Park green space planning scheme deduction method based on multi-agent quarterly simulation

Abstract

The application discloses a park green space planning scheme deduction method based on multi-agent quarterly simulation, which comprises four main steps of park historical data collection, crowd seasonal travel feature extraction, park green space multi-agent visit modeling and iterative optimization, park layout deduction and scheme output. The application captures the relationship between the seasonal travel mode of the park green space and the urban built environment, constructs a deducible park green space multi-agent travel model, and then connects with multi-scenario park green space planning scheme data to form quarterly park green space passenger flow deduction to assist intelligent evaluation of the planning scheme. The application can realize quarterly park green space planning passenger flow deduction and intelligent evaluation, and optimize the existing park green space planning scheme evaluation method.

Description

TECHNICAL FIELD

[0001] The present application relates to a kind of urban park green space passenger flow deduction technology, belong to urban planning and green space system planning technical field, in particular to a kind of park green space passenger flow condition and evaluation method of multi-agent prediction. BACKGROUND

[0002] Urban green space is an important part of urban system, has ecological, economic and social value, has important significance to promote urban sustainable development, improve ecological link quality and improve the happiness index of citizen life.In actual work, urban green space system planning needs to plan urban existing landscape resources, residents daily life needs and overall construction requirements under the premise, to the existing green space development index, to multiple green space layout scheme is evaluated and scrutinized.

[0003] At present, the evaluation of green space system scheme mainly concentrates on post-use evaluation, i.e.in the selected and built park green space, according to the passenger flow data and resident evaluation of park green space, the existing problems are summarized to guide the next step of park green space construction and optimization, it is difficult to effectively predict the passenger flow situation before scheme determination, and the evaluation method nowadays often ignores the difference of passenger flow number and tourist type composition between each quarter, and cannot realize effective evaluation of the use of each quarter of park green space.In the scheme scrutiny stage, the evaluation of specific scheme often needs to be completed by planning and design personnel, which needs to manually organize existing landscape resource data and predict passenger flow situation.The process has the problems of high labor cost, economic cost and time cost, and the evaluation result is subjective, lacks quantitative basis.

[0004] The emergence and continuous maturity of multi-agent simulation technology provide technical methods for passenger flow deduction and evaluation of urban park green space scheme. SUMMARY

[0005] The purpose of the present application is to construct a quarterly park green space travel multi-agent simulation model based on multi-agent simulation technology, to deduce passenger flow and evaluate scheme of each park green space scheme under each quarterly scene, which can effectively solve the problems of strong subjectivity of park green space layout scheme design in original green space system planning and not considering the difference of passenger flow number and tourist type composition between each quarter.

[0006] To achieve the above purpose, the present application is realized by the following technical scheme:

[0007] The park green space planning scheme deduction method based on multi-agent quarterly simulation includes:

[0008] Step S1: park historical data acquisition

[0009] Based on the target city area, basic spatial data, park and green space climate and environmental data, and urban spatiotemporal trajectory big data are acquired; the urban spatiotemporal trajectory big data is preprocessed by data cleaning and stop point identification, and the mobile phone user trajectory is matched with park and green space through coordinate data. The names of the park and green space passed by each user are associated with the corresponding user data to generate a quarterly historical usage dataset of park and green space.

[0010] Step S2: Extraction of seasonal travel characteristics of the population

[0011] S2-1 is the basic feature variable set of multi-agents. From the quarterly historical park and green space usage dataset described in S1, all user IDs and their location attributes are extracted, along with the date distribution d and time period distribution t for each user's park and green space trips. Based on the seasonal anchoring classification method, the user feature indicators are classified, generating a total of 12 types of multi-agents.

[0012] S2-2 Multi-agent Park and Green Space Travel Probability Set. For each type of multi-agent user ID, extract their quarterly historical park and green space usage dataset. Use the cumulative travel frequency method to extract the daily park and green space travel probability set {P1, P2, P3, ..., P} for each type of multi-agent in the four quarters. 24}, where P = {probability of travel to parks and green spaces, park number}, and 24 is a time slice with a slice length of one hour.

[0013] Construction of the "Quarterly-Multi-Agent Subject" Feature Set for Parks and Green Spaces S2-3. The park and green space travel probability sets of various multi-agent subjects in S2-2 are summarized by time and space and then distributed to each park and green space to generate the "Quarterly-Multi-Agent Subject" feature set for each park and green space. This "Quarterly-Multi-Agent Subject" feature set includes the daily multi-agent type composition and effective dwell intensity feature set {H1, H2, H3, ..., H...} for each park and green space in each quarter. u}, where H u = {Effective dwell probability of the u-th type of multi-agent, total dwell time of the u-th type of multi-agent}; and adopt a sampling method to examine and verify the actual resident composition and usage of the selected sample parks and green spaces in different seasons and time periods.

