Smart city update planning method and system based on artificial intelligence

Abstract

The invention relates to the technical field of smart city planning, in particular to a smart city update planning method and system based on artificial intelligence. The invention discloses a smart city update planning method based on artificial intelligence. The method comprises the following steps: S1, acquiring space attribute data and city element associated data of scattered parcels of a target city; s2, processing and integrating the spatial attribute data and the city element associated data to generate a feature description sequence of the target city sporadic plots; and S3, constructing a purpose planning data set according to the excellent planning cases of the regions, and establishing a purpose planning model of the urban sporadic plots by using a random forest algorithm based on the purpose planning data set. According to the invention, long-term idle urban sporadic plot resources can be activated and converted into micro spaces for convenience service, ecological restoration, exercise and fitness or culture display which are matched with community requirements, the urban living environment is improved, the space utilization efficiency is improved, and urban refined governance and continuous update are promoted.

Description

Technical Field

[0001] This invention relates to the field of smart city planning technology, and in particular to a smart city renewal planning method and system based on artificial intelligence. Background Technology

[0002] Current urban planning primarily focuses on the development of large-scale areas, lacking effective planning methods for the numerous scattered plots of land, such as marginal, interspersed, and marginal lots, within cities. These plots are small, irregularly shaped, and scattered, lacking independent development potential, and have long been in a management vacuum, becoming "leftovers" of urban space. As urbanization enters a phase of improving the quality of existing land, land resources are becoming increasingly scarce, making the meticulous use of every inch of land an inevitable requirement. At the same time, residents' demand for public spaces such as community convenience services, ecological recreation, and cultural and sports facilities continues to rise.

[0003] Traditional methods of planning scattered plots based on human experience are costly, inefficient, and highly subjective, making it difficult to systematically explore the potential value of these plots to meet diverse and dynamic public needs. Therefore, there is an urgent need to develop a smart city renewal planning method and system based on artificial intelligence that can conduct land use planning based on the characteristics of scattered plots and take into account public needs, in order to activate these "dormant" spaces. Summary of the Invention

[0004] To overcome the shortcomings of existing technologies that neglect the land use planning of scattered urban plots, this invention provides a smart city renewal planning method and system based on artificial intelligence.

[0005] The technical implementation scheme of this invention is: a smart city renewal planning method based on artificial intelligence, comprising the following steps: S1: Obtain spatial attribute data and urban element association data of scattered plots in the target city; S2: Process and integrate the spatial attribute data and urban element association data to generate a feature description sequence of scattered plots in the target city; S3: Based on excellent planning cases from various regions, construct a land use planning dataset. Based on the land use planning dataset, use the random forest algorithm to establish a land use planning model for scattered urban plots. S4: Input the feature description sequence of the scattered plots in the target city into the land use planning model for planning, and obtain the recommended land use result sequence; S5: Conduct a public demand evaluation on the recommended use result sequence to obtain the final use result sequence.

[0006] Preferably, the acquisition of spatial attribute data and urban element association data of scattered plots in the target city includes: the spatial attribute data includes spatial coordinate codes, plot area, perimeter and geometric shape, and the urban element association data includes building attributes, service facility types and distribution within a preset radius.

[0007] Preferably, the step of processing and integrating the spatial attribute data and urban element association data to generate the feature description sequence of the scattered plots in the target city includes: performing morphological analysis on the spatial attribute data to obtain compactness and minimum bounding rectangle fill rate; processing the urban element association data to obtain service diversity entropy value, per capita green space area and use conflict factor; and encoding the above results into fixed-length feature vectors according to the spatial coordinates of the plots to form the feature description sequence of the scattered plots in the city.

[0008] Preferably, the step of constructing a land use planning dataset based on excellent planning cases from various regions, and establishing a land use planning model for scattered urban plots using the random forest algorithm based on the land use planning dataset, includes: collecting planning schemes for scattered plots in urban renewal cases, labeling feature description sequences and final use tags, wherein the final use tags include category tags and specific uses, and the category tags include community convenience type, ecological restoration type, sports and fitness type, and cultural display type, forming a land use planning dataset that maps feature description vectors to use tags.

