Network point management method and device based on dynamic grid division, electronic equipment, medium and program product

Abstract

The invention provides a website management method and device based on dynamic grid division, electronic equipment, a medium and a program product, and can be applied to the technical field of artificial intelligence, the technical field of block chains and the technical field of Internet of Things. The method comprises the following steps: acquiring service area grids and multi-source real-time data of a plurality of network points; calculating the real-time service load of the service area grid based on the multi-source real-time data, and generating a grid load weight; in response to the condition that the grid load weight of any network point meets a preset adjustment condition, updating the grid boundary of the service area grid; and performing customer demand prediction and / or resource scheduling strategy generation for the plurality of network points based on the updated service area grid.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of artificial intelligence, the technical field of blockchain and the technical field of Internet of Things, and more particularly to a grid management method and device based on dynamic grid division, equipment, medium and program product. BACKGROUND

[0002] With the gradual introduction of intelligent management mechanisms in the digital transformation of bank outlets, fine division and intelligent optimization of outlet service areas have become a key requirement.

[0003] However, the existing grid management technology still has many technical bottlenecks in multi-dimensional data modeling, dynamic strategy generation and system collaborative processing, etc. First, the current grid division method relies on static rules or geographical boundaries, lacks modeling capability for multi-source dynamic data such as time-series transaction data, population heat migration and urban function evolution, and is difficult to realize real-time reconstruction and flexible adjustment of the grid. Second, customer behavior modeling is mainly based on historical statistics, lacks prediction algorithms and scene adaptation mechanisms for user behavior trends, and cannot drive the adaptive evolution of service strategies or resource allocation. Third, grid coordination relies on manual or rule triggering and has not formed a linkage mechanism, making it difficult to identify associated risks or resource scheduling coordination paths. Finally, there is a lack of scheduling algorithm system with feedback learning ability, resulting in poor decision granularity and real-time performance of personnel scheduling, cash supply and equipment control. SUMMARY

[0004] In view of the above problems, the present application provides a grid management method, device, equipment, medium and program product based on dynamic grid division.

[0005] According to a first aspect of the present application, a grid management method based on dynamic grid division is provided, the method comprising: acquiring service area grids of a plurality of outlets and multi-source real-time data; calculating real-time service load of the service area grids based on the multi-source real-time data, and generating grid load weights; in response to any grid load weight of the outlets meeting a preset adjustment condition, updating the grid boundary of the service area grid; and based on the updated service area grid, performing customer demand prediction and / or resource scheduling strategy generation for the plurality of outlets.

[0006] According to an embodiment of the present application, the method further comprises constructing service area grids of a plurality of outlets, comprising:

[0007] generating a plurality of initial service units based on geographic location information of the plurality of outlets; calculating outlet coverage weights based on service demand information and service supply capability information of the plurality of initial service units; and adjusting the boundaries of the plurality of initial service units based on the outlet coverage weights to form the service area grids.

[0008] According to an embodiment of the present application, the multi-source real-time data of the plurality of service points is obtained, including: obtaining internal data by using an Internet of Things sensor, the internal data including at least one of customer flow, equipment use time length, and queuing state; obtaining external data by using an external data interface, the external data including at least one of regional population flow, economic fluctuation index, and weather change information; and time aligning and space mapping the internal data and the external data to obtain the multi-source real-time data.

[0009] According to an embodiment of the present application, the real-time service load of the service area grid is calculated based on the multi-source real-time data to generate a grid load weight, including: encoding the multi-source real-time data into a load feature vector, constructing a service area graph based on the load feature vector and a spatial adjacency relationship of the service area grid, wherein a node of the service area graph represents the service area grid, and an edge of the service area graph represents a spatial relationship and / or a business relationship between the service area grids; performing information propagation on the service area graph by using a graph convolution network to generate an initial load embedding representation of each service area grid; and inputting the initial load embedding representation into a graph attention network to generate the grid load weight.

[0010] According to an embodiment of the present application, the initial load embedding representation is input into the graph attention network to generate the grid load weight, including: obtaining an adjacent grid corresponding to a target service area grid, calculating an attention weight between the target service area grid and the adjacent grid based on the initial load embedding representation and an attention mechanism in the graph attention network, wherein the target service area grid represents any service area grid; performing weighted aggregation on the initial load embedding representation of the adjacent grid based on the attention weight to obtain a context representation vector of the target service area grid; and inputting the context representation vector into a nonlinear transformation layer in the graph attention network to generate a grid load weight corresponding to the target service area grid.

[0011] According to an embodiment of the present application, the customer demand prediction for the plurality of service points is performed based on the updated service area grid, including: obtaining historical service load data corresponding to the updated service area grid, constructing a spatiotemporal feature tensor based on the updated service area grid and the historical service load data; and inputting the spatiotemporal feature tensor into a customer demand prediction network to output customer future service demand information.

[0012] According to an embodiment of the present application, based on the updated service area grid, resource scheduling strategy generation for multiple points is performed, including: obtaining real-time service load corresponding to the updated service area grid and corresponding resource allocation constraint information; constructing a resource demand parameter set based on the real-time service load; and inputting the resource demand parameter set and the resource allocation constraint into a resource allocation optimization model, and generating a resource scheduling strategy based on the output result of the resource allocation optimization model.

[0013] According to an embodiment of the present application, the method further comprises: storing the real-time service load in the updated service area grid, the customer demand prediction result and the resource scheduling result into a consortium blockchain network; and in response to any updated service area grid detecting an abnormal behavior event, generating a data announcement containing an event feature summary, and broadcasting based on the data announcement in the consortium blockchain network.

[0014] According to an embodiment of the present application, the method further comprises: obtaining a pre-constructed three-dimensional visualization interface; and in response to a user selecting a target service area grid, dynamically displaying at least one of the structural layout, historical load trend, business demand prediction information and resource scheduling information of the target service area grid based on the three-dimensional visualization interface.

[0015] The second aspect of the present application provides a point management device based on dynamic grid division, the device comprising: a data acquisition module, configured to acquire service area grids of multiple points and multi-source real-time data; a grid load weight generation module, configured to calculate real-time service load of the service area grid based on the multi-source real-time data, and generate a grid load weight; a grid boundary updating module, configured to update the grid boundary of the service area grid in response to any grid load weight of a point meeting a preset adjustment condition; and a point management module, configured to execute customer demand prediction and / or resource scheduling strategy generation for the multiple points based on the updated service area grid.

[0016] The third aspect of the present application provides an electronic device, comprising: one or more processors; a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0017] The fourth aspect of the present application further provides a computer-readable storage medium having a computer program or instructions stored thereon, wherein the computer program or instructions are executed by a processor to implement the steps of the above method.

[0018] The fifth aspect of the present application further provides a computer program product comprising a computer program or instructions, wherein the computer program or instructions are executed by a processor to implement the steps of the above method.

[0019] According to the embodiments of the present application, by calculating the real-time service load of the service area grid, the grid load weight is generated, which can realize the dynamic division and intelligent management of multiple point service areas, evaluate the grid service load condition in combination with multi-source real-time data, and flexibly update the grid boundary based on the preset adjustment condition, effectively adapt to the service pressure changes caused by customer mobility and business fluctuations; at the same time, through the updated service area grid, the customer demand prediction and resource scheduling strategy generation are executed, the prediction accuracy and resource allocation efficiency are improved, thereby realizing the linkage optimization of service area structure and operation and maintenance strategy, and enhancing the overall response capability and service level of the system. BRIEF DESCRIPTION OF DRAWINGS

[0020] The above and other objects, features and advantages of the present application will become more apparent from the following description of the embodiments of the present application taken in conjunction with the accompanying drawings, in which:

[0021] Figure 1 An application scenario diagram of a point management method, device, equipment, medium and program product based on dynamic grid division according to embodiments of the present application is schematically shown;

[0022] Figure 2 A flowchart of a point management method based on dynamic grid division according to embodiments of the present application is schematically shown;

[0023] Figure 3 A flowchart of a method of calculating the real-time service load of the service area grid according to some exemplary embodiments of the present application is schematically shown;

[0024] Figure 4 A multi-layer system architecture of point management based on dynamic grid division according to some exemplary embodiments of the present application is schematically shown;

[0025] Figure 5 A structural block diagram of a point management device based on dynamic grid division according to embodiments of the present application is schematically shown; and

[0026] Figure 6 A block diagram of an electronic device suitable for implementing a point management method based on dynamic grid division according to embodiments of the present application is schematically shown. DETAILED DESCRIPTION

[0027] In the following, embodiments of the present application will be described with reference to the drawings. It should be understood, however, that the description going with the drawings is merely exemplary and not intended to limit the scope of the present application. In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of embodiments of the present application. However, it will be apparent to one skilled in the art that one or more embodiments can be practiced without these specific details. In other instances, well-known structures and

[0028] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application. As used herein, the term "includes" and tautological expressions thereof, means the inclusion of the stated features, steps, operations, and / or elements but not to the exclusion of one or more other features, steps, operations, and / or elements.