[0014] S2-4 Construction of the "quarterly multi-agent space-time environment" feature set of the park green space. From the basic spatial data described in S1, the internal and surrounding built environment feature indicators of the park green space in each season and each time slice are extracted, and through the correction of the measured environment data, the "quarterly multi-agent space-time environment" feature set of the park green space is generated. The space-time environment feature set includes measured environment, park green space range and scale, surrounding built environment POI density, and surrounding built environment road network density indicators. The surrounding built environment refers to the urban built space within 1 km from the park green space.

[0015] Step S3: Multi-agent visit modeling and iterative optimization of park green space

[0016] Using a large language model (LLM) as a decision-making agent, the "quarterly multi-agent subject" feature set, the "quarterly multi-agent space-time environment" feature set, and the space-time trajectory big data in the quarterly park green space historical use data set are used as machine learning labels to capture the relationship between multi-agent model variables and travel trajectories. The seasonal travel characteristics of the crowd are matched with the learned travel model to construct a multi-agent travel plan. Through self-evaluation and recursive reasoning, the agent continuously compares the generated trajectory with the real-world situation to improve its output, and finally obtains a visit multi-agent simulation model of the park green space.

[0017] Step S4: Park layout deduction and scheme output

[0018] Input multi-scenario park green space layout schemes for multi-agent simulation to generate "park-multi-agent" pairing results for each quarter under different layout schemes, and construct a park green space scheme evaluation module by combining social integration assessment to output the optimal scheme for displaying park green space layout scenarios.

[0019] Further, in the step S1, the basic spatial data in the quarterly park green space historical use data set is obtained from the Open street map website, specifically including park green space vector data, urban road network vector data, building vector data, and industry POI data. Park green space climate environment data is obtained from the local meteorological bureau website of the research area, including wind environment data, humidity data, temperature data, and solar radiation data. The urban space-time trajectory big data is LBS data obtained from a mobile information push service provider, including LBS effective stay point data and user attribute data.

[0020] The quarterly park green land historical use data set refers to the spatiotemporal trajectory big data and the park green land climate environment data in each park boundary respectively summarized according to four quarters, forming a spatiotemporal trajectory big data set and a park green land climate environment data set; wherein, the spatiotemporal trajectory big data needs to be matched by time range and stay threshold value, and the park green land climate environment data selects the 24-hour meteorological data corresponding to the day with the most effective stay number in each quarter, the 24-hour meteorological data including wind environment data, humidity data and temperature data.

[0021] Further, the seasonal anchoring classification method in step S2-1 is:

[0022] Step S2-11 Quarterly Park Green Land Travel Feature Division. The proportion of the number of times of the user's i-th quarterly park green land travel to the total number of times of travel in the year is denoted as F i Then the distribution entropy of the user's quarterly park green land travel is:

[0023]

[0024] If H(F) is greater than or equal to 0.7ln4, the user is marked as a four-season traveler, otherwise, the user is marked as a single-season traveler.

[0025] Step S2-12 Daily Park Green Land Travel Feature Division. 6:00-18:00 (including 6:00, not including 18:00) is marked as daytime, and the rest of the time is marked as night. According to the daily travel characteristics of the user, the user whose number of park green land travel times during the day is greater than or equal to 80% of the total number of travel times in the year is marked as a daytime traveler; the user whose number of park green land travel times at night is greater than or equal to 80% of the total number of travel times in the year is marked as a night traveler; the rest of the users are marked as all-day travelers.

[0026] Step S2-13 Territorial Feature Division. According to the user's territorial feature, the user whose park green land travel trajectory is in the same territory as his / her territory is marked as a local person; the rest of the users are marked as non-local people.

[0027] Step S2-14 Multi-agent Classification. According to the quarterly travel feature, the daily travel feature, and the territorial travel feature, the multi-agent is divided into 12 types.

[0028] Further, the travel frequency accumulation method in step S2-2 is:

[0029] Step S2-21 Park Green Land Travel Information Summary. The park green lands in the target city are numbered, and the park numbers h and the total stay time s of each park visited by each type of multi-agent are summarized according to the quarter.

[0030] Step S2-22 park green space travel probability calculation. The daily park green space effective stay probability set of each type of multi-agent per quarter is:

[0031]

[0032] wherein a is the time slice number, b is the park green space number with the most effective stay times of this type of multi-agent in this time slice, n is the total number of users in this time slice, m is the number of users with effective stay records in park green space b in this time slice, and s is the total stay duration in park green space b in this time slice.

[0033] Further, the method for checking and verifying the actual resident composition and use status of the selected sample park green space in different seasons and time periods in step S2-3 is:

[0034] Step S2-31 determines the sampling object. In the qth quarter, select time slice a, and sample and inspect the ten park green spaces with the most visitors in the qth quarter in time slice a in the target city range, and mark the park green spaces as b1, b2, b3…b10. 10 .

[0035] Step S2-32 constructs a field verification set. In time period a, obtain the dynamic video data of the ten park green spaces by means of unmanned aerial vehicle shooting, obtain the actual effective stay number k of park green space b in time slice a by means of video semantic recognition ab , and obtain the crowd type composition by field investigation (the crowd type number division result is the same as the multi-agent type division table in step S2-14). According to the size of the value k, obtain the ranking R1 of the ten parks.

