Urban planning method and system based on digital twinning

Abstract

The invention relates to the technical field of urban planning, in particular to an urban planning method and system based on digital twinning, and the method comprises the following steps: collecting people flow density data in a public space monitoring camera, carrying out the classified statistics of the people flow density data according to the space types of squares, parks and pedestrian streets, and obtaining a space activity statistical value. According to the method, the people flow density data of the public space are collected in real time, classified statistics is carried out in combination with the space types, and people flow activity characteristics of different areas are accurately obtained. And on the basis of statistical data superposition calculation, space use intensity distribution is formed, so that the dynamic analysis capability is provided for use conditions of different areas. On the basis, geographic coordinate information of rest facilities, landscape facilities and barrier-free facilities is collected, the facility distribution density is calculated, in combination with space use intensity data, overlapping analysis is conducted on the facility coverage range and the activity hot spot area, and the reasonability of the facility layout is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of urban planning, and particularly to an urban planning method and system based on digital twin. Background Art

[0002] Urban planning methods belong to the technical field of urban planning, aiming to provide systematic and scientific planning strategies and implementation plans to guide the rational layout of urban space and the efficient use of resources. This method is widely used in fields such as urban renewal, infrastructure optimization, traffic network design, environmental governance, and public space construction, ensuring that the city can achieve sustainable development, improve urban livability, and enhance the scientificity and accuracy of urban governance under the background of economic growth, population expansion, and environmental changes.

[0003] The existing technologies mainly rely on fixed data statistics and planning models for urban layout design, lacking real-time evaluation of the usage of public spaces, resulting in a low degree of adaptation between the planning and actual needs. At the same time, it fails to fully combine the pedestrian flow characteristics and facility distribution in different regions, with low utilization rate of some spaces and overloaded operation of facilities in some areas, affecting the balanced distribution of space resources. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of the present invention is to solve the deficiencies in the existing technologies and propose an urban planning method and system based on digital twin.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions. The urban planning method based on digital twin includes the following steps: Collect the pedestrian flow density data in the surveillance cameras of public spaces, classify and count the pedestrian flow density data according to the spatial types of squares, parks, and pedestrian streets to obtain the spatial activity statistical values; based on the spatial activity statistical values, superimpose the statistical data under each spatial type to generate the spatial usage intensity distribution values; Collect the geographical coordinate information of rest facilities, landscape facilities, and barrier-free facilities in various types of public spaces, calculate the facility distribution density to obtain the facility distribution density data; based on the facility distribution density data and the spatial usage intensity distribution values, analyze the facility coverage range and activity hotspots, count the proportion of the overlapping area, and generate the facility service coverage index; Based on the facility service coverage index, collect the activity trajectory data of resident groups and the interaction frequency between groups in public spaces to obtain the group interaction data set; based on the group interaction data set, perform spatial clustering on the interaction data, identify the hot and cold regions of social activities, and establish a spatial interaction heat network; Based on the spatial interaction thermal network, collect the group expression features in the public space to obtain a group emotion feature matrix. Based on the group emotion feature matrix, count the proportion of group emotion distribution in different regions to generate spatial emotion distribution data.

[0006] Preferably, the steps for obtaining the spatial activity statistical value are as follows: Collect the pedestrian flow density data from the surveillance cameras in the public space, classify the data according to the spatial type, and the spatial types include squares, parks, and pedestrian streets, to obtain the classified pedestrian flow density data; Based on the classified pedestrian flow density data, count the degree of pedestrian flow aggregation, the staying duration, and the activity range of each type of space to obtain the pedestrian flow activity feature data; Based on the pedestrian flow activity feature data, calculate the spatial activity statistical value, and the calculation formula is: Wherein, is the spatial activity statistical value, is the pedestrian flow density of the jth activity feature, is the average pedestrian flow density of all activity features, is the duration of the jth activity feature, is the average duration of all activity features, and k is the number of activity features.

[0007] Preferably, the steps for obtaining the spatial usage intensity distribution value are as follows: Based on the spatial activity statistical value, calculate the spatial usage intensity distribution value of each spatial type, and the calculation formula is: Wherein, is the spatial usage intensity distribution value, is the spatial activity statistical value of the ith spatial type, is the average activity duration of the ith space, is the number of spatial types, is the arithmetic mean of the sum of all spatial activity ranges.

[0008] Preferably, the steps for obtaining the facility distribution density data are as follows: Based on the spatial usage intensity distribution value, collect the geographical coordinate information of the rest facilities, landscape facilities, and barrier-free facilities in each type of public space to obtain the facility geographical coordinate data; Based on the facility geographical coordinate data, calculate the facility distribution density, and the calculation formula is: Wherein, is the facility distribution density, is the The geographical coordinates of a facility, which are the central coordinates of the crowd gathering point, is the peripheral crowd density of the th facility, Based on the distribution density of the facilities, normalize the distribution of different facility types to obtain facility distribution density data.

[0009] Preferably, the steps for obtaining the facility service coverage index are as follows: Based on the facility distribution density data and the spatial usage intensity distribution value, determine the service coverage range of each facility. Taking the facility geographical coordinates as the center, set the influence radius according to the facility type to generate facility coverage range data; Based on the facility coverage range data, identify the activity hotspots, and judge the spatial overlapping area between the facility service area and the activity hotspots to obtain the overlapping area data of the facilities and the activity hotspots; Based on the overlapping area data, calculate the facility service coverage index, and the calculation formula is: wherein, is the facility service coverage index, is the overlapping area of the th facility and the activity hotspots, is the spatial usage intensity distribution value of the area where the th facility is located, is the service coverage range of the th facility, is the arithmetic mean of the overlapping ratios of the coverage ranges of all facilities and the activity hotspots,

[0010] Preferably, the steps for obtaining the group interaction dataset are as follows: Based on the facility service coverage index, determine the influence range of the facilities on the activities of the resident group in the public space, collect the individual trajectory data of the residents within the influence range, and perform temporal sorting on the trajectory data in combination with the timestamp information to generate resident group activity trajectory data; Based on the resident group activity trajectory data, identify the activity intersection areas of different groups, extract the stay duration and proximity of the group members in the intersection areas, and calculate the number of interaction events between individuals in combination with time segments to generate group interaction frequency data; Based on the group interaction frequency data, classify the interaction events according to the time series, remove the outliers and standardize the data, and perform clustering analysis according to the group size and interaction intensity to generate the group interaction dataset.

[0011] Preferably, the steps for obtaining the spatial interaction heat map network are as follows: Based on the group interaction data set, extract the geographical location, interaction time, and frequency information of group interaction events, classify the data according to spatial coordinates, and perform segmented statistics in combination with the time dimension to generate interaction spatial distribution data; Based on the interaction spatial distribution data, perform spatial clustering, analyze the change trend of interaction density in different regions, and divide the regions into high-interaction-frequency areas, medium-interaction-frequency areas, and low-interaction-frequency areas according to the concentration degree of interaction events to generate data on social activity hot spots and cold spots; Based on the data on social activity hot spots and cold spots, construct the connection relationship of spatial interaction nodes to generate a spatial interaction heat map network.

[0012] Preferably, the steps for obtaining the spatial emotion distribution data are as follows: Based on the spatial interaction heat map network, determine the group interaction active areas in each public space, collect the facial expression feature data of the groups in the interaction active areas, and perform facial expression classification through a convolutional neural network model to generate group facial expression feature data; Based on the group facial expression feature data, organize it according to the time series and spatial position, perform aggregated analysis on the individual facial expression features, extract the main emotion categories and their proportions of the groups in different regions to generate a group emotion feature matrix; Based on the group emotion feature matrix, perform statistics according to the spatial regions, calculate the proportions of various emotions in different regions, integrate the information of the region range, interaction activity, and emotion categories to generate spatial emotion distribution data.

