A smart early warning event distribution and disposal method, system, device and medium

Abstract

The wisdom early warning event allocation and disposal method, system, device and medium of the application belong to the technical field of event processing. It includes: acquiring monitoring area perception data and real-time analysis, generating early warning events when abnormal, including event type, location, time and preliminary description; associating risk portrait for early warning events, calculating expected completion time limit, and determining basic reward points; screening candidate nodes according to risk portrait, expected completion time limit and occurrence location; transmitting event offer encapsulated by expected completion time limit, basic reward points and the like to target nodes to drive them to generate commitment vectors; after the validity period of the offer expires, collect commitment vectors, select the optimal node, send task binding notification to it, and broadcast the decision result notification to the remaining target nodes. The application has the beneficial technical effects of reducing disposal cost and decision error rate.

Description

Technical Field

[0001] This application relates to the technical field of event handling, and in particular to a method, system, device and medium for intelligent early warning event allocation and handling. Background Technology

[0002] In today's society, the demand for safety early warning and incident handling is increasing across all sectors. Whether it's urban management, business operations, or public safety assurance, timely and accurate detection of abnormal situations and effective handling are essential.

[0003] Traditional methods for allocating and handling early warning events typically rely on analysis of a single data source, such as image monitoring or sensor data alone. After an anomaly is detected, an early warning event is generated through manual or simple rule-based judgment, and then assigned to the appropriate personnel according to a fixed process. This allocation process often only considers the scope of responsibility, leading to blind and delayed event allocation, difficulty in quickly matching the most suitable response point, and increased handling costs and decision-making error rates. Summary of the Invention

[0004] To reduce handling costs and decision-making error rates, this application provides a smart early warning event distribution and handling method, system, equipment, and medium.

[0005] Firstly, this application provides a method for intelligent early warning event allocation and handling, employing the following technical solution:

[0006] A method for intelligent early warning event allocation and handling includes:

[0007] Acquire sensing data within the monitored area, including images, audio, and sensor data;

[0008] The sensed data is analyzed in real time. When an anomaly is detected, an early warning event is generated. The early warning event includes the event type, the location of occurrence, the reporting time, and a preliminary description.

[0009] Associate risk profiles with the warning events and calculate the expected completion timeframes for the warning events;

[0010] The basic reward points are determined based on the risk profile and the expected completion timeframe.

[0011] Candidate nodes are selected based on the responsibility area matrix of the handling nodes and the location of the incident to obtain a set of candidate nodes;

[0012] The event offer is passed to the target node, which then generates a commitment vector based on its own private state. The event offer is encapsulated by the expected completion time, the basic reward score, the candidate node set, the offer creation time, and the offer validity period.

[0013] After the offer expires, all commitment vectors are collected, the optimal node is selected, a task binding notification is sent to the optimal node, and the adjudication result notification is broadcast to the remaining target nodes.

[0014] By adopting the above technical solutions, the bottlenecks of data isolation and delayed analysis in traditional early warning models are broken. Real-time multi-source data fusion analysis can more accurately and quickly capture anomalies and generate standardized early warning events, making event handling more sensitive and reliable. Risk profiling and the calculation of expected completion deadlines establish a priority benchmark for event handling. Combined with candidate node screening using a responsibility-region matrix, the blindness and lag of traditional task assignment are completely eliminated, enabling the immediate matching of node resources with the best handling capabilities and geographical advantages. The two-way selection mechanism based on event offer and commitment vectors introduces market-based resource allocation logic. It fully mobilizes the enthusiasm of handling nodes through basic reward points and ensures a high degree of fit between handling tasks and node capabilities through the autonomous response of node private states. Finally, through the selection of optimal nodes and result broadcasting, efficient and accurate allocation of handling resources is achieved. This not only improves the response speed and matching accuracy of early warning event handling but also reduces handling costs and decision-making error rates.

[0015] Optionally, the specific steps for passing the event offer to the target node include:

[0016] The offer urgency is determined based on the expected completion time limit, and a broadcast mode is selected based on the offer urgency. The broadcast mode includes mode one and mode two.

[0017] For Mode 1: If the offer urgency is low, then broadcast to all nodes in the candidate node set simultaneously;

[0018] For Mode 2: If the urgency of the offer is medium or high, a progressive broadcast is initiated. First, the offer is sent only to the node in the candidate node set that is closest to the location of the event. After waiting for a preset time, if no valid commitment is received, the broadcast range is expanded to the three closest nodes in the candidate node set, until the entire candidate node set is covered or a valid commitment is received. The preset time is less than the offer validity period.

[0019] By adopting the above technical solution and determining the urgency of the offer, the broadcast mode is divided into two paths: synchronous full coverage and progressive reach. This approach not only takes into account the handling needs of events with different urgency levels but also achieves refined scheduling of handling resources. For low-urgency events, the synchronous broadcasting of all candidate nodes maximizes the range of choices for task acceptance. Through the autonomous response competition of multiple nodes, it ensures that the task is completed with the most suitable cost and efficiency. For medium or high-urgency events, the progressive broadcast mode precisely addresses the core requirement of emergency handling—rapid response. By prioritizing the reach of the nearest node, it can compress the time cost of handling response with the shortest physical distance. At the same time, the preset waiting time and progressive expansion mechanism build a balanced bridge between "rapid response" and "guaranteed acceptance," avoiding handling delays caused by the inability of a single node to respond and preventing resource redundancy and information interference caused by large-scale broadcasting. This not only makes the transmission of event offers more targeted and flexible, but also improves the efficiency and success rate of handling different levels of early warning events through dynamic adaptation of resource scheduling. While ensuring rapid response to emergencies, it also minimizes the resource consumption caused by the transmission of ineffective information.

