in Figure 4 The average land surface temperature of the concentric ring excluding water and high DEM areas was calculated, and the decay process of the average land surface temperature in each buffer zone was fitted using an inverse S-shaped equation.

[0119] The inverse S-shaped equation is expressed as follows:

[0120]

[0121] The second derivative of the equation f(r) with respect to the variable r is:

[0122]

[0123]

[0124] where f(r) is the average land surface temperature of the rth multi-ring buffer zone, f"(r) is the second derivative of f(r), t, c, a, D are constants to be fitted, e is the natural constant, and h is an intermediate variable in the derivation process of the equation.

[0125] Since the second derivative of the function describes the change in the decay rate of the land surface temperature from the city to the surrounding area, the average temperature of the city core and the rural background temperature can be determined as the land surface temperatures corresponding to the extreme points of the second derivative f"(r), i.e.

[0126]

[0127] At this time, there are two solutions r1 and r2, and f(r1) and f(r2) represent the city core and rural background temperatures, respectively, i.e.

[0128]

[0129]

[0130] Based on the average land surface temperature of each equal-area buffer zone, the four parameters of the inverse S-shaped equation were solved using the nonlinear least squares method of MATLAB R2021a, as shown in Table 1. Among them, parameters t and c are two horizontal asymptotes, when r approaches infinity, f(r) tends to c, which is the average land surface temperature of the urban fringe area; when r = 0, since the value of a is usually between 2 and 6, f(r) = (t-c) / (1+e -α ) + c tends to t, t represents the average land surface temperature of the urban core area. Since r1+r2=D, parameter D represents the approximate urban range.

[0131] The solved parameters of the inverse S-shaped equation are shown in Table 1, and the adjusted R 2 of the fitting result is 0.98, indicating that the inverse S-shaped equation can well fit the attenuation process of urban land surface temperature.

[0132] Table 1 Fitting parameters of urban and rural land surface temperature attenuation

[0133]

[0134] As Figure 5 shown, the land surface temperature of Wuhan city gradually decreases with distance from the city center, and the land surface temperature first decreases slowly, then decreases rapidly, and finally decreases slowly again. Substituting the equation parameters in Table 1 into the inverse S-shaped equation, the rural background temperature of Wuhan city is obtained as 30.70℃.

[0135] Step 5, calculate the urban heat island intensity based on the land surface temperature of the city and the countryside, and divide it into different levels.

[0136] The calculation method of urban heat island intensity ΔT is as follows:

[0137] ΔT = T Urban -T Rural (13)

[0138] In the formula, T Urban represents the urban background temperature, which is the average land surface temperature in the low-density built-up area of the city, and T Rural represents the rural background temperature.

[0139] According to the low-density built-up area boundary of Wuhan city obtained in step 3, the average land surface temperature of the city is obtained as 32.85℃, and combined with the rural background temperature obtained in step 4, the urban-rural temperature difference of Wuhan city is calculated as 2.15℃. As shown in Table 2, the rural background temperature to the urban background temperature is defined as the first level of urban heat island, and each increase of one time of the urban-rural temperature difference above the urban background temperature is added by 1, and the area greater than four times the temperature difference is considered as the fifth level of heat island. The heat island intensity level of Wuhan city is shown in Figure 6 .

[0140] Table 2 Definition of heat island intensity level

[0141]

[0142] Step 6, establish the mapping relationship between the multi-type ant colony algorithm and the heat island cooling land use optimization problem, and determine the objective function.

[0143] The multi-type ant colony algorithm process is shown in Figure 7 The mapping relationship between the multi-type ant colony algorithm (Multi-type Ant Colony Optimization, MACO) and the heat island cooling land use optimization problem is established as follows: the number of ant colonies is the same as the number of land use types, the number of initial ants in each ant colony is determined by the number of pixels of each land use type on the remote sensing image, each pixel in the initial state is randomly occupied by an ant, and the final ant type of each pixel is derived as the best land use layout scheme. The current search space of each pixel is K, K is the number of ant types, which is also the number of land use types. In this embodiment, the types of land to be optimized are LNDVI, MNDVI and HNDVI, so K is 3. In the iteration process, if the objective function of the pixel in this iteration is greater than that in the last iteration, the pixel selects the k type of ant with the maximum selection probability in this iteration, and the land use type of the pixel becomes k; otherwise, the pixel will keep the current land use type. The selection probability P ijk of the ant colony k (0 < k < K) to the pixel (i, j) at iteration t times can be defined as follows:

[0144]

[0145] Suit ijk (k, i, j, t) represents the heuristic value of the ant colony k at the pixel (i, j), τ ijk (k, i, j, t) represents the pheromone intensity of the ant colony k at the pixel (i, j), η ijtype (k, i, j, t) represents the heuristic value of the ant colony k at the pixel (i, j), τ ijtype (k, i, j, t) represents the pheromone intensity of the ant colony k at the pixel (i, j), and the constants α and β are used to adjust the relative importance of the two factors. In this embodiment, they are set to 0.75 and 1.25, respectively.

[0146] The heuristic value η ijk is defined as follows:

[0147]

[0148] Suit ijk (k, i, j, t) represents the suitability of the ant colony k (land use type k) at the pixel (i, j), and ∑ (i,j)∈x Suitijk Suitability is the sum of land suitability in the whole study area, and x represents the whole study area.

[0149] Pheromone intensity τ ijk is the only factor guiding the implementation of the multi-type ant colony algorithm to optimize land use, which is initialized as:

[0150]

[0151] In the formula, G is the total number of pixels in the whole study area.

[0152] The principle of updating the pheromone intensity at each iteration is: if the land use type at pixel (i, j) changes in this iteration, the pheromone intensity of the land use type after the change at pixel (i, j) is enhanced at the end of this iteration, and the pheromone intensity of other land use types at pixel (i, j) is reduced (i.e. volatilized), that is:

[0153] τ ijk (t+1) = τ ijk (t) (1-ρ) + Δτ ijk (t) (17)

[0154]

[0155] In the formula, τ ijk (t), τ ijk (t+1) are the pheromone intensities of the tth and t+1th iterations, respectively, ρ is the coefficient controlling the volatilization rate, r is the coefficient adjusting the pheromone additional value, and Δτ ijk () refers to the pheromone left by the ant of population k at pixel (i, j), which is determined by the initialized τ ijk (0).

[0156] After multiple iterations, the pheromone of a specific ant colony at different pixels will dominate according to the pheromone updating process, so that a stable and optimal land use allocation is output in the last stage.

[0157] The goal of the multi-type ant colony algorithm has two, which are to maximize land suitability and to minimize the difference between land surface temperature and human comfort temperature, and the objective function F ij is defined as follows:

[0158] F ij = max k (μ×Suit ijk -v×abs (LST ijhtt -LST ijk )) (19)

[0159] where μ and v are the weights of land suitability and LST, respectively, and are set to be equal in this example, i.e., μ = v = 0.5, Suit ijk represents the land suitability of the population of k ants at pixel (i,j), LST ijhtt represents the human comfort temperature, LST ijk is the LST of land use type k after the change of pixel in each iteration.

[0160] Since the thermal infrared band data and the surface emissivity are two important factors for the inversion of LST, the values of the thermal infrared band and the surface emissivity need to be updated after the change of land use type. Considering that the surface emissivity of different land types is only related to the NDVI index, the present application uses different strategies to update the values of the thermal infrared band and NDVI of the pixel after the change of land use type at different stages: 1) after the end of each iteration, the minimum value of the thermal infrared band and the maximum value of NDVI corresponding to the changed land use type k in the original image are used, and this method is used in the process of algorithm iteration in order to obtain the maximum cooling potential; 2) a buffer area is made with the current pixel as the center, and the average value of the thermal infrared band and the average value of NDVI corresponding to the changed land type k in the buffer area are used, and this method tries to simulate the actual cooling effect by considering regional heterogeneity, and is used for the output of the final result after the end of all iterations.

[0161] Step 7: The pheromone intensity updating process based on thermal environment perception is coupled into the multi-type ant colony algorithm to construct a thermal knowledge guided multi-type ant colony algorithm.

[0162] The pheromone intensity updating process based on thermal environment perception includes two aspects of self-perception and environment evaluation, which are defined as follows:

[0163] 1) Enhance the self-perception ability of ants.