[0013] The present invention provides an urban planning system, including: A pedestrian flow density analysis module, which collects the pedestrian flow density data in the public space monitoring cameras, classifies and statistics it according to squares, parks, and pedestrian streets, calculates the number of pedestrians under different spatial types, and statistics the activity intensity of each type of space to obtain the spatial activity statistical value; A facility distribution evaluation module, based on the spatial activity statistical value, collects the geographical coordinates of the rest facilities, landscape facilities, and barrier-free facilities in squares, parks, and pedestrian streets, calculates the distribution density of various facilities to obtain the facility distribution density data; A facility service analysis module, based on the facility distribution density data and the spatial activity statistical value, conducts a comparative analysis on the facility coverage range and the hot spots of pedestrian flow activities, statistics the overlapping area of the two, calculates the coverage index, and generates the facility service coverage index; A social interaction analysis module, based on the facility service coverage index, collects the activity trajectory data of resident groups and the interaction frequencies between groups, performs spatial clustering on the interaction data, identifies social hotspots and cold spots, and establishes a spatial interaction heat network; An emotional feature analysis module, based on the spatial interaction heat network, collects the group expression features in public spaces, obtains the emotional feature matrix of the group, counts the emotional distribution ratios of groups in different regions, and generates spatial emotion distribution data.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: The present invention accurately obtains the characteristics of human flow activities in different regions by collecting real-time human flow density data in public spaces and classifying and counting them according to spatial types. Based on the superposition calculation of statistical data, a spatial usage intensity distribution is formed, enabling dynamic analysis of the usage conditions in different regions. On this basis, the geographical coordinate information of rest facilities, landscape facilities, and barrier-free facilities is collected, the facility distribution density is calculated, and combined with the spatial usage intensity data, an overlapping analysis of the facility coverage range and activity hotspots is carried out to ensure the rationality of facility layout. Combining the group activity trajectory data and interaction frequency information, a group interaction system is constructed, which can not only analyze individual behaviors but also identify crowd aggregation patterns, improving the adaptability of urban spatial functions. Based on spatial clustering technology, high-density and low-density regions of social activities are identified, and a spatial interaction heat network is established to provide data support for optimizing the configuration of public spaces. Combining the expression features of groups in public spaces, a group emotional feature matrix is constructed, and the distribution of different emotions in the region is accurately calculated, enabling spatial planning to take into account the psychological perception of residents. Brief Description of the Drawings

[0015] Figure 1 It is a schematic diagram of the steps of the present invention. Detailed Embodiment

[0016] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0017] Please refer to Figure 1 , the present invention provides a technical solution, a digital twin-based urban planning method, including the following steps: Collect the human flow density data in the public space monitoring cameras, classify and count the human flow density data according to the spatial types of squares, parks, and pedestrian streets to obtain spatial activity statistical values; based on the spatial activity statistical values, superimpose the statistical data under each spatial type to generate spatial usage intensity distribution values; Collect the geographical coordinate information of rest facilities, landscape facilities, and barrier-free facilities in various types of public spaces, calculate the facility distribution density, and obtain the facility distribution density data; based on the facility distribution density data and the spatial usage intensity distribution value, analyze the facility coverage range and activity hotspots, and count the proportion of the overlapping area to generate the facility service coverage index; Based on the facility service coverage index, collect the activity trajectory data of resident groups and the interaction frequency between groups in public spaces to obtain the group interaction dataset; based on the group interaction dataset, perform spatial clustering on the interaction data to identify the social activity hotspots and cold spots, and establish a spatial interaction heat network; Based on the spatial interaction heat network, collect the group expression features in public spaces to obtain the group emotion feature matrix, and based on the group emotion feature matrix, count the proportion of group emotion distribution in different regions to generate the spatial emotion distribution data.

[0018] The steps for obtaining the spatial activity statistical value are as follows: Collect the crowd density data from the surveillance cameras in public spaces, classify the data according to the spatial type, and the spatial types include squares, parks, and pedestrian streets, to obtain the classified crowd density data; Based on the classified crowd density data, count the crowd aggregation degree, stay duration, and activity range of each type of space to obtain the crowd activity characteristic data; Based on the crowd activity characteristic data, calculate the spatial activity statistical value, and the calculation formula is: where, is the spatial activity statistical value, is the crowd density of the jth activity characteristic, is the average crowd density of all activity characteristics, is the duration of the jth activity characteristic, is the average duration of all activity characteristics, and k is the number of activity characteristics.

[0019] Specifically, continuous frame information in the public space surveillance camera footage is collected. Combining the timestamps and geographical coordinate data recorded by the surveillance equipment, the initial pedestrian flow statistics data is obtained by identifying the human silhouettes in each frame and calibrating the total number of visible people. The surveillance equipment captures multiple frames per second and extracts the number of people information in the corresponding frames. Subsequently, these number data are associated with the surveillance area location. The monitored number of people whose location falls within the square area is recorded in the square type dataset, the monitored number of people whose location falls within the park area is recorded in the park type dataset, and the monitored number of people whose location falls within the pedestrian street area is recorded in the pedestrian street type dataset. Then, a comparative analysis is carried out based on the records of different time periods of each day. Further inspection is performed on some frames with low recognition confidence or severe occlusion to check for misjudgment or omission. The misjudged and omitted parts are corrected by comparing with the previous and subsequent frames to ensure the correct calibration of the number of people in each period. Then, based on the confirmed number data and the actual geographical area of the region, the number value is divided by the area of the corresponding space to calculate the real-time density value. Next, the surveillance data for the whole day or a longer time range is extracted and integrated. The density values are identified and classified and accumulated hour by hour in the same way. Finally, the surveillance records and space types for different time periods are correspondingly superimposed and aligned, and the data in the invalid range (such as the number of people being greater than the upper limit of the single-frame image recognition or the items confirmed to be repeatedly calibrated) are screened out. After the statistics are completed, the classified pedestrian flow density data is obtained.

[0020] After obtaining the classified pedestrian flow density data in the previous step, a specific time interval is selected as the statistical period. The number of people in different regions within the period is accumulated, and the start and end times of continuous stay are recorded. The stay duration is estimated by recording the entry time and departure time of each monitored object in the same region. At the same time, the movement trajectory of the object within the region is tracked to determine the actual activity range and the distribution profile of the crowd activities. The central distribution position and boundary position of the actual occupancy are calculated. Then, the degree of pedestrian flow aggregation in the corresponding space is measured by the ratio of the number of people to the area. The target data with a stay duration greater than zero and less than several day-night total durations is marked as valid stay. The abnormal records with a stay duration far exceeding the monitoring expected range are checked. The cases of true long-term stay or not leaving the monitoring range are reconfirmed and retained in the statistical results. Then, in combination with the degree of pedestrian flow aggregation in each region and the main movement radius within the region, a data comparison is formed. The data significantly exceeding or lower than the radius is carefully reviewed to ensure that the error outliers do not affect the overall statistics. Finally, each region is quantified in three dimensions: the degree of pedestrian flow aggregation, the stay duration, and the activity range. Then, after summarization, the pedestrian flow activity characteristic data is integrated.

[0021] The benefit of the formula is that, firstly, it depicts the fluctuation degree of the pedestrian flow density among multiple activity characteristics, and at the same time introduces It is used to measure the cumulative effect of dwell time in the overall activity statistics, thereby taking into account the combined impact of crowd distribution and length of stay in the same indicator. In the current system or method, it can more comprehensively reflect the attractiveness and usage level of various types of spaces for crowd activities.

[0022] The steps to obtain are: This parameter indicates the The crowd density corresponding to the activity feature is first located in the crowd activity feature data obtained above. The activity characteristics are in a specific time period and spatial segment within the monitoring cycle, and then the corresponding number of people per unit area is extracted from the crowd monitoring records of the segment. The preliminary value of the crowd density is obtained according to the frame acquisition frequency and the number of recognized human outlines set in the early stage. Then, the records with large deviations that may be affected by factors such as strong light or low brightness at night are compared again, and the trend of the number of people and the spatial range at adjacent moments are compared. If the recognition result shows continuous jumps, it is reconfirmed, and finally the confirmed crowd density value is archived in the first In this implementation, the number of similar activity features actually collected is usually between 50 and 200, and the software can analyze them in turn. In this example, three types of activity features A, B, and C can be selected. The mean density recorded in each activity feature is 2.1 people / m2, 1.7 people / m2, and 2.8 people / m2, respectively. Therefore, we can get , , Equal values.

