Cross-regional supervision risk prediction method and system based on urban functions

Abstract

The invention provides a cross-regional supervision risk prediction method and system based on urban functions. The method comprises the following steps: 1, data preprocessing and linkage region screening: screening an adjacent region as a linkage analysis region; step 2, feature engineering and regional relationship modeling; step 3, abnormal region identification and data cleaning: carrying out anomaly detection on transportation data between regions based on a DBSCAN density clustering algorithm; and 4, risk quantification and comprehensive assessment, which relates to the technical field of market supervision, and comprises the following steps of: screening adjacent areas to establish a linkage combination, calculating category supervision similarity and personnel flow intensity, eliminating transportation abnormal areas, modeling by using an intelligent algorithm, and combining with a dynamic weight to realize risk comprehensive assessment. The system breaks through the limitation of traditional single-region analysis, constructs a cross-region supervision system, remarkably improves the risk prediction accuracy and response efficiency, and is suitable for supervision, prevention and control in the cross-region circulation fields of food, medicines, industrial products and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of market supervision, and particularly to a cross-regional supervision risk prediction method and system based on urban functions. Background Art

[0002] Under the background of the rapid development of globalization and regional economic integration today, the cross-regional nature of market activities has become increasingly prominent. Market supervision departments are facing huge challenges. How to accurately predict cross-regional market supervision risks has become a key issue for ensuring the stable and healthy development of the market.

[0003] Most traditional market supervision risk prediction methods are limited within a single region and mainly analyze based on the historical data and current market performance of that region itself. However, this method ignores the mutual relevance and influence between regions and cannot comprehensively and accurately evaluate cross-regional market supervision risks. In traditional prediction methods, when analyzing the market supervision risk of the food market in a certain city, only the production and operation data, sales data, etc. of food enterprises within that city are concerned, without considering the inflow and outflow of food from surrounding cities, the impact of personnel exchanges on food consumption habits and safety risks, and various risk factors that may occur during the transportation process.

[0004] With the continuous expansion of the market scale and the increasing frequency of economic exchanges between regions, the linkage effect between regions has become more obvious. Market fluctuations, policy changes, emergencies, etc. in one region may quickly affect surrounding regions or even more distant regions. A sudden food safety incident in a certain area may lead to a sharp reduction in the demand for the same type of food in surrounding areas, and at the same time trigger a comprehensive investigation of the relevant food supply chain by the supervision department, thus affecting the entire cross-regional food market pattern. Existing prediction methods fail to fully consider these complex regional linkage relationships, resulting in large deviations in the prediction results and being unable to provide timely and accurate decision-making support for market supervision departments.

[0005] In addition, traditional prediction methods also have limitations in data collection and processing. They often focus on collecting a single type of data, such as only paying attention to market transaction data while ignoring other important information, such as personnel flow data, transportation situation data, market address transfer data, etc. These multi-dimensional data are crucial for comprehensively understanding market dynamics and accurately predicting supervision risks. Moreover, traditional methods lack effective means of integrating and analyzing these multi-source heterogeneous data and cannot fully explore the potential information behind the data.

[0006] In summary, the existing market supervision risk prediction methods are difficult to meet the complex requirements of cross-regional market supervision. There is an urgent need for a cross-regional prediction method and system that can fully consider regional linkage relationships and comprehensively analyze multi-dimensional data to improve the accuracy of market supervision risk prediction and provide more powerful decision-making basis for market supervision departments. Summary of the Invention

[0007] To overcome the existing problems, the embodiments of the present application provide a cross-regional supervision risk prediction method and system based on urban functions. By selecting multiple neighboring regions as linkage analysis regions in the region to be measured, counting the introduction amounts of the same category of items in the region to be measured and the linkage analysis regions, calculating the supervision similarity of each category of items to screen and mark each item category, processing the linkage purchase similarity amounts of various marked items between each neighboring region and the region to be measured to obtain the comprehensive linkage characteristics of items, counting the personnel flow interaction amounts between different neighboring regions and the region to be measured, classifying the linkage relationships between the region to be measured and each neighboring region, collecting the transportation anomalies of each neighboring region and excluding some neighboring regions, marking the remaining regions and then selecting different prediction methods to analyze and generate analysis results for single-region monitoring and supervision, collecting the market address transfer frequencies and local market item production amounts of each marked region, setting the supervision risk weights, and comprehensively evaluating the supervision risks of the region to be measured based on the single-region monitoring and supervision analysis results and the supervision risk weights, so as to improve the accuracy of the cross-regional prediction method.