[0020] Optionally, the specific steps for generating a commitment vector based on its own private state include:

[0021] Each node that receives the offer queries its own historical average processing time for similar early warning events;

[0022] Calculate the estimated travel time based on the current location of its own node, the location of the event, and real-time traffic data;

[0023] Treat the warning event as a task to be inserted, and calculate the impact on the completion time of all tasks of the node after inserting it into different positions in the current task queue.

[0024] Select the insertion position that meets the expectations, and calculate the estimated completion time after the warning event is inserted;

[0025] The promised completion time is calculated based on the estimated completion time, estimated travel time, and historical average processing time.

[0026] A commitment vector is generated based on the commitment completion time.

[0027] By adopting the above technical solutions, nodes are allowed to autonomously query the historical average processing time of similar events, providing a practice-based reference benchmark for task completion time estimation. Combining current location, event location, and real-time traffic data to calculate travel time further incorporates dynamic environmental factors, making time estimation more closely aligned with real-world scenarios. Furthermore, by simulating the insertion of early warning events as tasks into different positions in the current task queue and analyzing their impact on overall task completion time, the existing workload of nodes is taken into account, avoiding delays in existing tasks caused by blindly accepting new tasks, and achieving collaborative optimization between new and old tasks. By integrating multi-dimensional data to calculate the promised completion time and generate a commitment vector, the risk of nodes defaulting on their commitments due to over-commitment is reduced, and the event initiator can make informed choices when selecting handling nodes, thereby improving the overall credibility and operational efficiency of the entire intelligent early warning event handling process.

[0028] Optionally, the specific steps for generating the commitment vector based on the commitment completion time include:

[0029] Calculate the time difference between the commitment completion time and the offer creation time, and determine whether the time difference exceeds a set threshold, wherein the set threshold = expected completion time × conditional commitment trigger threshold;

[0030] If so, further analyze whether the node can advance the completion time of the warning event by releasing an existing low-priority task. If so, the node generates a conditional commitment; otherwise, abandon the response.

[0031] The conditional commitment, commitment completion time, and resource bid integral are encapsulated into a commitment vector, where the resource bid integral = node fatigue × basic reward integral;

[0032] If not, an unconditional commitment is generated, and the commitment completion time and resource quotation integral are encapsulated into a commitment vector.

[0033] By adopting the above technical solution, a clear benchmark for judging the node's response behavior is established by comparing the difference between the promised completion time and the offer creation time with a set threshold. When the difference exceeds the threshold, the node does not directly abandon the response, but further guides the node to assess the possibility of completing the warning event ahead of schedule by releasing low-priority tasks. This design not only avoids nodes missing emergency handling opportunities due to existing task saturation, but also retains flexibility in task acceptance through conditional commitments, balancing the handling priorities of new and old tasks. This mechanism effectively reduces the risk of default caused by nodes blindly committing, and improves the utilization rate of emergency resources through the flexible design of conditional commitments.

[0034] Optionally, the specific steps for collecting all commitment vectors and selecting an optimal node from them include:

[0035] Collect all commitment vectors and filter valid commitments, which are commitment vectors whose promised completion time does not exceed twice the expected completion time.

[0036] Calculate the standardized utility value for each valid commitment, which includes time utility and cost utility;

[0037] The comprehensive adjudication score is calculated based on the standardized utility value and the node's reputation score;

[0038] The node with the highest overall decision score is selected as the optimal node.

[0039] By adopting the above technical solution and screening effective commitments, commitments that clearly exceed reasonable processing timeframes are excluded, ensuring the basic processing capabilities of candidate nodes from the outset. Through standardized calculations of time utility and cost utility, core indicators such as commitment completion time and resource quotation points are transformed into quantifiable and comparable utility values. This approach balances the core requirement of timeliness in emergency response with a full consideration of the rationality of resource investment, avoiding cost waste caused by simply pursuing speed or response delays caused by simply controlling costs. Furthermore, a weighted calculation of the comprehensive adjudication score based on node reputation score and standardized utility value is introduced, further incorporating the node's historical processing performance into the decision-making process. This allows nodes with good reputations and stable processing capabilities to receive more opportunities, effectively improving the reliability and success rate of task processing and incentivizing nodes to continuously improve their processing capabilities and service quality.

[0040] Optionally, the specific steps for selecting the node with the highest overall decision score as the optimal node include:

[0041] If the valid commitment with the highest overall adjudication score is an unconditional commitment, then the corresponding node is the optimal node.

[0042] If the valid commitment with the highest overall adjudication score contains a conditional commitment, then a new secondary event offer is created based on the conditional commitment;

[0043] Within the set offer time window, a broadcast is initiated to the nodes in the neighborhood. The neighborhood is defined as the area centered on the node itself with a set value as the radius.

[0044] If a secondary commitment is received from another node, the condition of the node itself is met, the node determines that its own valid commitment is valid, and at the same time locks the secondary commitment and sends a task binding notification to the corresponding node.

[0045] If no secondary commitments are received from other nodes, the node's own valid commitments are deemed invalid and removed.

[0046] After resolving all conditional commitments, among the remaining valid commitments, the node corresponding to the commitment with the highest overall adjudication score is selected as the optimal node.