[0029] All terms used herein including technical and scientific terms have the meanings commonly understood by one of ordinary skill in the art unless otherwise defined. It should be noted that the terms used herein are defined as having a meaning that is consistent with the context of the specification in which the terms are utilized and the terms should not be interpreted in an idealized or overly formal sense.

[0030] In situations where similar terminology is used for similar items, one of ordinary skill in the art would understand the terminology to have a meaning in accordance with the description of the specification and would interpret the terminology in the idealized or overly formal sense.

[0031] The terms described herein will be explained and described below.

[0032] Mesh: refers to a regular or irregular spatial division unit of an area in a physical or logical space, each mesh represents an independent spatial section, and is used as a basic spatial granularity unit for calculation, prediction, scheduling, etc.

[0033] Alliance blockchain network: refers to a blockchain system maintained by multiple trusted subjects, each node has consensus capability, supports sharing and notarization of data between service area meshes, and verification of traceability, and is used to ensure the tamper-proof nature and full-process audit capability of scheduling data and load information.

[0034] Graph convolutional network (GCN): refers to a neural network for information propagation of graph structure data, which realizes node representation learning by aggregating feature vectors of adjacent nodes, and can capture spatial dependence and topological structure features between service areas, and is used to generate embedding representation of each mesh unit.

[0035] Graph Attention Network (GAT): refers to a deep learning model that introduces attention mechanism on the basis of graph structure, dynamically learns the influence weight of each adjacent node on the center node, realizes embedding aggregation and feature enhancement under context awareness, and is used in the present scheme to generate more accurate grid load weight.

[0036] With the acceleration of the digitalization and intelligentization of bank outlets, the traditional operation mode centered on physical outlets is gradually shifting to an intelligent service system with data-driven and algorithmic decision-making as the core. In this context, the fine division, dynamic optimization and multi-dimensional linkage of outlet service areas have become the key to improving the overall operational efficiency and risk response capability of banks. However, the current grid management mode used by most banks still has significant technical limitations.

[0037] Firstly, in terms of grid division, existing methods are mostly based on static administrative divisions, fixed historical transaction heat maps or manual rule configuration, lacking the ability to fuse and model multi-source heterogeneous data such as geographic heat, customer flow, transaction frequency, event density, etc. This makes it difficult for the system to respond to the spatio-temporal changes of regional economic activities, and unable to support real-time reconstruction or flexible adjustment of outlet service range. The lack of dynamic grid redrawing models supported by time series prediction and transfer learning is the key to the lack of scalability of current systems.

[0038] Secondly, in terms of customer behavior modeling and strategy generation, traditional outlet strategy design mainly relies on static customer profiling, focusing on rule engines or manual stratification analysis, lacking prediction algorithms for future behavior trends and service combination recommendation capabilities based on context awareness. Especially under the trend of online and cross-regional customer behavior, relying solely on historical interaction data or channel preference statistics makes it difficult to drive forward-looking adaptation of service capabilities and resource allocation. How to build user behavior prediction models with time series modeling, reinforcement learning or graph attention mechanism is a technical bottleneck for service strategy evolution.

[0039] Thirdly, in terms of cross-grid coordination mechanism, the current system lacks modeling capabilities based on structured data graph, business transmission path or risk propagation model, resulting in a heavy reliance on manual configuration or preset processes for inter-grid linkage response, making it difficult to capture cross-regional, cross-outlet event linkage, risk spread or resource redundancy issues in a timely manner. Especially in the face of complex risk situations such as regional enterprise chain guarantees, cross-region risk flow, etc., there is a lack of early warning and joint defense mechanisms based on causal reasoning, knowledge graph or graph neural network.

[0040] Finally, in terms of resource scheduling and adaptive optimization, existing systems often use fixed scheduling templates or empirical rules to allocate human resources, cash, and equipment, and cannot dynamically adjust in real time based on business load, environmental changes, and customer behavior. The scheduling process lacks a strategy training mechanism based on reinforcement learning, multi-objective optimization, or simulation feedback, resulting in significant resource bottlenecks during peak periods and severe idle problems during low periods, affecting overall service capacity and operational cost control.

[0041] Based on this, the embodiments of the present application provide a grid point management method based on dynamic grid division, which comprises: acquiring service area grids of a plurality of grid points and multi-source real-time data; based on the multi-source real-time data, calculating real-time service load of the service area grids to generate grid load weights; in response to any grid point grid load weight meeting a preset adjustment condition, updating the grid boundary of the service area grid; and based on the updated service area grid, performing customer demand prediction and / or resource scheduling strategy generation for the plurality of grid points. According to the embodiments of the present application, by calculating the real-time service load of the service area grid to generate the grid load weight, dynamic division and intelligent management of the service area of the plurality of grid points can be achieved, the multi-source real-time data is combined to evaluate the grid service load, and the grid boundary is flexibly updated based on the preset adjustment condition to effectively adapt to changes in service pressure caused by customer mobility and business fluctuations; at the same time, by updating the service area grid to perform customer demand prediction and resource scheduling strategy generation, the prediction accuracy and resource allocation efficiency are improved, thereby realizing the linkage optimization of the service area structure and operation strategy, and enhancing the overall response capability and service level of the system.

[0042] It should be noted that the grid point management method, device, equipment, medium and program product based on dynamic grid division determined by the present application can be used in the fields of artificial intelligence technology, blockchain technology and Internet of Things technology, and can also be used in various fields other than the fields of artificial intelligence technology, blockchain technology and Internet of Things technology. The application field of the grid point management method, device, equipment, medium and program product based on dynamic grid division provided by the embodiments of the present application is not limited.

[0043] In the technical solutions of the present application, the user information (including but not limited to user personal information, user image information, user equipment information such as location information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved are information and data authorized by the user or authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards, necessary security measures are taken, do not violate public order and good customs, and provide corresponding operation portal for user to choose authorization or refusal.

[0044] In the scenario of making automated decisions by using personal information, the method, device and system provided by the embodiments of the present application all provide corresponding operation entrances for the user to select to agree or reject the automated decision result; if the user selects to reject, the expert decision process is entered. The expression "automated decision" herein refers to the activity of making decisions by automatically analyzing and evaluating the behavior habits, interests and hobbies, or economic, health and credit conditions of a person by using a computer program. The expression "expert decision" herein refers to the activity of making decisions by personnel who are engaged in a certain field of work, have special experience, knowledge and skills and have reached a certain professional level.

[0045] Figure 1 An application scenario diagram of the grid point management method, device, equipment, medium and program product based on dynamic grid division according to the embodiments of the present application is schematically shown.

[0046] As shown in Figure 1 The application scenario 100 according to the embodiments can include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104 and a server 105. The network 104 is a medium for providing a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103 and the server 105. The network 104 can include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0047] The user can use the first terminal device 101, the second terminal device 102 and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102 and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0048] In the embodiments of the present application, the first terminal device 101 can be taken as an example of the first device, and the second terminal device 102 and / or the third terminal device 103 can be taken as an example of at least one second device. The first device and the second device can communicate with each other through a client internal mechanism to implement the data distribution and rendering logic in the grid point management method based on dynamic grid division.

[0049] In some embodiments, the first device and the at least one second device can be different display modules, windows or screens on the same computing terminal (such as a host computer), or can be multiple physical devices working cooperatively through a network, such as different client instances deployed on a desktop computer, a tablet terminal or a mobile device.

[0050] The first terminal device 101, the second terminal device 102, and the third terminal device 103 can be various electronic devices with displays and support web browsing, including but not limited to smart mobile terminals, tablet computers, laptop computers, and desktop computers.

[0051] Server 105 can be a server that provides various services, such as a backend management server that supports websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (this is just an example). The backend management server can analyze and process data such as received user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal devices.