[0164] The perception information domain of ants is expanded using population thermal data and heat island intensity level data, and the cooling urgency is determined by evaluating whether the pixel is located in a densely populated area. Specifically, the population-dense areas and high heat island intensity areas have a higher demand for cooling, while the temperature in the sparsely populated areas (such as rural and industrial parks) and low heat island intensity areas can remain unchanged. When the pixel is located in a densely populated area, the pheromone intensity of the pixel (i,j) now with the type of land will be reduced, i.e.:

[0165]

[0166] where τ ijk (t) is the pheromone intensity of the tth iteration, pop ij is the population density at pixel (i,j), and in this example, the population density is represented using population thermal data, popmean is the average value of population density in the study area, pop max is the maximum value of population density.

[0167] If the heat island intensity level is high, the pheromone intensity of the cell (i,j) now with land type information will be reduced, which can be expressed as:

[0168]

[0169] where τ ijk (t) is the pheromone intensity of the tth iteration, θ l is the pheromone concentration decay coefficient based on heat island intensity, θ l 0.9, 0.8, and 0.7, respectively, in this embodiment.

[0170] 2) Enhance the ability of ants to evaluate the environment.

[0171] Due to human activities and unreasonable land allocation, some areas in the city may have excessive temperature difference with the surrounding areas. Therefore, during the iteration process of the optimization algorithm, the ants should consider the temperature difference between themselves and the surrounding environment to evaluate whether the cell (i,j) is located in an abnormally high temperature area. The optimization process of the abnormal temperature area can be expressed as follows, by constructing a perception neighborhood U for each ant:

[0172]

[0173]

[0174]

[0175] where τ ijk (t) is the pheromone intensity of the tth iteration, θ z is the pheromone intensity decay coefficient controlled by the temperature difference, U ijp is the perception neighborhood of the ant within a p x p range centered at the cell (i,j), is the average land surface temperature of the perception neighborhood, LST ij is the land surface temperature of the cell (i,j), DIFLST ij is the temperature difference of the cell (i,j), DIFLST abnormal and DIFLST max are constants, representing the abnormal temperature difference and the maximum temperature difference, respectively.

[0176] In addition, with urban construction, some areas with high-level heat islands gradually become heat core areas, and due to the lack of cooling facilities in these areas, heat hazard events may occur, therefore, the ant needs to evaluate the aggregation of high-level heat islands in the perception neighborhood, and disintegrate the heat core area by optimizing the land use type, and the process can be represented as:

[0177]

[0178] In the formula, τ ijk (t) is the pheromone intensity of the tth iteration, HI xy is the number of high-level heat islands in the perception neighborhood with the pixel (i, j), U ijp is the perception neighborhood in the range of ×p centered on the pixel (i, j), N is the entire image, n is the total number of pixels of the three types of land use (LNDVI, MNDVI, HNDVI) to be optimized, θ c is the pheromone intensity decay coefficient controlled by the heat core area, which is set to 0.8 in the embodiment.

[0179] The pheromone intensity is updated each iteration, and the pheromone intensity update formula needs to be selected according to the actual situation, and if multiple conditions are met at the same time, the formulas that meet the conditions are multiplied, for example, when the pixel is located in a densely populated area, a high heat island intensity area and an abnormally high temperature area at the same time,

[0180] Step 8, the multi-type ant colony algorithm based on heat knowledge guidance optimizes the land use layout to cool the identified urban heat island, and outputs the heat island cooling result and the land use layout optimization result.

[0181] The heat island cooling result displays the change of the number of heat islands of each level, the number of people exposed to high-level heat islands, etc. in the form of a table, and displays the spatial distribution of each level heat island after cooling in the form of a spatial graph. The land use layout optimization result is the number structure and spatial distribution of each type of land after optimization.

[0182] Table 3 shows the comparison of the heat island cooling effect of the traditional multi-type ant colony algorithm and the multi-type ant colony algorithm based on heat knowledge guidance proposed by the present application, from Table 3, it can be seen that the multi-type ant colony algorithm based on heat knowledge guidance achieves better heat island cooling effect, reduces the number of heat islands of level 4 and level 5 by 7.2% and 5.1% respectively, and reduces the risk of residents exposed to high-level heat islands.