[0047] By adopting the above technical solution, the priorities of unconditional and conditional commitments are distinguished. High-scoring commitments without additional conditions are directly locked to the optimal node, ensuring decision-making efficiency in simple scenarios. For the highest-scoring conditional commitments, a secondary event offer is created to transform the preconditions of the commitment into actionable subtasks. Broadcasting to specific neighboring nodes establishes a resource connection channel for fulfilling the preconditions. The receiving and locking mechanism of secondary commitments not only quickly verifies the feasibility of conditional commitments but also enables coordinated scheduling between the main task and subtasks through synchronous binding of subtasks. This transforms conditional commitments that originally posed execution risks into reliable and effective fulfillments, fully tapping the processing potential of nodes and avoiding task delays caused by unmet preconditions. Furthermore, the design of promptly deeming and removing a conditional commitment invalid when a secondary commitment cannot be obtained effectively avoids idle waiting and waste of resources, ensuring the efficient advancement of the decision-making process. After all the issues related to the implementation of conditional commitments are resolved, the optimal node is selected from the remaining valid commitments. This makes the entire task allocation process more flexible and fault-tolerant, providing a solid mechanism guarantee for emergency response in complex scenarios and effectively improving the overall reliability and operational efficiency of the intelligent early warning event handling system.

[0048] Optionally, the intelligent early warning event distribution and handling method further includes:

[0049] When the optimal node begins to handle the early warning event, receive the task status reported by the optimal node;

[0050] When the task status is "processing completed", record the actual completion time;

[0051] Calculate the time deviation rate based on the actual completion time, the promised completion time, and the offer creation time;

[0052] The performance result is determined based on the time deviation rate, and the node's reputation score is updated based on the performance result.

[0053] By adopting the above technical solution, and receiving real-time task status reports during the optimal node processing, the pace of task progress can be grasped in a timely manner. The design of recording the actual completion time after task completion and calculating the time deviation rate based on this provides a quantitative basis for node performance, transforming abstract performance quality into measurable and concrete indicators. Determining the performance result and updating the node's reputation score based on the time deviation rate directly links the node's historical performance with future task opportunities, forming a positive incentive and a negative constraint: for nodes with excellent performance and low time deviation rates, an improved reputation score allows them to obtain higher priority in subsequent task competition, thereby gaining more processing resources and benefits; conversely, for nodes with poor performance and excessive time deviation rates, a decreased reputation score directly affects their subsequent competitiveness, forcing nodes to continuously improve their processing capabilities and time management levels. It can not only effectively improve the nodes' awareness of fulfilling their obligations and their efficiency in handling matters, but also continuously optimize the quality of the entire handling node pool through a dynamically updated credit score system. This allows high-quality nodes to receive more resource support, while inefficient nodes are gradually eliminated or achieve self-improvement, thereby continuously promoting the iterative upgrading of the overall operational efficiency and service quality of the intelligent early warning event handling system.

[0054] Secondly, this application provides an intelligent early warning event distribution and handling system, which adopts the following technical solution:

[0055] A smart early warning event distribution and handling system includes:

[0056] The data acquisition module is used to acquire sensing data within the monitored area, including images, audio, and sensor data.

[0057] The data analysis module is used to perform real-time analysis of the sensed data;

[0058] The early warning module is used to generate an early warning event when the data analysis module detects an anomaly. The early warning event includes the event type, the location of occurrence, the reporting time, and a preliminary description.

[0059] The early warning and dispatching module is used to associate risk profiles with the early warning events and calculate the expected completion time of the early warning events; to determine the basic reward points based on the risk profiles and the expected completion time; to filter candidate nodes based on the responsibility region matrix and the location of the incident to obtain a candidate node set; to transmit the event offer to the target node and drive the target node to generate a commitment vector based on its own private state, wherein the event offer is encapsulated by the expected completion time, the basic reward points, the candidate node set, the offer creation time, and the offer validity period; and to collect all commitment vectors after the offer validity period expires, select the optimal node from them, send a task binding notification to the optimal node, and broadcast the adjudication result notification to the remaining target nodes.

[0060] Thirdly, this application provides a computer device that adopts the following technical solution:

[0061] A computer device includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the intelligent early warning event allocation and handling method as described in the first aspect.

[0062] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution:

[0063] A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described in the first aspect of the intelligent early warning event allocation and handling method.

[0064] In summary, this application includes at least one of the following beneficial technical effects:

[0065] This application introduces a market-based resource allocation logic based on a two-way selection mechanism of event offer and commitment vectors. It fully motivates disposal nodes through basic reward points and ensures a high degree of matching between disposal tasks and node capabilities through the autonomous response of nodes' private states. Finally, through the selection of the optimal node and the broadcasting of results, it achieves efficient and precise allocation of disposal resources. This not only improves the response speed and matching accuracy of early warning event handling but also reduces disposal costs and decision-making error rates. Attached Figure Description

[0066] Figure 1 This is a first flowchart of an embodiment of the method of this application;

[0067] Figure 2 This is a second flowchart of an embodiment of the method of this application;

[0068] Figure 3 This is a third flowchart of an embodiment of the method of this application;

[0069] Figure 4 This is the fourth flowchart of an embodiment of the method of this application;

[0070] Figure 5 This is the fifth flowchart of an embodiment of the method of this application;

[0071] Figure 6 This is the sixth flowchart of an embodiment of the method of this application. Detailed Implementation

[0072] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figures 1-6The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0073] The first embodiment of this application discloses a method for intelligent early warning event allocation and handling. (Refer to...) Figure 1 The intelligent early warning event allocation and handling method includes S110-S170:

[0074] S110, acquire sensing data within the monitored area, including images, audio, and sensor data;

[0075] S120 performs real-time analysis of the sensed data. When an anomaly is detected, an early warning event is generated. The early warning event includes the event type, location of occurrence, reporting time, and preliminary description.

[0076] S130: Create a risk profile for the warning event and calculate the expected completion time of the warning event;

[0077] S140, Determine the basic reward points based on the risk profile and expected completion time;

[0078] S150, Based on the responsibility area matrix of the handling nodes and the location of the incident, candidate nodes are selected to obtain a set of candidate nodes;

[0079] S160, the event offer is passed to the target node and the target node is driven to generate a commitment vector based on its own private state. The event offer is encapsulated by the expected completion time limit, the basic reward score, the candidate node set, the offer creation time and the offer validity period.