[0052] It should be noted that the site management method based on dynamic grid partitioning provided in this application embodiment can generally be executed by server 105. Correspondingly, the site management device based on dynamic grid partitioning provided in this application embodiment can generally be located in server 105. The site management method based on dynamic grid partitioning provided in this application embodiment can also be executed by a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105. Correspondingly, the site management device based on dynamic grid partitioning provided in this application embodiment can also be located in a server or server cluster that is different from server 105 and capable of communicating with the first terminal device 101, the second terminal device 102, the third terminal device 103, and / or server 105.

[0053] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0054] The following will be based on Figure 1 The described scene, through Figures 2 to 4 The site management method based on dynamic grid partitioning according to the disclosed embodiments is described in detail.

[0055] Figure 2 A flowchart illustrating a site management method based on dynamic grid partitioning according to an embodiment of this application is shown.

[0056] like Figure 2 As shown, the point management method 200 based on dynamic grid division in this embodiment includes operations S210 to S240.

[0057] The S210 is used to acquire service area grids and multi-source real-time data from multiple network points.

[0058] In the embodiments of the present application, the plurality of network points can include bank business offices, automatic teller machine deployment sites, logistics distribution sites, medical service stations, government service windows, and the like service nodes with spatial distribution attributes. Each network point corresponds to an initially delineated service area, which can be preliminarily defined by a geographic information system or a historical service radiation range.

[0059] In the embodiments of the present application, in actual implementation, the service area grid can be implemented by dividing the service area of each network point into a plurality of spatial units. The service area grid can have a uniform resolution, or can be adaptively divided according to the regional business density to improve the accuracy of data representation and modeling. Each service area grid can be updated regularly in the time dimension to reflect the change trend of the service radiation range.

[0060] In some embodiments, multi-source real-time data can be constructed in combination with internal collection and external acquisition. For example, internal data can come from local systems or deployed Internet of Things sensors of each network point; external data can access third-party data interfaces or open platforms. Multi-source real-time data can be fused through spatio-temporal tags to input subsequent models in a unified structure.

[0061] In some embodiments, to improve data processing efficiency and scalability, data access and preprocessing modules can be deployed on a cloud computing platform to clean, standardize, and normalize the collected multi-source real-time data, and build a unified data view for downstream modules. In some scenarios, differential privacy protection mechanisms or federated learning frameworks can be used for the accessed multi-source data to ensure that sensitive information is not leaked while still being used for modeling analysis.

[0062] For example, in the government service network point management scenario of a certain city, the system can obtain more than 20 district-level government halls and their covered areas, divide the service area grid of each government hall, obtain the number of people from video monitoring, the handling time of the government system, the queuing information, and the traffic congestion index of the traffic department and the temperature data of the weather platform, to form a comprehensive multi-source real-time data set; for example, in the bank business network point scenario, different types of network points including self-service teller machines, counters, and business halls can be obtained, and customer transaction logs, customer flow monitoring, equipment operating status, and external economic data can be combined to build the corresponding service area grid.

[0063] In operation S220, based on the multi-source real-time data, the real-time service load of the service area grid is calculated to generate a grid load weight.

[0064] In the embodiments of the present application, after obtaining multi-source real-time data, the business pressure state of each service area grid can be calculated to obtain a quantifiable load index.

[0065] In some embodiments, to improve the generalization ability and local feature extraction ability of service load calculation, multi-source data can be encoded into a service load feature vector, and a graph structure representation can be constructed in combination with the spatial adjacency relationship between the service area grids. Subsequently, the system can perform message propagation and feature fusion based on a graph neural network to generate a load embedding vector for each grid, and further map it to a grid load weight for quantifying the current service pressure level.

[0066] In some embodiments, the load calculation of the service area grid can be achieved by constructing a feature vector, that is, different types of data in the multi-source real-time data are mapped into unified numerical features, and a comprehensive load indicator is obtained through weighted superposition or normalization processing. For example, in the scenario of a bank outlet, the customer visit volume, the counter service processing volume, and the self-service machine usage frequency can be used as the main feature dimensions; in the scenario of a government service outlet, the number of accepted work orders, the average processing time, and the queuing situation during peak hours can be used as the load features. The system can generate a grid load weight according to the changes in these multi-dimensional features, thereby reflecting the service pressure differences of each service area in different time periods.

[0067] In some embodiments, the generation of the grid load weight can also use a combination of time series analysis and machine learning methods. For example, the system can use a sliding time window to perform real-time statistics on sensor data, and output the load level of each grid at the current time through a regression model, a neural network model, or a time series prediction model trained on historical data.

[0068] In some embodiments, the load calculation of the service area grid can also introduce spatial correlation. For example, an increase in the load of a grid in a commercial area often has an overflow effect on adjacent grids. The traffic intensity and customer distribution trend of the adjacent area can be used as additional factors when calculating the grid load weight, a service area graph can be constructed, and a load embedding representation can be generated through a graph structure propagation mechanism.

[0069] In some embodiments, the system can also achieve minute-level or second-level load calculation by fusing high-frequency data sources such as video surveillance, infrared sensors, and intelligent terminal logs. For example, in the scenario of a logistics distribution point, the system can receive the express throughput, the sorting equipment running rate, and the frequency of delivery vehicle entry and exit from sensors in real time, and quickly generate dynamic load weights through data aggregation, thereby providing real-time support for grid division and transportation allocation.

[0070] In operation S230, in response to the grid load weight of any outlet meeting a preset adjustment condition, the grid boundary of the service area grid is updated.

[0071] In some embodiments, the preset adjustment condition can be implemented by threshold decision rules. For example, when the real-time service load weight of a certain service area grid continuously exceeds the set threshold of 80% for more than 15 minutes, the system determines that the service pressure of the grid is too high, thereby triggering the boundary expansion operation. Conversely, when the load weight of the service area grid is below the threshold of 20% for a long time, the shrinkage or merging operation is triggered to avoid resource waste.

[0072] In some embodiments, the preset adjustment condition can be set in combination with time series trend analysis. The system predicts the historical data of the grid load, and when the customer demand shows a significant upward trend in the future period of time, the boundary adjustment operation is triggered in advance. For example, during the lunch peak on weekdays or before holidays, the system can pre-expand the service boundary of some business district grid according to the prediction result to share the upcoming passenger flow pressure.

[0073] In some embodiments, the preset adjustment condition can also introduce a spatial adjacency comparison mechanism. That is, not only the load of the target grid itself is concerned, but also the load of the adjacent grid is considered for comprehensive judgment. For example, when the load weight of the target grid is significantly higher than that of its adjacent grid (such as the difference exceeds the set proportion threshold), part of the area is divided into the low-load grid by adjusting the boundary to achieve load balancing.

[0074] In the embodiments of the present application, the updating process can be implemented by both the method based on geometric division and the optimization strategy driven by data.

[0075] Illustratively, the grid boundary updating can be realized by dynamic adjustment of Voronoi diagram. The system generates Voronoi cells as service areas according to the geographical location of the net points in the initial stage, and then updates the weight center of each grid in the running process. When the weight of a certain grid is significantly improved, the corresponding Voronoi boundary will shift outward, thereby automatically expanding the service coverage.

[0076] Illustratively, the updating of the grid boundary can be realized based on clustering or partitioning algorithm. For example, the system can combine historical load and real-time data to redivide the customer group distribution by clustering method, and adjust the service grid boundary accordingly, so that the new division can better reflect the customer aggregation area and business hotspots.

[0077] Illustratively, the system can also utilize the optimization model to redraw the boundary. This method converts the grid boundary updating problem into a constrained optimization problem, taking the minimization of the load difference across the grid or the resource scheduling cost as the objective function, and introducing constraints such as personnel, equipment and service delay. Through iterative solution of the optimization model, the system can obtain a more reasonable grid division scheme, thereby ensuring customer experience while reducing operating costs.

[0078] In some embodiments, the grid boundary update not only relies on a single indicator, but also can be comprehensively judged in combination with multi-dimensional conditions. For example, the system can trigger a boundary update operation when it monitors that multiple external factors such as high-frequency customer flow, regional economic fluctuations, and weather mutations are superimposed, so as to avoid potential service bottlenecks in advance. In addition, the system can also use a predictive triggering strategy, that is, when the customer demand prediction result shows that there will be a significant load increase in a certain area in the future, the grid boundary is adjusted in advance to ensure that future service demand is effectively met.

[0079] In operation S240, based on the updated service area grid, customer demand prediction and / or resource scheduling strategy generation for the plurality of points are performed.

