Urban renewal district intelligent decision-making method and system based on machine learning and genetic algorithm

Abstract

The present application relates to the technical field of machine learning, and more particularly to a city renewal area intelligent decision-making method and system based on machine learning and genetic algorithm, comprising collecting historical city renewal data of a city to be decided; screening key factors in the historical city renewal data based on features; using the key factors and city renewal types as training data of a first machine learning model and a second machine learning model respectively, predicting the area to be updated by the trained first machine learning model, predicting the update type of the area to be updated by the trained second machine learning model, and generating an implementation time sequence scheme of the city renewal area by a genetic algorithm. The present application realizes intelligent decision-making in the whole process, effectively solving the problems of strong subjectivity, low efficiency and insufficient system in traditional planning methods.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a method and system for intelligent decision-making in urban renewal areas based on machine learning and genetic algorithms. Background Technology

[0002] Big data technology has been gradually applied to urban research, but existing technical solutions are mostly limited to the shallow use of single-type data sources. They lack in-depth integration and mining of multimodal data such as building entities, socio-economic data, and environmental behavior data. They are unable to systematically learn and extract patterns from a large number of historical update cases, thus restricting the scientific nature and accuracy of the decision-making process.

[0003] Currently, machine learning models are being used to identify inefficiently used land, and deep learning algorithms are assisting in the initial determination of whether to retain, modify, or demolish urban renewal units. However, these technological applications exhibit a clear "fragmented" characteristic, mostly focusing only on a single isolated link in the entire planning process, such as current potential identification or automatic scheme generation. They fail to construct a complete technological chain covering everything from intelligent understanding of the current situation and configuration of renewal methods to time-series optimization decisions. The breaks and gaps between these links make it difficult for intelligent analysis results to directly and systematically guide the preparation of holistic and implementable regional urban renewal plans. Summary of the Invention

[0004] To overcome the shortcomings of the prior art, such as strong subjectivity and low data utilization, this invention provides an intelligent decision-making method and system for urban renewal areas based on machine learning and genetic algorithms.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A smart decision-making method for urban renewal areas based on machine learning and genetic algorithms includes the following steps:

[0007] Collect historical urban renewal data for the city to be decided, including urban renewal data of different dimensions and different renewal types;

[0008] Key factors are obtained from urban renewal data of different dimensions based on feature selection; the historical urban renewal data corresponding to the key factors are used as training data to train the first machine learning model; the first machine learning model is trained to predict the renewal potential of the city to be decided, and the areas within the city to be decided that need to be renewed are obtained.

[0009] The second machine learning model is trained using the different update types and their corresponding historical urban update data as training data; the updated type is predicted for the area that needs to be updated using the trained second machine learning model.

[0010] For the area that needs to be updated, a multi-objective optimization model is constructed with multiple preset indicators as the target. Under preset constraints, the optimization model is solved based on a genetic algorithm to obtain the implementation sequence plan of the urban renewal area.

[0011] As a preferred option, the urban renewal data of different dimensions includes spatial morphology data, socio-economic data, and planning management data.

[0012] As a preferred embodiment, the step of obtaining key factors from the urban renewal data of different dimensions based on feature screening includes:

[0013] Extract feature variables from data of different dimensions, use any one feature variable as the dependent variable and the remaining feature variables as independent variables to perform linear regression, and calculate the variance inflation factor (VIF) value of each feature variable; and remove feature variables that are greater than the preset VIF threshold to obtain a set of feature variables.

[0014] Select the feature variable set whose cumulative VIF variance contribution rate reaches a preset contribution rate threshold. m Each principal component serves as a key factor; its expression is as follows:

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021] in, for p One characteristic variable, For the first j One characteristic variable, For the first j VIF values ​​of each feature variable For the first j Coefficient of determination under linear regression of individual characteristic variables Let be the fitting coefficients for each independent variable. For random error, This represents the original data matrix after standardization. For this is the first k The score vector of each principal component. It is the first k 1 eigenvector It is a matrix The k 1 eigenvalue, Indicates the first k The percentage of variance of each principal component relative to the total information content of the original data. For the front k The cumulative variance contribution rate of each principal component.