[0023] The steps to obtain are: This parameter represents the average density of all activity features. The sequence is summed and divided by the number of features To ensure that the average value reflects a more realistic level of traffic, it is necessary to exclude the crowd data that cannot be identified in place or has ghosting accumulation in extreme cases, and only retain one valid value for the part that is repeatedly recorded. In practice, Sort and check the values ​​that differ greatly from the median within a certain range, and then confirm whether these values ​​correspond to abnormal peaks of traffic. If they are real peaks, they should be retained, otherwise they should be eliminated. In this way, The calculation is closer to the actual monitoring situation. When the final confirmation is made, all the reserved values ​​are added up and divided by Get , the example can be found in , , Under the conditions of 。

[0024] The obtaining steps of This parameter represents the duration of the th activity feature. It needs to be statistically calculated based on the specific residence duration of people monitored in the previously obtained pedestrian activity feature data. It can be obtained by accumulating the difference between the entry and exit times of each independent group or individual within the monitoring range. To prevent double counting caused by the same person entering the same range multiple times, it is necessary to identify duplicate identities in the preliminary trajectory tracking data, eliminate overly overlapping residence periods, and then summarize all the valid duration records under the same feature, obtain the total duration by summing, and correspond to the number of occurrences in the same feature. Calculate the average value, which can be regarded as 、 、 and other values are substituted into subsequent operations.

[0025] The obtaining steps of This parameter represents the comprehensive value of the average duration of all activity features. It is necessary to summarize the obtained previously and divide by the total number of activity features . Since some features may come from statistics in different time periods, in order to avoid deviating too much from the actual situation due to the presence of data with extremely long or short stays under a certain feature in extreme cases, obvious abnormal records can be eliminated before calculation, and then the sum of the remaining durations is divided by . Similar to , this step also includes reconfirming and filtering outliers. In the example, according to the aforementioned 、 、 , can be obtained.

[0026] The obtaining steps of This parameter represents the number of activity features. Usually, when the monitoring area and monitoring period are fixed, the total number of different activity features can be directly determined through the previous statistics. Each activity feature can be divided by the main activity type and the corresponding monitoring period. For example, in a certain space, there are morning exercise activities, short lunch breaks, evening art performances, etc. These independent activity situations are regarded as different activity features respectively. After all activity features are established and invalid data are screened out, , in the example, statistics are carried out through the above three activity features A, B, and C. Therefore, the value of

[0027] Calculation process: The first step is to calculate , substitute , , and , and get: Then perform a square operation and divide by : The second step is to calculate , substitute , , and , and get: The third step is to calculate : Finally, multiply the two to get: This result indicates that the spatial activity statistical value is approximately 1.0776. The higher the value, the more obvious the fluctuation of the pedestrian flow density in the space during the monitoring period and the greater the impact of the staying time on the overall activity level. When this value is greater than 1, it can usually be regarded as a medium or above activity intensity level. If it is in the range exceeding 2 or 3, it often indicates that there is a relatively dense and continuous aggregation in the area during the monitoring period. On the contrary, if this value is lower than about 0.5, it means that the overall activity is relatively dispersed or the staying time is relatively short. Therefore, in the subsequent steps, this value can be combined with other statistical indicators for more in-depth urban planning analysis.

[0028] The steps to obtain the spatial usage intensity distribution value are as follows: Based on the spatial activity statistical value, calculate the spatial usage intensity distribution value of each space type. The calculation formula is: Among them, is the spatial usage intensity distribution value, is the spatial activity statistical value of the i-th spatial type, is the average activity duration of the i-th space, is the number of spatial types, is the arithmetic mean of the sum of all spatial activity ranges.

[0029] Specifically, based on the obtained spatial activity statistical values, by classifying and processing the recorded data within the same monitoring period and confirming its validity, first group these data according to different types such as public squares, parks, and pedestrian streets and archive them in the time dimension. Then set a comparison interval for each grouped data, retrieve the crowd aggregation level and the continuous active situation of the crowd within the corresponding interval. If duplicate counting is found, cross-check and exclude it by combining the records of the previous and subsequent time periods. Then count the population scale at different time periods in each spatial type and divide it into several levels in combination with the activity area to identify the distribution state and relative density of crowd activities within this type of space. Subsequently, segment the monitored activity trajectories on this basis, mark the start time and end time of each segment of the trajectory, so as to obtain the average residence duration of each spatial type at each time period. Then compare these average residence durations with the coverage range of the activity trajectories, check each record with an abnormally long or short stay time item by item and confirm that the personnel of this record have indeed entered or left the monitoring range. Finally, perform duplicate removal and correction processing on all the confirmed data, and statistically obtain the crowd distribution range covered by each spatial type during the monitoring period and the corresponding average activity duration.

[0030] The advantage of the formula is that it incorporates the spatial activity statistical value, the average activity duration, and the overall scale of the spatial activity range into the operation together. Through the cooperation of exponential operation and arctangent function, the result can more three-dimensionally display the utilization level of each spatial type based on the existing crowd activity level.

[0031] The acquisition steps of are as follows: This parameter represents the average activity duration of the -th space. Based on all the stay records of this space during the monitoring period, calculate the difference by subtracting the entry and exit times statistically obtained before, and correspond the difference to the occurrence frequency to obtain the overall average residence duration of the crowd in this space. During data collection, the data integrity will also be further verified. For example, check and exclude the obviously erroneous parts of the stay records during the period with more night noise in the monitoring video. Then sum up all the valid stay times and divide by the number of available records to obtain this average activity duration. For example, for the square space, it is calculated that minutes, the park space is calculated to obtain minutes, etc. For a complete example of obtaining: Set a sensing device in the square area to automatically record the entry time and departure time. By performing a time difference operation on each recorded data, removing the situation of repeated entry and exit within one minute or staying overnight, a large number of residence time values are obtained. Finally, the average value is summarized to obtain the .

[0032] The obtaining steps are as follows: This parameter represents the number of spatial types, which is determined by clarifying how many types of public space actually exist within the monitoring range. For example, if there are both squares and parks, as well as pedestrian streets or green landscape areas, after classifying and labeling all monitoring areas in the early stage, the number of label types is counted as , and it needs to be updated when a new spatial type is included in the monitoring and the data is reorganized. In practice, common urban areas usually contain 3 to 5 main types of public space. It is confirmed in combination with the official spatial classification list given by the regional planning department. For example, if three types of public space are statistically obtained locally, namely . For a complete example of obtaining: Obtain a list of public squares, parks, and pedestrian streets in a certain area from the municipal management agency in advance. If it is confirmed that the three are the main types of public space, then directly let .

[0033] The obtaining steps are as follows: This parameter represents the arithmetic mean of the sum of all spatial activity ranges. Usually, first vectorize the population activity areas of each spatial type during the monitoring period, obtain its occupied boundary and calculate the area, then process other spatial types in the same way and add the area values, and finally divide by the number of spatial types , and then can be obtained. To improve the accuracy, it is usually necessary to measure and summarize at multiple different time periods, excluding the ineffective areas of the flow of people caused by temporary construction or closure. In actual urban planning, each public space may correspond to an effective activity area of tens of thousands to hundreds of thousands of square meters. After zoning and overlaying these values in the geographic information system and then averaging, the obtained can be used as a key parameter to measure the available range of the overall space. For example, according to the actual measurement results in the early stage, after calculating the areas of the effective activity areas of the three types of squares, parks, and pedestrian streets respectively, the total is 160,000 square meters, and the average value is square meters. For a complete example of obtaining: Use drones to photograph and compare with geographic information for the three types of public space respectively to obtain the exact activity area boundaries, calculate the area values in blocks, add all the blocks and then divide by 3 to obtain .

