A city event grading early warning disposal resource scheduling method fusing edge computing

Abstract

The application discloses a kind of urban event grading early warning disposal resource scheduling methods of fusion edge computing, belong to city management, intelligent emergency scheduling field.Rely on internet of things terminal and city sensor, realize the real-time collection of multidimensional event data such as traffic, environment, security, etc., weighted adaptive fusion and anomaly detection are carried out by edge node, realize data front-end intelligent preprocessing.Adopt fusion feature hierarchical inference algorithm, combined with rule discrimination and deep learning model, complete event automatic classification.According to event level, edge-cloud cooperation realizes grading early warning and resource scheduling, ordinary event is responded in situ by edge node, and major event and above are uploaded to cloud simultaneously, drive global decision and resource optimization scheduling.Through whole-process performance feedback, dynamically optimize grading, early warning and scheduling strategy, improve the intelligentization and efficiency of city emergency response.

Description

Technical Field

[0001] This invention belongs to the field of urban management and intelligent emergency dispatch, and more specifically relates to a resource dispatching method for hierarchical early warning and response of urban events that integrates edge computing. Background Technology

[0002] With the acceleration of urbanization, cities are expanding in scale, and population density and traffic flow are continuously increasing, posing increasingly complex challenges to urban public safety and emergency management. Traditional urban event monitoring and emergency dispatch methods mainly rely on centralized data collection and processing, which suffers from problems such as poor information timeliness, slow response speed, and untimely resource allocation, making it difficult to meet the needs of modern smart cities for refined management and efficient collaborative response to urban events.

[0003] In recent years, the widespread deployment of the Internet of Things (IoT) and smart sensors has significantly improved the real-time acquisition capabilities of multi-source data in cities. However, data often needs to traverse multiple network links before being uploaded to the central cloud, resulting in high bandwidth pressure and transmission latency. Furthermore, with the diversification of urban event types, including traffic congestion, environmental pollution, and public security emergencies, event identification and classification require the integration of multi-dimensional information and intelligent data analysis. Relying solely on traditional rule-based or single deep learning-based classification methods makes it difficult to balance accuracy and real-time performance.

[0004] Meanwhile, a single cloud-based response mechanism may lead to imbalances in emergency resource allocation or localized response delays when facing large-scale emergencies. The rise of edge computing technology offers a new solution for intelligent handling of urban events. By performing data fusion and preliminary intelligent analysis at edge nodes close to the data source, the real-time performance and resilience of emergency dispatch can be significantly improved. However, how to effectively integrate edge computing and cloud-based intelligent capabilities to achieve multi-level event classification, joint early warning, and collaborative resource dispatch, thereby improving the efficiency and intelligence of urban event handling, remains a key technical challenge that urgently needs to be addressed in the field of smart city management.

[0005] Therefore, developing a hierarchical early warning and resource allocation method for urban events that combines the advantages of multi-source heterogeneous data perception, edge intelligent reasoning, and cloud-based global scheduling has become an important technological direction for the modernization and intelligent upgrading of urban management. Summary of the Invention

[0006] This invention aims to address the technical problems existing in current urban event management, such as untimely fusion of multi-source data, insufficient accuracy in event classification and discrimination, slow emergency resource scheduling and response, and imperfect edge-cloud collaboration mechanisms. By constructing a hierarchical early warning and resource scheduling system that integrates edge computing and cloud-based intelligent decision-making, it enables efficient and intelligent identification, hierarchical response, and collaborative handling of multi-dimensional urban events, thereby improving the real-time performance, accuracy, and resource utilization efficiency of event handling.