[0022] As a preferred option, the first machine learning model is a random forest classification model, and the hyperparameters within the first machine learning model are adjusted through grid search and 10-fold cross-validation.

[0023] As a preferred option, the second machine learning model is a random forest classification model, and the hyperparameters within the second machine learning model are adjusted using Bayesian optimization algorithm and five-fold cross-validation.

[0024] As a preferred option, the second machine learning model outputs the probability of different renewal types for the urban renewal area, takes the renewal type with the highest probability as the prediction result for the urban renewal area, and performs spatial optimization of the urban renewal area based on morphological opening and closing operations.

[0025] As a preferred embodiment, the step of solving the problem based on a genetic algorithm under preset constraints includes:

[0026] An initial timeline scheme is generated based on the urban renewal area; non-dominated sorting and congestion calculation are performed using the preset indicators, and iterative optimization is carried out under the constraints until the preset iteration conditions are met, so as to obtain the implementation timeline scheme of the urban renewal area.

[0027] As a preferred option, the multiple preset indicators include economic, social, and environmental objectives; wherein, the economic objectives include land appreciation revenue and renewal costs; the social objectives include the balance of public services and social inclusion; and the environmental objectives are quantified based on the equivalent value of ecosystem services and thermal environment simulation.

[0028] As a preferred embodiment, the land appreciation revenue is determined based on benchmark land price, plot ratio, land use coefficient, and location coefficient; the renewal cost is determined based on comprehensive cost per unit area, renewal area, and adjustment coefficients based on renewal type and location, and its expression is as follows:

[0029]

[0030]

[0031] in, For land revenue, , , These are plot ratio, land use coefficient, and location coefficient, respectively. , These are the benchmark land price and the current land price, respectively. To cover replacement costs, , These represent the comprehensive cost per unit area and the land area of ​​the renewal zone, respectively. This is a differential adjustment coefficient based on update type and location.

[0032] This invention also proposes an intelligent decision-making system for urban renewal areas based on machine learning and genetic algorithms, the system comprising:

[0033] The data acquisition module is used to collect historical urban renewal data of the city to be decided. The historical urban renewal data includes urban renewal data of different dimensions and different renewal types.

[0034] The update potential prediction module is used to obtain key factors from urban renewal data of different dimensions based on feature screening; train the first machine learning model with the historical urban renewal data corresponding to the key factors as training data; and predict the renewal potential of the city to be decided through the trained first machine learning model to obtain the areas within the city to be decided that need to be updated.

[0035] The update type prediction module is used to train the second machine learning model using the different update types and their corresponding historical urban update data as training data; and to predict the update type of the area to be updated using the trained second machine learning model.

[0036] The time-series planning module is used to construct a multi-objective optimization model for the area to be updated with multiple preset indicators as targets, and solve the optimization model based on a genetic algorithm under preset constraints to obtain the implementation time-series plan for the urban renewal area.

[0037] Compared with the prior art, the beneficial effects of the technical solution of the present invention are as follows:

[0038] This invention utilizes a first machine learning model to evaluate and delineate urban renewal potential areas. Based on this, a second machine learning model intelligently configures the most suitable renewal type for each urban renewal area, enhancing the rationality of the plan. Finally, a multi-objective genetic algorithm is employed to optimize the project implementation sequence, achieving dynamic equilibrium under multiple objectives. This realizes intelligent decision-making throughout the entire process, from potential identification to timeline scheduling, effectively solving the problems of strong subjectivity, low efficiency, and insufficient systematicity in traditional planning methods. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms in Example 1.

[0040] Figure 2 This is an architecture diagram of an intelligent decision-making system for urban renewal areas based on machine learning and genetic algorithms, as shown in Example 2. Detailed Implementation

[0041] In the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, they do not mean that the applicant has used or necessarily used the solution.

[0042] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent.

[0043] To better illustrate this embodiment, some parts in the accompanying drawings may be omitted, enlarged, or reduced, and do not represent the actual product dimensions.

[0044] It will be understood by those skilled in the art that certain well-known structures and their descriptions may be omitted in the accompanying drawings.