[0034] Calculation process: In the first step, first calculate , take in the example, let , , , , and according to what is measured in another space , , then In the second step, divide this result by : In the third step, perform a cube root operation: In the fourth step, combine to execute 's calculation. In the example , when calculating , it usually needs to be transformed in a numerical analysis tool. First, it can be approximately processed by looking up a table or converting 53333 to radians during the calculation to obtain , In the fifth step, multiply the two: This result shows that the spatial usage intensity distribution value is approximately 5.984. When this value is in the range of 2 to 6, in most cases, it represents that the activity distribution of this space type during the monitoring period is relatively balanced and has a certain ability to attract people. If it exceeds 10, it often indicates extremely high-intensity use or large-scale continuous concentration. If the value is too low, it shows that the utilization status of this space is weak. The obtained through this formula is helpful for subsequent spatial resource allocation and management strategy planning.

[0035] Among them is the spatial usage intensity distribution value. Based on this result, the current usage status of different public spaces can be further monitored and compared. The specific approach is to classify and categorize the obtained spatial usage intensity distribution values. First, a list of numerical intervals is set to define the reference ranges for each category. For example, in the interval list, it is defined that 0 to 2 represents a weak usage intensity, 2 to 6 represents a general usage intensity, 6 to 10 represents a high usage intensity, and greater than 10 is recorded as an extremely high usage intensity. Then, each monitored space type is matched and statistically analyzed with its corresponding distribution value. The results falling into different intervals after matching are screened one by one. By checking the control level and actual traffic guidance ability of this area in the urban management system, inappropriate records are excluded. Manually correct and re-verify the data points with suspected extremely low concentration. Mark the part exceeding the upper limit of the preset interval as a high saturation value in the corresponding record. Summarize all these matching results and view them item by item according to the space type. Determine the usage intensity level of each space type with reference to the classification statistical table and record its corresponding duration and number of days. During this process, the geographical outline of the area exceeding the specified density area can be marked with actual geographical coordinate information as needed. The accuracy of the monitoring results is verified by measuring the boundary coordinates and verifying the area of the effective activity area. Finally, summarize the space types and their usage intensity distribution values within each category to obtain the final classification record.

[0036] The steps for obtaining the facility distribution density data are as follows: Based on the spatial usage intensity distribution value, collect the geographical coordinate information of the rest facilities, landscape facilities, and barrier-free facilities in each type of public space to obtain the facility geographical coordinate data; Based on the facility geographical coordinate data, calculate the facility distribution density. The calculation formula is: Where, is the facility distribution density, is the geographical coordinate of the th facility, is the central coordinate of the crowd gathering point, is the peripheral crowd density of the

[0037] th facility,

[0038] Specifically, the monitoring area range is divided according to the space utilization intensity distribution values obtained previously. Combining the lists of rest facilities, landscape facilities, and barrier-free facilities provided by the urban management department, the identification numbers of each type of facility are matched with the corresponding public space categories and their locations are retrieved. According to the previously established coordinate acquisition rules, the longitude and latitude of each facility are recorded in sequence. Accurate geographical coordinates are collected by setting up measuring devices capable of recording location information at the facility site. When the coordinates recorded by the measuring device deviate significantly from the pre-set reference coordinate range, they are marked as suspicious data and compared again in subsequent steps. Subsequently, a basic information table for rest facilities, landscape facilities, and barrier-free facilities is established, and the measured geographical coordinates are associated with the corresponding facility numbers in this basic information table. In case of duplicate or missing coordinates, manual comparison is required to determine whether they are the same facility or the coordinate records are incomplete, and then re-measurement and confirmation are carried out. After confirming the valid data, the collected coordinates are converted through a unified coordinate system to be consistent with the urban map system, and then the coordinates of all facilities are stored separately and incorporated into the subsequent operation preparation. For facilities deployed in remote or underground areas, additional positioning methods are used to confirm their specific longitude and latitude to ensure the integrity of the records. For areas with occlusion or signal obstruction, multiple measurements are accumulated and cross-calculated. The results of each measurement are aligned with the reference geographical benchmark and re-field surveys are carried out when the range exceeds the specified value. Finally, all facility coordinate records are numbered and summarized according to the classification of rest facilities, landscape facilities, and barrier-free facilities to obtain the facility geographical coordinate data.

[0039] The advantage of the formula is that it incorporates the coordinate differences between facilities and human flow aggregation points and the surrounding human flow density into a single metric. It can not only reflect the simple geographical distance but also dynamically measure the impact of human flow distribution on the density of facility layout, thus providing a distribution assessment closer to the actual usage situation for urban planning departments on a unified scale.

[0040] The acquisition steps are as follows: This parameter represents the longitude value of the Among them, the example acquisition method is to lay a positioning device in the center of the square for a certain rest facility, record its longitude value as 114.386752, and after comparing with the geographical reference, confirm that the value is valid, and is assigned as 114.386752.

[0041] The acquisition steps of This parameter represents the latitude value of the th facility in the geographical coordinate system. The acquisition process is similar to , and it is necessary to read the corresponding latitude value from the facility's geographical coordinate data. If the latitude results of the same facility measured multiple times show significant inconsistencies, on-site positioning checks need to be carried out again to exclude interference errors. Usually, cross-comparisons are made according to the urban surveying and mapping database or high-precision remote sensing images to determine the final valid latitude value and write it into . For example, the latitude value 30.512907 can be measured at the location of the same rest facility, and after error verification, it is used as the of this facility. A complete acquisition example is to record the latitude of the facility at three time periods: early, middle, and late in a day. If the difference degree of the three recorded values is lower than 0.000010, the final value is confirmed by comparing with the map reference data and registered in .

[0042] The acquisition steps of This parameter represents the longitude of the center coordinate of the crowd gathering point. It is necessary to first identify the area of the activity hotspots monitored in the early stage. By viewing the set of geographical locations with the highest population density within a period of time, calculate the weighted center of these hotspots and use the weighted population scale as a reference. Perform a combined operation on the weighted results of multiple hotspots, and finally select the center longitude value that best represents the personnel gathering as . In the actual situation, the crowded location distribution of a certain square or commercial street can be obtained from the urban monitoring system, and then the overall gathering center longitude is determined by accumulating and weighting the longitude set. In the example, it is detected that the peak points of the crowd in a certain area of the square are concentrated around 114.389120. After multiple statistics, it is found that its fluctuation is small, so 114.389120 is registered as .

[0043] The acquisition steps of This parameter represents the latitude of the center coordinate of the crowd gathering point, which is the same as The acquisition method is similar. It is necessary to extract the average position of the peak area based on the hotspot distribution within the same time or continuous time periods, weight it with the pedestrian flow scale, and then compare it with the actual surveying and mapping data to obtain the most representative central latitude. In the example, the peak area of the pedestrian flow in the aforementioned square can be taken near latitude 30.513300. After verification in multiple time periods, it is found that the fluctuations are within the range of 30.513290 to 30.513310. Therefore, 30.513300 is used as .

[0044] The acquisition steps of This parameter represents the pedestrian flow density around the th facility. It is necessary to periodically count the number of people within a certain radius around the facility while measuring the facility coordinates, identify the number of people in this range through installing counting devices on-site or through urban monitoring, take the number of people per unit area as the benchmark of the pedestrian flow density, and at the same time combine the changes in the pedestrian flow at different times of the day, and then take their average value to form . In the example, the number of people in the area within a radius of 20 meters around a barrier-free facility is counted multiple times. After each record, the number of people is divided by the area to obtain the pedestrian flow density, and then the sum is accumulated and averaged over several time periods in a day. Assuming that the final recorded density result is 2.5 people per square meter, then can be obtained.

[0045] The acquisition steps of This parameter represents the total number of facilities. It is necessary to classify and number all the rest facilities, landscape facilities and barrier-free facilities within the monitoring range, and the total number of the numbers can be obtained to get . Usually, first use the urban facility management database to obtain all the recorded public facilities, then supplement and record the newly added facilities that are not recorded and merge them into the final list to confirm the total number of facilities. If duplicate numbers or redundant registrations are found during the confirmation process, they will be cleaned up, and finally this number is determined. For example, the facilities in three types of areas, namely squares, parks and pedestrian streets, can be summarized. If a total of 150 facilities are counted, then can be recorded.