[0036] Step S2-33 constructs a sampling feature set. From the “quarterly-multi-agent main body” feature set, extract the sampling feature set of the park green space according to the park number and the quarterly feature q and time slice a. The sampling feature set contains the one-day multi-agent type composition and its effective stay intensity feature set H; mark each multi-agent type in each park green space b as , and summarize the effective stay probability of each type of multi-agent, and then calculate the number of each park green space; the specific formula is as follows:

[0037]

[0038] In the formula, P i represents the effective stay probability of the ith type of multi-agent, n i represents the total number of samples of the ith type of multi-agent in the historical use data set of the quarterly park green space.

[0039] Step S2-34 verifies the data between the real verification set and the sampling feature set. Based on the type number and the number of people in the real verification data set and the sampling feature set, the high-quality data set with more than half of the actual coincidence degree is screened out. The specific rules are as follows: according to the size of the value l, the ranking R2 of the ten parks is obtained, and the comparison between R1 and R2, the type composition of the crowd and the type composition of the multi-agent are made. If they are the same, it is considered that the simulation situation of the park green land is consistent with the actual situation, otherwise, it is considered that the simulation situation of the park green land is not consistent with the actual situation. In this way, the ten park green lands detected by sampling are tested. If more than half of the park green lands with simulation situation consistent with the actual situation, the data set is considered to pass the test.

[0040] Further, the "quarterly-multi-agent space-time environment" feature set in step S2-4 includes public transportation facility coverage, job-housing ratio, urban land function information entropy, and physiological equivalent temperature.

[0041] Further, the social integration degree is evaluated by applying the analytic hierarchy process in step S4, and the specific steps are as follows:

[0042] Step S4-1: crowd diversity dimension evaluation. The total number of multi-agent categories with effective stay in the simulation result of the park green land to be evaluated is denoted as u, the total number of effective stays is denoted as n, and the effective stay number of the i-th multi-agent is denoted as m i The crowd diversity distribution entropy G of the layout scheme is:

[0043]

[0044] The crowd diversity distribution entropy G is taken as the evaluation standard of the crowd diversity dimension. The average value of the crowd diversity distribution entropy G of the park green land in four seasons is obtained The higher the value is, the better the annual comprehensive performance of the scheme in the crowd diversity dimension is. The crowd diversity distribution entropy of the i-th quarter of the park green land is denoted as G i Then:

[0045]

[0046] If H(G) < 0.5ln4, it is considered that the quarter difference of the park green land in the crowd diversity dimension is obvious, and the best quarter and the worst quarter of the performance in the crowd diversity dimension are recorded.

[0047] Step S4-2: Park green space capacity suitability dimension evaluation. The "Park Design Specification" (GB 51192-2016) stipulates that the average land area per person for comprehensive park visitors is 30-60 square meters, the average land area per person for special park visitors is 20-30 square meters, the average land area per person for community park visitors is 20-30 square meters, and the average land area per person for garden visitors is 30-60 square meters. The total number of effective stops of multiple agents in the simulation results of the park green space to be evaluated is denoted as n, and the park green space area g is obtained from the step S1 basic space data. Then, the park green space capacity suitability index C is:

[0048]

[0049] The park green space capacity suitability index C is used as the park green space capacity suitability evaluation standard. The average value of the park green space capacity suitability index C for the four quarters of the park green space is calculated as The smaller the value, the better the annual comprehensive performance of the scheme in the park green space capacity suitability dimension. The park green space capacity suitability index for the i-th quarter of the park green space is denoted as C i Then:

[0050]

[0051] If H(C) < 0.5ln4, it is considered that the park green space has obvious quarter difference characteristics in park green space capacity suitability, and the quarter with the best performance and the quarter with the worst performance in the comprehensive environmental dimension are recorded.

[0052] Step S4-3: Comprehensive environmental dimension evaluation. The public transportation facility coverage x1, the job-housing ratio x2, the urban land function information entropy x3, and the physiological equivalent temperature x4 of the park green space to be evaluated are obtained from the step S2-4 "quarterly multi-agent space-time environment". The same type of data is normalized in all evaluation schemes to obtain y1, y2, y3, and y4. The normalization method is:

[0053]

[0054] Then, the comprehensive environmental dimension index E of the layout scheme is:

[0055]

[0056] The comprehensive environmental dimension index E is used as the comprehensive environmental dimension evaluation standard. The average value of the comprehensive environmental dimension index E for the four quarters of the park green space is calculated as The higher the value, the better the annual comprehensive performance of the scheme in the comprehensive environmental dimension. The comprehensive environmental dimension index for the i-th quarter of the park green space is denoted as E i Then:

[0057]

[0058] If H(E) < 0.5ln4, it is considered that the park green space has significant quarterly differences in the suitability of park green space capacity. At the same time, the best and worst quarters in terms of its comprehensive environmental performance are recorded.

[0059] Step S4-4: Scheme Evaluation and Output. Based on the evaluation dimensions of the selected multi-scenario park and green space layout schemes, scores and evaluations are performed on each dimension, and the park and green space layout schemes and quarterly characteristic evaluations under that dimension are output.