[0045] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0046] Example 1

[0047] This embodiment proposes an intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms, such as... Figure 1 The diagram shown is a flowchart of an intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms, according to this embodiment.

[0048] A smart decision-making method for urban renewal areas based on machine learning and genetic algorithms includes the following steps:

[0049] S1. Collect historical urban renewal data of the city to be decided, including urban renewal data of different dimensions and different renewal types;

[0050] S2. Based on feature filtering, key factors are obtained from the urban renewal data of different dimensions; the historical urban renewal data corresponding to the key factors are used as training data to train the first machine learning model; the first machine learning model is trained to predict the renewal potential of the city to be decided, and the areas within the city to be decided that need to be renewed are obtained.

[0051] S3. The second machine learning model is trained using the different update types and their corresponding historical urban update data as training data; the update type is predicted for the area that needs to be updated using the trained second machine learning model.

[0052] S4. For the area that needs to be updated, a multi-objective optimization model is constructed with multiple preset indicators as the target. Under preset constraints, the optimization model is solved based on a genetic algorithm to obtain the implementation sequence plan of the urban renewal area.

[0053] In this embodiment, an automated and intelligent urban renewal decision-making framework from data to decision is constructed. Using feature filtering and a first machine learning model, urban renewal areas with potential are identified from multi-dimensional data. Next, a second machine learning model predicts the renewal type of each area, ensuring the matching degree between the renewal strategy and the area's characteristics. Finally, a genetic algorithm is used for global optimization under multi-objective constraints to generate a timeline of implementation plans that considers multiple benefits, significantly improving the scientific rigor, systematic nature, and operability of urban renewal decisions.

[0054] In one optional embodiment, the urban renewal data of different dimensions includes spatial morphology data, socio-economic data, and planning management data.

[0055] As an example, spatial morphology data is obtained through remote sensing image interpretation, oblique photogrammetry, and building surveys, including building renewal patches, building outlines before and after renewal, number of floors, building age, building structure, building density, floor area ratio, road network data, and current land use. Socioeconomic data is obtained through publicly available data channels, including housing price levels, nighttime light intensity, population density, POI (Point of Interest) data, and social media data. Planning and management data includes historical urban renewal project patches, regulatory detailed plans, and urban design guidelines. All data undergoes preprocessing operations such as cleaning, format conversion, and coordinate unification to ensure data quality and consistency.

[0056] More specifically, the integrated data undergoes standardization and spatialization processing to establish a unified spatial reference system, registering all vector and raster data. Secondly, based on research needs, standardized spatial analysis units are defined; as an example, a regular grid system covering the entire research area is established. Then, various types of data are assigned to each spatial analysis unit through spatial association and attribute connections, forming a spatial grid database containing rich attribute information.

[0057] In this embodiment, data from three dimensions—spatial morphology, socio-economic factors, and planning management—are incorporated to construct a comprehensive and three-dimensional data foundation for urban renewal analysis, thus avoiding decision-making biases caused by a single data dimension.

[0058] More specifically, based on the theoretical framework of "architectural form-spatial structure-socioeconomic factors," this step systematically constructs a multi-dimensional system of urban renewal driving factors. The architectural form dimension includes indicators such as building area, building density, building shape index, compactness, and number of floors, reflecting the physical characteristics and spatial layout of buildings. The spatial structure dimension includes indicators such as distance to the city center, distance to transportation hubs, distance to commercial centers, and land use type, reflecting the accessibility and functional complexity of the location. The socioeconomic dimension includes indicators such as nighttime light intensity, housing price levels, and building age, reflecting the region's economic vitality and development stage.

[0059] In an optional embodiment, the step of obtaining key factors from the urban renewal data of different dimensions based on feature filtering includes:

[0060] Extract feature variables from data of different dimensions, use any one feature variable as the dependent variable and the remaining feature variables as independent variables to perform linear regression, and calculate the variance inflation factor (VIF) value of each feature variable; and remove feature variables that are greater than the preset VIF threshold to obtain a set of feature variables.

