Policy selection and intelligent grading method for city updating units driven by time context pedigree

Abstract

The invention discloses a strategy selection and intelligent grading method for a city updating unit driven by a time context pedigree, and the method comprises the following steps: forming a plot sample and a strategy template through multi-source data collection and standardization; utilizing a multi-target benefit model to calculate updating potential, screening to-be-updated land parcels, and clustering to select updating pilot points; through a randomized allocation strategy and period monitoring, constructing a time venation spectrum; optimal pilot points are screened, updating classification is formed, similarity matching and space connectivity analysis are combined to be popularized to non-pilot-point plots, and updating units are automatically delimited; the strategy effect is identified, and dynamic strategy combination and feedback adjustment are realized through multi-objective optimization; and finally, visual display is carried out to assist the planner and resident in immersive decision making. According to the method, city updating intelligent grading and strategy collaboration driven by a time spectrum system are realized, and the method has the advantages of comprehensive data, systematic method and efficient decision, and can be widely applied to city updating and space optimization practice.

Description

Technical Field

[0001] This invention belongs to the field of urban planning and urban renewal technology, and in particular relates to a strategy selection and intelligent classification method for urban renewal units driven by time-series spectrum. Background Technology

[0002] As urbanization deepens, the traditional model of large-scale demolition and overall reconstruction is increasingly unable to meet the needs of contemporary urban development. On the one hand, this approach often leads to enormous economic costs and social disruption, drastically disrupting residents' living environment and community networks in a short period, resulting in social conflicts and damage to the urban fabric. On the other hand, urban development has gradually shifted from "incremental expansion" to "stock optimization," with improving the efficiency of existing space utilization, enhancing environmental quality, and meeting diverse lifestyle needs becoming the primary objectives. Against this backdrop, small-scale, gradual urban renewal is becoming the mainstream approach. This model, through meticulous intervention, embeds renewal measures into the daily operation of communities. For example, through the construction of pocket parks, optimization of pedestrian and cycling systems, improvement of the interface environment, and functional replacement, it achieves quality improvement in local spaces and gradual improvement in residents' well-being without altering the overall urban structure.

[0003] However, existing small-scale, incremental redevelopment methods still have significant shortcomings in practice. First, the acquisition and integration of multi-source data is inadequate. Information from various sources, such as remote sensing imagery, street views, pedestrian monitoring, and policy texts, lacks a unified coordinate system and standardized processing, making it difficult to compare plot characteristics. Second, the methods for identifying and classifying redevelopment potential are relatively static, often relying on static indicators such as plot ratio, building age, and population density, neglecting dynamic feedback from changes in resident vitality, environmental quality, and spatial quality over time, thus failing to reflect the timeliness and actual needs of redevelopment. Third, the mechanism for identifying the effectiveness of strategies is not scientific enough, lacking inferences about the causal effects of single strategies in complex environments, making it difficult to accurately quantify the effectiveness and replicability of redevelopment measures. Finally, the existing system lacks a generalizable mapping mechanism, making it difficult to efficiently transfer the experience and results of pilot redevelopment to other plots, hindering the large-scale and institutionalized development of small-scale redevelopment. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to provide a time-based spectrum-driven method for strategy selection and intelligent classification of urban renewal units, which can automatically classify urban renewal units and match strategies.