[0007] To achieve the above objectives, the present invention employs the following technical solution: the method comprises: The system enables real-time collection of multi-dimensional urban event data and edge fusion preprocessing, relying on IoT terminals and urban sensors to collect multi-dimensional urban event-related data in traffic, environment, and security. The intelligent classification and identification of urban events introduces a fusion feature hierarchical reasoning algorithm, which combines rule reasoning and deep learning to map and analyze the key features of events in multiple dimensions and perform hierarchical analysis. It uses edge nodes to make real-time inferences and realizes automatic classification of events. Based on event-level edge-cloud joint early warning, for general events, the edge nodes issue early warnings locally and execute basic responses; when a major event is detected, the edge nodes synchronously upload the early warning signal and key characteristics to the cloud, and initiate advanced decision-making and resource scheduling. Tiered response and coordinated resource scheduling optimization: Based on the event level, scope of impact and resource availability, coordinated dispatch of urban emergency resources at the edge and cloud levels to achieve intelligent matching and priority allocation of resources; The system handles performance feedback, collects multi-dimensional performance indicators of the handling process and results at the edge and cloud, and dynamically optimizes the grading, early warning and scheduling models.

[0008] In one approach, the multi-dimensional data collection and edge fusion preprocessing for urban events includes: collecting multi-dimensional status information such as traffic congestion, environmental anomalies, and public security incidents through IoT terminals covering key areas of the city; dynamically fusing the raw data by edge computing nodes based on a weighted adaptive aggregation algorithm, with the fusion weights determined by the historical validity of the nodes and the confidence level of the data source; and performing preliminary risk assessment on the aggregated data after processing by an adaptive threshold anomaly detection module, thereby improving the accuracy of data fusion and the efficiency of edge response.

[0009] In one scheme, the intelligent classification and discrimination of urban events includes: a fusion feature hierarchical reasoning algorithm, which performs semantic deconstruction on multidimensional data from edge fusion nodes through a feature extraction module to generate key feature vectors; The key feature vectors are sequentially input into the first-level rule discrimination layer and the deep learning layer. The first-level rule discrimination layer adopts the expert system threshold or pattern discrimination rule, and the deep learning layer adopts the multilayer perceptron or convolutional neural network to output the event level probability distribution. The fusion reasoning layer integrates the outputs from the rule layer and the deep layer according to a weighted decision structure, and finally provides the event classification results, which are used for rapid early warning and intelligent resource scheduling of subsequent urban events.

[0010] In one approach, the event-level-based edge-cloud joint early warning includes: based on the intelligent classification and discrimination results, edge nodes perform real-time traffic diversion according to the classification and preset threshold strategy; For general-level events, edge nodes adopt a local self-circulating early warning model, responding through local broadcasting, local push and linked terminals, and only initiating early warning signals and activating local resource scheduling at the edge layer.

[0011] In one scheme, the hierarchical disposal collaborative resource scheduling optimization includes: based on the intelligent hierarchical results, coverage and impact range, and emergency resource availability status, under the edge and cloud collaborative system, comprehensively considering event level, impact, and spatiotemporal accessibility, and using a multi-objective weight combination optimization objective function to model the scheduling problem. The scheduling process simultaneously satisfies resource over-allocation, scheduling requirements, and geographical constraints. Through an improved particle swarm optimization distributed solver based on edge intelligent agents, global information retrieval and local resource optimization are integrated to dynamically optimize resource allocation.

[0012] In one approach, the performance feedback includes: automatically collecting multi-dimensional performance data from edge nodes and the cloud throughout the entire process to form an edge-cloud all-time-space performance feedback stream; designing a standardized performance objective function, setting ideal targets for each type of indicator, and providing comprehensive feedback with indicator weights as a driving signal for model optimization. An adaptive optimization strategy based on historical performance gradients is adopted, with the latest events and performance feedback serving as additional training samples. Edge nodes periodically feed back raw and analyzed performance data to the cloud through a data channel secured by blockchain.

[0013] In one scheme, the aforementioned event-level-based edge-cloud joint early warning includes: for major and catastrophic events, triggering an edge-cloud collaborative alarm chain, reporting key characteristic data packets to the cloud platform through a bandwidth dynamic management mechanism, triggering an advanced collaborative scheduling and decision-making module on the cloud based on the event level and data packets, executing local emergency plans synchronously on the edge, generating global handling suggestions or scheduling instructions on the cloud, and setting handling delay requirements using a dynamic feedback scheduling function for different event levels.

[0014] Beneficial effects of this invention: This invention integrates edge computing and cloud-based decision-making in urban event management, enabling real-time fusion analysis and intelligent hierarchical early warning of multi-source heterogeneous data, significantly improving the timeliness and accuracy of event identification and response. Utilizing edge nodes for rapid preliminary processing and judgment of events greatly reduces data transmission latency and the computational burden on the central platform.