[0061] Select the feature variable set whose cumulative VIF variance contribution rate reaches a preset contribution rate threshold. m Each principal component serves as a key factor; its expression is as follows:

[0062]

[0063]

[0064]

[0065]

[0066]

[0067]

[0068] in, for p One characteristic variable, For the first j One characteristic variable, For the first j VIF values ​​of each feature variable For the first j Coefficient of determination under linear regression of individual characteristic variables Let be the fitting coefficients for each independent variable. For random error, This represents the original data matrix after standardization. For this is the firstk The score vector of each principal component. It is the first k 1 eigenvector It is a matrix The k 1 eigenvalue, Indicates the first k The percentage of variance of each principal component relative to the total information content of the original data. For the front k The cumulative variance contribution rate of each principal component.

[0069] In this embodiment, by using variance inflation factor (VIF) analysis and principal component analysis (PCA) to map the high-dimensional feature space to the low-dimensional key factor space while preserving the original information, the dimensionality and noise of the subsequent machine learning model input are reduced, allowing the model to focus more on learning the core driving factors, thereby improving the accuracy and robustness of the updated potential prediction.

[0070] In one optional embodiment, the first machine learning model is a random forest classification model, and the hyperparameters within the first machine learning model are adjusted through grid search and 10-fold cross-validation.

[0071] More specifically, grid units that have undergone urban renewal in the past are used as positive samples, and grid units that have not undergone renewal are used as negative samples. Based on the prediction results of the first machine learning model, the natural breakpoint method is used to scientifically divide the output probability values ​​into different renewal potential levels, and this determines whether an area is an urban renewal area requiring a certain type of renewal. Then, a density-based spatial clustering algorithm is used, considering spatial proximity, similarity of renewal potential values, and scale thresholds determined in planning practice, to aggregate discrete grid units into several spatially contiguous areas, which are then considered as new independent areas.

[0072] In one optional embodiment, the second machine learning model is a random forest classification model, and the hyperparameters within the second machine learning model are adjusted using a Bayesian optimization algorithm and five-fold cross-validation.

[0073] As an example, update types include "demolition and reconstruction", "comprehensive renovation" and "functional replacement".

[0074] Furthermore, based on the three-dimensional theoretical framework of "building form - property rights status - functional attributes," a more refined discrimination index system is constructed for different types. The building form dimension includes building density, average number of floors, structural type, year of construction, and plot ratio, reflecting the physical state and quality of the building; the property rights status dimension covers the clarity of property rights and the complexity of land ownership, reflecting new legal and policy obstacles; and the functional attributes dimension includes the dominant functional type and business diversity, reflecting the functional positioning and vitality of the area. A combination of variance inflation factor analysis and principal component analysis is used to eliminate redundant indicators and retain the core indicators with the strongest discriminative power, forming the optimal subset of indicators.

[0075] In one optional embodiment, the second machine learning model outputs the probability of different renewal types for the urban renewal area, takes the renewal type with the highest probability as the prediction result for the urban renewal area, and performs spatial optimization of the urban renewal area based on morphological opening and closing operations.

[0076] In this embodiment, morphological opening and closing operations are used to spatially optimize the preliminary discrimination results, eliminate isolated "noise" units, smooth the boundaries, and finally form a continuous update partitioning scheme with clear spatial boundaries.

[0077] In an optional embodiment, the step of solving the problem based on a genetic algorithm under preset constraints includes:

[0078] An initial timeline scheme is generated based on the urban renewal area; non-dominated sorting and congestion calculation are performed using the preset indicators, and iterative optimization is carried out under the constraints until the preset iteration conditions are met, so as to obtain the implementation timeline scheme of the urban renewal area.

[0079] As an example, the genetic algorithm used is the NSGA-II algorithm.

[0080] As a preferred option, the multiple preset indicators include economic, social, and environmental objectives; wherein, the economic objectives include land appreciation revenue and renewal costs; the social objectives include the balance of public services and social inclusion; and the environmental objectives are quantified based on the equivalent value of ecosystem services and thermal environment simulation.

[0081] In this embodiment, a multi-objective system that coordinates "economy, society, and environment" is constructed.