[0005] Technical Solution: To achieve the above objectives, the urban renewal unit strategy selection and intelligent classification method of the present invention includes the following steps: S1. Source data collection and standardization: Obtain basic information on population activity, environmental quality, and spatial quality for each plot through remote sensing imagery, street view, and pedestrian video surveillance; Align data from different sources to the same base map through coordinate unification and data cleaning techniques to form a plot sample library that can be directly compared; Collect bidding information on update measures from the government procurement website, and use text mining models to sort out the "update strategy terminology" to form an update strategy list and parameter template; S2. Renewal potential evaluation and pilot selection: The renewal potential score of the plots is calculated by the multi-objective benefit evaluation model, and the set of plots to be renewed is selected by the threshold screening rule; The plots to be renewed are divided into 5 categories by the morphological-target feature clustering algorithm, and the 3 plots with the lowest potential scores in each category are selected as pilots, forming a test list of 15 pilots in total. S3. Pilot update monitoring and timeline genealogy construction: Input the update strategy list and parameter template collected in step S1, and allocate update measures to each pilot through a restricted randomization algorithm; record the changes in multi-objective benefits of each pilot through fixed-period monitoring, which can be once a week, to form a monitoring dataset; integrate the update monitoring dataset through genealogy modeling methods, and mark branches and merges when encountering strategy superposition, replacement or adjacent linkage, to construct a timeline genealogy of "event-version-divide-merge tree"; S4. Benefit Frontier Assessment and Renewal Unit Promotion and Division: Using an efficiency frontier model, an efficiency frontier diagram is drawn for the pilot projects between multi-objective benefit improvement and renewal costs. The pilot projects with the highest efficiency within each class are selected as potential optimal cases, and their efficiency is used as a representative value to rank and assign levels to form 5 categories of renewal levels. A similarity matching algorithm is used to match non-pilot plots of the same type with plots that have similar conditions to the potential optimal cases, generating a priority development list. Through spatial connectivity analysis and community division algorithms, adjacent, similar, and connected plots are automatically delineated into the same renewal unit, and morphological rules are used to correct the boundaries to ensure that the boundaries are continuous and the shapes are reasonable. S5. The causal inference of strategy effectiveness and strategy matching are coordinated. The difference-in-differences (DID) model is used to compare the benefit changes of pilot and non-implemented plots to identify the average treatment effect of a single strategy. For different plot types and heterogeneous characteristics, the regression discontinuity (RDD) and propensity score matching (PSM) methods are used for supplementary verification to ensure the robustness of the results. The effectiveness results of all strategies are input into a multi-objective combination optimization model to automatically generate the optimal strategy combination for updated plots. The deviation between the actual benefits and the model's expectations is monitored monthly. If the expected results are not met, the time context spectrum is used for feedback adjustment and strategy reconfiguration. S6. Visual presentation and decision release.

[0006] Optionally, the multi-objective benefit evaluation model in step S2 consists of three major categories of evaluation indicators: comprehensive population vitality, ecological environment, and spatial quality. A combined weighting method is used to determine the indicator weights, and a comprehensive benefit score is synthesized through linear weighting. The combined weighting method refers to a combination of the analytic hierarchy process (AHP) and the entropy weighting method. The specific indicator framework is as follows: comprehensive population vitality indicators include pedestrian density, nighttime vitality index, and social interaction intensity; ecological environment indicators include green coverage rate, air pollution index, and heat island intensity; spatial quality indicators include public service facility density, street interface permeability, and architectural style integrity. The calculation formula for the multi-objective benefit evaluation model is as follows: , in For comprehensive benefit scoring, These are standardized values ​​for comprehensive indicators of population vitality, ecological environment, and spatial quality, specifically using Min-Max standardization. The weighting coefficients are calculated using a combination of the Analytic Hierarchy Process (AHP) and the entropy weighting method, with AHP accounting for 40% of the subjective weight and the entropy weighting method accounting for 60% of the objective weight.

[0007] Optionally, the threshold screening rule in step S2 adopts the dynamic quantile threshold method, and the specific steps are as follows: First, calculate the comprehensive benefit score of all plots. The scores are sorted in descending order. Then, a window-based variable point detection algorithm is used to identify natural breakpoints in the scoring sequence, which serve as the basis for threshold adjustment. Next, the top 30% of the plots are selected as the set of plots to be updated. If a natural breakpoint is located near the 30% quantile (near means within ±5%), then the natural breakpoint is used as the actual threshold. Finally, the set of plots to be updated is output, and a potential distribution map is generated.

[0008] Optionally, the morphology-target feature clustering algorithm in step S2 is an improved clustering method that integrates multi-scale morphological features and target benefit features, specifically including the following steps: First, morphological features and target benefit features are extracted. Morphological features include building density, plot ratio, street width, and plot shape index. Target benefit features include population vitality score, ecological environment score, spatial quality score, and comprehensive benefit score. After data standardization, the K-means++ algorithm is used for preliminary clustering, dividing the data into 5 categories, and the morphological-target feature outlines of each category are labeled. Finally, the 3 plots with the lowest comprehensive benefit scores from each cluster are selected as update pilots, forming a test list of 15 pilots.

[0009] Optionally, in step S3, the timeline is defined as follows: each update measure is treated as an "event," and after an event takes effect, a corresponding "version" is generated and linked together chronologically; when strategies are superimposed or replaced on the same plot, a "split" is generated, and the rest are "continuous continuations" that advance over time, forming a directional version hierarchy; multi-objective benefits are recorded simultaneously. and the increase in benefits in this period Simultaneously record edge type indication , ;pass The minimal set of fields provides standardized input for subsequent efficiency frontier assessment and causal inference.