[0015] Meanwhile, the cloud-edge collaborative resource scheduling mechanism can dynamically optimize the allocation and scheduling of emergency resources according to the event level, achieving efficient linkage between multiple regions and departments, and effectively avoiding problems such as resource duplication and untimely allocation. Overall, this invention improves the intelligence and automation level of urban public safety management and emergency response, enhances the city's ability to respond to emergencies, and has significant practical significance and application value for promoting smart city construction and improving urban operation management. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a flowchart of the edge-cloud joint early warning process based on event levels in this invention. Detailed Implementation

[0017] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0018] Unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. To facilitate understanding, the invention will now be described more fully with reference to the accompanying drawings. Typical embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to make the disclosure of the invention more thorough and complete.

[0019] like Figure 1 As shown, a resource scheduling method for hierarchical early warning and response of urban events integrating edge computing is specifically included: Step 1: Multi-dimensional data collection and edge fusion preprocessing for urban events Firstly, relying on IoT terminals and urban sensors, real-time collection of multi-dimensional urban event-related data, including traffic, environment, and security, is achieved. Unlike traditional centralized transmission, edge nodes are used to complete preliminary data fusion, achieving noise reduction, summarization, and local aggregation of heterogeneous data. A weighted adaptive aggregation algorithm is designed—that is, the fusion weights are dynamically adjusted according to the historical validity of nodes and data sources to improve data quality and complete the initial event judgment in a timely manner at the edge nodes.

[0020] Firstly, by utilizing IoT terminals covering key areas of the city, such as video surveillance cameras, traffic flow detectors, gas detectors, and mobile positioning devices, various types of sensors are employed to collect real-time information on multi-dimensional urban conditions, including traffic congestion, environmental anomalies, and security incidents. The collected raw data suffers from issues such as inconsistent formats, latency fluctuations, and noise interference. Directly transmitting this data back to the cloud center for processing would waste bandwidth and hinder real-time response. Therefore, edge computing nodes within the city undertake the initial data fusion task. An innovative "weighted adaptive aggregation algorithm" is introduced to improve the accuracy and robustness of data fusion.

[0021] Specifically, suppose there are n edge nodes in a certain region, and each node i at time... The event data sensed is The fusion weight corresponding to this node is Historical validity of each node (e.g., anomaly detection accuracy) And the confidence level of the data source (such as sensor brand level, network latency, packet loss rate, etc.). The weights are jointly determined by the factors of weight and other factors. The formula for dynamically adjusting the weights is: in, These are weighting coefficients obtained through experience or dynamic optimization, ensuring that validity and source confidence jointly influence the results. The fused multidimensional event data is obtained by the following weighted summation: During data aggregation, edge nodes also utilize adaptive threshold anomaly detection modules (such as z-score criteria based on sliding windows, first-order mean, or dynamic adjustment of variance) to denoise and remove invalid or pseudo-data, thereby ensuring the authenticity of the summary data. After completing weighted adaptive fusion and local anomaly filtering, edge nodes use rule matching or lightweight machine learning models to perform preliminary event risk assessment on the aggregated data, such as determining whether the event reporting threshold has been reached.

[0022] Step Two: Intelligent Classification and Judgment of Urban Events A hierarchical reasoning algorithm based on integrated features is introduced, combining rule-based reasoning and deep learning to map and analyze the key features of events in multiple dimensions. This hierarchical algorithm not only dynamically adapts to the city's operational status but also utilizes edge nodes for real-time inference, enabling automatic event classification (e.g., general, major, major), laying the foundation for subsequent early warning and resource allocation.

[0023] Receive preprocessed multidimensional data from the edge fusion node The data is then fed into the hierarchical discrimination module. The fusion feature hierarchical reasoning algorithm combines rule-based semantic reasoning and deep neural network representation learning capabilities to achieve automatic discrimination across multiple features and levels.