[0080] S170: After the offer expires, collect all commitment vectors, select the optimal node from them, send a task binding notification to the optimal node, and broadcast the adjudication result notification to the remaining target nodes.

[0081] Specifically, in step S110, full coverage is achieved by deploying a multi-type sensor network in the monitored area: video cameras are used to capture high-definition images and video streams; microphone arrays are used to capture ambient audio; and various IoT sensors, including temperature and humidity sensors, smoke detectors, vibration sensors, and pressure sensors, are embedded in key facility nodes to upload structured sensor data in real time. The raw data stream is uploaded to the regional edge server via 5G or LoRaWAN wireless communication protocols. To ensure data synchronization, all devices employ an NTP timestamp alignment mechanism to ensure that cross-modal data is spatiotemporally aligned within milliseconds. For example, when crowds gather at a subway station entrance, the visual module captures changes in crowd density, the audio module detects an increasing trend in noise, and the pressure tile sensor reports a sudden increase in ground load; these three elements together constitute the initial sensing input.

[0082] Step S120 relies on an AI inference engine platform based on a microservices architecture. The platform is deployed at edge computing nodes or regional cloud centers. After receiving data streams from S110, the computer vision model uses YOLOv8 or Swing Transformer as the backbone network and performs transfer training on a self-built labeled dataset to output target categories, bounding boxes, and confidence scores. The self-built labeled dataset includes scenarios such as illegal parking, pedestrians entering restricted areas, and blocked fire lanes. The audio recognition model is fine-tuned based on the Wav2Vec 2.0 pre-trained model. A classifier is built for abnormal sounds such as explosions, broken glass, and continuous cries for help. Features are extracted using Mel spectrograms and then fed into a 1D-CNN-LSTM hybrid structure to complete the recognition. For IoT sensor data, a time series anomaly detection model such as Temporal Fusion Transformer (TFT) or LSTM-Autoencoder is used. During training, historical normal operating data is used as a benchmark, and a sliding window (such as the past 6 hours) is set to extract statistical features (mean, variance, mutation rate, etc.). When the prediction residual exceeds 3 times the standard deviation, it is judged as a potential fault or risk exceeding the standard. The outputs of the three types of models are semantically aligned via a knowledge graph middleware. For example, if the visual model detects that a distribution box door is open and the infrared image shows a localized high temperature, while the temperature sensor reading suddenly rises, a composite event of "abnormal equipment status" is triggered. Similarly, if a sharp sound of breaking glass suddenly occurs in an area at night and there is no record of legitimate personnel activity, it is marked as "suspected intrusion." When the confidence level of any model output exceeds a preset threshold, such as >0.85, a risk is determined, and the system automatically generates a structured early warning event record, including the event type code (e.g., EVT_013 indicating "illegal intrusion"), the GPS coordinates or indoor location tag of the location (e.g., B3-Floor2-ZoneA), the UTC timestamp of the reporting time, and an AI-generated natural language description (e.g., "Suspected climbing behavior was observed on the south wall, video confidence level 92%), and pushes it to the next processing step via a message queue (Kafka).

[0083] In step S130, the system queries the object feature profile database using the location as the key to obtain information such as the historical risk level of the associated object, the average handling time of similar past events, and the required professional skills. Specifically, a graph database-driven object profiling system (such as Neo4j or JanusGraph) is established, where each entity node represents a geographical unit (such as a power distribution room or a group of manhole covers on a road section) or a management responsibility entity (such as an inspection team). Edge relationships characterize attributes such as historical interactions, fault frequency, maintenance difficulty, and skill dependencies. After receiving an early warning event, the system retrieves the complete profile file of the corresponding facility / area using the event's geographical location as the primary key. This profile is then attached to the original early warning event. Profile updates employ an incremental learning mechanism, automatically backfilling the actual handling results after each new event loop closure, forming a dynamically evolving knowledge graph. For example, if the alarm occurs in a transformer room in an old residential area, the profile might show: a historical risk level of "high" (due to 6 repair reports in the past year), an average handling time of 47 minutes, typically requiring an electrician with a license and high-voltage work permit, and the most recent maintenance was 3 months ago. This metadata not only comes from the historical work order system (ERP integration interface), but also incorporates external context such as weather influencing factors (e.g., a 30% increase in failure rate during the rainy season) and surrounding population density (fewer people at night facilitates rapid entry).

[0084] In step S140, a three-dimensional decision matrix designed with the participation of domain experts is constructed. This matrix uses event type as rows, risk level as columns, and time period as layers. Event types can be set to 50 categories, covering everything from minor disturbances to major safety hazards. Risk levels are divided into low, medium, and high levels, and time periods are divided into daytime (7:00-19:00) and nighttime (19:00-7:00). Each cell stores the corresponding expected completion time limit (in minutes). For example, the standard response time limit for "fire smoke alarm + high risk + nighttime" is set to 8 minutes; "streetlights off + low risk + daytime" is allowed to be resolved within 90 minutes. This rule base is stored in the configuration management center in JSON Schema format, supporting version control and canary releases. During system runtime, the base time limit value is obtained by looking up the table based on the current event's triple attribute index. Furthermore, to cope with emergencies, the rule base supports dynamic weighted adjustments. For example, during a typhoon emergency response, the system can temporarily increase the time limit priority of all outdoor facility events, forcibly shortening the response window to ensure priority protection for critical infrastructure.