[0009] Preferably, the step of establishing a land use planning model for scattered urban land parcels based on the land use planning dataset using a random forest algorithm includes: using a sampling with replacement strategy to extract a preset number of samples from the land use planning dataset for each decision tree as a training subset, and using the remaining samples as an out-of-bag dataset for model validation; in the single decision tree construction stage, when splitting each node, a subset of features equal to the square root of the total number of features is randomly selected from all features as candidate features, using Gini impurity as the splitting criterion, traversing all splitting points of the candidate features and selecting the feature that causes the Gini index to decrease the most and the threshold for splitting, setting a preset minimum number of splitting samples and not limiting the maximum depth of the decision tree until the node is pure, and constructing multiple structurally differentiated decision trees to form a random forest.

[0010] Preferably, the remaining samples are used as out-of-bag datasets for model validation, including: in the model validation stage, out-of-bag error is calculated using out-of-bag data as a generalization ability evaluation index; when the out-of-bag error is stably lower than a preset error threshold, the model is determined to have converged; at the same time, the feature permutation method is used to quantify the importance of features: after randomly permuting the feature values, the out-of-bag error is recalculated, and the ratio of the difference between the original out-of-bag error and the original out-of-bag error is the importance score of the feature.

[0011] Preferably, the step of inputting the feature description sequence of scattered plots in the target city into the land use planning model for planning to obtain a recommended land use result sequence includes: traversing all decision trees in the land use planning model to perform parallel prediction and obtain the probability distribution value of land use categories; summarizing the probability distribution values ​​of land use categories output by each tree to generate a recommended land use result sequence arranged in descending order of probability distribution values, and using the probability distribution value as a confidence score.

[0012] Preferably, the step of conducting public demand evaluation on the recommended use result sequence to obtain the final use result sequence includes: conducting public demand evaluation on the recommended use result sequence and displaying the location of the plots on a 3D map; introducing the entropy weight-TOPSIS comprehensive evaluation algorithm to construct an evaluation index matrix, determining the index weights based on information entropy, and calculating the closeness ranking of each scheme to the ideal solution; determining the final score based on the closeness ranking, and selecting schemes whose final scores exceed the score threshold to enter the final use result sequence.

[0013] Preferably, the step of introducing the entropy weight-TOPSIS comprehensive evaluation algorithm, constructing an evaluation index matrix, determining index weights based on information entropy, and calculating the closeness ranking of each scheme to the ideal solution includes: the evaluation index matrix, where the matrix rows correspond to the recommended use result sequence, and the columns include scheme suitability, public demand matching degree, and public demand urgency. The scheme suitability is taken from the confidence score output by the use planning model, and the public demand matching degree and the public demand urgency degree are taken from the public demand evaluation; the entropy weight method is executed to determine the index weights, the information entropy value is calculated for each index column, the difference coefficient is calculated based on the information entropy value, and finally, the objective weights of each index are normalized.

[0014] Preferably, an artificial intelligence-based smart city renewal planning system further includes: Data acquisition module: Acquires spatial attribute data of scattered plots of land in the target city and correlation data of urban elements; Data processing module: processes and integrates the spatial attribute data and urban element association data to generate a feature description sequence of scattered plots in the target city; Model building module: Based on excellent planning cases from various regions, a land use planning dataset is constructed. Based on the land use planning dataset, a land use planning model for scattered urban plots is established using the random forest algorithm. Land Use Planning Module: Input the feature description sequence of scattered land parcels in the target city into the land use planning model for planning, and obtain a recommended land use result sequence; Public voting module: The recommended usage result sequence is evaluated by the public to obtain the final usage result sequence.

[0015] The beneficial effects of this invention are: This invention constructs a deep feature description sequence that integrates spatial attributes of land parcels with data related to urban elements, and establishes an intelligent land use planning model based on the random forest algorithm to learn from excellent planning cases. This model can automatically and accurately generate land use recommendations for scattered urban land parcels, improving the scientific nature, efficiency, and coverage of planning. Furthermore, it introduces a public demand evaluation mechanism that combines 3D map display, ensuring that the planning results align with the actual needs of the public and enhancing the effectiveness of public participation and the transparency of decision-making.

[0016] This invention can revitalize long-term idle or inefficiently used urban scattered land resources, transforming them into micro-spaces that match the needs of the community, such as convenient services, ecological restoration, sports and fitness, or cultural displays, thereby improving the urban living environment, enhancing space utilization efficiency, and promoting refined urban governance and continuous renewal. Attached Figure Description

[0017] Figure 1 This is a flowchart of the AI-based smart city renewal planning method of the present invention; Figure 2 This is a structural diagram of the AI-based smart city renewal planning system of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1: A smart city renewal planning method based on artificial intelligence, such as Figure 1 As shown, it includes the following steps: S1: Obtain spatial attribute data and urban element association data of scattered plots in the target city; Spatial attribute data includes spatial coordinate codes, plot area, perimeter, and geometric shape. The urban element association data includes building attributes, service facility types, and distribution within a preset radius.