[0024] First, through the feature extraction module... Semantic deconstruction is performed to obtain a set of key feature vectors. These features cover dimensions such as traffic density, environmental indicators, and security parameters. A Level 1 Rule Layer is established, employing a set of threshold or pattern discrimination rules defined by experts. Its form is: in, For feature importance weights, For the rule threshold, For indicator functions, The number of features involved in the rule-based judgment. If If the event exceeds the set classification threshold, the event will be initially classified into a classification level.

[0025] To further enhance the adaptability and generalization ability of the hierarchical model, a Level 2 DNN layer is introduced. Input a pre-trained multilayer perceptron (MLP) or lightweight convolutional neural network (CNN), and the network outputs a rank probability distribution of the occurrence of events. Category tag set The specific model is defined as follows: in, and These are network parameters, is the activation function. This deep model learns hierarchical patterns from historical event data and can dynamically adapt to changes in the city's real-time operational status.

[0026] Finally, the fusion inference layer combines the two-level results and determines the final classification result through a weighted decision structure. Specifically, the scoring vectors from the rule layer and the deep layer are combined linearly with weights: in, Assign weights to the standards for each level. This is a dynamically adjustable fusion coefficient. Final event classification. By maximizing Implementation of associated category weights: This hierarchical reasoning algorithm can not only make rapid judgments at the edge, but also combine historical semantics with the adaptive capabilities of deep learning to achieve dynamic and accurate event classification, which greatly improves the level of intelligence in urban situational awareness and decision-making, and lays a solid foundation for rapid early warning and efficient resource scheduling.

[0027] Step 3: Edge-Cloud Joint Early Warning Based on Event Level Based on the tiered results, for general events, edge nodes issue early warnings and execute basic responses locally. When a major event is detected, the edge nodes simultaneously upload the early warning signal and key characteristics to the cloud, initiating advanced decision-making and resource allocation. By guiding the early warning chain through event tiers, the hierarchical nature and efficiency of urban response are improved.

[0028] like Figure 2 As shown, based on the aforementioned intelligent classification results of urban events... This enables hierarchical response to early warning signals and edge-cloud collaborative triggering.

[0029] The specific process is as follows: First, the edge nodes are classified according to the final hierarchy. Real-time traffic diversion is implemented based on preset threshold strategies. For "general" level events, a "local self-circulating early warning model" is adopted, where edge nodes respond quickly via local early warning broadcasts, local push notifications, and linked terminals (such as traffic lights and electronic displays). The algorithm uses rule-based judgment; if... Then execute: Early warning signals are only issued at the edge layer, and local resource scheduling is activated.

[0030] When the event classification result is "Major" or "Extremely Serious," an edge-cloud collaborative alarm is triggered. This mechanism guides data reporting decisions based on event level, employing the following triggering strategy: Meanwhile, edge nodes The key feature data packets are packaged and reported to the cloud through a dynamic bandwidth management mechanism. It can be defined as: in For geographic tags, This is a summary of the event context. After receiving it, the cloud platform... and The "Advanced Cloud-based Collaborative Scheduling and Decision-Making Module" is triggered. At this point, the local emergency plan is executed synchronously at the edge, and the cloud platform generates global handling suggestions or mandatory scheduling instructions according to the "event priority-resource availability" model.

[0031] To address the varying delay requirements for handling different types of events, the mechanism also includes a "dynamic feedback scheduling function": in For local processing delay, For data upload latency, This reduces delays in cloud-based decision-making. By appropriately setting threshold levels and edge-cloud layered response discrimination algorithms, not only is the network and cloud computing resource consumption of general events greatly reduced, but the collaborative efficiency of handling major and catastrophic events and the accuracy of city-level responses are also guaranteed.

[0032] The cloud platform will also flexibly adjust based on the persistence or multi-source cascading of received alerts. Thresholds and warning broadcast levels enable dynamic adaptive warning enhancement, optimizing the entire city's event response chain.

[0033] Step 4: Optimize Tiered Disposal and Collaborative Resource Scheduling Based on the event severity, scope of impact, and resource availability, the algorithm coordinates edge and cloud-based dispatch of urban emergency resources to achieve intelligent resource matching and prioritization. The algorithm incorporates an "edge blockchain recording and dispatch traceability mechanism" to ensure the dispatch process is traceable and tamper-proof, guaranteeing fairness and transparency in urban event handling.