[0010] Optionally, the efficiency frontier model in step S4 is used to evaluate the trade-off between the multi-objective benefits and costs of the pilot update project, specifically including the following steps: Firstly, regarding the cost and benefit indicators for renovation, the cost indicators include the cost per unit area, time cost, and coordination cost. The benefit indicators include a comprehensive score of the improvement in population vitality, the improvement in the ecological environment, and the improvement in spatial quality. The overall improvement rate of multi-objective benefits is then calculated using the following formula: , in, These represent indicators of population vitality, ecological environment, and spatial quality, respectively. These represent "before" and "after" respectively. The weighting coefficients are determined using the Analytic Hierarchy Process (AHP). Finally, using the update cost as the horizontal axis and the comprehensive improvement rate of multi-objective benefits as the vertical axis, a scatter plot of all pilot projects was drawn, i.e., an efficiency frontier plot.

[0011] Optionally, the potential optimal case in step S4 is selected by combining the efficiency frontier model with cluster analysis, specifically including the following steps: First, based on the efficiency frontier plot, identify the pilots located on the efficiency frontier in each cluster category; then calculate the overall efficiency value for each pilot. The specific calculation formula is as follows: , in, Indicates the first The input metric is the update cost; Indicates the first The output indicator is the efficiency improvement rate; and These are the weights of the input and output metrics, respectively. To input the number of indicators, The number of output metrics; Next, the overall efficiency values ​​are sorted from high to low, and the sorted pilots are assigned levels to form 5 categories of update levels, from level 1 to level 5, where level 1 represents the highest efficiency and level 5 represents the lowest efficiency. Finally, the top-ranked pilot in each category is selected as the potential optimal case, representing the highest update efficiency level for that category.

[0012] Optionally, the similarity matching algorithm in step S4 is a matching method that integrates multi-feature similarity calculation and nearest neighbor search, used to identify plots with similar conditions to potential optimal cases among non-pilot plots of the same type, specifically including the following steps: The feature similarity between non-pilot plots and potential optimal cases is calculated using the following formula: , in, For feature similarity, For Euclidean distance, For cosine similarity, These are the weighting coefficients, and the feature similarity is standardized. For each potential optimal case, search for the top-ranking most similar non-pilot plots of the same type. For each plot of land, a priority development list is generated. The plots in the list are sorted from high to low according to their similarity scores, and their respective categories and the best matching cases are marked.

[0013] Optionally, the spatial connectivity analysis and community division algorithm in step S4 is used to identify the spatial relationships between plots and automatically delineate update units, specifically including the following steps: First, spatial connectivity analysis is performed, which uses a spatial adjacency matrix to describe the adjacency relationship between plots, where adjacent plots are assigned a value of 1, and otherwise 0; Then, the community partitioning algorithm, namely the Girvan-Newman algorithm, is used for community discovery. By gradually removing the edge with the highest betweenness, the community is identified and the plot is divided into multiple communities, which are the update units. The final update unit delineation involves designating plots within the same community as the same update unit, ensuring that plots within the unit are spatially adjacent, functionally similar, and highly connected.

[0014] Optionally, the morphological rule correction boundary in step S4 is achieved through a combination of geometric morphological operations and spatial constraints, used to ensure the continuity and rationality of the update unit boundary, specifically including the following steps: First, compactness is calculated, that is, the compactness of the update unit is calculated using the Cole compactness algorithm. The calculation formula is: , in, To update the actual area of ​​the cell, To update the minimum circumcircle area of ​​the cell, the compactness threshold is set to 0.3; cells with a value lower than this need to be shaped. Next is the minimum area constraint, which sets a minimum area threshold for the update unit. Units below the minimum area threshold need to be merged with adjacent units. Units with too small an area can be merged into adjacent units through morphological dilation operations. Finally, boundary correction is performed, which involves using Gaussian filtering to smooth the cell boundaries, avoiding jagged edges, and filling the holes within the cells with morphological closing operations to ensure boundary continuity.