[0085] In step S140, the basic reward integral is determined by looking up a table based on the urgency and complexity of the event. Basic reward integral = f(urgency, complexity) = Where Ttarget is the expected completion time obtained from S140, and Cc is taken from the comprehensive score of the risk profile in S130, calculated using the weighted formula: 0.4 × historical risk level code + 0.3 × normalized average handling time + 0.3 × number of skill thresholds. The weights can be adjusted according to the actual situation. Coefficient k1, K2 is determined by the operations team based on a human resource cost model to ensure that tasks with high timeliness requirements and high skill thresholds receive higher rewards. For example, a task to investigate the potential danger of falling objects from heights, which needs to be completed within 10 minutes, has a basic reward score of 80 due to its urgency and professionalism; while a routine cleaning inspection has a basic reward score of 20. The aim is to encourage capable individuals to proactively take on difficult and urgent tasks, avoiding the phenomenon of "picking up the easy tasks."

[0086] In step S150, all nodes qualified to handle the event within their responsibilities and geographical scope are selected. This is achieved using a predefined spatial responsibility allocation table. Organized with an R-tree index, this table records the service coverage (GeoJSON polygon), functional category (e.g., water and electricity, security, municipal), and a list of certified qualifications and their validity periods for each handling node (which can be an individual, team, or robot). Upon receiving the event location, the system performs a spatial query to determine which nodes fall within the responsibility grid to which that node belongs, and further verifies whether their functions match the event type. For example, a gas leak alarm will only be pushed to certified technicians with "gas repair" qualifications located within a 5-kilometer radius. This set, called the initial candidate node set, typically does not exceed 10 nodes to prevent broadcast storms. To improve flexibility, the system supports temporary authorization extensions, such as requests for cross-regional support from personnel in neighboring areas, and records permission change trajectories via blockchain to ensure audit traceability.

[0087] In S160, the specific steps for transmitting the event offer to the target node include:

[0088] The urgency of the offer is determined based on the expected completion time, and the broadcast mode is selected based on the urgency of the offer. The broadcast modes include Mode 1 and Mode 2.

[0089] For Mode 1: If the offer urgency is low, then broadcast to all nodes in the candidate node set simultaneously;

[0090] For Mode 2: If the urgency of the offer is medium or high, a progressive broadcast is initiated. First, the offer is sent to the node closest to the location in the candidate node set. After waiting for a preset time, if no valid commitment is received, the broadcast range is expanded to the three closest nodes in the candidate node set until the entire candidate node set is covered or a valid commitment is received. The preset time is less than the offer validity period.

[0091] Specifically, in step S160, for low-urgency events, such as garbage accumulation in green belts, the system activates Mode 1: simultaneously sending standardized event offer packets to all nodes in the candidate node set via MQTT topic broadcast. For medium- to high-level events, such as structural collapse warnings, Mode 2, progressive broadcasting, is activated; initially, the offer is only sent via private message to the candidate node closest to the incident location; a waiting time is set, such as 15 seconds, typically 0.3 times the offer validity period, during which the system listens for a valid response from that node. If no valid response is received within the timeout period, the broadcast circle is expanded to the three closest nodes, and the waiting period continues; this process continues until all candidates are covered or a commitment is successfully obtained. This "funnel-shaped wake-up" mechanism significantly reduces the probability of unrelated personnel being disturbed, and is particularly suitable for sensitive nighttime periods. The entire offer message body is encapsulated using Protobuf serialization and includes a unique ID (UUIDv4), structured event details, profile summary (such as "historical average time taken is 45 minutes, electrician qualification required"), expected completion time limit, basic reward points, creation time and offer validity period (default is 1.5 times the expected time limit), and a set of candidate nodes, and is protected against tampering by digital signature.

[0092] Reference Figure 2 The specific steps for generating a commitment vector based on its own private state include S210-S260:

[0093] S210, Each node that receives the offer queries its own historical average processing time for similar early warning events;

[0094] S220 calculates the estimated travel time based on its current node location, the location of the event, and real-time traffic data;

[0095] S230: Treat the warning event as a task to be inserted, and calculate the impact on the completion time of all tasks of the node after inserting it into different positions of the current task queue.

[0096] S240, Select the insertion position that meets the expectations, and calculate the estimated completion time after the warning event is inserted;

[0097] S250, calculate the promised completion time based on the estimated completion time, estimated travel time, and historical average processing time;

[0098] S260, Generate a commitment vector based on the commitment completion time.

[0099] Specifically, the node is equipped with an intelligent agent client, one of whose core functions is to maintain a local task queue (TaskQueue), using a priority queue structure, such as implemented using Python heapq. When a node, such as an operations engineer holding a tablet terminal, receives an offer, it first executes step S210: querying the historical average processing time is done by accessing the local SQLite database, with the table structure task_history(task_type, avg_duration_min, success_rate). The weighted average processing time of similar tasks is queried through SQL, with recent tasks having higher weights.

[0100] In step S220, the calculation of the estimated travel time relies on the integrated map service API. The starting point, i.e., the current GPS coordinates of the node itself, and the destination, i.e. the location where the event occurs, are input to obtain the estimated travel time from the driving / walking route planning results, and the weather impact coefficient is added, such as multiplying by a delay factor of 1.4 times in case of heavy rain.

[0101] In steps S230 and S240, task insertion simulation occurs: the node prioritizes its currently incomplete work queue, imports it into a local scheduling simulator, attempts to insert the new task into every possible position (front, middle, and rear of the queue), and runs heuristic algorithms such as genetic algorithms or simulated annealing to evaluate the change in the total delay of all tasks under each permutation. The optimal insertion point is selected that minimizes the overall delay without violating hard constraints (such as the original task deadline). Based on this, the estimated completion time of the task in the current workflow is determined.