[0008] The technical solutions adopted by the embodiments of the present application to solve their technical problems are as follows:

[0009] A cross-regional supervision risk prediction method and system based on urban functions, including:

[0010] Step 1: Data preprocessing and linkage region screening:

[0011] Screen neighboring regions as linkage analysis regions. The method for screening neighboring regions is to call the two-dimensional data in the structure database to calculate the area of the region to be measured, expand the neighboring regions based on the side length, form a linkage combination, calculate the supervision similarity of the same category of items and mark the high-correlation item categories;

[0012] The step of screening neighboring regions as linkage analysis regions in Step 1 includes:

[0013] Call the two-dimensional data of the region to be measured in the structure database to calculate the area of the region to be measured. Use the length or width of the region to be measured as the common side to expand outward until the expanded area of each side is the same as the area of the region to be measured. Set the expanded region of each side as a neighboring region, use each neighboring region as multiple different linkage analysis regions of the region to be measured, and use the combination of the region to be measured and different linkage analysis regions as the linkage combination;

[0014] Call the inventory management systems of the areas to be measured and the linkage analysis areas within each linkage combination to obtain the introduction quantities of various categories of items in the two areas. Calculate the difference between the introduction quantities of the same category of items in the two areas within the linkage combination to obtain the supervision similarity of the corresponding category of items within the linkage combination. Take the average value of the supervision similarities of all categories of items within each linkage combination as the supervision similarity threshold corresponding to the linkage combination;

[0015] Screen out the item categories with supervision similarity lower than the supervision similarity threshold within the corresponding linkage combination and mark them;

[0016] Among them, the calculation formula for supervision similarity in step one is:

[0017] S c =|Q t -Q a | / max(Q t ,Q a )

[0018] Among them, S c is the category supervision similarity, Q t is the introduction quantity of the area to be measured, and Q a is the introduction quantity of the adjacent area;

[0019] Step two: Feature engineering and regional relationship modeling:

[0020] Integrate the item circulation data and the supervision records of highly correlated categories to generate an item comprehensive linkage feature vector;

[0021] Among them, the formula for the item comprehensive linkage feature vector in step two is:

[0022]

[0023] Among them, w i is the category weight, and S ci is the supervision similarity of the i-th marked item;

[0024] Combine the personnel flow volume data between regions, classify the regional linkage relationship through a clustering algorithm, and quantify the collaborative effect between regions;

[0025] The calculation formula for the personnel flow volume data in step two:

[0026]

[0027] Among them, P t is the population size of the area to be measured, P a is the population size of the adjacent area, d is the geographical distance, and k is the correction coefficient determined according to the regional economic relevance;

[0028] Step three: Identification of abnormal areas and data cleaning:

[0029] Based on the DBSCAN density clustering algorithm, perform anomaly detection on inter-regional transportation data such as frequency, path, and timeliness, and identify transportation anomaly regions; eliminate the anomaly regions and retain the normal circulation regions as subsequent modeling samples;

[0030] Step 4: Risk quantification and comprehensive evaluation:

[0031] Set dynamic risk weights based on the frequency of market address transfer and key indicators of regional output, and combine historical risk data; input the item linkage characteristics, regional relationship classification results, and risk weights into the comprehensive evaluation model, calculate the risk index of the area to be measured, and output the risk prediction result;

[0032] Among them, the formula for the dynamic risk weight in Step 4 is:

[0033]

[0034] Among them, M is the frequency of address transfer, Q l is the regional output, is the proportion of local output, is the single-region risk score, and the weight coefficients satisfy α + β + γ = 1, which are determined through training with historical risk data, and α is positively correlated with M, β is negatively correlated with α, β, and γ are weight coefficients, which are determined through training with historical data.