[0020] It should be further explained that, through the spatial database interface of the urban geographic information system, spatial attribute data and urban element association data of all scattered plots within the administrative boundary of the target city are obtained. The spatial attribute data includes unique spatial coordinate codes issued by the surveying and mapping department, plot area values ​​verified by the land resources department, perimeter values ​​calculated from the coordinates of the boundary line nodes, and geometric vector data composed of boundary point coordinate sequences. The urban element association data comes from the urban infrastructure thematic layer, specifically including: statistical data on the classification and quantity of building attributes, and the spatial distribution density values ​​of park facilities, educational facilities, medical facilities, and commercial facilities within a preset radius buffer zone around the geometric center point of the scattered plots.

[0021] S2: Process and integrate the spatial attribute data and urban element association data to generate a feature description sequence of scattered plots in the target city; Morphological analysis is performed on the spatial attribute data to obtain compactness and minimum bounding rectangle fill rate. The urban element association data is processed to obtain service diversity entropy value, per capita green space area and use conflict factor. The above results are encoded into fixed-length feature vectors according to the spatial coordinates of the plots to form a feature description sequence of scattered urban plots.

[0022] It should be further explained that morphological analysis is performed on the spatial attribute data: based on the coordinate sequence of boundary points of the geometric shape, a compactness index reflecting the regularity of the plot shape is calculated. The compactness calculation formula is as follows: In the formula, For compactness, The area of ​​the land parcel. The area is the square of the plot's perimeter. The minimum bounding rectangle fill rate index is calculated simultaneously, defined as the ratio of the actual plot area to the area of ​​the minimum bounding rectangle, which is generated by iteratively rotating the plot's geometric vector data. Multi-dimensional quantitative analysis is performed on the city element association data: based on the type and distribution of service facilities, the service diversity entropy index is calculated. The formula for calculating the service diversity entropy is: In the formula, To serve the diversity entropy value, For the first The ratio of the number of similar service facilities to the total number of service facilities. The number and types of service facilities are determined. Based on the total area of ​​surrounding green space and the surrounding permanent population, the per capita green space area is calculated. The attributes of surrounding buildings and the distribution of service facilities are analyzed, and the use conflict factor index is calculated. The use conflict factor is the minimum Euclidean distance between service facilities of the same type. The above processing results are standardized and integrated: using the spatial coordinate code of each plot as a unique identifier, five core indicators, including compactness index, minimum bounding rectangle fill rate index, service diversity entropy value index, per capita green space area index, and use conflict factor index, are encoded into fixed-dimensional feature vectors in a preset order, ultimately forming a machine-recognizable feature description sequence of scattered urban plots.

[0023] S3: Based on excellent planning cases from various regions, construct a land use planning dataset. Based on the land use planning dataset, use the random forest algorithm to establish a land use planning model for scattered urban plots. Planning schemes for scattered plots in urban renewal cases are collected, and feature description sequences and final use labels are marked. The final use labels include category labels and specific uses. Category labels include community convenience type, ecological restoration type, sports and fitness type, and cultural display type, forming a use planning dataset that maps feature description vectors to use labels.

[0024] A sampling strategy with replacement is used to extract a preset number of samples from the purpose planning dataset for each decision tree as a training subset, and the remaining samples are used as out-of-bag datasets for model validation. In the single decision tree construction stage, when each node splits, a subset of features equal to the square root of the total number of features is randomly selected as candidate features. The Gini impurity is used as the splitting criterion. All splitting points of the candidate features are traversed and the feature that causes the Gini index to decrease the most and the threshold are selected for splitting. A preset minimum number of splitting samples is set and the maximum depth of the decision tree is not limited until the node is pure. Multiple decision trees with different structures are constructed to form a random forest.

[0025] During the model validation phase, out-of-bag error is calculated using out-of-bag data as an evaluation index of generalization ability. When the out-of-bag error is stably lower than a preset error threshold, the model is determined to have converged. At the same time, the feature permutation method is used to quantify the importance of features: after randomly permuting the feature values, the out-of-bag error is recalculated, and the ratio of the difference between the original out-of-bag error and the original out-of-bag error is the importance score of the feature.