[0060] Beneficial effects:

[0061] 1. This invention enables the quarterly projection of visitor flow in parks and green spaces. Steps S2 and S3 are based on multi-agent simulation technology and seasonal anchoring feature extraction. They innovatively propose a visitor flow projection model based on the spatiotemporal characteristics of parks and green spaces in each quarter, which breaks through the limitation of traditional green space system planning that cannot predict the quarterly visitor flow characteristics before construction.

[0062] 2. This invention considers the different emphases of planning schemes in the work and the applicability of intelligent inspection. In step 4, based on the prediction results achieved by this invention and the actual work requirements, an evaluation is conducted on three dimensions: population diversity, capacity suitability, and comprehensive environment, and the evaluation results are output according to the work requirements.

[0063] 3. This invention addresses the prediction accuracy and process scientificity of multi-scheme planning and evaluation. In terms of the time consumption of multi-scenario planning and evaluation, it transforms the traditional method of organizing a large number of human and financial resources to organize expert review meetings into a method of quarterly intelligent extrapolation and evaluation based on historical datasets. This greatly reduces time costs and significantly improves extrapolation dimensions, decision-making efficiency, and accuracy. Attached Figure Description

[0064] Figure 1 This is a flowchart illustrating the method for deriving and evaluating park and green space planning schemes according to the present invention. Detailed Implementation

[0065] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0066] like Figure 1 As shown, the technical solution of the present invention will be described in detail using a multi-scenario planning scheme in Gulou District, Nanjing City, Jiangsu Province:

[0067] Step S1: Collection of Historical Park Data

[0068] According to the target city range, obtain basic spatial data, park green space climate environment data and city space-time trajectory big data; and preprocess the city space-time trajectory big data through data cleaning and stay point identification, and match the mobile phone user trajectory and the park green space through coordinate data, and associate the park green space name passed by each user with the corresponding user data, to generate a quarterly park green space historical use data set. Among them, the basic spatial data is obtained from the Open streetmap website, including park green space vector data, city road network vector data, building vector data and industry POI data; the park green space climate environment data is obtained from the local meteorological bureau website of the research area, including wind environment data, humidity data, temperature data and solar radiation data; the city space-time trajectory big data is the LBS data obtained from the mobile information push service provider, including LBS effective stay point data and user attribute data. The specific attribute description of various data is as follows:

[0069] Data type and attribute description

[0070]

[0071] The quarterly park green space historical use data set refers to the space-time trajectory big data and park green space climate environment data within the boundary of each park are summarized respectively according to four seasons, to form a space-time trajectory big data set and a park green space climate environment data set; wherein the space-time trajectory big data needs to be matched by time range and stay threshold value by season, and the park green space climate environment data selects the 24-hour meteorological data corresponding to the date with the most effective stay number in each season, which includes wind environment data, humidity data and temperature data. The specific attribute description of various data is as follows:

[0072] Quarterly park green space historical use data set attribute description

[0073]

[0074] Step S2: crowd seasonal travel feature extraction

[0075] S2-1 basic characteristic variable set of multi-agent. Extract all user IDs and their attributes from the quarterly park green space historical use data set described in S1, as well as the date distribution d and time period distribution t of each user's each park green space travel, and classify the user's feature index according to the seasonal anchoring classification method, to generate 12 types of multi-agent.

[0076] The seasonal anchoring classification method in step S2-1 is:

[0077] Step S2-11 Quarterly park green space travel feature division. The proportion of the number of times of the user's i-th quarterly park green space travel to the total number of times of travel in the year is denoted as F i , and the distribution entropy of the quarterly park green space travel of the user is:

[0078]

[0079] If H(F)≥0.7ln4, the user is marked as a four-season traveler, otherwise, the user is marked as a single-season traveler.

[0080] Step S2-12 Daily park green space travel feature division. 6:00-18:00 (including 6:00 and excluding 18:00) is marked as daytime, and the rest of the time is marked as night. According to the daily travel characteristics of the user, the user whose number of park green space travels in the daytime in the year is greater than or equal to 80% of the total number of travels in the year is marked as a daytime traveler; the user whose number of park green space travels at night in the year is greater than or equal to 80% of the total number of travels in the year is marked as a night traveler; the rest of the users are marked as all-day travelers.

[0081] Step S2-13 Local feature division. According to the local feature of the user, the user whose park green space travel trajectory is in the same local as the user is marked as a local person; the rest of the users are marked as non-local people.

[0082] Step S2-14 Multi-agent category division. According to the quarterly travel feature, the daily travel feature, and the local travel feature, the multi-agent is divided into 12 types, and the specific features are as follows.

[0083] Multi-agent category division

[0084]

[0085] S2-2 Park green space travel probability set of multi-agent. For each type of multi-agent user ID, the quarterly park green space historical use data set is extracted, and the travel frequency accumulation method is used to extract the one-day park green space travel probability set {P1, P2, P3, …, P 24} of each type of multi-agent in four quarters, where P={park green space travel probability, park number}, 24 is the time slice, and the slice length is one hour.