[0035] Preferably, the cross-regional supervision risk prediction system based on urban functions includes an intelligent perception layer, a data processing layer, a linkage analysis layer, a risk prediction layer, and a cloud management layer;

[0036] Among them, the intelligent perception layer is used to collect data on regional area, item import volume, and personnel flow volume;

[0037] The intelligent perception layer includes a data collection module and an interface module;

[0038] Among them, the data collection module calls the structure database to calculate the area of the area to be measured, expands the adjacent areas based on the side length, and forms a linkage combination;

[0039] The interface module: docks with the inventory management system and the personnel flow monitoring platform, such as transportation hub data and logistics tracking systems, to obtain data on item import volume, personnel interaction volume, transportation trajectories, etc. in real time;

[0040] The data processing layer includes a supervision similarity calculation unit and an anomaly filtering unit;

[0041] The data processing layer includes a supervision similarity calculation unit and an anomaly data filtering unit;

[0042] The regulatory similarity calculation unit calculates the difference in the import volume of items of the same category within the linkage combination;

[0043] The abnormal data filtering unit identifies transportation anomalies based on the DBSCAN algorithm, such as a route deviation threshold > 15% and a detention time > 24 hours;

[0044] The linkage analysis layer is used to generate comprehensive item linkage features and classify regional linkage relationships;

[0045] The linkage analysis layer is implemented through the following steps:

[0046] S31. Combine the personnel flow data collected by the intelligent perception layer to classify the regional linkage relationships;

[0047] S32. Generate comprehensive item linkage features based on the regulatory data of highly correlated product categories and personnel flow characteristics;

[0048] The risk prediction layer is used to implement single-region modeling and comprehensive risk assessment;

[0049] The risk prediction layer implements comprehensive risk assessment through the following steps:

[0050] S41. Set risk weights based on the frequency of market address transfer and production volume;

[0051] S42. Input the comprehensive item linkage features and the classification results of regional linkage relationships into the evaluation model to calculate the risk index of the area to be measured;

[0052] S43. The cloud management layer generates visual warning information based on the risk index and outputs regulatory policies based on a preset policy library;

[0053] Meanwhile, for the remaining marked areas, a regulatory risk model is established using the random forest algorithm, and the input parameters include: comprehensive item linkage features F, personnel flow intensity G, and number of transportation anomalies E;

[0054] The cloud management layer is used to provide visual warnings and generate regulatory policies;

[0055] The cloud management layer includes a visual warning platform and a regulatory policy engine;

[0056] Among them, the visual warning platform generates a cross-regional risk heat map and marks high-risk linkage combinations;

[0057] The regulatory policy engine automatically generates a cross-regional joint inspection plan;

[0058] The early warning priority calculation formula of the cloud management layer is:

[0059]

[0060] Among them, T is the risk duration, and a red warning is triggered when P≥0.7.

[0061] The advantages of the embodiments of this application are as follows:

[0062] By comprehensively considering various correlation factors between regions, the present invention realizes accurate prediction of cross-regional market supervision risks. It not only counts the introduction volume of the same category of items in the area to be measured and its neighboring areas, deeply analyzes the similarity of item supervision, but also comprehensively combines multi-dimensional data such as the volume of personnel flow interaction, the frequency of market address transfer, and the local market item output. Through in-depth mining and organic combination of these data, it can more comprehensively and accurately grasp the internal connection and change trend of market supervision risks between regions.

[0063] According to the classification results of the linkage relationship between different neighboring regions and the area to be measured, the present invention specifically selects a prediction method for single-region supervision risk analysis, avoiding the inadaptability of traditional single prediction methods in different scenarios, greatly improving the analysis efficiency. At the same time, through an automated data collection and processing process, it reduces manual intervention and data processing time.

[0064] The determination of neighboring regions, multi-dimensional data fusion analysis, and flexible prediction method selection in the present invention can better cope with complex and changeable market environments, whether facing sudden market fluctuations, policy changes, or adjustments in the inter-regional economic cooperation model, and make accurate predictions in a timely manner.