[0046] Calculation process: In the first step, process the absolute coordinate difference and the surrounding density. Let in the example , the coordinates of facility 1 are , , and its surrounding pedestrian flow density is . The coordinates of facility 2 are , , and the surrounding pedestrian flow density is . The coordinates of facility 3 are , , and the surrounding pedestrian flow density is ; Central longitude , Central latitude ; Second step, substitute the differences in the formula in sequence: ; ; ; Third step, divide it by : Fourth step, square each result and sum them up: Fifth step, take the square root of the sum result: This result indicates that the facility distribution density is approximately . The higher this value is, it usually means that there are more facilities near the central gathering point and the surrounding pedestrian flow density is relatively low. If this value is small, it means that the distance between the facilities and the pedestrian flow hot spot is large or the surrounding pedestrian flow density is high. In the urban planning process, this result can be combined with the space usage situation to identify whether the public facility configuration matches the pedestrian flow demand.

[0047] Based on the completed facility distribution density calculation process, compare the calculated distribution densities of facilities in each public space with each other, and distinguish them by facility type. List the distribution density values calculated for rest facilities, landscape facilities, and barrier-free facilities separately. First, check item by item whether there are extremely high or low distribution density values under the same type. By comparing with the existing local urban basic database, if it is found that a certain distribution density deviates from the normal range of the same type of facilities in the same period, then call the coordinates and pedestrian flow monitoring information collected on-site to review again to exclude the influence of accidental observation errors or duplicate registration records. Subsequently, sort and summarize all the confirmed distribution density values, and set corresponding numerical ranges for each type of facility. For example, select 0.001 to 0.005 as the medium distribution density for rest facilities, 0.0005 to 0.003 as the medium distribution density for landscape facilities, and 0.0002 to 0.001 as the medium distribution density for barrier-free facilities. Then mark the records below this range as sparse distribution and the records above this range as dense distribution. Finally, normalize these records according to a unified standard. By calculating the relative density ratio within each facility type, map the distribution density values to the interval of 0 to 1, and record them in the corresponding normalization result table. Finally, summarize and record the relative density values of each type of facility in the form of a graph or list to obtain the facility distribution density data.

[0048] The steps to obtain the facility service coverage index are as follows: Based on the facility distribution density data and the spatial usage intensity distribution values, determine the service coverage range of each facility. Taking the facility geographical coordinates as the center, set the influence radius according to the facility type to generate the facility coverage range data; Based on the facility coverage range data, identify the activity hotspots, and judge the spatial overlapping area between the facility service area and the activity hotspots to obtain the overlapping area data of the facility and the activity hotspots; Based on the overlapping area data, calculate the facility service coverage index. The calculation formula is: Among them, is the facility service coverage index, is the th overlapping area of the facility and the activity hotspots, is the th spatial usage intensity distribution value of the area where the th facility is located, is the service coverage range of the th facility, is the arithmetic mean of the overlapping ratios of all facility coverage ranges and the activity hotspots,

[0049] Specifically, based on the facility distribution density data and spatial usage intensity distribution values obtained previously, combined with the urban supporting facility type registration list, determine the corresponding influence radius for each type of facility. The setting of the influence radius can refer to the public service radius standard formulated by the urban planning department. First, read the type information of the rest facilities and query their normal usage coverage range, and then perform the same query process for the landscape facilities and barrier-free facilities. Record the influence radius corresponding to each facility type in the facility basic information. Subsequently, retrieve the location of each facility in the coordinate system, and combine the influence radius information of this facility type to draw a circular range or a polygon approximation range with the facility location as the center and a radius equal to the registered value in the map system. If it is found that there is a significant discrepancy between the original influence radius and the actual site conditions, call the on-site measurement records for comparison and revise the value of the influence radius. Register all the revised ranges and accumulate the coverage range data of all facilities. When segmentally displaying the drawn coverage ranges, layer management can be performed according to categories. By checking the overlapping situation between the coverage circles (or polygons) and the ground feature boundaries, confirm whether the service range shown by each facility on the map conforms to the initial positioning information. For facilities with marked offsets or coordinate drifts, re-calibrate and update the coverage range information by means of repeated measurement and manual verification. Correlate all the coverage ranges with the original facility coordinates one by one and record them in the database. Finally, list the coverage ranges of all facilities in the map display interface or text list.

[0050] Based on the facility coverage range data, summarize and display the distribution maps of the ranges of each facility and the activity hotspots formed during the monitoring period. Mark the boundaries of each hotspot one by one according to the coordinate shapes extracted previously, and overlay them with the coverage ranges of each facility. Then, calculate the intersection areas between the facility ranges and the boundary lines of the hotspot areas item by item. Perform area calculations on the polygons existing in the hotspot areas and the facility coverage circles or polygons. For the case of multiple overlaps, accumulate the actual areas of the overlapping parts by blocks, and compare the statistical results with the hotspot boundary areas previously recorded in the urban basic database to confirm whether there are situations such as excessive coverage or low coincidence degree. If it is found that the overlapping area exceeds the predetermined standard, mark it as a high coincidence degree in the corresponding coverage range data and conduct subsequent manual analysis. For the cases of significantly low overlap or non-overlap, check whether the values of the facility coverage range are set unreasonably or there are deviations in the delineation of the hotspot areas. Check the time periods when the hotspots appear and the available time periods of the facilities in the comparison records to identify whether the low spatial overlap degree is caused by mismatched time segments. If it is confirmed to be correct, directly archive the overlapping area values obtained this time. If there are still doubts, retrieve more detailed monitoring information for supplementary statistics. Finally, list or visually display the overlapping areas between all facility ranges and hotspot areas one by one, and obtain the overlapping area data between the facilities and the activity hotspots.

[0051] The advantage of the formula is that it simultaneously introduces the overlapping area and the spatial usage intensity distribution value for weighting, and through the joint operation of the service coverage range and the overlapping ratio, comprehensively measures the coverage matching degree of each facility to the activity hotspots, making it easier for urban planning to grasp the balance between the facility layout and the population needs.

[0052] The acquisition steps of This parameter represents the th facility's overlapping area with the activity hotspot. It needs to be quantified based on the superposition of the generated facility coverage range data and the activity hotspot area boundary. The monitoring personnel can perform geometric operations on the facility coverage map and the hotspot map, calculate the area of the overlapping polygon range, and usually use numerical integration or the discrete grid method to obtain the exact value. If an irregular boundary is encountered, it needs to be disassembled into several smaller sub-polygons and calculate the area separately, and then add the areas of the sub-polygons to obtain the final overlapping area. In real urban planning, the facility coverage range often presents as a circle or an irregular polygon, and the activity hotspot area may also be irregular. Therefore, the polygon Boolean operation of the map system can quickly obtain the overlapping polygon and output the corresponding area, which is finally recorded as . For example, when measuring the overlapping situation of the pedestrian street rest facility with the afternoon crowd hotspot, if the overlapping polygon area is 150 square meters, it can be recorded as .

[0053] The acquisition steps of are: According to the previously obtained spatial usage intensity distribution value, use this value as

[0054] The acquisition steps of This parameter represents the th facility's service coverage range, that is, the area value obtained based on the influence radius or the actual coverage boundary set by the facility type before. Through geometric methods, calculate the area of the circle or polygon formed by the coverage range. If the actual layout of the facility is not an ideal circle, it is equivalently segmented into multiple regular regions for area superposition or the rasterization method is used to subdivide and count the edges, and the sum of the areas of all subdivided blocks can be used to obtain . After the summarization is completed, mark this value in the service coverage range parameter of the th facility. For example, if the coverage radius of a landscape facility is set to 40 meters, if it is simplified to a circle, the circle area square meters can be directly calculated, and it is recorded as .

