An abnormality processing method and system for low-altitude air route based on five-dimensional space-time information

Abstract

The application discloses an abnormality processing method and system for low-altitude air route based on five-dimensional space-time information, when a flight intelligent agent is abnormal and needs to force landing or parachute, five-dimensional space-time information and state information of the abnormal flight intelligent agent are acquired, the five-dimensional space-time information includes an identification code ID, longitude, latitude, height and residual flyable time, the state information includes speed, attitude and health state, a five-dimensional space-time grid covering a current air route is constructed, each grid unit includes longitude, latitude, height, time constraint and digital twin IP address, a force landing or parachute address is determined based on the five-dimensional space-time grid model and constraint conditions, wherein the constraint conditions include aircraft residual energy, reachable time, communication link, flight trajectory and ground safety coefficient, finally, force landing or parachute is performed, and the force landing or parachute address is recalculated and updated in real time with a specified time interval as a period. The force landing or parachute address is determined based on the five-dimensional space-time information, and the safety is high and the real-time performance is good.

Description

Technical Field

[0001] This invention relates to the field of aircraft technology, and in particular to an anomaly handling method and system for low-altitude routes based on 5D spatiotemporal information. Background Technology

[0002] With the rapid development of the low-altitude economy, low-altitude flight paths are becoming increasingly dense, and the number of aircraft such as eVTOLs and drones is growing rapidly, leading to an increasingly complex flight environment. In emergency scenarios such as power failure, communication interruption, flight control anomalies, sudden weather changes, collision risks, collision avoidance requirements, and warnings of unauthorized flights, aircraft need to quickly choose appropriate handling methods, such as returning to base, avoiding collisions, making a safe emergency landing, or having to crash in a specific safe area. If returning to base is chosen, pre-planned or real-time planned flight paths need to be matched to establish an independent return route; if avoiding collisions is chosen, flight parameters need to be changed, especially changes in position, heading, or speed; if making a safe emergency landing is chosen, a specific safe area needs to be found. According to regulations, the true altitude of low-altitude flight includes different spatial layers such as 120 meters, 300 meters, 600 meters, or a future 1000 meters. Landing time is short, and avoidance requires consideration of the impact of changes in position, heading, and speed on different flight layers. Returning to base requires reasonable safety settings between the aircraft and existing planned flight paths. The decision-making process for selecting a safe landing site requires rapid decision-making and judgment within a very short time. Similarly, in the event of an unavoidable crash, the same challenge arises: selecting a safe crash site on the ground. Since the descent speed of an aircraft is typically set at 5 meters per second, a landing maneuver can be completed in one minute at an altitude of 300 meters, and only two minutes at 600 meters. Therefore, the selection and decision-making regarding a safe landing or crash site must be repeatedly and dynamically updated in real-time within milliseconds, culminating in the final safe decision and the termination of the forced landing.

[0003] Chinese patent application CN121281330A discloses a low-altitude flight edge computing-based intelligent scheduling and monitoring system, method, electronic device, and storage medium. It generates control signals based on aircraft setting status data, behavior, and early warning information to control flight. The early warning information is generated by an early warning unit and includes various risk warnings, such as weather warnings, flight agent malfunction warnings, airspace foreign object warnings, flight path collision risk warnings, and public security system anti-organized crime warnings. However, it does not address the behavioral patterns during the handling process, such as the dynamic real-time selection of safe ground locations for forced landings and crashes.

[0004] Existing methods for selecting landing sites or crash locations are mostly based on two-dimensional or three-dimensional space, lacking the temporal dimension and the unique attribute of digital twin IP network addresses to map unique aircraft identification codes for accurate identification. They also lack information on remaining energy, remaining loiter time, and the dynamic trajectory of normal aircraft, making it impossible to pinpoint specific identifiable aircraft and potentially leading to unreachable or aerial collisions. Current aircraft decision-making methods are mostly based on single-aircraft autonomous decision-making, lacking multi-agent collaboration with edge cloud, normal flight intelligent agents, and ground rescue units. This increases the risk of multi-aircraft collisions or secondary ground risks, and also fails to provide real-time dynamic route updates. When the environment changes abruptly, malfunctions worsen, or airspace occupancy changes, millisecond-level replanning is not possible. Furthermore, existing aircraft decision-making systems lack sufficient integration with new infrastructure such as 6G low-altitude intelligent networks, integrated sensing, computing, and intelligent control, and digital twins. This results in incomplete perception, delayed decision-making, low reliability, and the inability to identify specific spacecraft identification codes, making it difficult to achieve "visibility, tangibility, and control." Furthermore, it is impossible to achieve real-time synchronization of the physical status of the aircraft with its digital twin, resulting in a significant reduction in the accuracy of airspace control and the effectiveness of emergency decision-making. Existing aircraft decision-making systems lack modern AI large-scale model (LLM), intelligent human-machine interaction for emergency disaster management, and an agile and rapid response mechanism for handling forced landings or accidental crashes. Summary of the Invention