[0102] In step S250, the promised completion time = current time + estimated travel time + queuing time after insertion + historical average processing time. This time must be a UTC timestamp and take into account time zone conversion.

[0103] Reference Figure 3 The specific steps for generating a commitment vector based on the commitment completion time include S310-S340:

[0104] S310, calculate the time difference between the commitment completion time and the offer creation time, and determine whether the time difference exceeds the set threshold. The set threshold = expected completion time × conditional commitment trigger threshold.

[0105] S320, if yes, then further analyze whether the node itself can advance the completion time of the warning event by releasing an existing low-priority task. If yes, then the node itself generates a conditional commitment. If no, then abandon the response.

[0106] S330 encapsulates the conditional commitment, commitment completion time, and resource bid integral into a commitment vector, where the resource bid integral = node fatigue × basic reward integral;

[0107] S340, if not, then generate an unconditional commitment, and encapsulate the commitment completion time and resource quotation integral into a commitment vector.

[0108] Specifically, in step S310, the threshold is defined as the expected completion time multiplied by a configurable "conditional commitment trigger threshold," which defaults to 1.2. For example, if the expected time is 30 minutes, then if the difference between the committed completion time and the creation time is greater than 36 minutes, it is considered impossible to fulfill the commitment unconditionally. At this point, step S320 is initiated: the system queries whether the node can free up time by releasing an existing low-priority task. This is achieved by running a "task replacement simulation" in the local scheduler, traversing the current task queue, attempting to remove every non-urgent task (i.e., task with a priority lower than the current offer), recalculating the new committed completion time, and checking if the threshold requirement is met. If so, a conditional commitment is generated in the form: "If someone else can take over Task #789 within XX time, then I can complete this task within YY time"; otherwise, the response is abandoned.

[0109] In steps S330 and S340, for conditional commitments, the resource bid integral is calculated using the formula Node Fatigue × Base Reward Integral. Node fatigue is a dynamic indicator that comprehensively considers factors such as continuous working time, the number of tasks completed in the past 24 hours, and sleep quality feedback, with a value ranging from 0.8 to 1.5, reflecting the individual's current workload. Higher fatigue implies a higher compensation demand. This integral, along with the commitment completion time and conditional statements, is packaged into a structured vector, i.e., the commitment vector. For unconditional commitments, the current commitment completion time and resource bid integral are directly encapsulated without any additional preconditions.

[0110] Reference Figure 4 The specific steps for collecting all commitment vectors and selecting an optimal node from them include S410-S440:

[0111] S410, collect all commitment vectors and filter valid commitments. Valid commitments are those whose completion time does not exceed twice the expected completion time.

[0112] S420, calculate the standardized utility value for each valid commitment, which includes time utility and cost utility;

[0113] S430, calculates the comprehensive adjudication score based on the standardized utility value and the node's reputation score;

[0114] S440 selects the node with the highest overall decision score as the optimal node.

[0115] Specifically, in step S410, all commitment vectors are aggregated to the central arbitration service via the AMQP protocol. The system removes invalid responses with "commitment completion time > 2 × expected completion time" to prevent fraud, and retains the remaining commitment vectors as valid commitments.

[0116] In step S420, two standardized utility values ​​are calculated for each valid commitment: time utility is calculated using an inverse proportional normalization function. ,in For each node's committed completion time, Tmin and Tmax represent the minimum and maximum time taken among all valid commitments, ensuring the fastest commitment receives full marks (1). Similarly, for cost-utility, ucost = max(0,1 - (score - Smin) / (Smax - Smin)), where score is the resource bid integral, and Smin and Smax represent the minimum and maximum resource bid integrals among all valid commitments, respectively; a lower integral results in a higher score. Both are dimensionless values ​​within the interval [0,1].

[0117] In step S430, the overall adjudication score = The weight The parameter is adjustable and can be initially set to 0.5, 0.3, or 0.2. The reputation score comes from the behavior evaluation system (S610-S640) accumulated by the node over a long period of time; Xx represents the reputation score.

[0118] Reference Figure 5 The specific steps for selecting the node with the highest overall decision score as the optimal node include S510-S560:

[0119] S510, if the valid commitment with the highest comprehensive adjudication score is an unconditional commitment, then the corresponding node is the optimal node;

[0120] S520, if the valid commitment with the highest overall adjudication score includes a conditional commitment, then a new secondary event offer is created based on the conditional commitment;

[0121] S530, within the set offer time window, initiates a broadcast to nodes in the neighborhood, where the neighborhood represents the area centered on the node itself with a set value as the radius;

[0122] S540: If a secondary commitment is received from another node, the condition of the node itself is met, the node determines that its own valid commitment is valid, and at the same time locks the secondary commitment and sends a task binding notification to the corresponding node.

[0123] S550: If no secondary commitment is received from other nodes, its own valid commitment is deemed invalid and removed.

[0124] S560, after resolving all conditional commitments, selects the node corresponding to the commitment with the highest overall adjudication score from the remaining valid commitments as the optimal node.

[0125] Specifically, if the highest score corresponds to an unconditional commitment, the task binding process proceeds directly. If it is a conditional commitment, a secondary coordination mechanism must be initiated: the system creates a new secondary event offer based on the condition, stating "Please assist in taking over Node_X's original task Task_Y," and broadcasts this offer to other qualified nodes within a radius r (e.g., 2km) centered on that node, attempting to "reduce its burden." If other nodes make commitments to the secondary task within the specified time window, the original condition is met, the original commitment becomes valid, and the two-way task binding relationship is locked, while relevant parties are notified; conversely, if no support is provided, the condition is deemed to have failed, and the original commitment is voided.