[0055] The acquisition steps of This parameter represents the arithmetic mean of the overlapping ratio between the coverage of all facilities and the active hotspots. First, it is necessary to calculate the ratio of the area overlapping with the hotspot area within the coverage of each facility to the total area of its own coverage. Then, sum up the overlapping ratios of all facilities and divide by the total number of facilities. , forming an average value, which can comprehensively reflect the proportion of the overall facilities covering the hotspot area. For example, if the coverage areas of two facilities are 400 square meters and 500 square meters respectively, and the overlapping parts with the hotspot are 100 square meters and 150 square meters respectively, then the overlapping ratios are 0.25 and 0.30 respectively. If there are two items in total, then Registered in Among them.

[0056] The acquisition steps of This parameter represents the total number of facilities, which is used to identify in the calculation formula how many facilities' coverage areas and related data need to be included in the statistics. Usually, all rest facilities, landscape facilities, and barrier-free facilities have been counted and the total has been obtained previously. One can refer to the numbered quantity in the facility basic database or conduct a full-scale scan of the database records in large-scale scenarios to avoid duplicate statistics or omission of records. For example, if the number of monitorable facilities registered in a city center area is 45, then it is recorded as .

[0057] Calculation process: The first step is to calculate , for example, take , let square meters, , square meters, , square meters, , then: The second step is to calculate , for example, the coverage area of Facility 1 is square meters, the coverage area of Facility 2 is square meters, and the coverage area of Facility 3 is square meters, then: The third step is to perform a division operation on the two and take the square root: The fourth step, The calculation of , corresponding to the previously obtained , then: Step 5: Multiply the two: The results show that the facility service coverage index It is approximately 0.007226. The larger the index value is, the more efficient the spatial overlap between facilities and activity hotspots is, and the surrounding people are more likely to enjoy the services of various public facilities. When the value is small, it means that the overall coverage is low or the hotspots and facility layouts are misaligned, and the actual service effect is limited. If the value is found to be above 0.01 in the later statistics during use, it can usually be regarded as good coverage. When it is lower than 0.005, it shows that the coverage is weak.

[0058] The steps for obtaining the group interaction dataset are: Based on the facility service coverage index, determine the impact range of facilities in public spaces on residents’ group activities, collect individual trajectory data of residents within the impact range, organize the trajectory data in time series based on timestamp information, and generate trajectory data of residents’ group activities; Based on the activity trajectory data of resident groups, the activity intersection areas of different groups are identified, the length of stay and proximity of group members in the intersection area are extracted, and the number of interaction events between individuals is calculated in combination with time segments to generate group interaction frequency data; Based on the group interaction frequency data, the interaction events were classified by time series, outliers were removed and the data were standardized, and cluster analysis was performed according to group size and interaction density to generate a group interaction dataset.

[0059] Specifically, based on the numerical records corresponding to the facility service coverage index, all resident activity identifiers within the coverage range in the urban public space are retrieved, the location information of these residents and the relevant personnel identification codes are extracted, and the identification codes are associated with the time stamp sequences during the monitoring process. Subsequently, based on the facility distribution information registered in the urban planning department previously, the coverage boundaries to which each facility belongs and the labels of the current public area are determined, and the resident individuals who are active within this label range are screened out and their entry times and exit times are marked. If the individual's stay time is less than the threshold within the pre-set minimum stay duration threshold, subsequent analysis is excluded. This threshold can be determined by the accumulation of urban management experience. For example, after long-term monitoring, it is found that the average effective stay duration in some areas is more than 3 minutes, so 3 minutes is used as the minimum stay duration threshold. Then, the complete trajectories of the individuals retained through screening within the influence range are recorded in chronological order, and the location information is calibrated to reduce multipath interference. For the individual records with too fast coordinate change rate or exceeding the limit speed of the monitoring area, they are reviewed by the speed judgment method of adjacent coordinate points. The trajectories with suspected excessive speed are classified as abnormal records and are corrected by comparing with the surrounding population distribution in the subsequent statistics. The records that still do not make sense are excluded. Then, the remaining trajectories are sorted by time stamps and the locations with concentrated stays are detected segment by segment. Each location segment is recorded in the form of longitude and latitude and is accompanied by the time length. If the same individual appears in the coverage range of the same facility multiple times, their multiple time periods are respectively checked and whether there is an overlap in time is examined. The overlapping parts are merged to prevent double counting. After integrating the resident activity trajectories within all facility ranges, the trajectory of each individual is segmented into continuous spatio-temporal segments and the entry time and departure time are marked. For the cases that may cross day and night or time zones, unified conversion is performed to ensure data consistency. All the processed trajectories are grouped according to the individual identifiers and stored in the database. Finally, all the grouped trajectories are summarized to form the resident group activity trajectory data.

[0060] Based on the trajectory data of the resident groups, retrieve the time-series distribution maps of the spaces occupied by different groups, and list separately the coordinate areas where there are intersections. Associate the identification codes of the people who appear in multiple groups within the same time segment in these areas, calculate the start and end times when these people stay in the intersection areas, and summarize the results in the form of time differences as the stay duration. If the stay duration exceeds a certain preset threshold, add it to the queue for subsequent calculation of interaction events. The threshold can be set by referring to the average value of the pedestrian flow stay distribution obtained previously or according to the shortest social contact duration confirmed by the urban management department. For example, set 5 minutes as the stay threshold. Then, measure the detection distance of the people who meet the stay duration requirement in all intersection areas. By comparing the movement trajectories within the intersection coordinates, calculate the degree of proximity between each individual and other individuals. A distance range can be set as the judgment basis. For example, in a pedestrian street area, consider the actual distance between individuals less than 1 meter as highly proximate, less than 3 meters as moderately proximate, and greater than 3 meters as lowly proximate. After statistical analysis for each time period, obtain multiple proximity degree data, and then record the duration of proximity between individuals in combination with specific time segments, so as to count the number of interaction events between each group of individuals. After establishing the association between the number of times and the corresponding individual identifiers, classify and summarize them. If some individuals continuously maintain a proximate state and are similar in multiple time segments, annotate them as high-frequency contact objects in subsequent records. Finally, sort the interaction event information in all time segments according to the location coordinates, time sequence, and personnel identifiers, and remove invalid or interfering data to obtain the group interaction frequency data.

[0061] Based on the group interaction frequency data, read the timestamps and personnel identifier information of each interaction event line by line and classify them according to hours or finer time intervals. Then, match the identifiers to the group size information to judge the number of members in different groups, and calculate the total number of interactions that occur among multiple members in the same time period. If there are a very small number of data that deviate significantly from the normal range, mark them as outliers and check them. If it is confirmed that there are mistakes in the coordinate or timestamp records, exclude this piece of data. When standardizing, the interaction event frequency can be mapped using linear normalization or normalization range division methods. For example, map the part with the number of events less than 5 to a lower value range, and map the part greater than 20 times to a higher value range. Then, perform clustering analysis on the data according to the numerical values of group size and interaction intensity. The details of the clustering can be used to judge the similarity between groups based on the distance measurement criteria. If the interaction frequency of a certain group exceeds 20 in most time periods and the group size is more than 10 people, it can be separately classified into the high-interaction large-scale group. Then, classify the small-scale low-interaction groups into the corresponding categories according to the same classification logic. Repeat the clustering process until all data are divided into appropriate groups. Finally, attach clustering labels to each group and record them in the group interaction event list to obtain the group interaction data set.

[0062] The steps for obtaining the spatial interaction heat network are as follows: Based on the group interaction dataset, extract the geographical location, interaction time, and frequency information of group interaction events, classify the data according to spatial coordinates, and conduct segmented statistics in combination with the time dimension to generate interaction spatial distribution data; Based on the interaction spatial distribution data, perform spatial clustering, analyze the change trend of interaction density in different regions, and divide the regions into high-interaction-frequency areas, medium-interaction-frequency areas, and low-interaction-frequency areas according to the concentration degree of interaction events to generate data on hot spots and cold spots of social activities; Based on the data of hot spots and cold spots of social activities, construct the connection relationship of spatial interaction nodes to generate the spatial interaction heat network.