[0086] The travel frequency accumulation method in step S2-2 is:

[0087] Step S2-21 Park green space travel information summary. The park green spaces in the target city are numbered, and the park numbers h and the total stay time s of each park visited by each type of multi-agent are summarized by quarter.

[0088] Step S2-22 park green space travel probability calculation. The set of valid stay probabilities of each type of multi-agent in a park green space per quarter is:

[0089]

[0090] wherein a is the time slice number, b is the park green space number with the most valid stay times of this type of multi-agent in this time slice, n is the total number of users in this time slice, m is the number of users with valid stay records in park green space b in this time slice, and s is the total stay duration in park green space b in this time slice.

[0091] S2-3 Construction of the "quarterly-multi-agent subject" feature set of park green space. The park green space travel probability set of each type of multi-agent in S2-2 is respectively aggregated in time and space in each park green space to generate the "quarterly-multi-agent subject" feature set of each park green space, which contains the daily multi-agent type composition and its valid stay intensity feature set {H1, H2, H3, …, H u} of each park green space in each quarter, wherein H u ={valid stay probability of the yth type of multi-agent, total stay duration of the u th type of multi-agent}; and a sampling method is adopted to verify and check the actual resident composition and usage of the selected sample park green spaces in different seasons and time periods.

[0092] The method for verifying and checking the actual resident composition and usage of the selected sample park green spaces in different seasons and time periods in step S2-3 is:

[0093] Step S2-31 Determine the sampling object. In the qth quarter, select time slice a, and sample and verify the ten park green spaces with the most visitors in the qth quarter a time slice in the target city range, and mark the park green spaces as b1, b2, b3, …, b 10 .

[0094] Step S2-32 Construction of the on-site verification set. In the a time period, obtain the dynamic video data of the ten park green spaces by means of unmanned aerial vehicle shooting, obtain the actual valid stay number k ab of the park green space b in the time slice a by video semantic recognition, and obtain the crowd type composition by on-site investigation (the crowd type number division result is the same as the multi-agent category division table in step S2-14). According to the size of the value k, obtain the ranking R1 of the ten parks.

[0095] Step S2-33 sample feature set construction. From the "quarterly multi-agent subject" feature set, the sample feature set of the park green space is extracted according to the park number and the quarterly feature q and the time segment a. The sample feature set contains the composition of multi-agent types and the effective stay intensity feature set H of a day; each multi-agent type in each park green space b is marked as and the effective stay probability of each type of multi-agent is summarized, and then the number of each park green space is calculated; the specific formula is as follows:

[0096]

[0097] In the formula, P i represents the effective stay probability of the i-th type of multi-agent, n i represents the total number of samples of the i-th type of multi-agent in the historical use data set of the quarterly park green space.

[0098] Step S2-34 data verification between the sample feature set and the field verification set. Based on the type number and the number of people sorting verification between the field verification data set and the sample feature set, the high-quality data set with more than half of the actual coincidence degree is screened out. The specific rules are as follows: according to the size of the value l, the ranking R2 of the ten parks is obtained, and R1 and R2, the composition of the crowd and the composition of the multi-agent are compared. If they are the same, it is considered that the simulation situation of the park green space is consistent with the actual situation, otherwise, it is considered that the simulation situation of the park green space is not consistent with the actual situation. In this way, the ten park green spaces detected by sampling are tested, and if more than half of the park green spaces with consistent simulation situation and actual situation, the data set is considered to pass the test.

[0099] S2-4 Construction of "quarterly multi-agent space-time environment" feature set of park green space. From the basic spatial data described in S1, the built environment feature indexes of the park green space and its surrounding environment in each season and each time slice are extracted, and the measured environment data is corrected to generate the "quarterly multi-agent space-time environment" feature set of the park green space. The space-time environment feature set includes measured environment, park green space range and scale, surrounding built environment POI density, surrounding built environment road network density index, and the surrounding built environment refers to the urban built space within 1 km straight distance from the park green space.

[0100] The "quarterly multi-agent space-time environment" feature set in step S2-4 includes public transportation facility coverage, job-housing ratio, urban land function information entropy and physiological equivalent temperature, and the specific indexes are as follows:

[0101] "Quarterly multi-agent space-time environment" feature set index

[0102]

[0103]

[0104] Step S3: Multi-agent visit modeling and iterative optimization of park green space

[0105] The large language model (LLM) is used as the decision-making agent. The "quarterly multi-agent subject" feature set, "quarterly multi-agent space-time environment" feature set, and space-time trajectory big data in the quarterly park green space historical use data set are used as the labels for machine learning to capture the relationship between the multi-agent model variables and the travel trajectory. The seasonal travel characteristics of the crowd are matched with the learned travel model to build a multi-agent travel plan. Through self-evaluation and recursive reasoning, the agent continuously compares the generated trajectory with the real-world situation to improve its output, and finally obtains the visit multi-agent simulation model of the park green space.