[0026] Further explanation is needed regarding the dataset construction. This involves collecting implemented planning schemes for scattered plots from successful urban renewal cases. For each case plot, the following operations are performed: Extracting the plot's feature description sequence; labeling the final use with two levels of information: Category label: selected from four predefined categories: community convenience, ecological restoration, sports and fitness, and cultural display; Specific use: recording the specific function of the implementation plan, such as park green space under the ecological restoration category; establishing a mapping relationship between feature description vectors and final use labels to form a structured land use planning dataset. Random forest model training: Training subset sampling: using a sampling strategy with replacement, a preset number of samples are drawn from the land use planning dataset for each decision tree as a training subset. The remaining unsampled samples automatically form the out-of-bag dataset. Single decision tree construction: when splitting a node, features equal to the square root of the total number of features are randomly selected from all features to form a candidate feature subset; Gini impurity is used as the splitting criterion. In the formula, For training subset The Gini index, for medium sample belongs to The probability of a class, for Each class is calculated and summed. All split points in the candidate feature subset are traversed; the feature that causes the largest decrease in the Gini index and its corresponding threshold are selected as the splitting rule; a preset minimum number of splitting samples is set, and splitting stops when the number of samples in a node falls below this value; the maximum depth of the decision tree is not limited until the node samples are pure or cannot be split. Forest formation: Multiple structurally differentiated decision trees are repeatedly constructed to form a random forest for planning purposes. Model validation and feature analysis, out-of-bag error validation: The model prediction error rate is calculated using the out-of-bag dataset as the out-of-bag error; when the out-of-bag error is continuously and stably lower than the preset error threshold, the model is considered converged and training is terminated; feature importance quantification is performed using the feature replacement method: the feature value of a certain feature in the out-of-bag dataset is randomly replaced; the out-of-bag error after replacement is recalculated; the ratio of the difference between the original out-of-bag error and the out-of-bag error after replacement is calculated as the feature importance score.

[0027] S4: Input the feature description sequence of the scattered plots in the target city into the land use planning model for planning, and obtain the recommended land use result sequence; Parallel prediction is performed by traversing all decision trees in the usage planning model to obtain the usage category probability distribution value; the usage category probability distribution values ​​output by each tree are summarized to generate a sequence of recommended usage results arranged in descending order of probability distribution value, and the probability distribution value is used as the confidence score.

[0028] Further explanation is needed regarding the parallel prediction execution: the feature description sequence of the land parcel to be planned is input into the trained land use planning model, and synchronous prediction is performed by traversing all decision trees in the forest; probability distribution value aggregation: the probability distribution value of the land use category output by each decision tree is calculated, and the average of the probability outputs of all trees is calculated; recommendation sequence generation: the land use categories are sorted from high to low according to the average probability value; a recommended land use result sequence is generated, and the sequence elements include the specific use and the probability distribution value; the output probability distribution value is used as the confidence score of the recommendation result.

[0029] S5: Conduct a public demand evaluation on the recommended use result sequence to obtain the final use result sequence.

[0030] The recommended use result sequence is evaluated by public demand, and the location of the plots is displayed on a 3D map. The entropy weight-TOPSIS comprehensive evaluation algorithm is introduced to construct an evaluation index matrix. The index weights are determined based on information entropy, and the closeness ranking of each scheme to the ideal solution is calculated. The final score is determined according to the closeness ranking, and the schemes with the final score exceeding the score threshold are selected to enter the final use result sequence.

[0031] The evaluation index matrix consists of rows corresponding to the recommended usage result sequence and columns including scheme suitability, public demand matching degree, and public demand urgency. The scheme suitability is taken from the confidence score output by the usage planning model, and the public demand matching degree and public demand urgency are taken from the public demand evaluation. The entropy weight method is used to determine the index weights. The information entropy value is calculated for each index column, and the difference coefficient is calculated based on the information entropy value. Finally, the objective weights of each index are normalized to obtain the objective weights.