[0015] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: (1) This invention integrates land vitality, environmental and spatial quality information through multi-source data collection and standardization, and generates a strategy list in conjunction with the renewal measures database; in the potential identification stage, multi-objective evaluation and clustering algorithms are used to classify land parcels and select pilot sites; in the process tracking stage, the time context spectrum method is used to record the iteration and benefit changes of renewal measures to form a dynamic spectrum structure; in the benefit evaluation stage, efficiency frontier and causal inference models are introduced to identify the effectiveness of strategies; in the promotion stage, based on similarity matching and spatial connectivity analysis, it is extended to non-pilot land parcels to form a replicable renewal unit division mechanism; it realizes the whole-process optimization of data-driven, strategy coordination and feedback loop, and can effectively support the intelligent and refined implementation of small-scale gradual urban renewal. (2) This invention achieves standardized integration of multi-source heterogeneous data, improving the accuracy and processing efficiency of land parcel information; it achieves accurate alignment of heterogeneous data in spatial and temporal dimensions, and the data matching accuracy is improved from about 85% of the traditional method to 98.6%; the optimized parallel data cleaning and feature extraction algorithm can complete the update of about 1,200 land parcel samples within 1 square kilometer within 30 minutes, and the data processing efficiency is improved by about 4.3 times compared with the manual method; it significantly shortens the data preparation cycle from the original average of 5 days to less than 1 day, providing a high-precision and low-latency data foundation for subsequent model analysis; (3) This invention constructs an automated path from potential identification to unit delineation, significantly improving the efficiency of update screening and spatial delineation; through multi-objective benefit evaluation and efficiency frontier analysis models, it achieves automated screening of land parcel update potential, with a classification accuracy of 94.2%; in the morphological clustering and spatial connectivity analysis stages, the boundary error of land parcel division is controlled within... Within a 100-meter radius, the speed of spatial unit delineation has increased by approximately 3.7 times, and the delineation cycle for each urban area has been reduced from the original 15 working days to 4 working days; it supports the rapid formation of a "priority renewal list" and renewal units with continuous boundaries under complex urban conditions, effectively improving the responsiveness and scientific nature of urban design and planning decisions; (4) This invention achieves accurate identification and dynamic optimization of strategy effectiveness based on causal inference, improving evaluation stability and feedback cycle; it uses a difference-in-differences (DID) model to identify the average treatment effect of a single strategy, with estimation error controlled within a certain range. Within a certain range; after robustness testing by combining Regression Discontinuity Decision Making (RDD) and Propensity Score Matching (PSM), the consistency of the results was improved to over 97%; in terms of model computation efficiency, the strategy effectiveness evaluation process can complete the full-process analysis of 500 plots within 12 hours, which is about 5 times faster than traditional statistical methods; through monthly monitoring and spectrum feedback mechanisms, the strategy adjustment cycle has been shortened from the original 90 days to within 30 days, realizing a closed-loop dynamic synergy between policy optimization and urban renewal. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the overall method of an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the updated potential evaluation and pilot selection process in this invention; Figure 3 This is a schematic diagram of the pilot update monitoring and temporal context spectrum construction in this invention; Detailed Implementation

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0018] like Figure 1 As shown, the present invention provides a strategy selection and intelligent classification method for urban renewal units driven by time-series data, comprising the following steps: S1. Source Data Acquisition and Standardization Basic information on population activity, environmental quality, and spatial quality for each plot of land is obtained through remote sensing imagery, street view, and pedestrian video surveillance. Coordinate unification and data cleaning techniques are used to align data from different sources to the same base map, creating a directly comparable sample database of plots. Information on winning bids for update measures from government procurement websites is collected, and a text mining model is used to compile an "update strategy thesaurus," forming a list of update strategies and parameter templates. Strategy types include pedestrian optimization, pocket parks, interface updates, and function replacements. The update strategy list and parameter templates include strategy type, update time, and update cost.

[0019] S2. Update potential assessment and pilot selection, such as Figure 2 As shown; The renewal potential score of the land parcels was calculated using a multi-objective benefit evaluation model, and a set of land parcels to be renewed was selected using a threshold screening rule. The land parcels to be renewed were divided into 5 categories using a morphological-target feature clustering algorithm, and the 3 land parcels with the lowest potential scores in each category were selected as pilot projects, forming a test list of 15 pilot projects.

[0020] Step S2 of the multi-objective benefit evaluation model consists of three major categories of evaluation indicators: comprehensive population vitality, ecological environment, and spatial quality. The weights of the indicators are determined using a combined weighting method, and a comprehensive benefit score is synthesized through linear weighting. The combined weighting method refers to a combination of the analytic hierarchy process (AHP) and the entropy weighting method. The specific indicator framework is as follows: Comprehensive population vitality indicators include pedestrian density, nighttime vitality index, and social interaction intensity; ecological environment indicators include green coverage rate, air pollution index, and heat island intensity; spatial quality indicators include public service facility density, street interface permeability, and architectural style integrity. The calculation formula for the multi-objective benefit evaluation model is as follows: ,in For comprehensive benefit scoring, These are standardized values ​​for comprehensive indicators of population vitality, ecological environment, and spatial quality, specifically using Min-Max standardization. The weighting coefficients are calculated using a combination of the Analytic Hierarchy Process (AHP) and the entropy weighting method, with AHP accounting for 40% of the subjective weight and the entropy weighting method accounting for 60% of the objective weight.