[0082] As a preferred embodiment, the land appreciation revenue is determined based on benchmark land price, plot ratio, land use coefficient, and location coefficient; the renewal cost is determined based on comprehensive cost per unit area, renewal area, and adjustment coefficients based on renewal type and location, and its expression is as follows:

[0083]

[0084]

[0085] in, For land revenue, , , These are plot ratio, land use coefficient, and location coefficient, respectively. , These are the benchmark land price and the current land price, respectively. To cover replacement costs, , These represent the comprehensive cost per unit area and the land area of ​​the renewal zone, respectively. This is a differential adjustment coefficient based on update type and location.

[0086] More specifically, the expressions for social and environmental goals are as follows:

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093] in, The Theil index, The larger the value, the more uneven the distribution of projects. The first in the site selection scheme The proportion of the area of ​​each plot in the study area to the total area of ​​all priority development plots in the study area. For the first The proportion of the population of each area to the total population of the study area. The value obtained after transforming T represents the degree of fairness. The larger the size, the more it reflects fairness; ( Thermal environment level as the cooling point ( =11 M), For the area and Euclidean distance between them; area and The weighted distance between them is It is assumed that each request point will be served by the nearest area. If the value of the region The nearest facility If yes, then it is 1; otherwise, it is 0. For other internal requirements, or The value is 0; otherwise it is 1.

[0094] Example 2

[0095] This embodiment proposes an intelligent decision-making system for urban renewal areas based on machine learning and genetic algorithms, applying the intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms proposed in Embodiment 1. For example... Figure 2 The diagram shown is an architecture diagram of an intelligent decision-making system for urban renewal areas based on machine learning and genetic algorithms, according to this embodiment.

[0096] This embodiment proposes an intelligent decision-making system for urban renewal areas based on machine learning and genetic algorithms, including:

[0097] The data acquisition module is used to collect historical urban renewal data of the city to be decided. The historical urban renewal data includes urban renewal data of different dimensions and different renewal types.

[0098] The update potential prediction module is used to obtain key factors from urban renewal data of different dimensions based on feature screening; train the first machine learning model with the historical urban renewal data corresponding to the key factors as training data; and predict the renewal potential of the city to be decided through the trained first machine learning model to obtain the areas within the city to be decided that need to be updated.

[0099] The update type prediction module is used to train the second machine learning model using the different update types and their corresponding historical urban update data as training data; and to predict the update type of the area to be updated using the trained second machine learning model.

[0100] The time-series planning module is used to construct a multi-objective optimization model for the area to be updated with multiple preset indicators as targets, and solve the optimization model based on a genetic algorithm under preset constraints to obtain the implementation time-series plan for the urban renewal area.

[0101] It is understood that the system in this embodiment corresponds to the method in Embodiment 1 above, and the options in Embodiment 1 above are also applicable to this embodiment, so they will not be described again here.

[0102] Example 3

[0103] This embodiment demonstrates an application of the intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms proposed in Embodiment 1.

[0104] Data Acquisition and Processing. In this embodiment, the entire city area is used as the research scope. First, data was collected from multiple sources, generating a 100m × 100m regular grid within the research area, totaling approximately 1.11 million grid cells. Subsequently, a spatial join tool was used to associate building data, POI data, and other data with each grid cell. For example, attributes such as total building area, average number of floors, POI density, and average house price within each grid were calculated. For raster data such as nighttime lights, a "zonal statistics" tool was used to calculate the average light intensity within each grid. Finally, a structured spatial grid database containing dozens of attribute fields was constructed, providing standardized input for subsequent machine learning models.