[0063] Specifically, based on the geographical location and time series information in the group interaction dataset, extract the specific coordinate points where each interaction event occurs, as well as the corresponding interaction time and frequency. By retrieving the longitude and latitude of each record and combining the previously stored personnel identification information, confirm which individuals appear at similar coordinate positions within the same time period. Subsequently, classify this information using the spatial coordinates as the index. If it is found that the interval between some coordinate data is too large or there are jumps in adjacent records, then use the geographical coordinate distance as the judgment basis to assign them to different coordinate groups. For the splitting of the time dimension, refer to the set upper limit of the time interval. For example, divide 24 hours a day into several fixed time periods, or separately consider the morning rush hour and evening rush hour as one segment according to urban management experience, and combine the non-peak hours into another segment. Compare all records with these segments in sequence to confirm which coordinate group and time period each interaction event specifically falls into. If some interaction events cross the segment boundary, then split them into multiple time slices and divide the frequency according to the time slices. For records with interaction frequencies significantly exceeding the normal level, such as group interactions occurring more than 100 times within 10 minutes at the same location, it can be regarded as a highly concentrated social gathering. It is necessary to further check whether the location has the characteristics of being able to accommodate a large number of people in the actual scenario. If not, further investigate whether there are duplicate data records or interference. If the data is confirmed to be correct, then mark this coordinate and the corresponding time period as high interaction degree. For records below the normal range or with extremely small data volume, confirm their validity by comparing the number of interaction events, and eliminate false interactions caused by positioning errors or duplicate markings. Finally, after summarizing all coordinate groups, form the interaction spatial distribution data integrated according to the coordinate position and segmented time.

[0064] Based on the interaction space distribution data obtained in the previous step, the interaction event statistics of each coordinate group at different time periods are arranged in sequence. By comparing the statistics, it can be identified which coordinate groups have accumulated more interaction events in a relatively short period of time. If the interaction event values of certain groups continuously exceed a certain established threshold, it can be considered that the interaction frequency is relatively intensive in urban planning practice. The corresponding threshold is usually set according to the data distribution obtained from previous monitoring. For example, take the average value of interaction events at the same time period every day in the previous observation plus twice the standard deviation as the threshold, and then use this threshold to determine which coordinate groups are in the high interaction event interval. The coordinate groups that meet or exceed this interval are classified as high interaction frequency areas, those near the average level are listed as medium interaction frequency areas, and those significantly lower than the average are listed as low interaction frequency areas. For individual abnormal points, if the value is too high but the time range is extremely short, further review is carried out and its reasonableness is judged by comparing the number of people that can be accommodated on-site with the space size. If no error is confirmed, it is retained in the high interaction frequency area. If it is excluded, it is marked as an outlier and not included in the formal statistics. Finally, all coordinate groups and the corresponding time periods are divided, and the high, medium, and low interaction areas are listed separately. If some areas maintain a high interaction frequency in multiple time periods, they are merged into a high interaction hot spot area, and the areas and time periods with continuous low interaction are marked as cold spot areas. Finally, all high interaction frequency areas and low interaction frequency areas are recorded and output to form social activity hot spot area and cold spot area data.

[0065] Based on the social activity hot spot area and cold spot area data, the coordinate groups in each hot spot area are numbered and registered. These numbers are connected to each other according to their geographical locations, and node connection relationships are established for the situations where there is continuous high interaction between adjacent coordinate groups or the same group repeatedly appears at multiple adjacent coordinate points. If it is found that the same group or multiple high interaction events appear in two adjacent hot spot areas, a strong connection is established between these two hot spots. If there are only individual event connections and the frequency values are low, a weak connection is established. According to the connection strength, lines of different thicknesses or colors can be presented on the map. If a hot spot has weak interactions with multiple cold spot areas, it is marked as an edge connection. Further, the node connection relationships between all hot spots are grouped and summarized to form a larger interaction network structure. When identifying nodes, the geographical coordinates can be aggregated into grid units and the node weights can be defined according to the total interaction events in the grid. If the cumulative event value of a node is higher than a certain statistical quantile, it is set as an important node. Then, the connection relationships between these important nodes and other surrounding nodes are divided into strong or weak connections. For cold spot areas where no significant interactions are detected, only simple records are made between them and other nodes, and no strong nodes or strong lines are established. Finally, all node information and line relationships are merged and the spatial interaction heat network is output.

[0066] The steps for obtaining the spatial emotion distribution data are as follows: Based on the spatial interaction thermal network, determine the group interaction active areas of each public space, collect the facial expression feature data of the groups in the active areas, and perform facial expression classification through a convolutional neural network model to generate group facial expression feature data; Based on the group facial expression feature data, organize it according to the time series and spatial position, conduct aggregated analysis on the individual facial expression features, extract the main emotion categories and their proportions of the groups in different areas, and generate a group emotion feature matrix; Based on the group emotion feature matrix, conduct statistics according to the spatial areas, calculate the proportions of various emotions in different areas, integrate the information of the area range, interaction activity, and emotion categories, and generate spatial emotion distribution data.

[0067] Specifically, based on the node connection relationships recorded in the spatial interaction thermal network, retrieve the coordinate positions of each group interaction active area in the public space, and mark the time period information of each active area. Perform face recognition and positioning in the camera monitoring records through these coordinates and time information. After obtaining the facial area of the corresponding individual, use the pre-collected and clearly labeled facial expression samples as the training set to provide supervised data for the classification process of the convolutional neural network. In the training stage, first perform unified grayscale and size processing on all labeled facial expression images and divide them into several sub-blocks for feature extraction. Then, define several convolutional layers and pooling layers in the neural network and extract local texture features layer by layer. Update the parameters learned in each convolutional kernel through multiple rounds of iteration, count the correct classification rate and loss function value in each round of training, and solidify the final parameters after convergence. After training is completed, use the same network to perform forward inference on the facial images collected in the active areas, output the corresponding facial expression category of each individual, such as happy, calm, nervous, etc., and then match the classification result with the time stamp and coordinates of the individual. If there are classification results of multiple individuals in the same active area and the same time period, record them in a grouped data at the same time. After grouping, recheck the accuracy of the facial expression classification against some manually labeled images. When the accuracy meets the requirements of urban monitoring, store all the recognized results uniformly and attach facial expression category labels to generate group facial expression feature data.

[0068] Based on the group expression feature data, screen and classify the expression categories and corresponding frequencies generated by different individuals at each time period and coordinate position. By listing and summarizing all the expression information in the same area within adjacent time periods item by item, identify multiple expression categories that appear in this area and the corresponding number of people. Then use time series to segment these expression records, set the time period division interval to a certain number of minutes or hours, and an average duration standard can be generated based on the data accumulated by urban monitoring to define the length of the time period. Combine the expression categories and the number of individuals within the same time period into a set, and calculate the ratio of the occurrence frequency of each expression category to the total frequency for each set. If the ratio of a certain expression in the same area and the same time period is significantly higher than other expression categories for many consecutive times, mark this expression category as the main emotion category, and indicate its proportion and duration in the record table. Arrange the main emotion categories and proportions of all areas and time periods in sequence. After processing, these data can be converted into a matrix structure with rows representing areas and columns representing emotion categories, and each cell stores the ratio occupied by the corresponding category. When the expression samples in individual areas are extremely few or invalid due to insufficient people flow, mark them as missing values or extremely low values to avoid affecting the overall analysis. Finally, complete the statistics of the main emotion categories of all areas and generate a group emotion feature matrix.