[0034] Based on the aforementioned intelligent classification results as the core decision-making basis, and taking into account the event level comprehensively. Scope of coverage and the current availability status of various emergency resources ( (Resource type numbering), initiating multi-objective resource optimization scheduling within an edge-cloud collaborative system. First, the algorithm models the event handling resource scheduling problem using "event classification - impact - spatiotemporal reachability" as core variables. Let the required resource amount at demand point i be... The available resources of type j in the resource pool are Each type of resource scheduling must meet objectives such as timeliness (e.g., minimizing response time), matching priority (prioritizing high-level events), and load balancing (avoiding over-scraping), while also taking into account actual service distance and resource constraints.

[0035] Therefore, an optimization objective function based on a combination of multi-objective weights is designed: in, For the decision variable of resource type j scheduled to demand point i, It is the estimated time for resources to be scheduled from the current location to location i. The weight of event i is determined by its level (e.g., "extraordinary" events have a higher weight than "normal" events). These are adjustable weight parameters that can be used to adjust timeliness, priority, and balance.

[0036] Constraints include not over-provisioning resources: The scheduling meets the minimum processing requirements: Furthermore, geographical constraints can be introduced to prevent unreasonable resource flow during scheduling. In the algorithm implementation, a distributed solver is employed, such as Edge-Aware Particle Swarm Optimization (EA-PSO) based on edge intelligent agents: each edge node represents a particle, sensing events and resource changes in its own domain and synchronizing with the cloud, integrating global retrieval and local optimization to iteratively find the optimal scheduling solution. After each iteration, resource allocation... Dynamic corrections are made based on feedback from the evaluation function.

[0037] To enhance the transparency and credibility of the scheduling process, an "edge blockchain recording and scheduling traceability mechanism" has been introduced into the scheduling system for the first time. Every scheduling strategy generation, actual resource scheduling action, and execution feedback is packaged into blocks by edge nodes, encrypted on the chain, and simultaneously broadcast to the authorized cloud and relevant scheduling nodes. The block content includes scheduling instructions, event classification criteria, resource type and quantity, timestamp, and responsible unit, achieving full-process tamper-proof and traceability. Blockchain smart contracts are used to constrain scheduling compliance and node behavior, ensuring that resource scheduling decisions remain traceable and accountable even in high-load, high-frequency scenarios, and preventing resource abuse or favoritism.

[0038] Step 5: Handling Performance Feedback Collect multi-dimensional performance indicators of the handling process and results at the edge and cloud, and dynamically optimize the hierarchical, early warning, and scheduling models. Utilize feedback-emergent patterns and shortcomings to automatically adjust weights, optimize strategies, and gradually achieve self-evolution, realizing a more efficient and intelligent method for urban event early warning and resource scheduling.

[0039] After each urban incident response is completed, multi-dimensional performance data of the entire process is automatically collected from edge nodes and the cloud, including key indicators such as response time, resource utilization rate, task completion rate, and public satisfaction, forming an edge-cloud all-time and space performance feedback flow.

[0040] Assume that for the nth event, the collected performance vector is... Where K is the number of evaluation dimensions, each It represents a certain type of indicator (such as average response delay, handling accuracy, resource matching rate, etc.).

[0041] First, a standardized performance objective function is designed, and ideal targets are set for each type of indicator. Calculate normalized performance deviation The total performance loss is a weighted sum: in The weights represent the indicators, reflecting their importance to the overall model optimization. This loss not only summarizes feedback but also serves as an automatic driving signal for subsequent model adjustments. In the edge-cloud "self-evolving learning engine," an adaptive tuning strategy based on historical performance gradients is adopted for sub-modules such as hierarchical discrimination, early warning thresholds, and resource scheduling weights. Taking the deep hierarchical model as an example, the latest event data and performance feedback are used as supplementary training samples. The model can be self-evolved step by step using the following parameter iteration rules: in For model parameters, For dynamic step size, Measure the impact of feedback on the objective. For rule-based criteria, such as warning thresholds. and resource scheduling weight Similarly, the following adaptive update is used: in For adaptive learning rate, this mechanism first aggregates real-time processing indicators and periodically (e.g., every minute or after event closure) feeds back raw and analyzed performance data to the cloud through a blockchain-secured data channel. When significant changes occur in business scenarios or service environments, new samples are quickly integrated through cloud-based incremental learning and transfer learning modules, achieving global deep adaptive optimization of the hierarchical model, early warning chain, and resource scheduling algorithm. Example: I. Implementation Scenarios A major traffic jam occurred on a main road in a city during the morning rush hour. The system needs to efficiently collect multi-dimensional data, achieve intelligent event classification and local response through edge nodes, synchronously upload major events to the cloud to initiate global resource scheduling, and continuously optimize based on feedback.