[0080] In some embodiments, customer demand prediction can be based on the updated grid unit historical data and real-time data to construct a spatio-temporal feature tensor, and input into a neural network model for future trend prediction. For example, in the application of bank outlets, the system can predict the customer visit volume, business peak period, and use demand of counters and self-service equipment in each grid unit in the future. The prediction result can be output as a quantitative indicator to assist managers to arrange staff scheduling or cash injection plan in advance, so as to improve customer experience and service efficiency.

[0081] In some embodiments, the generation of resource scheduling strategy can construct an optimization model based on the updated grid structure and resource constraint conditions. When the system identifies that some grid units have too high service pressure, it can automatically generate a resource adjustment scheme across the points through the scheduling model, such as increasing temporary clerks in terms of manpower, or temporarily enabling standby self-service equipment in terms of equipment, or increasing the frequency of automatic teller machine replenishment in terms of cash supply. The scheduling result can be directly issued to the operation system of each point to realize semi-automatic or automatic resource allocation execution.

[0082] In some embodiments, customer demand prediction and resource scheduling strategy can be jointly executed to form a closed-loop intelligent management mode. For example, the system first performs customer demand prediction based on the updated grid to obtain customer flow and business demand curves in different time windows in the future; then, in combination with these prediction results, a global-oriented resource scheduling optimization model is constructed to generate a scheduling strategy scheme across multiple grids. This joint process can complete resource configuration in advance before customer demand occurs, realizing forward-looking operation optimization.

[0083] In some embodiments, the system can also generate multiple versions of resource scheduling strategies according to the prediction results and introduce an adaptive selection mechanism. For example, the system can generate three types of scheduling schemes, conservative, balanced and aggressive, corresponding to different assumptions of customer demand growth; during actual execution, the system compares the deviation between the prediction and the actual situation in real time and automatically switches to the scheduling scheme that best fits the actual situation, thereby improving the flexibility and robustness of resource scheduling.

[0084] In some embodiments, comprehensive prediction and scheduling can also be combined with external environmental data. For example, when a business district in a city is predicted to have a large-scale increase in passenger flow during a major holiday, the system will not only adjust the resource allocation in that area, but also generate corresponding scheduling strategies for surrounding transportation hubs and residential areas to cope with potential customer diversion effects. This cross-grid collaborative prediction and scheduling can significantly improve the overall adaptability of the system in complex scenarios.

[0085] According to embodiments of the present application, by obtaining the service area grid of multiple sites and multi-source real-time data, the business operation status and environmental changes of each area can be comprehensively perceived; by calculating the real-time service load of the service area grid based on multi-source data and generating grid load weights, the business pressure level of different areas can be accurately described; by dynamically updating the boundaries of the service area grid when the grid load weight meets the preset adjustment condition, the service coverage can be matched with the actual demand, avoiding resource redundancy and service shortage; further, by performing customer demand prediction and / or resource scheduling strategy generation based on the updated service area grid, the business demand can be identified in advance and the resource allocation can be optimized intelligently, thereby effectively improving the adaptability, resource utilization and customer service level of the site system.

[0086] The following will specifically describe the site management method based on dynamic grid division of the embodiments of the present application in the preferred embodiment.

[0087] In traditional site operation management, the division of service areas usually relies on administrative division, historical experience or fixed geographical boundaries, such as determining the service range in units of streets, communities or postal areas. Although this division method is simple, it has obvious technical deficiencies: there is no direct correspondence between administrative division and actual customer flow trajectory, making it difficult to reflect the real service demand distribution; at the same time, fixed boundaries cannot dynamically adapt to population migration, changes in business districts, road traffic improvement and other external environmental changes, easily leading to situations of service overload in some areas and service vacancy in some areas, thereby reducing overall service efficiency. Especially in the scenario of multiple sites operating in parallel, without a scientific spatial division and boundary adjustment mechanism, resource allocation imbalance, customer experience degradation and risk control lag are likely to occur.

[0088] To solve the above problems, in the embodiments of the present application, the service area grid of multiple service points can be constructed in advance based on the following method. Specifically, multiple initial service units can be generated based on geographic location information of the multiple service points. For example, the geographic location information can include the latitude and longitude coordinates of the service points, administrative division location, road network topology, and surrounding geographic environment data, etc. By spatial mapping and segmentation of the geographic location information, the overall service area can be divided into multiple non-overlapping initial service units, which are used to represent the basic coverage range of each service point.

[0089] Further, the coverage weight of each service point can be calculated based on the service demand information and service supply capacity information of the multiple initial service units. For example, the service demand information can include customer density, number of business handling pens, time period traffic distribution, etc.; the service supply capacity information can include personnel configuration number, equipment carrying capacity, available business types, and historical service efficiency, etc. By fusing and calculating the demand side and supply side information, the coverage weight representing the service load capacity and actual service pressure of each service point can be obtained.

[0090] Subsequently, the boundaries of the multiple initial service units can be adjusted based on the service point coverage weight to form the final service area grid. For example, when the coverage weight of a certain service point is significantly higher than that of the surrounding units, its boundary can be expanded to share the service pressure; for another example, when the coverage weight is lower than the preset standard, it can be appropriately contracted or merged with adjacent units to achieve overall balance. The service area grid generated through the above process can be more consistent with the actual business distribution in terms of spatial structure, and effectively reflect the service supply and demand relationship in terms of function, thereby providing a more reasonable basis for subsequent load calculation, boundary dynamic adjustment, and generation of prediction and scheduling strategies.

[0091] According to the above embodiments, the service area grid formed thereby not only accurately reflects the spatial distribution characteristics, but also dynamically reflects the supply and demand relationship, providing a reasonable data structure and operation basis for subsequent load calculation, customer demand prediction, and resource scheduling, thereby improving the adaptability and global optimization capability of the system.

[0092] In the embodiments of the present application, obtaining the multi-source real-time data of the multiple service points can include comprehensive collection and fusion processing of internal data and external data. Specifically, the internal data can be collected in real time by deploying Internet of Things sensors at each service point, for example, the sensors can include passenger flow monitoring sensors, cameras, infrared detection devices, and smart terminals bound to business systems, etc., for obtaining data such as customer flow information entering the service point, customer stay time in different areas, usage time of various self-service equipment, and queuing and waiting state, etc., thereby reflecting the business operation and customer behavior characteristics inside the service point in real time.

[0093] Optionally, the system can also obtain external environment data related to the service area through an external data interface. For example, the intensity and direction of population flow in the area can be obtained through a data interface of a location service platform to represent the spatial distribution trend of customers at different times; macroeconomic fluctuation indicators, industrial operation conditions, and related financial activity data can be obtained through a government or industry big data platform to depict the influence of the economic environment on service demand; and weather change information such as temperature, rainfall, and wind can be obtained in real time through a weather interface provided by a meteorological department to reflect the potential influence of the external environment on customer travel and business handling demand.

[0094] To ensure effective fusion of internal data and external data, the multi-source data can be further time-aligned and spatially mapped. Specifically, the system can synchronize the data from different sensors and external platforms according to a unified timestamp to avoid deviations in the time sequence of the data; and the collected data can be mapped to corresponding service area grid cells according to geographic coordinates or service area identifiers, so that data from different sources can be integrated at the same spatial granularity.

[0095] In some embodiments, time alignment of multi-source data can be achieved using a time window mechanism. Specifically, the system can set a sliding window with a unit of seconds or minutes to uniformly sample and align data streams from different sensors and external platforms. For example, when a passenger flow sensor outputs flow data at a second level and a weather interface updates weather data once a minute, the system can map the two types of data to a unified minute-level time axis through interpolation or delay padding to avoid inconsistencies in time resolution during model calculation.

[0096] In some embodiments, time alignment can be achieved through an event-driven approach. For example, when a business handling peak event trigger is detected, the system will synchronously call an external population flow platform and an economic data interface to force the data from different sources to be unified to the time node of the event, thereby generating consistent time series data at critical business moments and enhancing the accuracy of event analysis and prediction.

[0097] In some embodiments, point-to-grid projection can be based on geographic coordinates. For example, geographic coordinate points output by a customer flow sensor can be quickly located to the corresponding service area grid cell through a spatial indexing method such as an R-tree or a quadtree, and accumulated as the flow feature value of the grid. This ensures that even if multiple service points have partially overlapping coverage, data from different sources can be accurately attributed to the appropriate grid cell.