[0183] Table 3 Comparison of heat island cooling effect of two algorithms

[0184]

[0185] Figure 8Compared with the traditional multi-type ant colony algorithm for optimizing the spatial distribution of pixels, the multi-type ant colony algorithm based on heat knowledge guidance provided by the present application can consider heat island grade information and identify heat anomaly areas, and the spatial distribution of the optimized pixels is highly consistent with the distribution of high-grade heat islands, avoiding the randomness of the spatial distribution of the pixels optimized by the multi-type ant colony algorithm.

[0186] Embodiment Two

[0187] Based on the same inventive concept, the present application also provides a heat island cooling system based on the multi-type ant colony algorithm guided by heat knowledge, which comprises a processor and a memory, the memory is used for storing program instructions, and the processor is used for calling the program instructions in the memory to execute the heat island cooling method based on the multi-type ant colony algorithm guided by heat knowledge as described above.

[0188] In specific implementation, the method provided by the technical scheme of the present application can be automatically run by computer software technology, and the system device of the method, such as the computer readable storage medium storing the corresponding computer program of the technical scheme of the present application and the computer equipment including the running corresponding computer program, should also be within the protection scope of the present application.

[0189] The specific embodiments described herein are merely illustrative of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or replace them with similar ways without departing from the spirit of the present application or exceeding the scope defined by the appended claims.

Claims

1. A heat island cooling method based on hot knowledge guided multi-type ant colony algorithm, characterized in that, The method comprises the following steps: Step 1, collecting remote sensing image data, population thermal data of the study area, and preprocessing the same, calculating a biophysical component index, and extracting impervious surfaces in the study area according to a set threshold value; Step 2, calculating the density of the impervious surfaces in the study area; Step 3, based on the different densities of the impervious surfaces calculated in step 2, using a city clustering algorithm to extract the main built-up area boundary of the study area; Step 4, according to the NDVI index and the K-means clustering algorithm, dividing the land use type of the study area, using the split window method to retrieve the land surface temperature of the study area, based on the urban gravity center to the periphery to make an equal area multi-ring buffer, after removing the water area and high DEM area in the buffer, using the inverse S-shaped equation to fit the attenuation process of the average land surface temperature in each buffer; Step 5, based on the land surface temperature of the city and the countryside, calculating the urban heat island intensity, and dividing it into different grades; Step 6, establishing a mapping relationship between the multi-type ant colony algorithm and the heat island cooling land use optimization problem, and determining the objective function; Step 7, coupling the pheromone intensity updating process based on the thermal environment perception into the multi-type ant colony algorithm, and constructing a thermal knowledge guided multi-type ant colony algorithm; Step 8, based on the thermal knowledge guided multi-type ant colony algorithm, identifying the urban heat island through optimizing the land use layout, and outputting the heat island cooling result and the land use layout optimization result.

2. The method of claim 1, wherein the method is based on a thermal knowledge guided multi-type ant colony optimization algorithm. The calculation method of the biophysical component index in step 1 is as follows: In the formula, BCI is the biophysical component index, H is the normalized TC1 or "high reflectivity", L is the normalized TC3 or "low reflectivity", and V is the normalized TC2 or "vegetation", which are represented by the following formula: wherein Tci represents the i-th component of the tuxedo, Tci max and Tci min are the maximum and minimum values of the i-th component of the tuxedo, respectively, i = 1, 2, 3. The threshold value of BCI is set when extracting the impervious surface, and the impervious surface of the study area is extracted.

3. The method of claim 1, wherein the method is based on a thermal knowledge guided multi-type ant colony optimization algorithm. The density calculation formula of the impervious surface in step 2 is as follows: where Density s (r) represents the impervious surface density within a radius r of the central pixel s i represents the distance between pixel i and the central pixel s within a radius r of the central pixel s, B si = 1 if pixel i is an impervious surface pixel, otherwise, B si = 0, and n is the number of pixels within a radius r of the central pixel s.

4. The method of claim 1, wherein the method is based on a thermal knowledge guided multi-type ant colony optimization algorithm. In step 3, the city clustering algorithm first randomly selects an impervious surface pixel as a center pixel, and calculates the connected region of the impervious surface in the Moore neighborhood, then defines a distance threshold L, merges the connected regions with a distance less than L, and determines an aggregation threshold S, removes the connected regions with a pixel number less than S, and finally obtains the built-up area boundary of different impervious surface density levels in the study area according to different impervious surface density input values.