[0005] To address some or all of the problems in existing technologies, and in order to improve the accuracy of airspace control and the efficiency of emergency decision-making in handling unexpected situations, thereby meeting the real-time, dynamic, high-density, and highly complex safety requirements of low-altitude flights, the first aspect of this invention provides an anomaly handling method for low-altitude routes based on 5D spatiotemporal information. This method determines the location of a forced landing or crash through multi-agent collaborative processing, including: When a flight agent malfunctions and needs to make an emergency landing or crash, the system acquires the 5-dimensional spatiotemporal information and status information of the malfunctioning flight agent. The 5-dimensional spatiotemporal information includes identification code ID, longitude, latitude, altitude, and remaining flight time. The status information includes speed, attitude, and health status. Construct a 5-dimensional spatiotemporal grid covering the current flight path, where each grid cell includes longitude, latitude, altitude, time constraints, and a digital twin IP address; The forced landing or crash location is determined based on the 5D spatiotemporal grid model and constraints, wherein the constraints include the aircraft's remaining energy, reachability time, communication link, flight trajectory, and ground safety factor. Perform a forced landing or crash, and during the forced landing or crash, recalculate and update the forced landing or crash address in real time at specified time intervals.

[0006] Furthermore, the time constraint includes: The time window that the abnormal flight agent's remaining energy can reach; A time conflict occurs when a normally flying intelligent agent passes through this airspace; The duration of stable availability of the communication link.

[0007] Furthermore, the exception handling method also includes: The edge agent synchronizes the status, constraints, and forced landing or crash address of the abnormal flight agent to the ground control agent for redundancy backup.

[0008] Furthermore, determining the forced landing or crash location based on the 5D spatiotemporal grid model and the constraints includes: The pre-trained prediction model, based on historical data or preset data, combined with the 5-dimensional spatiotemporal grid model and constraints, determines the location of a forced landing or crash.

[0009] Furthermore, determining the forced landing or crash location based on the 5D spatiotemporal grid model and the constraints also includes: The predicted landing or crash locations output by the model are evaluated to determine spatial accessibility, ground safety, airspace conflict risk, rescue accessibility, and time accessibility. Based on the aforementioned spatial accessibility, ground safety, airspace conflict risk, rescue accessibility, and time accessibility, the location of the forced landing or crash is iteratively optimized until the preset requirements are met.

[0010] Furthermore, the specified time interval is determined based on the computing power and communication rate of the multi-agent system, so that at least two calculations of the forced landing or crash address are performed during the forced landing or crash process.

[0011] Furthermore, the exception handling method also includes: The behavior pattern of the flight agent is determined based on the cause of the anomaly, wherein the behavior pattern includes returning to home, avoiding obstacles, forced landing, and crashing.

[0012] Furthermore, the exception handling method also includes: The ground-based control agent controls the normally flying agent to automatically avoid the emergency landing or crash based on the address of the crash, in order to open a safe temporary passage.

[0013] Furthermore, the exception handling method also includes: By connecting ground-based emergency intelligent agents with ground safety factor assessment and emergency resource dispatch.

[0014] A second aspect of the present invention provides an anomaly handling system for low-altitude air routes based on 5D spatiotemporal information, comprising: A multi-source sensing and communication module is used to acquire the status and environmental information of the flying intelligent agent; The 5D spatiotemporal grid modeling module is used to divide the airspace of flight routes into 5D grid cells with time windows. The forced landing or crash address prediction module is used to determine the forced landing or crash address based on constraints.

[0015] This invention provides a method and system for handling anomalies in low-altitude flight paths based on 5D spatiotemporal information. By introducing the time dimension, it ensures that the emergency landing or crash site is truly reachable within the remaining energy and time, significantly improving the success rate of emergency landings. Simultaneously, by mapping the digital twin IP address to the aircraft's unique identification code, the aircraft possesses a unique identity within the communication network, enabling precise location and identification. When selecting an emergency landing or crash site, based on historical or preset data, it prioritizes avoiding dangerous areas such as crowds, buildings, traffic, high-voltage lines, and chemical plants, significantly reducing ground risks and ensuring high safety.

[0016] By employing multi-agent collaboration—comprising intelligent agents for abnormal flight, edge agents, agents for normal flight, ground control agents, and ground emergency agents—the aforementioned anomaly handling method can automatically avoid agents for normal flight, prevent spatiotemporal conflicts, and prepare rescue forces in advance, thereby improving the safety of low-altitude emergency landings by an order of magnitude. Simultaneously, multi-agent collaboration reduces decision-making latency to less than 100 milliseconds, making it applicable to high-density low-altitude urban scenarios.

[0017] The aforementioned anomaly handling method continuously refreshes the forced landing or crash address at specified time intervals, thereby addressing dynamic scenarios such as fault deterioration, sudden weather changes, and communication fluctuations, ensuring that the entire forced landing process is safe and controllable.