[0042] 1. Multidimensional data collection and edge fusion preprocessing for urban events The various data are shown in the table below: Edge nodes perform weighted adaptive aggregation, for example: Fusion weight = 0.6 × data source confidence + 0.4 × node historical validity score 2. Edge intelligence classification process (classification results are shown in the table below) After fusing the original data, feature vectors are generated. These vectors are then subjected to a combination of first-level rules and deep learning for discrimination, resulting in the following hierarchical classification: 3. Edge-Cloud Joint Early Warning and Traffic Diversion Algorithm-based automatic decision-making: For general events (such as auxiliary road B), the edge node will issue a local broadcast warning (SMS / electronic screen) to dispatch nearby traffic police and temporary guides.

[0043] For major / critical events (main road A, intersection), edge nodes dynamically manage bandwidth and only upload critical data packets to the cloud, which then generates global scheduling suggestions.

[0044] 4. Collaborative resource scheduling optimization Multi-objective optimization scheduling: The system coordinates with traffic police, emergency tow trucks, and intelligent traffic signals, optimizing resource allocation based on event severity, location, and available resources.

[0045] The resources to be scheduled are summarized below: The scheduling optimization objective simultaneously constrains traffic police to avoid over-allocation and prioritizes geographical accessibility, while the scheduling function dynamically balances nearby resources and global resources.

[0046] 5. Handling performance feedback The performance data at each stage of the statistical processing is fed back in a timely manner to optimize subsequent models.

[0047] Edge nodes use blockchain-secured channels to automatically synchronize performance data to the cloud, and the cloud incrementally trains the model with the latest events and feedback samples.

[0048] In this embodiment, a combination of edge intelligence and cloud optimization is employed. While the global response time in the cloud for major and catastrophic events is slightly longer than that of a local single-point response, global resource scheduling and intelligent road network linkage significantly improve the recovery rate of secondary road segments and alleviate congestion on main roads. The system continuously adaptively optimizes edge and cloud-based hierarchical and scheduling strategies through multi-dimensional performance feedback. The average relief rate increased by 12%. Overall resource utilization increased by 15%; False alarms / missed alarms in the tiered early warning system decreased by 23%.

[0049] This embodiment demonstrates the method's efficient support capability for large-scale urban emergencies, and possesses practical scalability and intelligent collaborative advantages.

[0050] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0051] It should be understood that the above detailed description of the technical solutions of the present invention with reference to preferred embodiments is illustrative and not restrictive. Those skilled in the art can modify the technical solutions described in the embodiments or make equivalent substitutions for some of the technical features based on reading this specification; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A resource scheduling method for hierarchical early warning and response of urban events integrating edge computing, characterized in that: The method includes: The system enables real-time collection of multi-dimensional urban event data and edge fusion preprocessing, relying on IoT terminals and urban sensors to collect multi-dimensional urban event-related data in traffic, environment, and security. The intelligent classification and identification of urban events introduces a fusion feature hierarchical reasoning algorithm, which combines rule reasoning and deep learning to map and analyze the key features of events in multiple dimensions and perform hierarchical analysis. It uses edge nodes to make real-time inferences and realizes automatic classification of events. Based on event-level edge-cloud joint early warning, for general events, the edge nodes issue early warnings locally and execute basic responses; when a major event is detected, the edge nodes synchronously upload the early warning signal and key characteristics to the cloud, and initiate advanced decision-making and resource scheduling. Tiered response and coordinated resource scheduling optimization: Based on the event level, scope of impact and resource availability, coordinated dispatch of urban emergency resources at the edge and cloud levels to achieve intelligent matching and priority allocation of resources; The system handles performance feedback, collects multi-dimensional performance indicators of the handling process and results at the edge and cloud, and dynamically optimizes the grading, early warning and scheduling models.