[0126] After all conditional commitments have been parsed, the node with the highest overall score among the remaining valid commitments is selected as the optimal node. A formal task binding notification (including electronic signature confirmation) is sent to this node, while a rejection notification with a brief reason, such as "failed to be selected due to a slightly slow response," is broadcast to other participants to maintain transparency and fairness. Once the task is bound, it is written to a distributed ledger such as Hyperledger Fabric, forming an immutable operation log.

[0127] Reference Figure 6 The intelligent early warning event allocation and handling methods also include S610-S640:

[0128] S610 receives the task status reported by the optimal node when the optimal node begins to handle the early warning event;

[0129] S620: When the task status is "processing completed", record the actual completion time;

[0130] S630, calculate the time deviation rate based on the actual completion time, the promised completion time, and the offer creation time;

[0131] S640 determines the performance result based on the time deviation rate and updates the node's reputation score based on the performance result.

[0132] Specifically, in step S610, when the optimal node arrives at the site and clicks the "Start Processing" button, the task is marked to enter the execution phase. The system receives the status changes reported by the node to track the task progress. Status changes include start, in progress, and completed.

[0133] In step S620, after the optimal node has been processed, the "processing completed" status is submitted, and the system records the actual completion time.

[0134] In step S630, the time deviation rate is calculated as (actual completion time - promised completion time) / expected completion time. This is used to quantify performance, with positive values ​​indicating delays and negative values ​​indicating ahead of schedule.

[0135] In step S640, the performance result is determined based on the time deviation rate, and the node's reputation score is updated accordingly: if the time deviation rate is ≤0, it is considered excellent performance, and the reputation score is increased by 5; if the time deviation rate is ≤0.2, it is considered normal performance, and the reputation score remains unchanged; if the time deviation rate is >0.2, it is considered untrustworthy performance, and the reputation score is decreased by 3, and the number of untrustworthy acts is accumulated. If there are n consecutive untrustworthy acts (such as 2 times), the order acceptance permission will be temporarily frozen.

[0136] In addition, the system performs batch analysis periodically, such as weekly, to achieve strategy self-evolution, including correlation analysis and strategy generation. During the correlation analysis phase, the system uncovers implicit patterns in historical data; for example, through regression analysis, it discovers: "When..." When the weight is greater than 0.7, although the average response time is reduced by 5%, the overemphasis on speed leads to some nodes blindly accepting tasks, ultimately increasing the failure rate by 15%, revealing the trade-off between efficiency and reliability. Based on this insight, the system automatically generates a strategy report during the strategy generation phase, proposing parameter optimization suggestions: "Given the scarcity of manpower at night, it is recommended to optimize the nighttime hours..." The price was lowered from 0.75 to 0.65, while the price was raised. The reputation weight was set to 0.25 to encourage stable and reliable nodes to take on important responsibilities. The report was automatically pushed to the administrator via WeChat or email, and after manual review, the effect was tested and verified. The parameters of the entire network were gradually iterated to achieve continuous evolution and self-optimization of the system.

[0137] Based on the above method embodiments, the second embodiment of this application discloses a smart early warning event distribution and handling system. The smart early warning event distribution and handling system of this application embodiment can implement any of the above-described smart early warning event distribution and handling methods, and the specific working process of each module in the smart early warning event distribution and handling system can refer to the corresponding process in the above method embodiments.

[0138] For ease of understanding, an example is as follows: A smart early warning event distribution and handling system includes:

[0139] The data acquisition module is used to acquire sensor data within the monitored area, including images, audio, and sensor data.

[0140] The data analysis module is used to perform real-time analysis of the sensed data;

[0141] The early warning module is used to generate early warning events when the data analysis module detects anomalies. The early warning events include the event type, location of occurrence, reporting time, and preliminary description.

[0142] The early warning and allocation module is used to associate risk profiles with early warning events and calculate the expected completion time of the early warning events; to determine the basic reward points based on the risk profiles and expected completion time; to filter candidate nodes based on the responsibility area matrix and occurrence location of the handling nodes to obtain a candidate node set; to transmit event offers to target nodes and drive the target nodes to generate commitment vectors based on their own private state; the event offer is encapsulated by the expected completion time, basic reward points, candidate node set, offer creation time, and offer validity period; and to collect all commitment vectors after the offer validity period expires, select the optimal node, send a task binding notification to the optimal node, and broadcast the adjudication result notification to the remaining target nodes.

[0143] The third embodiment of this application provides a computer device, which may include a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement a smart early warning event allocation and handling method.

[0144] The memory can communicate with the processor via a communication bus, which can be an address bus, a data bus, a control bus, etc.

[0145] Additionally, the memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device.

[0146] Furthermore, the processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0147] The fourth embodiment of this application provides a computer-readable storage medium storing a computer program that can be loaded by a processor and executed as a smart early warning event allocation and handling method.

[0148] The computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device; the program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0149] It should be noted that the computer device and storage medium in the embodiments of this application are respectively electronic devices and storage media applying the above-described intelligent early warning event allocation and handling method. That is, all embodiments of the above-described intelligent early warning event allocation and handling method are applicable to the computer device and storage medium, and can achieve the same or similar beneficial effects. For the computer device / storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple; relevant details can be found in the descriptions of the method embodiments.

[0150] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce a good effect.

[0151] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A method for intelligent early warning event allocation and handling, characterized in that, include: Acquire sensing data within the monitored area, including images, audio, and sensor data; The sensed data is analyzed in real time. When an anomaly is detected, an early warning event is generated. The early warning event includes the event type, the location of occurrence, the reporting time, and a preliminary description. Associate risk profiles with the warning events and calculate the expected completion timeframes for the warning events; The basic reward points are determined based on the risk profile and the expected completion timeframe. Candidate nodes are selected based on the responsibility area matrix of the handling nodes and the location of the incident to obtain a set of candidate nodes; The event offer is passed to the target node, which then generates a commitment vector based on its own private state. The event offer is encapsulated by the expected completion time, the basic reward score, the candidate node set, the offer creation time, and the offer validity period. After the offer expires, all commitment vectors are collected, the optimal node is selected, a task binding notification is sent to the optimal node, and the adjudication result notification is broadcast to the remaining target nodes.