[0021] In step S2, the threshold screening rule adopts the dynamic quantile threshold method. The specific steps are as follows: First, calculate the comprehensive benefit score of all plots. The scores are sorted in descending order. Then, a window-based variable point detection algorithm is used to identify natural breakpoints in the scoring sequence, which serve as the basis for threshold adjustment. Next, the top 30% of the plots are selected as the set of plots to be updated. If a natural breakpoint is located near the 30% quantile (near means within ±5%), then the natural breakpoint is used as the actual threshold. Finally, the set of plots to be updated is output, and a potential distribution map is generated.

[0022] Step S2, the morphology-target feature clustering algorithm, is an improved clustering method that integrates multi-scale morphological features and target benefit features. Specifically, it includes the following steps: First, morphological features and target benefit features are extracted. Morphological features include building density, plot ratio, street width, and plot shape index. Target benefit features include population vitality score, ecological environment score, spatial quality score, and comprehensive benefit score. After data standardization, the K-means++ algorithm is used for preliminary clustering, dividing the data into 5 categories, and the morphological-target feature outlines of each category are labeled. Finally, the 3 plots with the lowest comprehensive benefit scores from each cluster are selected as update pilots, forming a test list of 15 pilots.

[0023] S3. Pilot update monitoring and temporal context genealogy construction, such as Figure 3 As shown; The update strategy list and parameter template collected in step S1 are input, and the update measures are allocated to each pilot project through a restricted randomization algorithm. The changes in the multi-objective benefits of each pilot project are recorded through fixed-period monitoring, which can be once a week, to form a monitoring dataset. The update monitoring dataset is integrated through a phylogenetic modeling method, and branches and merging are marked when strategies are superimposed, replaced or linked, to construct a time-line phylogenetic system of "event-version-branch-merge tree".

[0024] In step S3, the timeline is structured by treating each update as an "event," generating a corresponding "version" after each event takes effect, and then sequentially listing them over time. When strategies overlap or are replaced on the same plot of land, a "split" is generated, while the rest are "continuous connections" that progress over time, forming a directional version hierarchy. Multi-objective benefits are recorded simultaneously. and the increase in benefits in this period Simultaneously record edge type indication ;pass The minimal set of fields provides standardized input for subsequent efficiency frontier assessment and causal inference.

[0025] S4. Benefit Frontier Assessment and Update Unit Promotion Division Efficiency frontier modeling was used to plot the efficiency frontier between multi-objective benefit improvement and renewal cost of pilot projects. The pilot projects with the highest efficiency within each class were selected as potential optimal cases, and their efficiency was used as representative values ​​to rank and classify them into 5 renewal categories. Similarity matching algorithm was used to match non-pilot plots of the same type with plots with similar conditions to potential optimal cases, generating a priority development list. Spatial connectivity analysis and community division algorithm were used to automatically delineate adjacent, similar, and connected plots into the same renewal unit, and morphological rules were used to correct the boundaries to ensure that the boundaries are continuous and have reasonable shapes.

[0026] In step S4, the efficiency frontier model is used to evaluate the trade-offs between the multi-objective benefits and costs of the pilot renewal project, specifically including the following steps: Firstly, regarding the cost and benefit indicators for renovation, the cost indicators include the cost per unit area, time cost, and coordination cost. The benefit indicators include a comprehensive score of the improvement in population vitality, the improvement in the ecological environment, and the improvement in spatial quality. The overall improvement rate of multi-objective benefits is then calculated using the following formula: , in, These represent indicators of population vitality, ecological environment, and spatial quality, respectively. These represent "before" and "after" respectively. The weighting coefficients are determined using the Analytic Hierarchy Process (AHP). Finally, using the update cost as the horizontal axis and the comprehensive improvement rate of multi-objective benefits as the vertical axis, a scatter plot of all pilot projects was drawn, i.e., an efficiency frontier plot.

[0027] The potential optimal case in step S4 is selected by combining the efficiency frontier model with cluster analysis, specifically including the following steps: First, based on the efficiency frontier plot, identify the pilots located on the efficiency frontier in each cluster category; then calculate the overall efficiency value for each pilot. The specific calculation formula is as follows: , in, Indicates the first The input metric is the update cost; Indicates the first The output indicator is the efficiency improvement rate; and These are the weights of the input and output metrics, respectively. To input the number of indicators, The number of output metrics; Next, the overall efficiency values ​​are sorted from high to low, and the sorted pilots are assigned levels to form 5 categories of update levels, from level 1 to level 5, where level 1 represents the highest efficiency and level 5 represents the lowest efficiency. Finally, the top-ranked pilot in each category is selected as the potential optimal case, representing the highest update efficiency level for that category.