5. The method of claim 1, wherein the method is based on a thermal knowledge guided multi-type ant colony optimization algorithm, and wherein the method further comprises: The NDVI index calculation formula in step 4 is as follows: ​ In the formula, R NIR , R RED are reflectivity in near-infrared band and red band, respectively. The K-means clustering algorithm is used to divide the study area into four types of land cover, namely water area, low NDVI area, medium NDVI area, and high NDVI area, the split window method is used to retrieve the land surface temperature in the study area, the average land surface temperature in the concentric circular ring is calculated by removing the water body and high DEM area, and the inverse S-shaped equation is used to fit the attenuation process of the average land surface temperature in each buffer, and the inverse S-shaped equation is expressed as follows: Let the second derivative of the equation f(r) with respect to the variable r be: In the formula, f(r) is the average land surface temperature of the rth multi-ring buffer, f"(r) is the second derivative of f(r), t, c, alpha, D are constants to be fitted, e is a natural constant, and h is an intermediate variable in the derivative process of the equation; The average temperature of the urban core and the rural background temperature are the extreme points of the second derivative f"(r) of the surface temperature, i.e.: At this time, there are two solutions r1 and r2, and f(r1) and f(r2) represent the average temperature of the urban core and the rural background, respectively, i.e. Based on the average surface temperature of each equal-area buffer zone, the four parameters of the inverse S-shaped equation are solved using the nonlinear least squares method, in which the parameters t and c are the two horizontal asymptotes, f(r) approaches c when r approaches infinity, i.e. the average temperature of the urban fringe, and f(r) approaches t when r = 0, i.e. the average temperature of the urban core. Since r1 + r2 = 1, the parameter D represents the approximate urban range.

6. The method of claim 1, wherein the method is based on a thermal knowledge guided multi-type ant colony optimization algorithm. The calculation method of the urban heat island intensity ΔT in step 5 is as follows: ΔT = T Urban -T Rural (13) where T Urban represents the urban background temperature, which is the average land surface temperature in the low-density built-up area of the city, T Rural represents the rural background temperature; The obtained urban heat island intensity is divided into different grades.

7. The method of claim 1, wherein the method is based on a thermal knowledge guided multi-type ant colony optimization algorithm. The mapping relationship between the multi-type ant colony algorithm and the heat island cooling land use optimization problem in step 6 is as follows: the number of types of ants is the same as the number of land use types, the number of initial ants in each type of ant is determined by the number of pixels of each land use type on the remote sensing image, each pixel in the initial state remote sensing image is randomly occupied by an ant, the final ant type of each pixel is derived as the best land use layout scheme, the current search space of each pixel is K, K is the number of types of ants, which is also the number of land use types, in the iteration process, if the objective function of the pixel in this iteration is greater than the last iteration, the pixel selects the k types of ants with the maximum selection probability in this iteration, and the land use type of the pixel becomes k; otherwise, the pixel will keep the current land use type; the selection probability P of the ant of the population k to the pixel (i, j) at iteration t times is as follows: ijk (t) is defined as follows: where η ijk (t) denotes the pheromone intensity of ants of population k at pixel (i,j), η ijk (t) denotes the pheromone intensity of ants of population k at pixel (i,j), η ijtype (t) denotes the pheromone intensity of ants of population k at pixel (i,j), η ijtype (t) denotes the pheromone intensity of ants of population k at pixel (i,j), η Heuristic value η ijk is defined as follows: Suit ijk Suit (i,j)∈x Suit ijk Suit Pheromone intensity τ ijk (t) is the only factor guiding the implementation of the multi-type ant colony algorithm to optimize land use, which is initialized as: In the formula, G is the total number of pixels in the entire study area; The pheromone intensity is updated at each iteration: if the land use type at pixel (i, j) changes at this iteration, the pheromone intensity of the changed land use type at pixel (i, j) is increased at the end of this iteration, and the pheromone intensity of other land use types at pixel (i, j) is decreased, i.e. t ijk (t+1) = t ijk (t)(1 - p) + Δt ijk (t) (17) wherein τ ijk (t) and τ ijk (t+1) are the pheromone intensity of the tth and (t+1)th iteration, respectively, p is a coefficient that controls the evaporation rate, r is a coefficient that adjusts the pheromone addition value, and Δτ ijk (t) is the pheromone left by the kth ant in the population at the pixel (i, j), which is determined by the initialized τ ijk (0). After multiple iterations, the pheromone of a specific ant colony at different pixels will dominate according to the pheromone update process, and a stable and optimal land use allocation will be output in the final stage.