2. The intelligent early warning event allocation and handling method according to claim 1, characterized in that, The specific steps for transmitting an event offer to the target node include: The offer urgency is determined based on the expected completion time limit, and a broadcast mode is selected based on the offer urgency. The broadcast mode includes mode one and mode two. For Mode 1: If the offer urgency is low, then broadcast to all nodes in the candidate node set simultaneously; For Mode 2: If the urgency of the offer is medium or high, a progressive broadcast is initiated. First, the offer is sent only to the node in the candidate node set that is closest to the location of the event. After waiting for a preset time, if no valid commitment is received, the broadcast range is expanded to the three closest nodes in the candidate node set, until the entire candidate node set is covered or a valid commitment is received. The preset time is less than the offer validity period.

3. The intelligent early warning event allocation and handling method according to claim 1, characterized in that, The specific steps for generating a commitment vector based on its own private state include: Each node that receives the offer queries its own historical average processing time for similar early warning events; Calculate the estimated travel time based on the current location of its own node, the location of the event, and real-time traffic data; Treat the warning event as a task to be inserted, and calculate the impact on the completion time of all tasks of the node after inserting it into different positions in the current task queue. Select the insertion position that meets the expectations, and calculate the estimated completion time after the warning event is inserted; The promised completion time is calculated based on the estimated completion time, estimated travel time, and historical average processing time. A commitment vector is generated based on the commitment completion time.

4. The intelligent early warning event allocation and handling method according to claim 3, characterized in that, The specific steps for generating a commitment vector based on the commitment completion time include: Calculate the time difference between the commitment completion time and the offer creation time, and determine whether the time difference exceeds a set threshold, wherein the set threshold = expected completion time × conditional commitment trigger threshold; If so, further analyze whether the node can advance the completion time of the warning event by releasing an existing low-priority task. If so, the node generates a conditional commitment; otherwise, abandon the response. The conditional commitment, commitment completion time, and resource bid integral are encapsulated into a commitment vector, where the resource bid integral = node fatigue × basic reward integral; If not, an unconditional commitment is generated, and the commitment completion time and resource quotation integral are encapsulated into a commitment vector.

5. The intelligent early warning event allocation and handling method according to claim 4, characterized in that, The specific steps for collecting all commitment vectors and selecting an optimal node from them include: Collect all commitment vectors and filter valid commitments, which are commitment vectors whose promised completion time does not exceed twice the expected completion time. Calculate the standardized utility value for each valid commitment, which includes time utility and cost utility; The comprehensive adjudication score is calculated based on the standardized utility value and the node's reputation score; The node with the highest overall decision score is selected as the optimal node.

6. The intelligent early warning event allocation and handling method according to claim 5, characterized in that, The specific steps for selecting the node with the highest overall decision score as the optimal node include: If the valid commitment with the highest overall adjudication score is an unconditional commitment, then the corresponding node is the optimal node. If the valid commitment with the highest overall adjudication score contains a conditional commitment, then a new secondary event offer is created based on the conditional commitment; Within the set offer time window, a broadcast is initiated to the nodes in the neighborhood. The neighborhood is defined as the area centered on the node itself with a set value as the radius. If a secondary commitment is received from another node, the condition of the node itself is met, the node determines that its own valid commitment is valid, and at the same time locks the secondary commitment and sends a task binding notification to the corresponding node. If no secondary commitments are received from other nodes, the node's own valid commitments are deemed invalid and removed. After resolving all conditional commitments, among the remaining valid commitments, the node corresponding to the commitment with the highest overall adjudication score is selected as the optimal node.

7. The intelligent early warning event allocation and handling method according to claim 3, characterized in that, The intelligent early warning event allocation and handling method also includes: When the optimal node begins to handle the early warning event, receive the task status reported by the optimal node; When the task status is "processing completed", record the actual completion time; Calculate the time deviation rate based on the actual completion time, the promised completion time, and the offer creation time; The performance result is determined based on the time deviation rate, and the node's reputation score is updated based on the performance result.

8. A smart early warning event distribution and handling system, characterized in that, The method for intelligent early warning event allocation and handling as described in any one of claims 1 to 7 includes: The data acquisition module is used to acquire sensing data within the monitored area, including images, audio, and sensor data. The data analysis module is used to perform real-time analysis of the sensed data; The early warning module is used to generate an early warning event when the data analysis module detects an anomaly. The early warning event includes the event type, the location of occurrence, the reporting time, and a preliminary description. The early warning and dispatching module is used to associate risk profiles with the early warning events and calculate the expected completion time of the early warning events; to determine the basic reward points based on the risk profiles and the expected completion time; to filter candidate nodes based on the responsibility region matrix and the location of the incident to obtain a candidate node set; to transmit the event offer to the target node and drive the target node to generate a commitment vector based on its own private state, wherein the event offer is encapsulated by the expected completion time, the basic reward points, the candidate node set, the offer creation time, and the offer validity period; and to collect all commitment vectors after the offer validity period expires, select the optimal node from them, send a task binding notification to the optimal node, and broadcast the adjudication result notification to the remaining target nodes.

9. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the intelligent early warning event allocation and handling method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The system stores a computer program that can be loaded by a processor and executed as described in any one of claims 1 to 7 for intelligent early warning event allocation and handling.

Images