[0028] Step S4 uses a similarity matching algorithm that combines multi-feature similarity calculation with nearest neighbor search. This algorithm identifies plots of similar type among non-pilot plots that are similar to potential optimal cases. Specifically, it includes the following steps: The feature similarity between non-pilot plots and potential optimal cases is calculated using the following formula: , in, For feature similarity, For Euclidean distance, For cosine similarity, These are the weighting coefficients, and the feature similarity is standardized. For each potential optimal case, search for the top-ranking most similar non-pilot plots of the same type. One plot of land, Typically, 10-20 plots are selected to generate a priority development list. The plots in the list are sorted from high to low according to their similarity scores, and their respective categories and the best matching cases are marked.

[0029] In step S4, the spatial connectivity analysis and community delineation algorithm is used to identify the spatial relationships between plots and automatically delineate update units. Specifically, it includes the following steps: First, spatial connectivity analysis is performed, which uses a spatial adjacency matrix to describe the adjacency relationship between plots, where adjacent plots are assigned a value of 1, and otherwise 0; Then, the community partitioning algorithm, namely the Girvan-Newman algorithm, is used for community discovery. By gradually removing the edge with the highest betweenness, the community is identified and the plot is divided into multiple communities, which are the update units. The final update unit delineation involves designating plots within the same community as the same update unit, ensuring that plots within the unit are spatially adjacent, functionally similar, and highly connected.

[0030] In step S4, the morphological rule correction boundary is achieved through a combination of geometric morphological operations and spatial constraints. This ensures the continuity and rationality of the updated unit boundary, and specifically includes the following steps: First, compactness is calculated, that is, the compactness of the update unit is calculated using the Cole compactness algorithm. The calculation formula is:

[0031] in, To update the actual area of ​​the cell, To update the minimum circumcircle area of ​​the cell, the compactness threshold is set to 0.3; cells with a value lower than this need to be shaped. Next is the minimum area constraint, which sets a minimum area threshold for the update unit, usually 1 hectare. Units with an area below the minimum area threshold need to be merged with adjacent units. Units with too small an area can be merged into adjacent units through morphological expansion operations. Finally, boundary correction is performed, which involves using Gaussian filtering to smooth the cell boundaries, avoiding jagged edges, and filling the holes within the cells with morphological closing operations to ensure boundary continuity.

[0032] S5. Causal Inference of Strategy Effectiveness and Coordination of Strategy Matching The difference-in-differences (DID) model is used to compare the benefit changes of pilot and non-implemented plots to identify the average treatment effect of a single strategy. For different plot types and heterogeneous characteristics, Regression Discontinuity Disorder (RDD) and Propensity Score Matching (PSM) methods are used for supplementary verification to ensure the robustness of the results. The effectiveness results of all strategies are input into a multi-objective combination optimization model to automatically generate the optimal strategy combination for updated plots. The deviation between the actual benefits and the model's expectations is monitored monthly. If the expected results are not met, the time context spectrum is used for feedback adjustment and strategy reconfiguration.

[0033] In step S5, the difference-in-differences (DID) model is used to compare the benefit changes of pilot and non-implemented plots, eliminate interference from common time trends, and identify the average treatment effect of a single update strategy; the calculation formula is as follows:

[0034] in, Individual In time The outcome variable; Refers to grouped dummy variables; if individuals For the treatment group, the value is 1; for the control group, the value is 0. Refers to a time-based dummy variable; if time After the update, the value is 1; before the update, the value is 0. The average treatment effect; This is the baseline constant term, representing the average benefit level of the control group before implementation; This indicates the inherent mean difference between the pretreatment group and the control group; This indicates the common time trend of all plots of land; This is the random error term.

[0035] S6. Visual Presentation and Decision Release The system utilizes a large digital screen supporting 4K resolution to display the update timeline, update hierarchy, and priority list; combined with augmented reality (AR) devices that support spatial positioning and 3D overlay, updated plots are overlaid in real-world scenes to assist planners and residents in immersive experiences and assessments.