[0018] The described anomaly handling method and system are fully compatible with 6G sensing integration, various spatial positioning systems, self-organizing networks, and low-altitude intelligent network digital twins, enabling emergency landing decisions to have stable perception and highly reliable communication support across the entire link, making them suitable for future large-scale low-altitude flight scenarios. Specifically, the ground control intelligent agent monitors the overall situation in real time through the 6G communication network, quickly transferring the situation to the edge intelligent agent for handling in the event of a fault and coordinating with other flight intelligent agents to maintain normal flight order; it also collaboratively ensures that the abnormal flight intelligent agent dynamically and in real time selects and executes a 5D emergency landing or / or crash safety address within a short period. Ground management personnel, through LLM intelligent interaction, collaborate with the ground emergency intelligent agent to quickly assess rescue accessibility and simultaneously activate emergency resources such as fire, medical, and transportation, minimizing the risk of secondary disasters.

[0019] The aforementioned anomaly handling method features adjustable constraints and a flexible architecture, making it suitable for various low-altitude flight paths, including eVTOL cargo and manned drones and general aviation aircraft. It can be widely applied in urban air traffic (UAM), low-altitude inspection, and short-haul general aviation transportation, and has significant promotional value. Attached Figure Description

[0020] To further illustrate the above and other advantages and features of the various embodiments of the present invention, a more specific description of the various embodiments of the present invention will be presented with reference to the accompanying drawings. It is to be understood that these drawings depict only typical embodiments of the invention and are therefore not intended to limit its scope. In the drawings, identical or corresponding parts will be indicated by identical or similar reference numerals for clarity.

[0021] Figure 1 This is a flowchart illustrating an anomaly handling method for low-altitude air routes based on 5D spatiotemporal information according to an embodiment of the present invention. Figure 2 This diagram illustrates a process for managing the transition from normal flight state to abnormal flight state according to an embodiment of the present invention. Figure 3 This diagram illustrates a process for determining the location of a forced landing or crash based on 5D spatiotemporal information according to an embodiment of the present invention. Figure 4 This diagram illustrates the AI ​​processing mechanism for the latest time point forced landing or crash address option in one embodiment of the present invention. Figure 5 This diagram illustrates the structure of a multi-agent cooperative system according to an embodiment of the present invention. Figure 6 This diagram illustrates the structure of an anomaly handling system for low-altitude air routes based on 5D spatiotemporal information, according to an embodiment of the present invention. Figure 7 This diagram illustrates the framework of a distributed route data information infrastructure base—a real-time edge computing security and intelligent management system for route data networks—and a municipal-level low-altitude service management center, according to an embodiment of the present invention. Figure 8 This diagram illustrates the global deployment of a three-tiered low-altitude service management center (national-regional-city level) according to an embodiment of the present invention. Detailed Implementation

[0022] In the following description, the invention is described with reference to various embodiments. However, those skilled in the art will recognize that the embodiments may be practiced without one or more specific details or in conjunction with other alternatives and / or additional methods or components. In other instances, well-known structures or operations are not shown or described in detail so as not to obscure the inventive points of the invention. Similarly, for illustrative purposes, specific numbers and configurations are set forth to provide a comprehensive understanding of embodiments of the invention. However, the invention is not limited to these specific details. Furthermore, it should be understood that the embodiments shown in the drawings are illustrative representations and are not necessarily drawn to scale.

[0023] In this specification, references to "an embodiment" or "this embodiment" mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment of the invention. The phrase "in one embodiment" appearing throughout this specification does not necessarily refer to the same embodiment in all instances.

[0024] It should be noted that the embodiments of the present invention describe the method steps in a specific order; however, this is only for illustrating the specific embodiment and not for limiting the order of the steps. On the contrary, in different embodiments of the present invention, the order of the steps can be adjusted according to actual needs.

[0025] To meet the real-time, dynamic, high-density, and highly complex safety requirements of low-altitude flights, and to dynamically and in real-time select and execute safe landing or crash site selection decisions based on ground safety after a warning, this invention provides an anomaly handling method and system for low-altitude routes based on 5D spatiotemporal information. It can accurately select and decide on reachable, safe, and rescuable safe landing or crash sites within a 5D spatiotemporal grid of longitude, latitude, altitude, time, and a digital twin network IP address representing a unique identifier for the physical flight agent. The method and system achieve distributed multi-agent collaborative decision-making through anomaly agents, edge agents, ground control agents, normal flight agents, and ground emergency agents, as well as the execution of safe address selection decisions associated with landing or crashes and real-time emergency handling.

[0026] Flight intelligent agents refer to aircraft with embodied intelligence based on the dedicated network on the flight path.