[0098] In some embodiments, the spatial mapping can also be assisted based on administrative divisions or road networks. For example, for the data of the nodes on both sides of the urban trunk roads, the system can divide the flow data collected by the sensors into different service grids on the left and right sides of the road centerline to reflect the differences in the flow of people in different directions; and in the scenario of cross-regional nodes, the system can preferentially divide the external data based on the administrative division boundaries and attribute it to the corresponding regional grid to keep consistent with the local business statistical caliber.

[0099] Further, the time alignment and spatial mapping can be combined for dynamic correction. For example, when a sudden increase in customer flow in a certain grid is detected, and the update time of the external population flow data is delayed, the system can use historical data and prediction models to compensate for the missing external data in the short term and make weighted corrections after mapping to the grid.

[0100] According to the embodiments of the present application, multi-source real-time data that can comprehensively reflect the service status of each node and the characteristics of the external environment can be obtained, providing high-precision data support for subsequent load calculation, grid boundary update, demand prediction, and resource scheduling.

[0101] In the embodiments of the present application, for real-time service load calculation of the service area grid, graph structure modeling and deep learning technology can be introduced under the support of multi-source real-time data to generate grid load weights that can dynamically reflect the business pressure level.

[0102] Figure 3 A flowchart of a method for calculating real-time service load of the service area grid according to some example embodiments of the present application is schematically shown.

[0103] As shown in Figure 3 The method for calculating real-time service load of the service area grid includes operations S310-S330.

[0104] In operation S310, the multi-source real-time data is encoded into a load feature vector, and a service area graph is constructed based on the load feature vector and the spatial adjacency relationship of the service area grid, wherein the nodes of the service area graph represent the service area grid, and the edges of the service area graph represent the spatial relationship and / or business relationship between the service area grids.

[0105] Specifically, the multi-source real-time data can be pre-processed first, for example, including data cleaning, normalization, feature extraction and other operations, to remove noise and outliers, and unify heterogeneous data to the same scale. Subsequently, the processed data can be encoded into a load feature vector. For example, the load feature vector can contain multi-dimensional features such as customer traffic, queuing waiting time, device usage rate, population flow trend, weather condition and economic indicators, thereby providing comprehensive input information for subsequent modeling.

[0106] In combination with the spatial adjacency relationship of the load feature vector and the service area grid, a service area graph can be constructed. The nodes in the service area graph are used to represent different service area grid units, and the edges represent the spatial adjacency relationship or potential business association relationship between the grids. For example, when two adjacent grids share the same main road or there is customer interaction flow, their connection can be established through the edge; for another example, when there is traffic volume coupling or cross-area resource scheduling history between different grids, a business relationship edge can also be established. By constructing the service area graph, the spatial and business dependencies between service areas can be included in the modeling range, rather than just calculating the load of a single area in isolation.

[0107] In operation S320, information propagation is performed on the service area graph using a graph convolution network to generate an initial load embedding representation of each service area grid.

[0108] After the graph structure is constructed, the GCN can be used to propagate information on the service area graph. The GCN can aggregate the feature information of the node neighbors through convolution operation, thereby generating an initial load embedding representation of each service area grid.

[0109] Specifically, when updating the node representation, the GCN not only considers the load feature vector of the node itself, but also combines the feature information of the adjacent grids, and after weighted summation and nonlinear transformation, a node representation with more contextual semantics is formed. Specifically, the embedding vector of a certain grid can reflect both its own load state and the business environment of the neighborhood, thereby improving the accuracy of the load representation as a whole.

[0110] In operation S330, the initial load embedding representation is input into a graph attention network to generate the grid load weight.

[0111] Specifically, it can include: obtaining the adjacent grid corresponding to the target service area grid, calculating the attention weight between the target service area grid and the adjacent grid based on the initial load embedding representation and the attention mechanism in the graph attention network, wherein the target service area grid represents any service area grid; based on the attention weight, the initial load embedding representation of the adjacent grid is weighted aggregated to obtain the context representation vector of the target service area grid; and the context representation vector is input into the nonlinear transformation layer in the graph attention network to generate the grid load weight corresponding to the target service area grid.

[0112] In this embodiment, the adjacent grid corresponding to the target service area grid can be obtained first, and the attention weight between the target service area grid and each adjacent grid is calculated based on the initial load embedding representation and the attention mechanism in the graph attention network. The target service area grid can represent any independent service area unit, and the attention weight thereof reflects the relative importance of different adjacent grids in load propagation and feature expression.

[0113] After the attention weight is calculated, the initial load embedding representation of the adjacent grid can be weighted aggregated based on the weight to generate the context representation vector of the target service area grid. Through the aggregation operation, the load features of the adjacent grid can be introduced into the representation of the target grid according to the contribution difference, thereby enhancing the modeling ability of the non-uniform influence relationship between regions. For example, when the customer flow of a certain commercial district grid surges and significantly affects the business of the surrounding residential district grid, the attention mechanism can automatically increase the weight of the commercial district grid, so that its features are strengthened in the aggregation process.

[0114] Subsequently, the obtained context representation vector is input into the nonlinear transformation layer in the graph attention network, and after the multi-head attention and nonlinear activation function processing, the grid load weight corresponding to the target service area grid is generated. The grid load weight can capture its own business features while integrating the dynamic influence of the surrounding area, thereby providing more accurate quantitative basis for subsequent grid boundary adjustment, customer demand prediction and resource scheduling optimization.

[0115] In the embodiments of the present application, GAT can assign different weights to different adjacent nodes by introducing an attention mechanism, thereby distinguishing the contribution degree of neighbors to the target node. For example, among the adjacent multiple service grids, the load change of some neighbors is highly related to the target grid, while the influence of some other neighbors is weak. The attention mechanism can automatically identify and strengthen the features of important neighbors, thereby realizing differentiated processing of information in aggregation.

[0116] In some embodiments, in order to improve the robustness and generalization ability of the model, the system can also introduce multi-layer graph convolution and multi-head attention mechanism. Multi-layer graph convolution can expand the receptive field of the node, so that the load embedding representation can cover the information of the farther neighbors, thereby capturing the cross-regional business conduction effect; the multi-head attention mechanism can model the neighbor contribution from multiple subspaces, improving the stability and expressiveness of weight calculation. In addition, the system can also introduce regularization terms and alignment constraints of historical load curves during the training stage to avoid model overfitting and ensure that the results are consistent with long-term business trends.

[0117] The grid load weight generated by the above process not only reflects the instantaneous business pressure of a single grid, but also embodies the relevance and dynamic influence between regions, providing high-precision, data-driven quantitative basis for subsequent grid boundary update, customer demand prediction and resource scheduling optimization. Compared with traditional load evaluation methods based on empirical threshold or simple statistics, the present embodiment introduces graph neural networks and attention mechanisms, making the load calculation results more accurate, flexible and interpretable, thereby improving the intelligent level of the system in a complex multi-source data environment.

[0118] In the embodiments of the present application, for the updated service area grid, customer demand prediction for multiple service points can be further performed to realize forward-looking analysis and resource preparation for future business trends. Specifically, the historical service load data corresponding to the updated service area grid can be first obtained, which can include customer visit volume, business handling number, device usage rate, average queuing time, etc. time series information, and is synchronized with external factors such as population flow trend, weather change and economic activity intensity, etc.

[0119] Further, a spatio-temporal feature tensor can be constructed based on the updated service area grid and the historical service load data. The spatio-temporal feature tensor can not only reflect the time series regularity within a single grid, but also embody the spatial adjacency relationship and business conduction effect between different grids. For example, when a commercial district grid experiences a surge in customer demand during the lunch peak period, its feature change may also be transmitted to the surrounding transportation hubs or residential area grids, forming a cross-regional service demand linkage. By integrating spatial and temporal features into a unified tensor structure, more global inputs can be provided for subsequent prediction.

[0120] Subsequently, the spatio-temporal feature tensor can be input into a customer demand prediction network, which can adopt recurrent neural networks, long short-term memory networks, spatio-temporal convolution networks or deep learning models with attention mechanisms to model spatio-temporal dependencies and output future customer service demand information. The prediction results can include customer flow curve, business demand distribution, peak period and demand proportion of different business types in a specific future time window, etc.

[0121] According to the embodiments of the present application, the system can accurately predict future customer service demand on the basis of the updated service area grid, thereby providing proactive resource preparation and scheduling support for branch operation. Compared with the traditional method based on static historical statistics, the present embodiment introduces spatiotemporal modeling and deep learning prediction mechanism, thereby not only improving the accuracy of prediction, but also enhancing the sensitivity of the model to external environmental changes and inter-regional linkage, thereby improving the self-adaptive ability and operation intelligent level of the branch system.