[0106] Step S4: Park layout deduction and scheme output

[0107] The multi-scenario park green space layout scheme is input for multi-agent simulation to generate the "park-multi-agent" pairing results for each quarter under different layout schemes. Combined with the social integration evaluation, a park green space scheme evaluation module is built to output the optimal scheme for displaying the park green space layout scenario.

[0108] In the step S4, the social integration evaluation is performed using the analytic hierarchy process. The specific steps are as follows:

[0109] Step S4-1: Crowd diversity dimension evaluation. Let u be the total number of multi-agent categories with effective stays in the simulation results of the park green space to be evaluated, n be the total number of effective stays, and m be the number of effective stays of the i-th multi-agent. i The crowd diversity distribution entropy G of this layout scheme is:

[0110]

[0111] The crowd diversity distribution entropy G is used as the evaluation standard for the crowd diversity dimension. The average value of the crowd diversity distribution entropy G of the four quarters of the park green space is calculated. The higher the value, the better the annual comprehensive performance of the scheme in the crowd diversity dimension. Let G be the crowd diversity distribution entropy of the i-th quarter of the park green space. i Then:

[0112]

[0113] If H(G) < 0.5ln4, it is considered that the park green space has obvious seasonal difference characteristics in the crowd diversity dimension, and the best and worst quarters in the crowd diversity dimension are recorded.

[0114] Step S4-2: Park green space capacity suitability dimension evaluation. The "Park Design Specification" (GB 51192-2016) stipulates that the average land area per person for comprehensive park visitors is 30-60㎡, the average land area per person for special park visitors is 20-30㎡, the average land area per person for community park visitors is 20-30㎡, and the average land area per person for garden visitors is 30-60㎡. The total number of effective stops of multiple agents in the simulation results of the park green space to be evaluated is denoted as n, and the park green space area g is obtained from the step S1 basic space data. Then, the park green space capacity suitability index C is:

[0115]

[0116] The park green space capacity suitability index C is used as the evaluation standard for the park green space capacity suitability. The average value of the park green space capacity suitability index C for the four quarters is: The smaller the value, the better the annual comprehensive performance of the scheme in the park green space capacity suitability dimension. The park green space capacity suitability index for the i-th quarter is denoted as C i Then:

[0117]

[0118] If H(C) < 0.5ln4, it is considered that the park green space has obvious quarter difference in park green space capacity suitability, and the quarter with the best and worst comprehensive environmental dimension performance is recorded.

[0119] Step S4-3: Comprehensive environmental dimension evaluation. The public transportation facility coverage x1, job-housing ratio x2, urban land function information entropy x3, and physiological equivalent temperature x4 of the park green space to be evaluated are obtained from the step S2-4 "quarterly multi-agent space-time environment". The same type of data is normalized in all evaluation schemes to obtain y1, y2, y3, and y4. The normalization method is:

[0120]

[0121] Then, the comprehensive environmental dimension index E of the layout scheme is:

[0122]

[0123] The comprehensive environmental dimension index E is used as the evaluation standard for the comprehensive environmental dimension. The average value of the comprehensive environmental dimension index E for the four quarters is: The higher the value, the better the annual comprehensive performance of the scheme in the comprehensive environmental dimension. The comprehensive environmental dimension index for the i-th quarter is denoted as E i Then:

[0124]

[0125] If H(E) < 0.5ln4, it is considered that the park green space has obvious quarter difference in the park green space capacity suitability degree, and the quarter with the best performance and the quarter with the worst performance in the comprehensive environmental dimension are recorded.

[0126] Step S4-4: scheme evaluation and output. According to the evaluation dimension of the selected multi-scenario park green space layout scheme, the score and evaluation are carried out for the dimension, and the park green space layout scheme and the quarter feature evaluation under the dimension are output.

[0127] The above shows and describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited to the above examples, and the above examples and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the claimed present application.