[0027] The ground control intelligent agent refers to a multi-modal perception and high-computing-power AI system based on the dedicated network on the flight route, responsible for managing the normal flight of numerous flight intelligent agents on the flight route. The edge intelligent agent refers to a multi-modal perception and high-computing-power AI system based on the dedicated network on the flight route, primarily used to receive and manage abnormal flight intelligent agents and execute related operational modes: return to base, avoidance, crash, and emergency landing. Flight behavior on the flight route can be divided into normal flight and abnormal flight. In the embodiments of this invention, the ground control intelligent agent is responsible for managing normal flight intelligent agents, and the edge intelligent agent is responsible for managing abnormal flight intelligent agents. Once any one or more flight intelligent agents in the normal flight intelligent agent group experience any unexpected malfunction, or external conditions require handling, or the competent authority specifically requests a change from normal flight status to abnormal flight status, the ground control intelligent agent immediately transfers the corresponding abnormal flight intelligent agent to the edge intelligent agent. The edge intelligent agent receives the abnormal flight intelligent agent transferred by the ground control intelligent agent in real time and performs corresponding handling. This allows the ground control intelligent agent to continue maintaining the normal management of the normal flight intelligent agent group on the flight route. In addition, ground control agents are required to assist or provide collaborative emergency response capabilities to agents in abnormal flight operations during normal flight management tasks.

[0028] Ground-based emergency intelligent agents refer to edge agents managing abnormal flight intelligent agents on the flight service center's unified multi-line emergency response AI system. These agents collaborate with the ground-based emergency intelligent agents at the flight service center to handle various behavioral patterns of abnormal flight intelligent agents, including but not limited to return to base, avoidance maneuvers, emergency landings, and crashes. If an emergency landing or crash is determined, a safe and reasonable choice and decision-making process for the landing or crash site is performed to minimize the harmfulness of the accident and avoid secondary disasters as much as possible. Simultaneously, the emergency intelligent agent enables real-time dynamic emergency command and response for various resources, such as emergency management of hospitals, fire departments, and transportation resources.

[0029] In one embodiment of the present invention, without affecting the time delay and urgency of the response, when the computing power of the abnormal flight agent, the ground control agent, and the edge agent is insufficient, they can further leverage the computing power of the ground emergency agent and the cloud-deployed LLM large model for collaborative computation to improve the decision-making quality and response capability in complex emergency scenarios. The public cloud AI large model LLM is connected with the ground emergency agent, the flight route-related agent, the edge agent, and the ground control agent in the low-altitude service management center. Through intelligent interaction of LLM information, managers can enable the ground emergency agent to quickly and in real-time reflect the manager's intentions into command and management applications, such as related hospital rescue, on-site potential fire fighting, and the opening of convenient emergency transportation green channels, to prevent secondary disasters.

[0030] The system integrates with new infrastructures such as 6G low-altitude intelligent sensing network or similar three-dimensional spatial information network, integrated sensing, computing and intelligent control, and digital twins. It has comprehensive perception, fast response and high reliability. It can realize real-time dynamic updates of flight routes to cope with sudden changes in the internal and external environment, including the deterioration of flight intelligent agent failure, unexpected changes in the occupation of flight airspace, and millisecond-level dynamic real-time replanning of flight route safety emergency landing and / or crash address in case of crash.

[0031] The technical solution of the present invention will be further described below with reference to the accompanying drawings of the embodiments.

[0032] Figure 1 This diagram illustrates a flowchart of an anomaly handling method for low-altitude air routes based on 5D spatiotemporal information, according to an embodiment of the present invention. Figure 1 As shown, an anomaly handling method for low-altitude air routes based on 5D spatiotemporal information includes: First, in step 101, status monitoring. During the flight of the intelligent flying vehicle, its status is monitored in real time. If an abnormality occurs, proceed to step 102 to determine the behavior pattern. Figure 2 This diagram illustrates a process for managing the transition from normal flight state to abnormal flight state according to an embodiment of the present invention. Figure 2As shown, in a 6G low-altitude intelligent network or similar three-dimensional information space network architecture, the normal flight intelligent agent interacts with the ground control intelligent agent in real time through an integrated air-space-ground-sea network. This includes: the flight intelligent agent reporting flight data, such as 5-dimensional spatiotemporal status [longitude X, latitude Y, altitude Z, remaining flight time T, corresponding unique digital twin IP address], attitude, speed, battery level, and flight path intention. The ground control intelligent agent authenticates the identity of the normal flight intelligent agent and shares meteorological information, collision risk information, public security query requirements, geographic information, and RTK precise positioning data. It manages normal flight according to a pre-set normal flight list, ensuring that the normal flight intelligent agent successfully completes its flight mission. All compliant aircraft are included in the ground control intelligent agent's "normal flight list" for unified management.

[0033] When the flight agent malfunctions due to its own faults or external environmental constraints, such as sudden severe weather, electromagnetic interference, foreign objects in the airspace, or sudden collision risks, or when required by authorities, such as temporary no-fly orders from public security departments, air traffic control, or special handling requirements from authorities, the ground control agent will dynamically reassess the flight agent's status in real time. If the assessment results determine that the flight agent no longer meets the conditions for normal flight, such as deviating from the approved flight path, entering a temporary no-fly zone, receiving a forced landing order, or encountering uncontrollable environmental disturbances, the system will trigger a transition from normal flight status to abnormal flight status.