[0105] This embodiment, based on the theoretical framework of "architectural form-spatial structure-socioeconomics," initially constructs a driving factor system comprising 15 indicators. The architectural form dimension includes: total building area within the grid, average number of building floors, building density, average construction year, and average building shape index. The spatial structure dimension includes: straight-line distance to the city center, road network density, and land use mix (based on the Shannon diversity index of POIs). The socioeconomic dimension includes: average house price and average nighttime light intensity. To optimize the system, the variance inflation factor (VIF) of all indicators was first calculated. It was found that the VIF values ​​for "total building area" and "building density" were too high, indicating collinearity; therefore, "total building area" was removed. Subsequently, principal component analysis (PCA) was performed on the remaining 14 indicators, extracting five principal components with eigenvalues ​​greater than 1. These five principal components, with a cumulative variance contribution rate of 85.2%, were used as the final key factors. The first machine learning model was then trained based on these five key factors to predict urban renewal areas. The training process used GridSearchCV for hyperparameter optimization. The trained first machine learning model was applied to all grid cells within the study area to predict the update probability of each grid cell. After obtaining the probability distribution map, the results were displayed using the "Natural Breakpoint Method" in ArcGIS, resulting in the final potential spatial distribution map. The DBSCOST optimization algorithm was then used to spatially cluster the identified high-potential, configured-update discrete grid cells. After clustering, preliminary areas requiring updating were obtained. Subsequently, the area boundaries were corrected and optimized by incorporating natural and artificial boundaries such as road networks and river systems, ultimately generating the areas to be updated.

[0106] Secondly, this embodiment mainly considers three main types of updates: demolition and reconstruction (large-scale demolition of existing buildings and redevelopment), comprehensive renovation (repair, renovation, and facility upgrades while retaining the original building's main framework), and functional replacement (changing the original building's function, such as converting an old factory into a cultural and creative park). Based on the three-dimensional theoretical framework of "building form - property rights status - functional attributes," this embodiment constructs a discrimination index system containing 12 indicators. The building form dimension includes: building density, average number of floors, proportion of structural types (brick-concrete, frame, etc.), year of construction, and plot ratio. The property rights status dimension includes: property rights clarity index (calculated based on cadastral data) and land ownership complexity (number of parcels / grid area). The functional attribute dimension includes: dominant function type (based on POI judgment) and business diversity (Shannon diversity index of POI). Similarly, a combination of VIF and PCA is used for feature optimization, ultimately retaining 8 core discrimination indicators. To address the sample imbalance problem, the SMOTE algorithm is used for oversampling of the relatively small number of "functional replacement" and "demolition and reconstruction" samples. Hyperparameter optimization was performed using BayesSearchCV. The trained second machine learning model was applied to all regions to be updated, predicting the update method for each grid. Based on the principle of maximum probability, each grid was assigned a dominant update type label. Subsequently, morphological opening and closing operations were used to spatially optimize the preliminary classification results, eliminating isolated "salt-and-pepper noise," connecting adjacent regions of the same type, and ultimately forming a spatially continuous and clearly defined update method partitioning scheme.

[0107] Finally, each area to be updated was treated as an independent decision-making unit. A multi-objective optimization system integrating "economy, society, and environment" was constructed. Based on the constraints of urban scale development, the final implementation sequence plan for the urban renewal area was generated through NSGA-II multi-objective optimization solution and scheme generation.

[0108] In different specific implementations, the methods or systems described in this application can be implemented in software, hardware, or a combination thereof. Furthermore, the order of the method steps can be changed, and various elements can be added, reordered, combined, omitted, or modified.

[0109] Obviously, the above embodiments of this application are merely examples for clearly illustrating this application, and are not intended to limit the implementation of this application, nor are they intended to limit this application. For those skilled in the art, other variations or modifications can be made based on the above description. The separate structural / functional modules or units can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part. The structure and function of the separate components can be implemented as a combined structure or component. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of the claims of this application.