Claims

1. A park green space planning scheme deduction method based on multi-agent quarterly simulation, characterized in that, The method comprises the following steps: Step S1: Park historical data collection According to the target city range, obtain basic spatial data, park green space climate environment data and city space-time trajectory big data; and perform data cleaning and stay point identification preprocessing on the city space-time trajectory big data, and pair the mobile phone user trajectory with the park green space through coordinate data, and associate each user's way through the park green space name with the corresponding user data to generate a quarterly park green space historical use data set; Step S2: crowd seasonal travel feature extraction S2-1 basic feature variable set of multi-agent; extract all user IDs and their attributes from the quarterly park green space historical use data set in S1, as well as the date distribution d and time period distribution t of each user's each park green space travel, and classify the feature indexes of the user according to the time anchor classification method to generate 12 types of multi-agents; S2-2 Park green space travel probability set of multi-agent; for each type of multi-agent user ID, extract its quarterly park green space historical use data set respectively, and extract the one-day park green space travel probability set of each type of multi-agent in four quarters {P1, P2, P3, …, P 24} respectively by using travel frequency accumulation method, wherein P={park green space travel probability, park number}, 24 is time slicing, and the slicing length is one hour; S2-3 Construction of the "quarterly multi-agent subject" feature set of the park green space; the park green space travel probability set of each type of multi-agent in S2-2 is respectively summarized in time and space to each park green space to generate the "quarterly multi-agent subject" feature set of each park green space, which contains the multi-agent type composition and its effective stay intensity feature set {H1, H2, H3, …, H u} of each park green space in each quarter, wherein H u ={the effective stay probability of the u-th type of multi-agent, the total stay duration of the u-th type of multi-agent}; and a sampling method is adopted to check and verify the actual resident composition and use condition of the selected sample park green space in different seasons and time periods; S2-4 construction of "quarterly-multi-agent space-time environment" feature set of park green space; from the basic spatial data in S1, extract the built environment feature indexes of the park green space and its surrounding area in each season and each time slice, and generate the "quarterly-multi-agent space-time environment" feature set of the park green space through the correction of the measured environment data; Step S3: multi-agent visit modeling and iterative optimization of park green space Use a large language model as a decision-making agent, use the "quarterly-multi-agent subject" feature set, "quarterly-multi-agent space-time environment" feature set of each park green space, and space-time trajectory big data in the quarterly park green space historical use data set as the label of machine learning to capture the relationship between the multi-agent model variable and the travel trajectory, match the crowd seasonal travel features with the learned travel model, and construct a multi-agent travel plan; through self-evaluation and recursive reasoning, the agent constantly compares the generated trajectory with the reality to improve its output, and finally obtains a visit multi-agent simulation model of the park green space; Step S4: park layout deduction and scheme output Input multi-scenario park green space layout schemes for multi-agent simulation to generate "park-multi-agent" pairing results for each quarter under different layout schemes, and construct a park green space scheme evaluation module by combining social integration evaluation to output the optimal scheme of the park green space layout scenario; The time anchor classification method in step S2-1 is: Step S2-11: Park green space travel feature division; the proportion of the number of times of the user's i quarter park green space travel to the total number of times of travel in the year is recorded as F i The distribution entropy of the user's quarter park green space travel is: If H(F) is greater than or equal to 0.7ln4, the user is marked as a four-season traveler, otherwise, the user is marked as a single-season traveler; Step S2-12 daily park green space travel feature division; mark 6:00-18:00 as daytime and the rest of the time as night; according to the daily travel features of the user, mark the users whose daytime park green space travel times in the statistical year are greater than or equal to 80% of the total travel times in the year as daytime travelers; mark the users whose night park green space travel times in the statistical year are greater than or equal to 80% of the total travel times in the year as night travelers; the rest of the users are marked as all-day travelers; Step S2-13 local feature division; according to the user local feature, the user whose park green space travel trajectory location is the same as the local is marked as a local person; the rest of the users are marked as outsiders; Step S2-14 multi-agent category division; according to the quarterly travel characteristics, daily travel characteristics and local travel characteristics, the multi-agent is divided into 12 types.

2. The park green space planning scheme deduction method based on multi-agent quarterly simulation according to claim 1, characterized in that, In the step S1, the basic space data in the quarterly park green space historical use data set is obtained from the Open streetmap website, specifically including park green space vector data, urban road network vector data, building vector data and industry POI data; the park green space climate environment data is obtained from the meteorological bureau website of the research area, including wind environment data, humidity data, temperature data and solar radiation data; the urban space-time trajectory big data is the LBS data obtained from the mobile information push service provider, including LBS effective stay point data and user attribute data; The quarterly park green space historical use data set refers to the space-time trajectory big data and park green space climate environment data within the boundary of each park in four quarters, forming a space-time trajectory big data set and a park green space climate environment data set; wherein, the space-time trajectory big data needs to be matched by time range and stay threshold in four quarters, and the park green space climate environment data selects the 24-hour meteorological data corresponding to the date with the most effective stayers in each quarter, which includes wind environment data, humidity data and temperature data.

3. The park green space planning scheme deduction method based on multi-agent quarterly simulation according to claim 2, characterized in that, The travel frequency accumulation method in step S2-2 is: Step S2-21 park green space travel information summary; number the park green spaces in the target city, and summarize the park number h and the total stay time s of each park visited by each type of multi-agent in each quarter; Step S2-22 park green space travel probability calculation; the daily park green space effective stay probability set of each type of multi-agent in each quarter is: Wherein, a is the time slice number, b is the park green space number with the most effective stayers of this type of multi-agent in this time slice, n is the total number of users in this time slice, m is the number of users with effective stay records in park green space b in this time slice, and s is the total stay time in park green space b in this time slice.