[0069] Based on the group emotion feature matrix, traverse the emotion categories in the matrix and their corresponding proportion values according to the pre-confirmed coordinate boundaries of the public space area, and record the main emotion distribution of each area at different time periods. Integrate these proportion values with the previously statistically interactive activity levels, and list the proportion of the main emotion categories of each area during peak or off-peak hours. By comparing the emotion characteristics of multiple areas, it can be determined whether the expression distribution of each area at different time periods falls within the set range. If there are proportion values that continuously deviate from the preset threshold, mark them in the corresponding record. This threshold is usually selected from the historical mean and fluctuation range of urban population statistics or social surveys. For example, based on the monitoring results of the past month, take the average proportion and upper and lower limits. If the currently measured value far exceeds this range, it is necessary to further check the source data and sample size and perform additional verification. If the samples are confirmed to be normal, include the emotion proportion data of this area in the overall summary. Finally, output a three-dimensional summary structure of time-space-emotion from the combined information of multiple areas and indicate the percentage occupied by each emotion category in the corresponding area. After integrating the area and population density of each area, spatial emotion distribution data can be formed.

[0070] The above are only the preferred embodiments of the present invention and do not limit the present invention in other forms. Any person skilled in the relevant art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. Urban planning method based on digital twins, characterized by: The following steps are involved: Collect crowd density data from public space surveillance cameras, classify and count the crowd density data according to the space types of squares, parks and pedestrian streets, and obtain spatial activity statistics; based on the spatial activity statistics, superimpose the statistical data under each space type to generate a spatial use intensity distribution value; Collect the geographic coordinate information of recreational facilities, landscape facilities and barrier-free facilities in various types of public spaces, calculate the distribution density of facilities, and obtain the facility distribution density data; based on the facility distribution density data and the spatial use intensity distribution value, analyze the facility coverage and activity hotspots, calculate the area ratio of overlapping areas, and generate the facility service coverage index; Based on the facility service coverage index, the activity trajectory data of resident groups in the public space and the interaction frequency between groups are collected to obtain a group interaction data set; based on the group interaction data set, the interaction data is spatially clustered to identify hot spots and cold spots of social activities, and a spatial interaction thermal network is established; Based on the spatial interactive thermal network, the facial expression features of groups in the public space are collected to obtain a group emotion feature matrix. Based on the group emotion feature matrix, the distribution ratios of group emotions in different areas are counted to generate spatial emotion distribution data.

2. The urban planning method based on digital twins according to claim 1 is characterized in that: The steps for obtaining the spatial activity statistics are as follows: Collect crowd density data from public space surveillance cameras and classify the data according to space types, including squares, parks, and pedestrian streets, to obtain classified crowd density data; Based on the classified crowd density data, the crowd concentration degree, stay time and activity range of each type of space are counted to obtain crowd activity characteristic data; Based on the characteristic data of human flow activities, the spatial activity statistics are calculated using the following formula: in, is the space activity statistics, is the crowd density of the jth activity feature, is the average crowd density of all activity characteristics, is the duration of the jth activity feature, is the average duration of all activity features, and k is the number of activity features.

3. The urban planning method based on digital twins according to claim 1, characterized in that: The steps for obtaining the spatial usage intensity distribution value are as follows: Based on the spatial activity statistics, the spatial usage intensity distribution value of each spatial type is calculated using the following formula: in, Use intensity distribution values ​​for space, is the spatial activity statistics of the i-th spatial type, is the average activity duration of the ith space, is the number of space types, It is the arithmetic mean of the sum of all spatial activity ranges.

4. The urban planning method based on digital twins according to claim 1 is characterized in that: The steps for obtaining the facility distribution density data are as follows: Based on the spatial use intensity distribution value, the geographic coordinate information of recreational facilities, landscape facilities and barrier-free facilities in various types of public spaces is collected to obtain facility geographic coordinate data; Based on the facility geographic coordinate data, the facility distribution density is calculated using the following formula: in, is the facility distribution density, For the The geographical coordinates of the facility, is the center coordinate of the crowd gathering point, For the The density of people around the facility, is the total number of facilities; Based on the facility distribution density, the distribution of different facility types is normalized to obtain facility distribution density data.

5. The urban planning method based on digital twins according to claim 1, characterized in that: The steps for obtaining the facility service coverage index are as follows: Based on the facility distribution density data and the spatial use intensity distribution value, determine the service coverage of each facility, take the facility geographic coordinates as the center, set the influence radius according to the facility type, and generate facility coverage data; Based on the facility coverage data, identifying the activity hotspot area, determining the spatial overlap area between the facility service area and the activity hotspot area, and obtaining the overlap area data between the facility and the activity hotspot; Based on the overlapping area data, the facility service coverage index is calculated using the following formula: in, is the facility service coverage index, For the The overlap area between facilities and activity hotspots, For the The spatial usage intensity distribution value of the area where the facility is located, For the The service coverage of each facility, is the arithmetic mean of the overlap ratios of all facility coverage areas to activity hotspots, is the total number of facilities.

6. The urban planning method based on digital twins according to claim 1, characterized in that: The steps for obtaining the group interaction dataset are as follows: Based on the facility service coverage index, determine the impact range of facilities in the public space on the activities of resident groups, collect individual trajectory data of residents within the impact range, organize the trajectory data in time series based on timestamp information, and generate resident group activity trajectory data; Based on the activity trajectory data of the resident groups, the activity intersection areas of different groups are identified, the length of stay and the degree of proximity of group members in the intersection areas are extracted, and the number of interaction events between individuals is calculated in combination with the time segments to generate group interaction frequency data; Based on the group interaction frequency data, the interaction events are classified by time series, outliers are removed and the data is standardized, cluster analysis is performed according to group size and interaction density, and a group interaction data set is generated.

7. The urban planning method based on digital twins according to claim 1 is characterized in that: The steps for obtaining the spatial interactive thermal network are: Based on the group interaction data set, the geographical location, interaction time and frequency information of the group interaction events are extracted, the data is classified according to the spatial coordinates, and segmented statistics are performed in combination with the time dimension to generate interaction spatial distribution data; Based on the interaction spatial distribution data, spatial clustering is performed to analyze the interaction density change trend in different areas, and the areas are divided into high interaction frequency areas, medium interaction frequency areas and low interaction frequency areas according to the concentration of interaction events, so as to generate data on hot spots and cold spots of social activities; Based on the data of the hot spots and cold spots of social activities, a spatial interaction node connection relationship is constructed to generate a spatial interaction thermal network.

8. The urban planning method based on digital twins according to claim 1, characterized in that: The steps for obtaining the spatial emotion distribution data are as follows: Based on the spatial interaction thermal network, the active group interaction areas of each public space are determined, the expression feature data of the groups in the active interaction areas are collected, and the expression classification is performed through the convolutional neural network model to generate group expression feature data; Based on the group expression feature data, the data is sorted according to time series and spatial position, individual expression features are aggregated and analyzed, the main emotion categories and proportions of groups in different regions are extracted, and a group emotion feature matrix is ​​generated; Based on the group emotion feature matrix, statistics are performed according to spatial regions, the proportion of each type of emotion in different regions is calculated, and the regional scope, interaction activity and emotion category information are integrated to generate spatial emotion distribution data.

9. The urban planning system according to any one of claims 1 to 8, characterized in that: include: The crowd density analysis module collects crowd density data from public space surveillance cameras, classifies and counts them by square, park and pedestrian street, calculates the number of people in different space types, counts the activity intensity of each type of space, and obtains the spatial activity statistics; The facility distribution assessment module collects the geographic coordinates of recreational facilities, landscape facilities and barrier-free facilities in squares, parks and pedestrian streets based on the statistical values ​​of spatial activities, calculates the distribution density of various facilities, and obtains the facility distribution density data; The facility service analysis module compares and analyzes the facility coverage and the hot spots of human activity based on the facility distribution density data and spatial activity statistics, counts the overlapping area of ​​the two, calculates the coverage index, and generates the facility service coverage index; The social interaction analysis module collects the activity trajectory data of resident groups and the interaction frequency between groups based on the facility service coverage index, performs spatial clustering on the interaction data, identifies social hotspots and cold spots, and establishes a spatial interaction thermal network; The emotion feature analysis module collects the expression features of groups in public spaces based on the spatial interactive thermal network, obtains the emotion feature matrix of the groups, counts the emotion distribution ratios of groups in different areas, and generates spatial emotion distribution data.

Images