8. The method of claim 7, wherein the method is based on thermal knowledge guided multi-type ant colony optimization algorithm. The objectives of the multi-type ant colony algorithm in step 6 are two, which are to maximize land suitability and to minimize the difference between land surface temperature and human comfortable temperature, and the objective function F ij are defined as follows: F ij = max k (μ x Suit ijk - v x abs (LST ijhtt - LST ijk )) (19) where μ and v are the weights of land suitability and land surface temperature, Suit ijk represents the land suitability of the ant of species k at pixel (i, j), LST ijhtt represents the land surface temperature of the land use type k after the change of pixel in each iteration; and ijk is the land surface temperature of the land use type k after the change of pixel in each iteration; and Since the thermal infrared band data and the surface emissivity are two important factors for LST inversion, considering that the surface emissivity of different land types is only related to the NDVI index, after the land use type changes, the following two strategies are used to update the values of the thermal infrared band and NDVI after the modification of the land use type at different stages: 1) After each iteration, the minimum value of the thermal infrared band and the maximum value of NDVI corresponding to the changed land use type k in the original image are used, this method is used in the process of ant colony algorithm iteration; 2) A buffer zone is created with the current pixel as the center, and the average value of the thermal infrared band and the average value of NDVI corresponding to the changed land type k in the buffer zone are used, this method is used for the final output after all iterations are completed.

9. The method of claim 1, wherein the method is based on a thermal knowledge guided multi-type ant colony optimization algorithm. In step 7, the pheromone intensity update process based on thermal environment perception is constructed, including self-perception and environmental assessment, where the self-perception ability enhancement process of the ant is represented as: where τ ijk (t) is the pheromone intensity of the tth iteration, pop ij is the population density at pixel (i, j), pop mean is the average value of the population density in the study area, pop max is the maximum value of the population density; If the heat island intensity level is very high, the pheromone intensity of the current land type at pixel (i, j) will be reduced, i.e. In the formula, τ ijk (t) is the pheromone intensity of the tth iteration, θ l is the pheromone concentration decay coefficient based on the heat island intensity, and m is the grade of the set high-grade heat island intensity. The environmental assessment ability of the ant is realized by constructing the perception neighborhood U, and the perception process of the abnormal temperature zone is represented as follows: where τ ijk (t) is the pheromone intensity of the tth iteration, θ z is the pheromone intensity decay coefficient controlled by temperature difference, U ijp is the perceived neighborhood of the ant in the p x p range centered on the pixel (i, j), is the average land surface temperature of the perceived neighborhood, LST ij is the land surface temperature of the pixel (i, j), DIFLST ij is the temperature difference of the pixel (i, j), DIFLST abnormal and DIFLST max are constants representing the abnormal temperature difference and the maximum temperature difference, respectively; The disintegration process of the thermal core region is defined as follows: where τ ijk (t) is the pheromone intensity of the tth iteration, HI xy is the number of high-level heat islands in the perception neighborhood of pixel (i, j), U ijp is the perception neighborhood of pixel (i, j) in the p x p range centered on the pixel, N is the entire image, n is the total number of pixels of the three types of land use to be optimized, θ c is the pheromone intensity decay coefficient controlled by the thermal core area; The pheromone intensity is updated at each iteration, and the pheromone intensity update formula is selected according to the actual situation, and the specific use is if multiple conditions are met at the same time, then the formulas that meet the conditions are multiplied.

10. A heat island cooling system based on hot knowledge guided multi-type ant colony algorithm, characterized in that, The processor and the memory, the memory is used to store program instructions, the processor is used to call the program instructions in the memory to execute the heat island cooling method based on the heat knowledge guided multi-type ant colony algorithm as claimed in any one of claims 1-9.

Images