[0034] In one embodiment of the present invention, when a normally flying intelligent agent switches from a normal flight state to an abnormal flight state due to various unforeseen circumstances, or when an unforeseen event occurs due to an internal capability failure, it will first report the abnormal state and abnormal level to the ground control intelligent agent, and its 5-dimensional status will be marked as abnormal instead of normal. Simultaneously, relevant information provided by the flight service center and the public network system regarding adverse external weather conditions leading to flight anomalies, or requests from national government authorities such as public security requiring a switch from a normal flight state to an abnormal flight state, will be shared with the flight intelligent agent to change its status. At this time, the ground control intelligent agent removes the abnormal flight intelligent agent from the "Normal Flight Route List" and transfers it to an edge intelligent agent for "Abnormal Flight Route List" management. The abnormal flight intelligent agent's interaction behavior changes from that of the ground control intelligent agent to that of the edge intelligent agent for corresponding information interaction. In one embodiment of the present invention, the information transferred to the edge intelligent agent includes surrounding airspace situation and global perception data, such as ground environment, weather, and obstacles. The abnormal flight intelligent agent is hosted by the edge intelligent agent, with surrounding normal flight intelligent agents providing collaborative support. Within the dedicated network on the flight path side, edge agents manage abnormal flight behavior, while ground control agents manage normal flight behavior. Simultaneously, edge agents and ground control agents perform mutual redundancy backups to improve system safety and stability. Edge agents synchronize and coordinate with ground control agents to perform redundant safety backups of the real-time status of abnormal flight management agents, emergency resource scheduling, and real-time periodic optimal 5D forced landing and / or crash safety addresses. Next, in step 102, the behavior pattern is determined. Based on the anomaly type, such as environmental constraints, permission intervention, or its own malfunction, the edge agent issues anomaly management operation instructions to the anomalous flight agent. In one embodiment of the invention, through multi-agent collaboration, it dynamically and in real-time determines whether a return to base or avoidance is possible and requires the anomalous flight agent to execute accordingly, or to perform a forced landing or crash. After ruling out return to base and / or avoidance, if a forced landing or crash is required, the process proceeds to step 103, where the anomalous flight agent's status is obtained, and a ground safety address selection is performed based on this status. Next, in step 103, the status of the abnormal intelligent flight entity is obtained. In one embodiment of the present invention, if an emergency landing or crash is required, a real-time dynamic periodic optimal 5D emergency landing or crash safety address selection and execution is performed. First, the 5D spatiotemporal information and status information of the abnormal intelligent flight entity need to be obtained. The 5D spatiotemporal information includes an identification code ID or its corresponding unique network IP address, longitude, latitude, altitude, and remaining flight time. The status information includes speed, attitude, and health status. In one embodiment of the present invention, the 5D status of the abnormal flight intelligent entity is obtained in real time through multimodal perception, a 6G sensory spatial information network or similar network, spatial positioning, and the aircraft's unique identification code ID. It interacts with the ground-based edge intelligent entity on the flight path side, and the ground-based control intelligent entity on the flight path side and the edge intelligent entity synchronously form a safety redundancy backup mechanism. Next, in step 104, the emergency landing or crash site is determined. As mentioned earlier, the descent time at an altitude of 300 meters is approximately 1 minute, and at 600 meters, it is only 2 minutes. In order to quickly pinpoint the final safe emergency landing and / or crash site within a short time, in one embodiment of the present invention, a 5-dimensional spatiotemporal grid is used, namely, three-dimensional spatial longitude X, latitude Y, altitude Z, and remaining emergency landing time T, and the unique IP address of the digital twin corresponding to the unique identification code ID of the flight agent, as a 5-dimensional spatiotemporal grid element to participate in the selection of the safe emergency landing and / or crash site in real time.