[0065] The present invention is not only applicable to the prediction of market supervision risks between cities and regions of different scales, but also can be applied to market supervision in various industry fields. Whether it is industries related to people's livelihood such as food and medicine, or industrial fields such as electronics and machinery, the data indicators and prediction methods can be appropriately adjusted according to the industry characteristics, and the methods and systems of the present invention can be applied to accurately predict cross-regional supervision risks, providing strong support for ensuring the stable and healthy development of the market. Brief Description of the Drawings

[0066] Figure 1 It is a schematic flow chart of the cross-regional supervision risk prediction method based on urban functions of the present invention;

[0067] Figure 2 It is a schematic flow chart of the cross-regional supervision risk prediction system based on urban functions of the present invention. Detailed Embodiments

[0068] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention. In addition, for the convenience of description below, the "upper", "lower", "left", "right", etc. cited are consistent with the upper, lower, left, right, etc. of the accompanying drawings themselves. The "first", "second", etc. in the following text are used for distinction in description and have no other special meanings.

[0069] Through the embodiments of the present application, a method and system for predicting cross-regional supervision risks based on urban functions are provided to solve the problems in the prior art. By comprehensively considering various correlation factors between regions, the present invention realizes accurate prediction of cross-regional market supervision risks. It not only counts the import volume of the same category of items in the area to be measured and its neighboring areas, deeply analyzes the similarity of item supervision, but also comprehensively considers multi-dimensional data such as the volume of personnel flow interaction, the frequency of market address transfer, and the local market item production. Through in-depth mining and organic combination of these data, the internal connection and change trend of market supervision risks between regions can be grasped more comprehensively and accurately. According to the classification results of the linkage relationship between different neighboring regions and the area to be measured, the present invention selectively uses prediction methods for single-region supervision risk analysis, avoiding the inadaptability of traditional single prediction methods in different scenarios, greatly improving the analysis efficiency. At the same time, through an automated data collection and processing process, manual intervention and data processing time are reduced. The determination of neighboring regions, multi-dimensional data fusion analysis, and flexible prediction method selection in the present invention can better cope with complex and changeable market environments, whether facing sudden market fluctuations, policy changes, or adjustments in the inter-regional economic cooperation model, and make accurate predictions in a timely manner. The present invention is not only applicable to the prediction of market supervision risks between cities and regions of different scales, but also can be applied to the market supervision of various industry fields. Whether it is industries related to people's livelihood such as food and medicine, or industrial fields such as electronics and machinery, the data indicators and prediction methods can be appropriately adjusted according to the industry characteristics, and the methods and systems of the present invention can be used for accurate cross-regional supervision risk prediction, providing strong support for ensuring the stable and healthy development of the market.

[0070] The overall idea of the technical solution in the embodiments of the present application to solve the above problems is as follows:

[0071] Embodiment 1

[0072] This embodiment provides a method for predicting cross-regional supervision risks based on urban functions. The method architecture is as shown in the accompanying drawings and includes:

[0073] Step 1: Data preprocessing and screening of linkage regions:

[0074] Screen the adjacent areas as the linkage analysis areas. The method for screening the adjacent areas is to call the two-dimensional data in the structure database to calculate the area of the area to be measured, expand the adjacent areas based on the side length to form a linkage combination, calculate the supervision similarity of the items of the same category and mark the high-correlation categories;

[0075] The steps of screening the adjacent areas as the linkage analysis areas in Step 1 include:

[0076] Call the two-dimensional data of the area to be measured in the structure database to calculate the area of the area to be measured. Use the length or width of the area to be measured as the common side and expand it outward until the expanded area of each side is the same as the area of the area to be measured. Set the expanded area of each side as the adjacent area, use each adjacent area as multiple different linkage analysis areas of the area to be measured, and use the combination of the area to be measured and different linkage analysis areas as the linkage combination;

[0077] Call the inventory management systems of the area to be measured and the linkage analysis areas within each linkage combination to obtain the import quantities of various categories of items in the two areas. Subtract the import quantities of the same category of items in the two areas within the linkage combination to obtain the supervision similarity of the corresponding category of items within the linkage combination, and use the average value of the supervision similarities of all categories of items within each linkage combination as the supervision similarity threshold of the corresponding linkage combination;

[0078] Screen out the item categories with supervision similarity lower than the supervision similarity threshold within the corresponding linkage combination and mark them;

[0079] Among them, the supervision similarity calculation formula in Step 1 is:

[0080] S c =|Q t -Q a | / max(Q t ,Q a )