[0122] In the embodiments of the present application, for the updated service area grid, resource scheduling strategy generation for multiple branches can be further performed to realize intelligent resource allocation at the branch level. Similarly, the real-time service load corresponding to the updated service area grid and the resource allocation constraint information related to the grid can be obtained first. Exemplarily, the real-time service load can include dynamic indicators such as customer flow intensity, business processing request volume, equipment usage saturation, etc.; the resource allocation constraint information can include personnel shift rules, equipment availability restrictions, upper limit of cash replenishment frequency, service time window, and cross-regional resource allocation restrictions, etc. By considering the load state and constraint condition at the same time, it can be ensured that the generated resource scheduling strategy meets the actual business demand while having executability and compliance.

[0123] Subsequently, a resource demand parameter set can be constructed based on the real-time service load. The resource demand parameter set can cover different types of resource demand quantities, such as the number of tellers, shift length, cash amount required for automatic teller machines, self-service terminal maintenance frequency, standby equipment activation timing, etc.

[0124] Further, the resource demand parameter set and the resource allocation constraint can be input into a resource allocation optimization model. For example, the resource allocation optimization model can adopt linear programming, integer programming, heuristic search algorithm or intelligent scheduling algorithm based on reinforcement learning, with the optimization objectives of minimizing overall waiting time, balancing regional service pressure or reducing operation cost. In the solving process of the optimization model, the system can automatically calculate the optimal or approximately optimal solution that meets the constraint condition, and map the solution result to a specific resource scheduling strategy, such as allocating additional tellers to a certain grid, temporarily activating standby equipment, increasing the cash replenishment frequency of automatic teller machines, or calling mobile service personnel across regions.

[0125] According to the embodiments of the present application, the generated resource scheduling strategy can be directly issued to the point operation system or the scheduling platform, supporting automatic execution or manual assisted execution, thereby improving the timeliness and accuracy of resource allocation. Compared with the traditional method of relying on manual experience to formulate a scheduling scheme, the present embodiment can quickly generate a reasonable scheduling scheme in a complex and variable business environment through the automatic calculation of real-time load, constraint conditions and optimization models, effectively improving the overall service efficiency and resource utilization of the point.

[0126] In the embodiments of the present application, the trusted storage and cross-regional linkage of multi-grid data based on the alliance blockchain network can also be included. Specifically, after completing the updated service area grid calculation, the system can store the real-time service load, customer demand prediction results and resource scheduling results corresponding to the grid into the alliance blockchain network. Due to the characteristics of multi-node common maintenance and data non-tamperability of the alliance blockchain, the above storage operation can ensure the integrity and traceability of the key business data, thereby avoiding service risks caused by single point failure or data tampering. At the same time, through the multi-party cooperation mechanism of the alliance chain, the data between different grids can be trusted shared within a controlled range to support cross-regional collaborative decision-making.

[0127] Further, when any updated service area grid detects an abnormal behavior event, such as a sudden increase in customer traffic, abnormal transaction behavior or equipment failure, the system can locally generate a data announcement containing an event feature summary and broadcast the data announcement through the alliance blockchain network. For example, the event feature summary can include information such as the time, geographic location, type of business involved, and key load indicators of the abnormal occurrence. Broadcasting through the blockchain network can ensure that all related grid nodes obtain the same abnormal information within a short time and trigger the corresponding linkage response mechanism. For example, the surrounding grids can immediately increase the risk control threshold or adjust the resource configuration after receiving the broadcast to prevent the spread of abnormal events or cause regional service paralysis.

[0128] In some embodiments, the storage method in the alliance blockchain network can be hierarchically processed according to the data type. For example, for high-frequency small-scale structured data (such as real-time service load indicators, prediction curve parameters), the system can be directly stored in the alliance chain main chain to ensure its non-tamperability and fast verification; for medium-frequency structured or semi-structured data (such as scheduling schemes, regional operation logs), side chain storage can be used, and cross-chain anchoring can be used to ensure data consistency; and for low-frequency but large-volume unstructured data (such as video monitoring records, voice logs, etc.), chain storage can be used and the data can be uploaded through a hash fingerprint, thereby ensuring data credibility while reducing the storage pressure of the blockchain system.

[0129] At the off-chain storage layer, distributed storage technology can be further combined to encrypt and fragment large-scale unstructured data and store them in a distributed manner, and the traceability and integrity of the data can be ensured through blockchain indexing. Each distributed storage node can also serve as a local cache of the service area grid to quickly access high-frequency data, while the data high availability and business continuity during cross-area grid collaboration can be ensured through multi-site redundancy and load balancing mechanisms. In addition, the system can also introduce a data hierarchical management strategy to divide data into high-frequency, medium-frequency and low-frequency categories according to access frequency, and store them in cache, distributed file storage and cold storage respectively, thereby ensuring access efficiency while effectively reducing overall storage costs.

[0130] In some embodiments, the consortium blockchain network can also be combined with an access control mechanism to realize multi-level data sharing. For example, different levels of management nodes can obtain different data access permissions based on smart contracts, and ordinary operating sites can only view the load information of themselves and adjacent grids, while regional management nodes can obtain cross-grid prediction and scheduling data, thereby realizing differentiated multi-party collaboration while ensuring data privacy and security. This mechanism not only avoids the risks brought by excessive data disclosure, but also supports rapid joint response across regions when necessary.

[0131] In some embodiments, the broadcast of abnormal events is not limited to information notification, but can also trigger an automated response mechanism. For example, when a certain grid detects abnormal customer traffic and generates a data notification, the smart contract can automatically call the resource scheduling module to allocate more personnel or temporarily enable standby equipment for the abnormal grid; at the same time, adjacent grids receiving the notification will also automatically upgrade the risk control strategy threshold by one level to prevent potential chain risks. Through this automatic triggering mechanism, the delay of manual intervention can be significantly reduced, and the emergency response efficiency can be improved.

[0132] According to embodiments of the present application, by storing and broadcasting grid operation data and abnormal event information to the consortium blockchain network, the security and transparency of data management are improved, and the collaboration of cross-grid risk prevention and emergency response is enhanced, thereby facilitating the stable operation and continuous optimization of the service area grid.

[0133] In some embodiments, dynamic display and interactive operation of the service area grid can also be realized based on a three-dimensional visualization interface. Specifically, the system can pre-build a visualization interface based on geographic information systems and three-dimensional modeling technology, and load the spatial boundaries, business indicators and related time series data of the service area grid in the interface. The three-dimensional visualization interface can intuitively present the distribution of each service area grid in the form of a map or a three-dimensional scene, enabling users to comprehensively grasp the operation status of the service area grid in a unified spatial view.

[0134] When the user selects the target service area grid, the system can dynamically display multi-dimensional information related to the grid based on the three-dimensional visualization interface. For example, the structural layout of the grid can be displayed, including the grid coverage, the location of the network point to which the grid belongs, and the relationship with adjacent areas; the historical load trend curve can be displayed to reflect the business pressure evolution of the grid at different time periods; future business demand information generated based on a prediction model can also be displayed to enable the user to identify potential peak periods and demand hotspots in advance; and resource scheduling information, such as the current personnel configuration, equipment operating status, and future scheduling plan, can also be displayed.

[0135] According to embodiments of the present application, the user can not only intuitively understand the current situation and history of the target service area grid, but also obtain forward-looking information such as prediction and scheduling, thereby completing the closed-loop management from monitoring to decision-making in the spatial interface. Compared with the traditional tabular or two-dimensional chart display method, the embodiments of the present application improve the intuitiveness and interactivity of information expression through three-dimensional visualization technology, enabling complex business operation data to be presented in a spatialized and stereoscopic form, and significantly enhancing the decision-making efficiency and response capability of management personnel.

[0136] Figure 4 A multi-layer system architecture for network point management based on dynamic grid division according to some example embodiments of the present application is schematically shown.

[0137] As shown in FIG. 1, the multi-layer system architecture can be divided into a data acquisition layer, a computing decision layer, a storage collaboration layer, and a visualization application layer in sequence, and the layers are connected to each other through data transmission and processing mechanisms to jointly complete the dynamic management and intelligent optimization of the bank network service area. Figure 4

[0138] At the data acquisition layer, the system collects data such as customer flow, equipment usage time, and queuing status in real time through internal Internet of Things sensors, and accesses information such as population flow, economic indicators, and weather data through external interfaces. All collected data are aggregated and preprocessed by a log aggregation module and uniformly transmitted to a message queue to achieve stable transmission with high throughput and low latency.