[0035] Figure 3 This diagram illustrates a process for determining the location of a forced landing or crash based on 5D spatiotemporal information, according to an embodiment of the present invention. Figure 3 As shown, determining the location of a forced landing or crash includes: First, a 5-dimensional spatiotemporal grid covering the current flight path is constructed. Each grid cell contains spatial coordinates of longitude (X), latitude (Y), and altitude (Z), a time window attribute, and a unique identification code ID or corresponding unique network IP address for the aircraft. In one embodiment of the invention, the time window attribute is used to constrain the time window within which the remaining energy of the abnormal flight agent can reach, the time conflict of the normal flight agent passing through the airspace, and the stable availability time of the communication link. The unique identification code representing the physical space flight agent in the 5-dimensional spatiotemporal grid can be mapped one-to-one with the unique IP attribute characteristics of the digital twin network, thereby clearly and accurately indicating the specific unique identification code belonging to the spacecraft, and accurately realizing effective and orderly management of low-altitude air traffic safety that is "visible, tangible, and manageable." By physically binding the network characteristic values ​​of the communication module embedded in the SIM and / or eSIM chip in the mobile network with its serial number, mobile phone number, mobile communication routing and switching network characteristic values, and the interface MAC address and / or SN serial number of the avionics computer with the unique identification code characteristic information of the aircraft as stipulated by the state, this serves as a unique attribute mapped one-to-one with the network IP address corresponding to the digital twin world. The digital twin uses a unique network IP address as its representative to map to a unique identifier for the physical flight agent, thus achieving a one-to-one mapping between the unique identifier bound to the physical flight agent and the network IP address represented by the digital twin in the mobile communication network. Furthermore, throughout the entire three-dimensional information space, the digital twin IP address can be used to map to the unique ID number of the corresponding physical space flight agent, achieving a one-to-one correspondence of unique attributes. Next, a 5D grid-based real-time dynamic process for selecting and executing emergency landing or crash locations is implemented, using historical data, preset data, or the latest contingency plan data generated by real-time AI analysis. A pre-trained predictive model, such as an AI algorithm model, determines the emergency landing or crash location based on the 5D spatiotemporal grid model and constraints, including the aircraft's remaining energy, reachability time, communication link, flight trajectory, and ground safety factor. In one embodiment of the invention, the accumulated emergency landing or crash location options generated during normal flight and the final historical safe addresses of emergency landings or crashes are combined to form a total historical big data set. This data is then analyzed by an AI agent to generate a contingency plan for the next latest time point, saving real-time dynamic selection and decision-making time in the process of edge agents managing abnormal flight situations, maximizing efficiency, and minimizing or avoiding the severity of disasters. By leveraging historical big data and AI applications to generate the latest time-based options for emergency landings or crashes, a closed-loop emergency response system can be built, encompassing multi-agent perception, collaboration, decision-making, execution, and emergency handling. This significantly improves the safety of emergency landings or crashes for low-altitude flight agents, minimizes environmental impact, and enables rapid and agile response, reducing the negative impact of emergencies to the greatest extent possible. Figure 4 This diagram illustrates the AI ​​processing mechanism for the latest time-node forced landing or crash address option formation according to an embodiment of the present invention. Figure 4 As shown, the AI ​​algorithm model is mainly implemented through the ground control agent's normal flight management and the edge agent's abnormal flight management. During the normal flight management process, the ground control agent internally absorbs historical forced landing or crash address information generated from past accidents, as well as historical forced landing or crash address information accumulated during its continuous safe flight. It then outputs the latest time-node contingency plan information as the initial input source for the abnormal flight management. During the abnormal flight management process, the edge agent uses a multi-agent collaborative mode to dynamically select 5-dimensional safe forced landing and / or crash addresses in real time. The generated 5-dimensional forced landing or crash addresses serve as input sources, entering the "Forming a Historical Safe Forced Landing and / or Crash Address Alternative Database" module in the ground control agent's normal flight management flowchart. This allows for continuous AI training and iteration, preparing the latest node's forced landing or crash address contingency plan information for the next abnormal flight. Finally, a 5D spatiotemporal grid is applied, and an AI model is used to quantitatively analyze the emergency landing or crash sites determined in the preceding steps. This includes, but is not limited to, safety factor assessment. The assessment dimensions include: spatial accessibility, ground safety, airspace conflict risk, rescue accessibility, and time accessibility. Based on these factors, the emergency landing or crash sites are iteratively optimized until preset requirements are met. The globally optimal 5D emergency landing or crash site (X0, Y0, Z0, T0, IP0) ​​is then output and sent to the abnormal flight agent aircraft to execute the emergency landing or crash. The IP0 address represents the aircraft's unique identification code ID0. Spatial accessibility refers to the remaining altitude and energy allowing gliding or drifting to the destination. Ground safety refers to the absence of buildings, crowds, traffic, high-voltage lines, chemical plants, or dangerous areas on the ground. Airspace safety refers to the absence of other flight agents occupying the airspace, and normal flight agents have already given way. Time accessibility refers to the ability to reach the destination within the remaining flight time. Rescue accessibility refers to the proximity of the emergency landing or crash site to a road, allowing for rapid arrival of emergency forces.