[0032] It should be further explained that the public voting platform releases a sequence of recommended land use results for target plots through the smart city public participation platform. This sequence includes all recommended schemes output by the land use planning model. The platform is linked to a city 3D map engine, using spatial coordinate encoding to locate scattered plots, allowing the public to click on plots to view details. Each recommended scheme displays the following core information: land use category tag and specific land use name, scheme suitability score, confidence score taken from the land use planning model output, and entry points for public demand matching degree and public demand urgency. Entropy Weight-TOPSIS comprehensive evaluation is implemented, and the evaluation index matrix is ​​constructed: the matrix rows correspond to each scheme in the recommended land use result sequence, and the matrix columns are fixed and contain three evaluation indicators: scheme suitability indicator (directly referencing the confidence score), public demand matching indicator (taking the average score from public voting), and public demand urgency indicator (taking the average score from public voting), forming an evaluation index matrix with 3 columns equal to the total number of schemes. Entropy weight method for determining indicator weights: Calculate the information entropy value for each indicator column; statistically analyze the probability distribution of all solutions under the indicator, calculate the disorder of the probability distribution using information entropy, and calculate the difference coefficient based on the information entropy value: the difference coefficient equals 1 minus the information entropy value. Normalize the difference coefficient: divide the difference coefficient of each indicator by the sum of the difference coefficients to obtain the objective weights of solution suitability, public demand matching degree, and public demand urgency. TOPSIS proximity ranking: Construct a weighted standardized matrix: multiply the values ​​of each column of the evaluation indicator matrix by the corresponding objective weight to determine the positive and negative ideal solutions: the positive ideal solution takes the maximum value of each column, and the negative ideal solution takes the minimum value of each column. Calculate the Euclidean distance between each solution and the positive and negative ideal solutions, and calculate the proximity: proximity equals the distance to the negative ideal solution divided by the sum of the distances to the positive and negative ideal solutions. Sort the solutions by proximity values ​​from largest to smallest using the algorithm. Final score and final sequence generation: convert the TOPSIS proximity score to a percentage score, and the final score equals the algorithm ranking score. End-use result sequence generation: Set a scoring threshold, select schemes whose final scores exceed the scoring threshold, and sort them from high to low to form the end-use result sequence.

[0033] Example 2: Based on Example 1, an artificial intelligence-based smart city renewal planning system, such as... Figure 2 As shown, it also includes: Data acquisition module: Acquires spatial attribute data of scattered plots of land in the target city and correlation data of urban elements; Data processing module: processes and integrates the spatial attribute data and urban element association data to generate a feature description sequence of scattered plots in the target city; Model building module: Based on excellent planning cases from various regions, a land use planning dataset is constructed. Based on the land use planning dataset, a land use planning model for scattered urban plots is established using the random forest algorithm. Land Use Planning Module: Input the feature description sequence of scattered land parcels in the target city into the land use planning model for planning, and obtain a recommended land use result sequence; Public voting module: The recommended usage result sequence is evaluated by the public to obtain the final usage result sequence.

[0034] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A smart city renewal planning method based on artificial intelligence, characterized in that, Includes the following steps: S1: Obtain spatial attribute data and urban element association data of scattered plots in the target city; S2: Process and integrate the spatial attribute data and urban element association data to generate a feature description sequence of scattered plots in the target city; S3: Based on excellent planning cases from various regions, construct a land use planning dataset. Based on the land use planning dataset, use the random forest algorithm to establish a land use planning model for scattered urban plots. S4: Input the feature description sequence of the scattered plots in the target city into the land use planning model for planning, and obtain the recommended land use result sequence; S5: Conduct a public demand evaluation on the recommended use result sequence to obtain the final use result sequence.

2. The smart city renewal planning method based on artificial intelligence according to claim 1, characterized in that, The acquisition of spatial attribute data and urban element association data of scattered plots in the target city includes: the spatial attribute data includes spatial coordinate codes, plot area, perimeter and geometric shape; the urban element association data includes building attributes, service facility types and distribution within a preset radius.

3. The smart city renewal planning method based on artificial intelligence according to claim 1, characterized in that, The process of processing and integrating the spatial attribute data and urban element association data to generate the feature description sequence of the scattered plots in the target city includes: performing morphological analysis on the spatial attribute data to obtain compactness and minimum bounding rectangle fill rate; processing the urban element association data to obtain service diversity entropy value, per capita green space area and use conflict factor; and encoding the above results into fixed-length feature vectors according to the spatial coordinates of the plots to form the feature description sequence of the scattered plots in the city.