Claims

1. A smart decision-making method for urban renewal areas based on machine learning and genetic algorithms, characterized in that, The method includes: Collect historical urban renewal data for the city to be decided, including urban renewal data of different dimensions and different renewal types; Key factors are obtained from urban renewal data of different dimensions based on feature selection; the historical urban renewal data corresponding to the key factors are used as training data to train the first machine learning model; the first machine learning model is trained to predict the renewal potential of the city to be decided, and the areas within the city to be decided that need to be renewed are obtained. The second machine learning model is trained using the different update types and their corresponding historical urban update data as training data; the updated type is predicted for the area that needs to be updated using the trained second machine learning model. For the area that needs to be updated, a multi-objective optimization model is constructed with multiple preset indicators as the target. Under preset constraints, the optimization model is solved based on a genetic algorithm to obtain the implementation sequence plan of the urban renewal area. The steps for obtaining key factors from urban renewal data of different dimensions based on feature filtering include: Extract feature variables from data of different dimensions, use any one feature variable as the dependent variable and the remaining feature variables as independent variables to perform linear regression, and calculate the variance inflation factor (VIF) value of each feature variable; and remove feature variables that are greater than the preset VIF threshold to obtain a set of feature variables. Select the feature variable set whose cumulative VIF variance contribution rate reaches a preset contribution rate threshold. m Each principal component serves as a key factor; its expression is as follows: in, for p One characteristic variable, For the first j One characteristic variable, For the first j VIF values ​​of each feature variable For the first j Coefficient of determination under linear regression of individual characteristic variables Let be the fitting coefficients for each independent variable. For random error, This represents the original data matrix after standardization. For this is the first k The score vector of each principal component. It is the first k 1 eigenvector It is a matrix The k 1 eigenvalue, Indicates the first k The percentage of variance of each principal component relative to the total information content of the original data. For the front k Cumulative variance contribution rate of each principal component; The preset indicators include economic, social, and environmental goals; wherein, the economic goals include land appreciation gains and renewal costs; the social goals include the balance of public services and social inclusion; and the environmental goals are quantified based on the equivalent value of ecosystem services and thermal environment simulation. The land appreciation revenue is determined based on benchmark land price, plot ratio, land use coefficient, and location coefficient; the renewal cost is determined based on comprehensive cost per unit area, renewal area, and adjustment coefficients based on renewal type and location, and its expression is as follows: in, For land revenue, , , These are plot ratio, land use coefficient, and location coefficient, respectively. , These are the benchmark land price and the current land price, respectively. To cover replacement costs, , These represent the comprehensive cost per unit area and the land area of ​​the renewal zone, respectively. This is a differential adjustment coefficient based on update type and location.

2. The intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms according to claim 1, characterized in that, The urban renewal data from different dimensions includes spatial morphology data, socio-economic data, and planning management data.

3. The intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms according to claim 1, characterized in that, The first machine learning model is a random forest classification model, and the hyperparameters within the first machine learning model are adjusted through grid search and 10-fold cross-validation.

4. The intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms according to claim 1, characterized in that, The second machine learning model is a random forest classification model, and the hyperparameters within the second machine learning model are adjusted using Bayesian optimization algorithm and five-fold cross-validation.

5. The intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms according to claim 4, characterized in that, The second machine learning model outputs the probability of different renewal types for the urban renewal area, takes the renewal type with the highest probability as the prediction result for the urban renewal area, and performs spatial optimization of the urban renewal area based on morphological opening and closing operations.

6. The intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms according to claim 1, characterized in that, The steps for solving the problem based on a genetic algorithm under preset constraints include: An initial timeline scheme is generated based on the urban renewal area; non-dominated sorting and congestion calculation are performed using the preset indicators, and iterative optimization is carried out under the constraints until the preset iteration conditions are met, so as to obtain the implementation timeline scheme of the urban renewal area.

7. An intelligent decision-making system for urban renewal areas based on machine learning and genetic algorithms, characterized in that, The system employing the intelligent decision-making method for urban renewal areas based on machine learning and genetic algorithms as described in any one of claims 1 to 6, comprises: The data acquisition module is used to collect historical urban renewal data of the city to be decided. The historical urban renewal data includes urban renewal data of different dimensions and different renewal types. The potential prediction module is used to predict the potential of the city to be decided based on the first machine learning model, and to obtain the areas within the city that need to be updated. The update type prediction module is used to predict the update type of the area that needs to be updated based on the second machine learning model; The time-series planning module is used to construct a multi-objective optimization model for the area to be updated with multiple preset indicators as targets, and solve the optimization model based on a genetic algorithm under preset constraints to obtain the implementation time-series plan for the urban renewal area.

Images