[0081] Among them, S c is the category supervision similarity, Q t is the import quantity of the area to be measured, and Q a is the import quantity of the adjacent area;

[0082] Step 2: Feature engineering and regional relationship modeling:

[0083] Integrate the item circulation data and the supervision records of high-correlation categories to generate an item comprehensive linkage feature vector;

[0084] Among them, the formula for the item comprehensive linkage feature vector in Step 2 is:

[0085]

[0086] Among them, wi is the category weight, S ci is the regulatory similarity of the tagged items in category i;

[0087] Combined with the data on inter-regional personnel mobility, the regional linkage relationships are classified through clustering algorithms to quantify the inter-regional synergy effects;

[0088] The formula for calculating the personnel flow data in step 2 is:

[0089]

[0090] Among them, P t is the population size of the area to be tested, P a is the population size of the neighboring region, d is the geographical distance, and k is the correction coefficient determined according to the regional economic correlation;

[0091] Step 3: Abnormal area identification and data cleaning:

[0092] Based on the DBSCAN density clustering algorithm, anomaly detection is performed on inter-regional transportation data, such as frequency, path, and timeliness, to identify abnormal transportation areas; abnormal areas are eliminated and normal circulation areas are retained as subsequent modeling samples;

[0093] Step 4: Risk quantification and comprehensive assessment:

[0094] Dynamic risk weights are set based on the market address transfer frequency, regional production key indicators, and historical risk data; item linkage characteristics, regional relationship classification results, and risk weights are input into the comprehensive assessment model to calculate the risk index of the area to be tested and output the risk prediction results;

[0095] The dynamic risk weight formula in step 4 is:

[0096]

[0097] Among them, M is the address transfer frequency, Q l is the regional output, is the proportion of local production, is the risk score of a single region, and the weight coefficient satisfies α+β+γ=1, which is determined by historical risk data training, and α is positively correlated with M, and β is positively correlated with Negative correlation, α, β, γ are weight coefficients, which are determined through historical data training.

[0098] By adopting the above technical solution:

[0099] Assume that the area to be tested is a medium-sized city A, and there are multiple cities of different sizes around it as potential linkage analysis areas.

[0100] Step 1: Data preprocessing and linkage area screening:

[0101] Determine the linkage analysis area: Obtain the two-dimensional geographical data of City A through its geographical information database, and calculate the area of City A as S a = 500 square kilometers. Expand outward with the long side of City A as the common side. Using geographical information system tools, three adjacent areas B, C, and D are obtained after expansion, and the expanded areas are all the same as the area of City A. Combine City A with B, C, and D respectively to form linkage combinations.

[0102] Calculate the supervision similarity and screen for marking: Obtain the data on the introduction volume of various categories of goods from the inventory management systems of City A and its adjacent areas B, C, and D. Taking the electronic product category as an example, City A introduced 1,000 mobile phones of a certain brand in the past month, and adjacent area B introduced 800. Then the supervision similarity of the mobile phone category within this linkage combination is |1000 - 800| = 200. Calculate the average value of the supervision similarities of various categories of items within all linkage combinations. Assuming that the supervision similarity threshold is obtained as 150, screen out the item categories with a supervision similarity lower than 150, such as a certain type of specialty food, and mark it.

[0103] Step 2: Feature engineering and regional relationship modeling:

[0104] Obtain the comprehensive linkage features of items: For the marked specialty food category, calculate the linkage purchase similarity between City A and adjacent area B. After normalizing the linkage purchase similarities of each marked item, use the principal component analysis method to obtain the comprehensive linkage feature vector of items between City A and adjacent area B.

[0105] Analyze and classify the linkage relationship: Through the data interface of the transportation department and the analysis of mobile phone signaling data, combined with the comprehensive linkage feature vector of items, use the K-means clustering analysis algorithm to classify the linkage relationship between City A and adjacent area B as general linkage.

[0106] Step 3: Abnormal area identification and data cleaning:

[0107] Collect transportation anomalies and exclude some areas: Connect with the transportation management platform of the local logistics association to obtain the transportation anomaly data of each adjacent area.