Claims

1. A method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series data, characterized in that, Includes the following steps: S1. Source data collection and standardization: Obtain basic information on population activity, environmental quality, and spatial quality for each plot through remote sensing imagery, street view, and pedestrian flow video surveillance; Align data from different sources to the same base map through coordinate unification and data cleaning techniques to form a plot sample library that can be directly compared; Collect bidding information on update measures from the government procurement website, and use text mining models to sort out the "update strategy terminology" to form an update strategy list and parameter template; S2. Renewal potential evaluation and pilot selection: The renewal potential score of the plots is calculated by the multi-objective benefit evaluation model, and the set of plots to be renewed is selected by the threshold screening rule; The plots to be renewed are divided into 5 categories by the morphological-target feature clustering algorithm, and the 3 plots with the lowest potential scores in each category are selected as pilots, forming a test list of 15 pilots in total. S3. Pilot update monitoring and timeline genealogy construction: Input the update strategy list and parameter template collected in step S1, and allocate update measures to each pilot using a restricted randomization algorithm; record the changes in multi-objective benefits of each pilot through fixed-period monitoring, which can be once a week, to form a monitoring dataset; integrate the update monitoring dataset through genealogy modeling methods, and mark branches and merges when encountering strategy superposition, replacement or adjacent linkage, to construct a timeline genealogy of "event-version-divide-merge tree"; S4. Benefit Frontier Assessment and Renewal Unit Promotion Division: Using the efficiency frontier model, draw the efficiency frontier diagram of the pilot projects between multi-objective benefit improvement and renewal costs. Select the pilot projects with the highest efficiency in the class to form potential optimal cases. Use their efficiency as representative values ​​to sort and assign grades to form 5 types of renewal classification. The similarity matching algorithm is used to match plots with similar conditions to potential optimal cases among non-pilot plots of the same type, and a priority development list is generated; through spatial connectivity analysis and community division algorithm, adjacent, similar and connected plots are automatically delineated into the same update unit, and the boundaries are corrected with morphological rules to ensure that the boundaries are continuous and the shapes are reasonable; S5. The causal inference of strategy effectiveness and strategy matching are coordinated. The difference-in-differences (DID) model is used to compare the benefit changes of pilot and non-implemented plots to identify the average treatment effect of a single strategy. For different plot types and heterogeneous characteristics, the regression discontinuity (RDD) and propensity score matching (PSM) methods are used for supplementary verification to ensure the robustness of the results. The effectiveness results of all strategies are input into a multi-objective combination optimization model to automatically generate the optimal strategy combination for updated plots. The deviation between the actual benefits and the model's expectations is monitored monthly. If the expected results are not met, the time context spectrum is used for feedback adjustment and strategy reconfiguration. S6. Visual presentation and decision release.

2. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series hierarchy as described in claim 1, characterized in that, The multi-objective benefit evaluation model in step S2 consists of three major categories of evaluation indicators: comprehensive population vitality, ecological environment, and spatial quality. The weights of the indicators are determined using a combined weighting method, and a comprehensive benefit score is synthesized through linear weighting. The combined weighting method refers to a combination of the analytic hierarchy process (AHP) and the entropy weighting method. The specific indicator framework is as follows: Comprehensive population vitality indicators include pedestrian density, nighttime vitality index, and social interaction intensity; ecological environment indicators include green coverage rate, air pollution index, and heat island intensity; spatial quality indicators include public service facility density, street interface permeability, and architectural style integrity. The calculation formula for the multi-objective benefit evaluation model is as follows: ， in For comprehensive benefit scoring, These are standardized values ​​for comprehensive indicators of population vitality, ecological environment, and spatial quality, specifically using Min-Max standardization. The weighting coefficients are calculated using a combination of the Analytic Hierarchy Process (AHP) and the entropy weighting method, with AHP accounting for 40% of the subjective weight and the entropy weighting method accounting for 60% of the objective weight.

3. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series genealogy as described in claim 1, characterized in that, The threshold screening rule in step S2 adopts the dynamic quantile threshold method. The specific steps are as follows: First, calculate the comprehensive benefit score of all plots. The scores are sorted in descending order. Then, a window-based variable point detection algorithm is used to identify natural breakpoints in the scoring sequence, which serve as the basis for threshold adjustment. Next, the top 30% of the plots are selected as the set of plots to be updated. If a natural breakpoint is located near the 30% quantile (near means within ±5%), then the natural breakpoint is used as the actual threshold. Finally, the set of plots to be updated is output, and a potential distribution map is generated.

4. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series hierarchy as described in claim 1, characterized in that, The morphological-target feature clustering algorithm in step S2 is an improved clustering method that integrates multi-scale morphological features and target benefit features, and specifically includes the following steps: First, morphological features and target benefit features are extracted. Morphological features include building density, plot ratio, street width, and plot shape index. Target benefit features include population vitality score, ecological environment score, spatial quality score, and comprehensive benefit score. After data standardization, the K-means++ algorithm is used for preliminary clustering, dividing the data into 5 categories, and the morphological-target feature outlines of each category are labeled. Finally, the 3 plots with the lowest comprehensive benefit scores from each cluster are selected as update pilots, forming a test list of 15 pilots.

5. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series genealogy as described in claim 1, characterized in that, The timeline in step S3 treats each update as an "event," generates a corresponding "version" after the event takes effect, and strings them together chronologically. When strategies overlap or are replaced on the same plot of land, a "split" is generated, while the rest are "continuous connections" that advance over time, forming a directional version hierarchy. Multi-objective benefits are recorded simultaneously. and the increase in benefits in this period Simultaneously record edge type indication , ;pass The minimal set of fields provides standardized input for subsequent efficiency frontier assessment and causal inference.

6. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series genealogy as described in claim 1, characterized in that, In step S4, the efficiency frontier model is used to evaluate the trade-off between the multi-objective benefits and costs of the pilot renewal project, specifically including the following steps: Firstly, regarding the cost and benefit indicators for renovation, the cost indicators include the cost per unit area, time cost, and coordination cost. The benefit indicators include a comprehensive score of the improvement in population vitality, the improvement in the ecological environment, and the improvement in spatial quality. The overall improvement rate of multi-objective benefits is then calculated using the following formula: ， in, These represent indicators of population vitality, ecological environment, and spatial quality, respectively. These represent "before" and "after" respectively. The weighting coefficients are determined using the Analytic Hierarchy Process (AHP). Finally, using the update cost as the horizontal axis and the comprehensive improvement rate of multi-objective benefits as the vertical axis, a scatter plot of all pilot projects was drawn, i.e., an efficiency frontier plot.

7. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series genealogy as described in claim 1, characterized in that, The potential optimal case in step S4 is selected by combining the efficiency frontier model with cluster analysis, specifically including the following steps: First, based on the efficiency frontier plot, identify the pilots located on the efficiency frontier in each cluster category; then calculate the overall efficiency value for each pilot. The specific calculation formula is as follows: ， in, Indicates the first The input metric is the update cost; Indicates the first The output indicator is the efficiency improvement rate; and These are the weights of the input and output metrics, respectively. To input the number of indicators, The number of output metrics; Next, the overall efficiency values ​​are sorted from high to low, and the sorted pilots are assigned levels to form 5 categories of update levels, from level 1 to level 5, where level 1 represents the highest efficiency and level 5 represents the lowest efficiency. Finally, the top-ranked pilot in each category is selected as the potential optimal case, representing the highest update efficiency level for that category.

8. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series genealogy as described in claim 1, characterized in that, The similarity matching algorithm in step S4 is a matching method that integrates multi-feature similarity calculation and nearest neighbor search. It is used to identify plots with similar conditions to potential optimal cases among non-pilot plots of the same type. Specifically, it includes the following steps: The feature similarity between non-pilot plots and potential optimal cases is calculated using the following formula: ， in, For feature similarity, For Euclidean distance, For cosine similarity, These are the weighting coefficients, and the feature similarity is standardized. For each potential optimal case, search for the top-ranking most similar non-pilot plots of the same type. For each plot of land, a priority development list is generated. The plots in the list are sorted from high to low according to their similarity scores, and their respective categories and the best matching cases are marked.

9. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series genealogy as described in claim 1, characterized in that, In step S4, the spatial connectivity analysis and community delineation algorithm is used to identify the spatial relationships between land parcels and automatically delineate update units. Specifically, it includes the following steps: First, spatial connectivity analysis is performed, which uses a spatial adjacency matrix to describe the adjacency relationship between plots, where adjacent plots are assigned a value of 1, and otherwise 0; Then, the community partitioning algorithm, namely the Girvan-Newman algorithm, is used for community discovery. By gradually removing the edge with the highest betweenness, the community is identified and the plot is divided into multiple communities, which are the update units. The final update unit delineation involves designating plots within the same community as the same update unit, ensuring that plots within the unit are spatially adjacent, functionally similar, and highly connected.

10. The method for strategy selection and intelligent hierarchical classification of urban renewal units driven by time-series genealogy as described in claim 1, characterized in that, In step S4, the morphological rule correction boundary is achieved through a combination of geometric morphological operations and spatial constraints. This ensures the continuity and rationality of the update unit boundary, and specifically includes the following steps: First, compactness is calculated, that is, the compactness of the update unit is calculated using the Cole compactness algorithm. The calculation formula is: ， in, To update the actual area of ​​the cell, To update the minimum circumcircle area of ​​the cell, the compactness threshold is set to 0.3; cells with a value lower than this need to be shaped. Next is the minimum area constraint, which sets a minimum area threshold for the update unit. Units below the minimum area threshold need to be merged with adjacent units. Units with too small an area can be merged into adjacent units through morphological dilation operations. Finally, boundary correction is performed, which involves using Gaussian filtering to smooth the cell boundaries, avoiding jagged edges, and filling the holes within the cells with morphological closing operations to ensure boundary continuity.

Images