[0036] In one embodiment of the present invention, such as Figure 5As shown, a multi-agent collaborative system is composed of an abnormal flight agent, an edge agent, a normal flight agent, a ground control agent, and a ground emergency agent. This system facilitates information exchange, conflict prediction, and collaborative avoidance. The ground emergency agent interfaces and collaborates with the LLM (Large-Scale Management) model. The abnormal flight agent reports abnormal flight management information to the edge agent, including but not limited to abnormal states, a 5D spatiotemporal grid, and the abnormal flight agent's remaining capabilities. The edge agent performs global situational awareness, airspace coordination, and generates emergency landing or crash locations. It maintains synchronization and redundancy with the ground control agent managing normal flight routes, which is configured with a safety redundancy mechanism. Simultaneously, the edge agent interfaces with the ground emergency agent to assess ground safety and schedule emergency resources. Under the management of the ground control agent, the normal flight agent provides automatic avoidance, opens temporary safe passages, or interacts with the abnormal flight agent as required by the edge agent's abnormal flight management. The ground control agent receives abnormal reports from flight agents, and / or the risk of severe weather and / or requests from authorities, and / or any other unexpected situations that cause normal flight to transition to abnormal flight. The ground control agent then handles the transition, transferring management to the edge agent. The ground control agent manages all other normal flight agents remaining on the flight path. A safety redundancy backup mechanism is established between the ground control agent and the edge agent for synchronized data backup. The ground emergency agent interfaces with the internal edge agent on the flight path, providing real-time dynamic 5D grid-based emergency landing or crash site selection, including address safety assessment and emergency resource scheduling. Emergency resource scheduling includes, but is not limited to, hospitals, fire departments, and transportation. Once the edge agent confirms the final selection of a safe 5D grid-based emergency landing and / or crash site, it completes information exchange, conflict prediction and coordinated avoidance, assessment of rescue accessibility, and synchronized rescue preparation. Furthermore, the ground emergency agent interfaces with the LLM (Limited Least Metric) model, enabling managers to achieve intelligent interaction and processing through LLM interaction. Finally, in step 105, the forced landing or crash address is dynamically updated. During the forced landing or crash process, the forced landing or crash address is recalculated and updated in real time at specified time intervals, forming a dynamic closed loop. In one embodiment of the present invention, the specified time interval is determined based on the computing power and communication rate of the multi-agent system, so that the forced landing or crash address is calculated at least twice during the forced landing or crash process. Under current hardware conditions, the specified time interval is preferably 50 to 100 milliseconds.

[0037] In the anomaly handling method provided by this invention, the edge agent provides the abnormal flight agent with the latest emergency landing or crash safety address plan at the latest time node during abnormal flight. The abnormal flight agent dynamically and in real time selects the optimal 5-dimensional emergency landing or crash safety address periodically according to the plan. The ground emergency agent performs ground safety scoring and resource readiness status assessment on the ground safety address selected by the abnormal flight agent in real time and feeds it back to the edge agent. The edge agent and the abnormal flight agent combine the above information to dynamically and in real time select the optimal 5-dimensional emergency landing or crash safety address periodically, determine the final safety address selection plan, and execute it. When the edge agent manages the abnormal flight agent and obtains the final 5-dimensional emergency landing or crash safety address conclusion, the abnormal flight agent performs an emergency landing or crash based on the safety address. The edge agent interacts and shares the emergency landing or crash safety address conclusion with the ground emergency agent, and the ground emergency agent performs emergency linkage processing, including but not limited to hospitals, fire departments, and transportation departments. Simultaneously, the ground emergency agent interacts and collaborates with the public network LLM model. Management personnel establish an interface with the LLM model and submit relevant handling instructions to the LLM model for execution in conjunction with the ground emergency agent. At the same time, the ground emergency agent feeds back relevant emergency handling status information to management personnel through the LLM model. It can be seen that a comprehensive collaborative relationship is formed between the normal flight agent and the ground control agent, the abnormal flight agent and the edge agent, the ground control agent and the edge agent, and the various agents within the dedicated network on the flight line interacting with the public network of the flight service center's ground emergency agent. Furthermore, if the computing power of the various agents within the flight line is insufficient, the computing power of the flight service center agent and the public network LLM model can be utilized. This multi-level collaborative decision-making can significantly reduce the probability of secondary accidents, and its real-time dynamic characteristics can achieve millisecond-level update speeds.

[0038] By linking the selection of the ground address for an emergency landing or crash with a 5D spatiotemporal grid model and exchanging information such as the aircraft's remaining energy, reachability time, communication link, and trajectory, the emergency landing or crash address can be reached within the actual remaining time window. The emergency landing airspace will not conflict with other aircraft in terms of time or airspace, and the emergency landing process can be dynamically updated in real time based on a comprehensive assessment of time and ground safety factors, thereby improving the realism and safety of the emergency landing.

[0039] The anomaly handling method provided by this invention adds stringent constraints to low-altitude flight routes for eVTOL cargo and manned aircraft, such as the national airspace classification true altitude divisions of 120 meters, 300 meters, 600 meters, and even the future 1000 meters. These constraints include, but are not limited to: avoiding densely populated areas, prohibiting landing in buildings, traffic, and high-voltage lines, mandatory scoring of rescue accessibility, and reducing the risk of secondary ground disasters. Thus, with stricter safety constraints, the decision-making process is more aligned with the actual needs of cargo and manned flight.

[0040] Figure 6 This diagram illustrates the structure of an anomaly handling system for low-altitude air routes based on 5D spatiotemporal information, according to an embodiment of the present invention. Figure 6 As shown, an anomaly handling system for low-altitude airways based on 5D spatiotemporal information includes a multi-source sensing and communication module 601, a 5D spatiotemporal grid modeling module 602, and an emergency landing or crash address prediction module 603. The multi-source sensing and communication module acquires the state and environmental information of the flight agent; the 5D spatiotemporal grid modeling module divides the airspace of the flight route into 5D grid cells with time windows; and the emergency landing or crash address prediction module determines the emergency landing or crash address based on constraints.