[0108] Single-region supervision risk analysis: For the remaining adjacent areas B and D, since the linkage relationship between B and A is general linkage, use a multiple linear regression model for single-region supervision risk analysis. Use the LSTM model for analysis. After sorting the relevant historical data in time series and inputting them into the LSTM model for training and prediction, obtain the single-region supervision risk analysis results of City A and adjacent areas B and D.

[0109] Step 4: Risk quantification and comprehensive evaluation:

[0110] Set regulatory risk weights: Collect the market address transfer frequencies and local market item production data of adjacent regions B and D through the data of market research institutions and the information released by government statistical departments, and use the analytic hierarchy process to construct a hierarchical structure model.

[0111] Comprehensively evaluate regulatory risks: Combine the regulatory risk weights of adjacent regions B and D and the single-region regulatory risk analysis results to comprehensively evaluate the regulatory risks of the area A to be measured.

[0112] Embodiment 2

[0113] This embodiment provides a cross-regional regulatory risk prediction system based on urban functions. The system architecture is as shown in the attached drawings, including an intelligent perception layer, a data processing layer, a linkage analysis layer, a risk prediction layer, and a cloud management layer;

[0114] Among them, the intelligent perception layer is used to collect data on regional area, item introduction volume, and personnel flow volume;

[0115] The intelligent perception layer includes a data collection module and an interface module;

[0116] Among them, the data collection module calls the structure database to calculate the area of the area to be measured, expands the adjacent areas based on the side length, and forms a linkage combination;

[0117] Interface module: Connect to the inventory management system and the personnel flow monitoring platform, such as traffic hub data and logistics tracking systems, to obtain real-time data on item introduction volume, personnel interaction volume, transportation trajectories, etc.;

[0118] The data processing layer includes a regulatory similarity calculation unit and an anomaly filtering unit;

[0119] The data processing layer includes a regulatory similarity calculation unit and an anomaly filtering unit. The data processing layer includes a regulatory similarity calculation unit and an abnormal data filtering unit;

[0120] The regulatory similarity calculation unit calculates the difference in the introduction volume of the same category of items within the linkage combination;

[0121] The abnormal data filtering unit identifies transportation anomalies based on the DBSCAN algorithm, such as a route deviation threshold > 15% and a detention time > 24 hours;

[0122] The linkage analysis layer is used to generate comprehensive item linkage features and classify regional linkage relationships;

[0123] The linkage analysis layer is implemented through the following steps:

[0124] S31. Combine the personnel flow volume data collected by the intelligent perception layer to classify the regional linkage relationships;

[0125] S32. Generate comprehensive linkage features of items based on regulatory data and personnel flow characteristics of highly correlated categories;

[0126] The risk prediction layer is used to implement single-region modeling and comprehensive risk assessment;

[0127] The risk prediction layer implements comprehensive risk assessment through the following steps:

[0128] S41. Set risk weights based on the frequency of market address transfer and production volume;

[0129] S42. Input the comprehensive linkage features of items and the classification results of regional linkage relationships into the evaluation model to calculate the risk index of the area to be measured;

[0130] S43. The cloud management layer generates visual warning information based on the risk index and outputs regulatory strategies based on a preset policy library;

[0131] At the same time, for the reserved marked areas, a regulatory risk model is established using the random forest algorithm, and the input parameters include: comprehensive linkage features F of items, personnel flow intensity G, and number of transportation anomalies E;

[0132] The cloud management layer is used to provide visual warnings and generate regulatory strategies;

[0133] The cloud management layer includes a visual warning platform and a regulatory strategy engine;

[0134] Among them, the visual warning platform generates a cross-regional risk heat map and marks high-risk linkage combinations;

[0135] The regulatory strategy engine automatically generates a cross-regional joint inspection plan;

[0136] The warning priority calculation formula of the cloud management layer is:

[0137]

[0138] Among them, T is the risk duration, and a red warning is triggered when P≥0.7.