2. The urban event hierarchical early warning and response resource scheduling method integrating edge computing as described in claim 1, characterized in that: The aforementioned multi-dimensional data collection and edge fusion preprocessing for urban events includes: collecting multi-dimensional status information on traffic congestion, environmental anomalies, and public security events through IoT terminals covering key areas of the city; dynamically fusing the raw data by edge computing nodes based on a weighted adaptive aggregation algorithm, with the fusion weights determined by the historical validity of the nodes and the confidence level of the data source; and performing preliminary risk assessment on the aggregated data after processing by an adaptive threshold anomaly detection module, thereby improving the accuracy of data fusion and the efficiency of edge response.

3. The urban event hierarchical early warning and response resource scheduling method integrating edge computing as described in claim 1, characterized in that: The aforementioned intelligent classification and discrimination of urban events includes: a fusion feature hierarchical reasoning algorithm, which uses a feature extraction module to semantically deconstruct multidimensional data from edge fusion nodes and generate key feature vectors; The key feature vectors are sequentially input into the first-level rule discrimination layer and the deep learning layer. The first-level rule discrimination layer adopts the expert system threshold or pattern discrimination rule, and the deep learning layer adopts the multilayer perceptron or convolutional neural network to output the event level probability distribution. The fusion reasoning layer integrates the outputs from the rule layer and the deep layer according to a weighted decision structure, and finally provides the event classification results, which are used for rapid early warning and intelligent resource scheduling of subsequent urban events.

4. The urban event hierarchical early warning and response resource scheduling method integrating edge computing as described in claim 1, characterized in that: The aforementioned event-level-based edge-cloud joint early warning includes: based on the intelligent classification and discrimination results, edge nodes perform real-time traffic diversion according to the classification and preset threshold strategy; For general-level events, edge nodes adopt a local self-circulating early warning model, responding through local broadcasting, local push and linked terminals, and only initiating early warning signals and activating local resource scheduling at the edge layer.

5. The urban event hierarchical early warning and response resource scheduling method integrating edge computing as described in claim 1, characterized in that: The aforementioned hierarchical disposal and collaborative resource scheduling optimization includes: based on intelligent hierarchical results, coverage and impact range, and emergency resource availability, under the edge and cloud collaborative system, comprehensively considering event level, impact, and spatiotemporal accessibility, and using a multi-objective weight combination optimization objective function to model the scheduling problem. The scheduling process simultaneously satisfies resource over-allocation, scheduling requirements, and geographical constraints. Through an improved particle swarm optimization distributed solver based on edge intelligent agents, global information retrieval and local resource optimization are integrated to dynamically optimize resource allocation.

6. The urban event hierarchical early warning and response resource scheduling method integrating edge computing as described in claim 1, characterized in that: The aforementioned performance feedback includes: automatically collecting multi-dimensional performance data from edge nodes and the cloud throughout the entire process to form an edge-cloud all-time and space-time performance feedback stream; designing a standardized performance objective function, setting ideal targets for each type of indicator, and providing comprehensive feedback with indicator weights as a driving signal for model optimization; An adaptive optimization strategy based on historical performance gradients is adopted, with the latest events and performance feedback serving as additional training samples. Edge nodes periodically feed back raw and analyzed performance data to the cloud through a data channel secured by blockchain.

7. The urban event hierarchical early warning and response resource scheduling method integrating edge computing according to claim 4, characterized in that: The aforementioned event-level-based edge-cloud joint early warning includes: for major and catastrophic events, triggering an edge-cloud collaborative alarm chain, reporting key characteristic data packets to the cloud platform through a bandwidth dynamic management mechanism, triggering an advanced collaborative scheduling and decision-making module on the cloud based on the event level and data packets, executing local emergency plans synchronously on the edge, generating global handling suggestions or scheduling instructions on the cloud, and setting handling delay requirements using a dynamic feedback scheduling function for different event levels.

Images