4. The smart city renewal planning method based on artificial intelligence according to claim 1, characterized in that, The process involves constructing a land use planning dataset based on excellent planning cases from various regions, and then using a random forest algorithm to establish a land use planning model for scattered urban plots. This includes: collecting planning schemes for scattered plots in urban renewal cases, labeling feature description sequences and final use tags, where the final use tags include category tags and specific uses. Category tags include community convenience, ecological restoration, sports and fitness, and cultural display, forming a land use planning dataset that maps feature description vectors to land use tags.

5. The smart city renewal planning method based on artificial intelligence according to claim 1, characterized in that, The process of establishing a land use planning model for scattered urban land parcels based on the land use planning dataset and using a random forest algorithm includes: using a sampling with replacement strategy to extract a preset number of samples from the land use planning dataset for each decision tree as a training subset, and using the remaining samples as an out-of-bag dataset for model validation; during the single decision tree construction stage, when splitting each node, a subset of features equal to the square root of the total number of features is randomly selected from all features as candidate features, and Gini impurity is used as the splitting criterion. All splitting points of the candidate features are traversed, and the feature that causes the Gini index to decrease the most and the threshold are selected for splitting. A preset minimum number of splitting samples is set, and the maximum depth of the decision tree is not limited until the node is pure, thereby constructing multiple structurally differentiated decision trees to form a random forest.

6. The smart city renewal planning method based on artificial intelligence according to claim 5, characterized in that, The remaining samples are used as out-of-bag datasets for model validation, including: in the model validation stage, out-of-bag error is calculated using out-of-bag data as a generalization ability evaluation index; when the out-of-bag error is stably lower than a preset error threshold, the model is determined to have converged; at the same time, the feature permutation method is used to quantify the importance of features: after randomly permuting the feature values, the out-of-bag error is recalculated, and the ratio of the difference between the original out-of-bag error and the original out-of-bag error is the importance score of the feature.

7. The smart city renewal planning method based on artificial intelligence according to claim 1, characterized in that, The step of inputting the feature description sequence of scattered plots in the target city into the land use planning model for planning and obtaining a recommended land use result sequence includes: traversing all decision trees in the land use planning model to perform parallel prediction and obtain the probability distribution value of land use categories; summarizing the probability distribution values ​​of land use categories output by each tree, generating a recommended land use result sequence arranged in descending order of probability distribution values, and using the probability distribution value as a confidence score.

8. The smart city renewal planning method based on artificial intelligence according to claim 1, characterized in that, The step of conducting a public demand evaluation of the recommended use result sequence to obtain the final use result sequence includes: conducting a public demand evaluation of the recommended use result sequence and displaying the location of the plots on a 3D map; introducing the entropy weight-TOPSIS comprehensive evaluation algorithm to construct an evaluation index matrix, determining the index weights based on information entropy, and calculating the closeness ranking of each scheme to the ideal solution; determining the final score based on the closeness ranking, and selecting schemes whose final scores exceed the score threshold to enter the final use result sequence.

9. The smart city renewal planning method based on artificial intelligence according to claim 8, characterized in that, The proposed entropy-weighted TOPSIS comprehensive evaluation algorithm constructs an evaluation index matrix, determines index weights based on information entropy, and calculates the ranking of the closeness of each scheme to the ideal solution. This includes: the evaluation index matrix, where rows correspond to the recommended use result sequence, and columns include scheme suitability, public demand matching degree, and public demand urgency. The scheme suitability is taken from the confidence score output by the use planning model, and the public demand matching degree and public demand urgency are taken from the public demand evaluation. The entropy-weighted method is executed to determine index weights, an information entropy value is calculated for each index column, a difference coefficient is calculated based on the information entropy value, and finally, the objective weights of each index are normalized.

10. A smart city renewal planning system based on artificial intelligence, comprising an intelligent diagnostic inspection method based on multi-source data fusion according to any one of claims 1-9, characterized in that, Also includes: Data acquisition module: Acquires spatial attribute data of scattered plots of land in the target city and correlation data of urban elements; Data processing module: processes and integrates the spatial attribute data and urban element association data to generate a feature description sequence of scattered plots in the target city; Model building module: Based on excellent planning cases from various regions, a land use planning dataset is constructed. Based on the land use planning dataset, a land use planning model for scattered urban plots is established using the random forest algorithm. Land Use Planning Module: Input the feature description sequence of scattered land parcels in the target city into the land use planning model for planning, and obtain a recommended land use result sequence; Public voting module: The recommended usage result sequence is evaluated by the public to obtain the final usage result sequence.

Images