[0139] By adopting the above technical solutions:

[0140] The area determination module is used to select multiple adjacent areas as linkage analysis areas in the area to be measured. Specifically, it calls the two-dimensional data of the area to be measured in the structure database to calculate the area of the area to be measured, extends the length or width of the area to be measured as a common side outward until the area extended by each side is the same as the area of the area to be measured, sets the extended area of each side as an adjacent area, takes each adjacent area as a different linkage analysis area of the area to be measured, and takes the combination of the area to be measured and different linkage analysis areas as a linkage combination;

[0141] The similarity calculation and screening module is used to count the import volume of items of the same category in the area to be measured and the linkage analysis area, call the inventory management systems of the area to be measured and the linkage analysis area within each linkage combination to obtain the import volume of items of various categories in the two areas, calculate the difference between the import volume of items of the same category in the two areas within the linkage combination to obtain the supervision similarity of the corresponding category of items within the linkage combination, take the average value of the supervision similarities of all categories of items within each linkage combination as the supervision similarity threshold of the corresponding linkage combination, screen out the item categories with supervision similarity lower than the supervision similarity threshold within the corresponding linkage combination and mark them;

[0142] The linkage feature and relationship analysis module is used to normalize the linkage purchase similarities of various marked items in each adjacent area and the area to be measured, reduce the dimensionality of the linkage purchase similarities of multiple marked items through the principal component analysis method to obtain the comprehensive linkage feature vectors of items in the corresponding adjacent area and the area to be measured, count the personnel flow interaction volume between different adjacent areas and the area to be measured, and use the clustering analysis algorithm to analyze and classify the linkage relationships between each adjacent area and the area to be measured in combination with the comprehensive linkage features of items and the personnel flow interaction volume, and classify the linkage relationships into tight linkage, general linkage, and loose linkage categories;

[0143] The exception handling and prediction module is used to collect the transportation exception situations in each adjacent area, eliminate some adjacent areas according to the transportation exception situations in the adjacent areas, mark the remaining adjacent areas, and select different prediction methods to conduct single-area supervision risk analysis on the area to be measured according to the classification results of the linkage relationships between the marked areas and the area to be measured, and obtain the single-area supervision risk analysis results of each marked area and the area to be measured;

[0144] The weight setting and comprehensive evaluation module is used to collect the market address transfer frequency and the local market item production volume of each marked area, set the supervision risk weights by using the analytic hierarchy process, and comprehensively evaluate the supervision risk of the area to be measured in combination with the supervision risk weights of each marked area and the single-area supervision risk analysis results of the area to be measured.

[0145] Finally, it should be noted that: Obviously, the above embodiments are only examples for clearly illustrating the present invention, rather than limiting the implementation manners. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation manners here. And the obvious changes or modifications derived therefrom are still within the protection scope of the present invention.

Claims

1. A cross-regional supervision risk prediction method based on urban functions, characterized in that, Including: Step 1: Data preprocessing and screening of linkage areas Screen adjacent areas as linkage analysis areas. The method for screening adjacent areas is to call the two-dimensional data in the structure database to calculate the area of the area to be measured, expand the adjacent areas based on the side length to form a linkage combination, calculate the supervision similarity of the same category of items, and mark the highly correlated product categories; Step 2: Feature engineering and regional relationship modeling Integrate the item circulation data and the supervision records of highly correlated product categories to generate an item comprehensive linkage feature vector; Combine the personnel flow volume data between regions, and classify the regional linkage relationship through a clustering algorithm to quantify the synergy effect between regions; Step 3: Identification of abnormal areas and data cleaning Based on the DBSCAN density clustering algorithm, perform anomaly detection on the transportation data between regions, such as frequency, path, and timeliness, to identify transportation abnormal areas; eliminate the abnormal areas and retain the normal circulation areas as subsequent modeling samples; Step 4: Risk quantification and comprehensive evaluation Set dynamic risk weights according to the frequency of market address transfer and key indicators of regional output, combined with historical risk data; Input the item linkage features, the classification results of regional relationships, and the risk weights into the comprehensive evaluation model, calculate the risk index of the area to be measured, and output the risk prediction results.