[0139] At the computing decision layer, the system first performs real-time processing on the collected data based on a stream computing framework, and inputs the results to subsequent analysis and modeling links. On the one hand, a dynamic division method based on Voronoi diagram can be used to adjust the boundary of the service area grid in real time, so that it can be flexibly updated according to the changes in business pressure and customer distribution; on the other hand, a spatiotemporal feature tensor can be constructed based on the model to predict the future business demand and load trend of the customers.

[0140] ​At the storage collaboration layer, the system introduces technologies such as blockchain, consortium chain and distributed file storage, and stores and shares real-time service load, customer prediction results and scheduling schemes. On the one hand, the tamper-proof nature of the blockchain ensures the credibility and traceability of the data; on the other hand, the consensus mechanism between the nodes of the consortium chain realizes safe collaboration across regions, and at the same time, the distributed file system effectively supports the storage and access of large-scale logs, audio and video and other unstructured data.

[0141] At the visualization application layer, the system constructs a three-dimensional visualization interface to intuitively display the service area grid after dynamic adjustment. Users can not only view the structural layout of the service grid in the interface, but also obtain real-time information such as historical load trends, future business predictions and resource scheduling schemes through interactive methods, realizing a complete closed loop from monitoring, analysis to decision-making.

[0142] Corresponding to the above-mentioned grid point management method based on dynamic grid division, the embodiment of the present application also provides a grid point management device based on dynamic grid division.

[0143] Figure 5 The structure block diagram of the grid point management device based on dynamic grid division according to the embodiment of the present application is schematically shown.

[0144] As shown in Figure 5 , the grid point management device based on dynamic grid division 500 of this embodiment includes a data acquisition module 510, a grid load weight generation module 520, a grid boundary update module 530 and a grid point management module 540.

[0145] The data acquisition module 510 can be used to acquire the service area grid of a plurality of grid points and multi-source real-time data. In an embodiment, the data acquisition module 510 can be used to perform the operation S210 described above, which will not be repeated here.

[0146] The grid load weight generation module 520 can be used to calculate the real-time service load of the service area grid based on the multi-source real-time data, and generate a grid load weight. In an embodiment, the grid load weight generation module 520 can be used to perform the operation S220 described above, which will not be repeated here.

[0147] The grid boundary update module 530 can be used to update the grid boundary of the service area grid in response to the grid load weight of any grid point meeting a preset adjustment condition. In an embodiment, the grid boundary update module 530 can be used to perform the operation S230 described above, which will not be repeated here.

[0148] The grid load weight generation module 520 can be configured to encode the multi-source real-time data into a load feature vector, construct a service area graph based on the load feature vector and a spatial adjacency relationship of the service area grid, wherein a node of the service area graph represents the service area grid, and an edge of the service area graph represents a spatial relationship and / or a business relationship between the service area grids; perform information propagation on the service area graph by using a graph convolution network to generate an initial load embedding representation of each of the service area grids; and input the initial load embedding representation into a graph attention network to generate the grid load weight.

[0149] According to embodiments of the present application, the data acquisition module 510 can also be configured to acquire internal data by using Internet of Things sensors, the internal data including at least one of customer flow, equipment usage duration, and queuing status; acquire external data by using an external data interface, the external data including at least one of regional population flow, economic fluctuation indicators, and weather change information; and perform time alignment and spatial mapping on the internal data and the external data to obtain the multi-source real-time data.

[0150] According to embodiments of the present application, the grid load weight generation module 520 can also be configured to encode the multi-source real-time data into a load feature vector, construct a service area graph based on the load feature vector and a spatial adjacency relationship of the service area grid, wherein a node of the service area graph represents the service area grid, and an edge of the service area graph represents a spatial relationship and / or a business relationship between the service area grids; perform information propagation on the service area graph by using a graph convolution network to generate an initial load embedding representation of each of the service area grids; and input the initial load embedding representation into a graph attention network to generate the grid load weight.

[0151] According to embodiments of the present application, the grid load weight generation module 520 can also be configured to acquire an adjacent grid corresponding to a target service area grid, calculate an attention weight between the target service area grid and the adjacent grid based on the initial load embedding representation and an attention mechanism in the graph attention network, wherein the target service area grid represents any service area grid; perform weighted aggregation on the initial load embedding representation of the adjacent grid based on the attention weight to obtain a context representation vector of the target service area grid; and input the context representation vector into a nonlinear transformation layer in the graph attention network to generate the grid load weight corresponding to the target service area grid.

[0152] According to embodiments of the present application, the grid load weight generation module 520 can also be configured to acquire an adjacent grid corresponding to a target service area grid, calculate an attention weight between the target service area grid and the adjacent grid based on the initial load embedding representation and an attention mechanism in the graph attention network, wherein the target service area grid represents any service area grid; perform weighted aggregation on the initial load embedding representation of the adjacent grid based on the attention weight to obtain a context representation vector of the target service area grid; and input the context representation vector into a nonlinear transformation layer in the graph attention network to generate the grid load weight corresponding to the target service area grid.

[0153] According to an embodiment of the present application, the net point management module 540 can also be configured to obtain real-time service load corresponding to the updated service area grid and corresponding resource allocation constraint information; construct a resource demand parameter set based on the real-time service load; and input the resource demand parameter set and the resource allocation constraint into a resource allocation optimization model, and generate a resource scheduling strategy based on an output result of the resource allocation optimization model.

[0154] According to an embodiment of the present application, the anomaly identification device 500 can also include a grid construction module. The grid construction module can be configured to generate a plurality of initial service units based on geographic location information of the plurality of net points; calculate net point coverage weights based on service demand information and service supply capability information of the plurality of initial service units; and adjust boundaries of the plurality of initial service units based on the net point coverage weights to form the service area grid.

[0155] According to an embodiment of the present application, the anomaly identification device 500 can also include a blockchain module. The blockchain module can be configured to store real-time service load, customer demand prediction results, and resource scheduling results in the updated service area grid into a consortium blockchain network; and in response to any updated service area grid detecting an abnormal behavior event, generate a data announcement containing an event feature summary, and broadcast the data announcement in the consortium blockchain network based on the data announcement.

[0156] According to an embodiment of the present application, the anomaly identification device 500 can also include a visualization module. The visualization module can be configured to obtain a pre-constructed three-dimensional visualization interface; and in response to a user selecting a target service area grid, dynamically display at least one of a structural layout, a historical load trend, business demand prediction information, and resource scheduling information of the target service area grid based on the three-dimensional visualization interface.

[0157] According to an embodiment of the present application, any multiple of the data acquisition module 510, the grid load weight generation module 520, the grid boundary update module 530 and the grid point management module 540 can be combined in one module, or any one of them can be split into multiple modules. Alternatively, at least part of the function of one or more of these modules can be combined with at least part of the function of other modules and implemented in one module. According to an embodiment of the present application, at least one of the data acquisition module 510, the grid load weight generation module 520, the grid boundary update module 530 and the grid point management module 540 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on chip, a system on board, a system on package, an application specific integrated circuit (ASIC), or any other reasonable way of hardware or firmware that can be integrated or packaged with a circuit, or any one of the three implementation ways of software, hardware and firmware or any appropriate combination of any of them. Alternatively, at least one of the data acquisition module 510, the grid load weight generation module 520, the grid boundary update module 530 and the grid point management module 540 can be at least partially implemented as a computer program module that can perform corresponding functions when it is run.

[0158] Figure 6 A block diagram of an electronic device suitable for implementing the grid point management method based on dynamic grid partitioning according to an embodiment of the present application is schematically shown.

[0159] As shown in Figure 6 The electronic device 600 according to an embodiment of the present application includes a processor 601 that can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 602 or loaded from a storage portion 608 into a random access memory (RAM) 603. The processor 601 can include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset, and / or a special-purpose microprocessor (such as an application specific integrated circuit (ASIC)), and the like. The processor 601 can also include an on-board memory for cache use. The processor 601 can include a single processing unit or multiple processing units for performing different actions of the method processes according to embodiments of the present application.

[0160] In the RAM 603, various programs and data required for the operation of the electronic device 600 are stored. The processor 601, the ROM 602, and the RAM 603 are connected to each other via the bus 604. The processor 601 performs various operations of the method flow according to the embodiments of the present application by executing the programs in the ROM 602 and / or the RAM 603. It should be noted that the programs can also be stored in one or more memories other than the ROM 602 and the RAM 603. The processor 601 can also perform various operations of the method flow according to the embodiments of the present application by executing the programs stored in the one or more memories.