[0041] Figure 7 This diagram illustrates the framework of a distributed route data information infrastructure base—a real-time edge computing security and intelligent management system for route data networks—and a municipal-level low-altitude service management center, according to an embodiment of the present invention. Figure 8 This diagram illustrates the global deployment of a three-tiered low-altitude service management center architecture (national-regional-city level) according to an embodiment of the present invention. It clearly demonstrates the logical connection system between the route-side architecture and the city-level low-altitude service management center, as well as the simplified three-tiered management architecture that connects the city-level low-altitude service management center to the regional-level low-altitude service management center, ultimately forming the national-level integrated low-altitude service management center. The core requirements for ensuring safe and orderly intelligent management and scheduling of low-altitude routes, as well as timely and safe intelligent response to any unforeseen events, are ensuring safe and orderly intelligent management and scheduling of low-altitude flight routes.

[0042] Although various embodiments of the invention have been described above, it should be understood that they are presented by way of example only and not as limitations. It will be apparent to those skilled in the art that various combinations, modifications, and alterations can be made without departing from the spirit and scope of the invention. Therefore, the breadth and scope of the invention disclosed herein should not be limited by the exemplary embodiments disclosed above, but should be defined solely by the appended claims and their equivalents.

Claims

1. An anomaly handling method for low-altitude air routes based on 5D spatiotemporal information, characterized in that, Determining the location of a forced landing or crash through multi-agent collaboration includes: When a flight agent malfunctions and needs to make an emergency landing or crash, the system acquires the 5-dimensional spatiotemporal information and status information of the malfunctioning flight agent. The 5-dimensional spatiotemporal information includes identification code ID, longitude, latitude, altitude, and remaining flight time. The status information includes speed, attitude, and health status. Construct a 5-dimensional spatiotemporal grid covering the current flight path, where each grid cell includes longitude, latitude, altitude, time constraints, and a digital twin IP address; The forced landing or crash location is determined based on the 5D spatiotemporal grid model and constraints, wherein the constraints include the aircraft's remaining energy, reachability time, communication link, flight trajectory, and ground safety factor. The aircraft is forced to land or crash, and during the forced landing or crash, the address is dynamically recalculated and updated in real time at specified time intervals.

2. The anomaly handling method as described in claim 1, characterized in that, The time constraints include: The time window in which the remaining energy of the abnormal flight agent can be reached; A time conflict occurs when a normally flying intelligent agent passes through the airspace where an abnormally flying intelligent agent is located. The stable availability time of the communication link.

3. The anomaly handling method as described in claim 1, characterized in that, Also includes: The status, constraints, and forced landing or crash addresses of abnormal flight agents are synchronized to the ground control agent through edge agents for redundancy backup.

4. The anomaly handling method as described in claim 1, characterized in that, Determining the forced landing or crash location based on the 5D spatiotemporal grid model and the constraints includes: The pre-trained prediction model, based on historical data or preset data, combined with the 5-dimensional spatiotemporal grid model and constraints, determines the location of a forced landing or crash.

5. The anomaly handling method as described in claim 4, characterized in that, Determining the forced landing or crash location based on the aforementioned 5D spatiotemporal grid model and the aforementioned constraints also includes: The predicted landing or crash locations output by the model are evaluated to determine spatial accessibility, ground safety, airspace conflict risk, rescue accessibility, and time accessibility. Based on the aforementioned spatial accessibility, ground safety, airspace conflict risk, rescue accessibility, and time accessibility, the location of the forced landing or crash is iteratively optimized until the preset requirements are met.

6. The anomaly handling method as described in claim 4, characterized in that, The specified time interval is determined based on the computing power and communication rate of the multi-agent system, so that at least two forced landing or crash addresses are calculated during the forced landing or crash process.

7. The anomaly handling method as described in claim 1, characterized in that, Also includes: The behavior pattern of the flight agent is determined based on the cause of the anomaly, wherein the behavior pattern includes returning to home, avoiding obstacles, forced landing, and crashing.

8. The anomaly handling method as described in claim 3, characterized in that, Also includes: Based on the location of the forced landing or crash, the ground control agent controls the normally flying agent to automatically avoid the crash and create a safe temporary passage.

9. The anomaly handling method as described in claim 1, characterized in that, Also includes: By connecting ground-based emergency intelligent agents with ground safety factor assessment and emergency resource dispatch.

10. An anomaly handling system for low-altitude air routes based on 5D spatiotemporal information, characterized in that, include: A multi-source sensing and communication module is configured to acquire the state and environmental information of the flying intelligent agent; The 5D spatiotemporal grid modeling module is configured to divide the airspace of flight routes into 5D grid cells with time windows. The forced landing or crash address prediction module is configured to determine the forced landing or crash address based on constraints.

Images