2. The cross-regional supervision risk prediction method based on urban functions according to claim 1, wherein The steps of screening adjacent areas as linkage analysis areas in Step 1 include: Call the two-dimensional data of the area to be measured in the structure database to calculate the area of the area to be measured, expand outward with the length or width of the area to be measured as the common side until the expanded area of each side is the same as the area of the area to be measured. Set the expanded area of each side as the adjacent area, use each adjacent area as multiple different linkage analysis areas of the area to be measured, and use the combination of the area to be measured and different linkage analysis areas as the linkage combination; Call the inventory management systems of the area to be measured and the linkage analysis areas within each linkage combination to obtain the import volume of various categories of items in the two areas. Subtract the import volume of the same category of items in the two areas within the linkage combination to obtain the supervision similarity of the corresponding category of items within the linkage combination. Take the average value of the supervision similarities of all categories of items within each linkage combination as the supervision similarity threshold of the corresponding linkage combination; Screen out the item categories with supervision similarity lower than the supervision similarity threshold within the corresponding linkage combination and mark them.

3. A cross-regional supervision risk prediction method based on urban functions according to claim 1, characterized in that The calculation formula for supervision similarity in Step 1 is: S c = |Q t - Q a | / max(Q t , Q a ) Among them, S c is the similarity of category supervision, Q t is the introduction quantity of the area to be measured, Q a is the introduction quantity of the adjacent area.

4. A cross-regional supervision risk prediction method based on urban functions according to claim 1, characterized in that The formula for the item comprehensive linkage feature vector in Step 2 is: Among them, w i is the category weight, and S ci is the regulatory similarity of the i-th category of marked items.

5. A cross-regional supervision risk prediction method based on urban functions according to claim 1, characterized in that, The calculation formula for the personnel flow volume data in Step 2: Among them, P t is the population scale of the area to be measured, and P a is the population scale of the adjacent area, d is the geographical distance, and k is the correction coefficient determined according to the regional economic relevance.

6. The cross-regional supervision risk prediction method based on urban functions according to claim 1, characterized in that The formula for the dynamic risk weight in Step 4 is: Among them, M is the address transfer frequency, and Q l is the regional output, is the proportion of local output, is the single-region risk score. The weight coefficients satisfy α + β + γ = 1, which are determined by training with historical risk data. Moreover, α is positively correlated with M, and β is negatively correlated with α, β, and γ are weight coefficients, which are determined by training with historical data.

7. A cross-regional supervision risk prediction system based on urban functions, characterized in that, This cross-regional supervision risk prediction system based on urban functions is applicable to a cross-regional supervision risk prediction method according to any one of claims 1-6. The cross-regional supervision risk prediction system based on urban functions includes an intelligent perception layer, a data processing layer, a linkage analysis layer, a risk prediction layer, and a cloud management layer; Among them, the intelligent perception layer is used to collect data on regional area, item import volume, and personnel flow volume; The data processing layer includes a supervision similarity calculation unit and an anomaly filtering unit; The linkage analysis layer is used to generate item comprehensive linkages and classify regional linkage relationships; The risk prediction layer is used to implement single-region modeling and comprehensive risk assessment; The cloud management layer is used to provide visual warning and regulatory strategy generation.

8. The cross-regional supervision risk prediction system based on urban functions according to claim 7, characterized in that, The intelligent perception layer includes a data acquisition module and an interface module; Among them, the data acquisition module calls the structure database to calculate the area of the area to be measured, expands the adjacent area based on the side length, and forms a linkage combination; The interface module: docks with the inventory management system and the personnel flow monitoring platform, such as transportation hub data and logistics tracking systems, to obtain data such as the quantity of imported items, the quantity of personnel interaction, and the transportation trajectory in real time.

9. The cross-regional supervision risk prediction system based on urban functions according to claim 7, characterized in that, The linkage analysis layer is implemented through the following steps: S31. Classify the regional linkage relationship in combination with the personnel flow data collected by the intelligent perception layer; S32. Generate the comprehensive linkage characteristics of items based on the regulatory data of high-correlation categories and the personnel flow characteristics.

10. The cross-regional supervision risk prediction system based on urban functions according to claim 7, characterized in that, The risk prediction layer realizes comprehensive risk assessment through the following steps: S41. Set the risk weight based on the market address transfer frequency and production volume; S42. Input the comprehensive linkage characteristics of items and the classification results of regional linkage relationships into the evaluation model to calculate the risk index of the area to be measured; S43. The cloud management layer generates visual warning information according to the risk index and outputs a regulatory strategy based on the preset policy library.

Images