[0161] According to the embodiments of the present application, the electronic device 600 can further include an input / output (I / O) interface 605, which is also connected to the bus 604. The electronic device 600 can further include one or more of the following components connected to the input / output (I / O) interface 605: an input part 606 including a keyboard, a mouse, and the like; an output part 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), and the like, and a speaker, and the like; a storage part 608 including a hard disk, and the like; and a communication part 609 including a network interface card such as a LAN card, a modem, and the like. The communication part 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as necessary. A removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, and the like is mounted on the drive 610 as necessary, so that a computer program read therefrom is installed in the storage part 608 as necessary.

[0162] The present application also provides a computer readable storage medium, which can be included in the device / apparatus / system described in the above embodiments; or can exist separately without being assembled into the device / apparatus / system. The above computer readable storage medium carries one or more programs, when the one or more programs are executed, the method according to the embodiments of the present application is implemented.

[0163] According to an embodiment of the present application, the computer readable storage medium can be a non-transitory computer readable storage medium, for example, can include but is not limited to: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In this application, a computer readable storage medium can be any tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present application, the computer readable storage medium can include one or more of the above-described ROM 602 and / or RAM 603 and / or memories other than the ROM 602 and the RAM 603.

[0164] Embodiments of the present application also include a computer program product, which includes a computer program containing program codes for executing the methods shown in the flowcharts. When the computer program product is run in a computer system, the program codes are used to make the computer system implement the dynamic grid partitioning based grid point management method provided by the embodiments of the present application.

[0165] The above-described functions defined in the system / device / apparatus of the embodiments of the present application are performed when the computer program is executed by the processor 601. According to an embodiment of the present application, the above-described system, device, module, unit, etc. can be implemented by computer program modules.

[0166] In one embodiment, the computer program can rely on tangible storage media such as optical storage media, magnetic storage media, etc. In another embodiment, the computer program can also be transmitted, distributed, and downloaded in the form of signals on network media. The computer program is downloaded and installed through the communication part 609 and / or installed from the detachable medium 611. The program codes contained in the computer program can be transmitted by any appropriate network media, including but not limited to: wireless, wired, etc., or any suitable combination of the foregoing.

[0167] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609 and / or installed from the detachable medium 611. When the computer program is executed by the processor 601, the above-described functions defined in the system of the embodiments of the present application are performed. According to an embodiment of the present application, the above-described system, device, apparatus, module, unit, etc. can be implemented by computer program modules.

[0168] According to embodiments of the present application, program code for implementing the computer programs provided by embodiments of the present application can be written in any combination of one or more programming languages, and can be implemented using high-level procedural and / or object-oriented programming languages, and / or assembly / machine languages. Program code can execute entirely on a user's computing device, partly on the user's device, as a stand-alone software package, partly on a remote computing device, or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computing device, such as through the Internet using an Internet Service Provider.

[0169] The computer program instructions can also be loaded onto a computer or other programmable information processing apparatus to cause a series of operations to be performed on the computer or other programmable information processing apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable information processing apparatus implement the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0170] Those skilled in the art will understand that features recited in the various embodiments of the present application can be combined and / or integrated in various ways, even if such combinations or integrations are not expressly noted in the present application. In particular, features recited in the various embodiments of the present application can be combined and / or integrated in ways that are not expressly noted in the present application, without departing from the spirit and teachings of the present application. All such combinations and / or integrations are within the scope of the present application.

[0171] The embodiments of the present application have been described above. However, these embodiments are merely for the purpose of illustration, and are not intended to limit the scope of the present application. Although the embodiments are described separately above, this does not mean that the measures in the various embodiments cannot be used advantageously in combination. Those skilled in the art can make various substitutions and modifications without departing from the scope of the present application, and these substitutions and modifications should all fall within the scope of the present application.

Claims

1. A site management method based on dynamic grid partitioning, characterized in that, The method includes: Acquire service area grids and multi-source real-time data from multiple network points; Based on the multi-source real-time data, the real-time service load of the service area grid is calculated to generate grid load weights; In response to any grid load weight at any point satisfying a preset adjustment condition, the grid boundary of the service area grid is updated; and Based on the updated service area grid, customer demand forecasting and / or resource scheduling strategy generation are performed for the multiple outlets.

2. The method according to claim 1, characterized in that, The method also includes constructing a service area grid of multiple locations, including: Multiple initial service units are generated based on the geographic location information of the multiple network points; Based on the service demand information and service supply capacity information of the multiple initial service units, the network coverage weight is calculated; and The boundaries of the multiple initial service units are adjusted based on the network coverage weights to form the service area grid.

3. The method according to claim 1, characterized in that, Acquire multi-source real-time data from multiple locations, including: The internal data is acquired using IoT sensors, and the internal data includes at least one of customer traffic, device usage time, and queue status. External data is acquired using an external data interface, wherein the external data includes at least one of regional population flow, economic fluctuation indicators, and weather change information; and The internal and external data are time-aligned and spatially mapped to obtain the multi-source real-time data.

4. The method according to claim 1 or 3, characterized in that, The step of calculating the real-time service load of the service area grid based on the multi-source real-time data and generating grid load weights includes: The multi-source real-time data is encoded into a load feature vector, and a service area graph is constructed based on the load feature vector and the spatial adjacency relationship of the service area grid. The nodes of the service area graph represent the service area grid, and the edges of the service area graph represent the spatial relationship and / or business relationship between the service area grids. The service region graph is propagated using a graph convolutional network to generate an initial load embedding representation for each service region grid; and The initial load is embedded into the input graph attention network to generate the grid load weights.

5. The method according to claim 4, characterized in that, The step of embedding the initial load into the input graph attention network to generate the grid load weights includes: Obtain the adjacent grids corresponding to the target service region grid, and calculate the attention weights between the target service region grid and the adjacent grids based on the initial load embedding representation and the attention mechanism in the graph attention network, wherein the target service region grid represents any service region grid; Based on the attention weights, the initial load embedding representations of the adjacent grids are weighted and aggregated to obtain the context representation vector of the target service region grid; and The context representation vector is input into the nonlinear transformation layer in the graph attention network to generate the grid load weights corresponding to the target service region grid.

6. The method according to claim 1, characterized in that, The process of performing customer demand forecasting for multiple locations based on the updated service area grid includes: Obtain the historical service load data corresponding to the updated service area grid, and construct a spatiotemporal feature tensor based on the updated service area grid and the historical service load data; as well as The spatiotemporal feature tensor is input into the customer demand prediction network, which outputs information on customers' future service needs.

7. The method according to claim 1, characterized in that, Based on the updated service area grid, resource scheduling policies for multiple sites are generated, including: Obtain the updated real-time service load and corresponding resource allocation constraints for the service area grid; Construct a resource requirement parameter set based on the real-time service load; as well as The resource demand parameter set and the resource allocation constraints are input into the resource allocation optimization model, and a resource scheduling strategy is generated based on the output of the resource allocation optimization model.

8. The method according to claim 1, characterized in that, The method further includes: The updated service area grid's real-time service load, customer demand forecast results, and resource scheduling results are stored in the consortium blockchain network. as well as In response to the detection of an anomalous behavior event by any updated service area grid, a data announcement containing a summary of the event characteristics is generated and broadcast on the consortium blockchain network based on the data announcement.

9. The method according to claim 1, characterized in that, The method further includes: Obtain a pre-built 3D visualization interface; and In response to a user selecting a target service area grid, the system dynamically displays at least one of the following based on the 3D visualization interface: the structural layout of the target service area grid, historical load trends, business demand forecast information, and resource scheduling information.

10. A network point management device based on dynamic grid partitioning, characterized in that, The device includes: The data acquisition module is used to acquire service area grids and multi-source real-time data from multiple network points. The grid load weight generation module is used to: calculate the real-time service load of the service area grid based on the multi-source real-time data, and generate grid load weights; The grid boundary update module is used to: update the grid boundary of the service area grid in response to any grid point's grid load weight meeting a preset adjustment condition; and The branch management module is used to: generate customer demand forecasting and / or resource scheduling strategies for the multiple branches based on the updated service area grid.

11. An electronic device, comprising: One or more processors; Memory, used to store one or more computer programs. The characteristic feature is that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 9.

13. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed by a processor, they implement the steps of the method according to any one of claims 1 to 9.

Images