4. The park green space planning scheme derivation method based on multi-agent quarterly simulation according to claim 3, characterized in that, The method for checking and verifying the actual resident composition and use status of the selected sample park green space in different seasons and time periods in step S2-3 is: Step S2-31 determines the sampling object; in the time slice a of the qth quarter, ten park green spaces with the most visitors in the target city in the qth quarter are selected for sampling inspection, and the park green spaces are marked as b1, b2, b3, …, b10. 10 ; Step S2-32, real-time verification set construction; within a time period, the dynamic video data of the ten parks is obtained by means of unmanned aerial vehicle shooting, the actual effective number of people k of the park b within the time period a is obtained by video semantic recognition ab , and the composition of the crowd type is obtained by field research; according to the size of the value k, the ranking R1 of the ten parks is obtained; Step S2-33: constructing a sample feature set; extracting a sample feature set of the park green space from the "quarterly multi-agent subject" feature set according to the park number and the quarterly feature q and the time segment a, wherein the sample feature set contains a daily multi-agent type composition and an effective stay intensity feature set H; marking each multi-agent type in each park green space b as and summarizing the effective stay probability of each type of multi-agent, and then calculating the number of people in each park green space; the specific formula is as follows: where P i represents the effective stay probability of the ith type of multi-agent, n i represents the total number of samples of the ith type of multi-agent in the historical use data set of the quarter park green space; Step S2-34 verifies the data between the real verification set and the sampling feature set; based on the type number and the number of people sorting verification between the real verification data set and the sampling feature set, the high-quality data set with more than half of the actual coincidence degree is screened out; the specific rules are as follows: according to the size of the value 1, the ranking R2 of the ten parks is obtained, and the comparison between R1 and R2, the composition of the crowd type and the composition of the multi-agent type are made, if they are the same, it is considered that the simulation situation of the park green space is consistent with the actual situation, otherwise, it is considered that the simulation situation of the park green space is not consistent with the actual situation, in this way, the ten park green spaces detected by sampling are tested, if more than half of the park green spaces with simulation situation consistent with the actual situation, the data set is considered to pass the test.

5. The park green space planning scheme deduction method based on multi-agent quarterly simulation according to claim 4, characterized in that, The surrounding built environment in step S2-4 refers to the urban built space within 1km straight-line distance from the park green space; the "quarterly-multi-agent space-time environment" feature set specifically includes public transportation facility coverage, job-housing ratio, urban land function information entropy and physiological equivalent temperature.

6. The park green space planning scheme derivation method based on multi-agent quarterly simulation according to claim 5, characterized in that, In step S4, the social integration degree is evaluated by using the analytic hierarchy process, and the specific steps are as follows: Step S4-1: crowd diversity dimension evaluation; the total number of classes of multi-agent that exist in the simulation result of the park green land to be evaluated is denoted as u, the total number of effective stays is denoted as n, and the number of effective stays of the i-th class of multi-agent is denoted as m i The crowd diversity distribution entropy G of the layout scheme is: The human diversity distribution entropy G is taken as the human diversity dimension evaluation criterion; and the average of the human diversity distribution entropy G of the four seasons of the park green land is obtained The higher the value is, the better the annual comprehensive performance of the scheme is in the human diversity dimension; the human diversity distribution entropy of the i season of the park green land is denoted as G i Therefore: If H(G) < 0.5ln4, it is considered that the park green space has obvious quarter difference characteristics in the diversity of the crowd, and the best quarter and the worst quarter in the diversity of the crowd are recorded; Step S4-2: evaluation of park green space capacity suitability dimension; according to the "Park Design Specification", the land area per capita of comprehensive park visitors is 30-60 square meters, the land area per capita of special park visitors is 20-30 square meters, the land area per capita of community park visitors is 20-30 square meters, and the land area per capita of garden visitors is 30-60 square meters; the total number of multi-agent effective stays in the simulation results of the park green space to be evaluated is n, and the park green space area g is obtained from the step S1 basic space data, then the park green space capacity suitability index C is: The park green space capacity suitability index C is taken as the evaluation standard of the park green space capacity suitability, and the average value of the four quarterly green space capacity suitability indexes C of the park green space is obtained The smaller the value is, the better the annual comprehensive performance of the scheme in the park green space capacity suitability dimension is; the c index of the i th quarter of the park green space is denoted as C i Then: If H(C) < 0.5ln4, it is considered that the park green space has obvious quarter difference characteristics in the park green space capacity suitability, and the best quarter and the worst quarter in the comprehensive environment dimension are recorded; Step S4-3: evaluation of comprehensive environment dimension; the public transportation facility coverage x1, the job-housing ratio x2, the urban land function information entropy x3 and the physiological equivalent temperature x4 of the evaluation park green space are obtained from the "quarterly-multi-agent space-time environment" in step S2-4, the same data in all evaluation schemes are normalized to obtain y1, y2, y3 and y4, and the normalization method is: The comprehensive environment dimension index E of the layout scheme is: The comprehensive environmental dimension index E is taken as the comprehensive environmental dimension evaluation standard; and the average value of the comprehensive environmental dimension index E of the four seasons of the park green land is obtained The higher the comprehensive environmental dimension annual comprehensive performance of the scheme is; the comprehensive environmental dimension index of the i season of the park green land is recorded as E i Then: If H(E) < 0.5ln4, it is considered that the park green space has obvious quarter difference characteristics in the park green space capacity suitability, and the best quarter and the worst quarter in the comprehensive environment dimension are recorded; Step S4-4: scheme evaluation and output; according to the evaluation dimension of the selected multi-scenario park green space layout scheme, the score and evaluation of the dimension are carried out, and the park green space layout scheme and the quarter feature evaluation under the dimension are output.

Images