A flight guarantee closed-loop management and control method and system based on petri net and dynamic weight

Abstract

This invention relates to a closed-loop management method and system for flight support based on Petri nets and dynamic weights, addressing the problems of existing technologies such as reliance on black boxes, inability to capture concurrent bottlenecks, and fragmented early warning and scheduling. The method includes: establishing a Petri net mapping of the physical support process and dynamically extrapolating time deviation propagation variables; constructing a nonlinear equation by integrating structural correlation, token congestion coefficient, and deviation sensitivity to dynamically track critical path weights; generating an adaptive early warning red line based on airport wave operation pressure to identify key bottlenecks when the comprehensive delay risk exceeds the limit; issuing control signals to the bottleneck to inject intervention tokens, and calculating the deviation reduction amount through a logarithmic negative feedback equation to perform closed-loop optimization. This invention achieves a shift from post-event prediction to in-event closed-loop control, improving the system's anti-disturbance capability.

Description

Technical Field

[0001] This invention relates to the field of civil aviation air traffic management and airport operation control technology. Specifically, it relates to a method for early warning and control of ground support delays for transit flights based on Petri net state evolution, dynamic critical path weight identification, and physical scheduling negative feedback closed loop. Background Technology

[0002] Ground handling services are the core link between flight arrivals and departures, directly affecting the overall operational efficiency and on-time performance of an airport. Modern civil aviation airport ground handling operations are a complex network composed of numerous highly coupled micro-physical components. These components are subject to strict safety logic and sequential constraints, as well as complex parallel and collaborative relationships.

[0003] In recent years, with the development of smart civil aviation, the industry has proposed several data-driven flight delay prediction methods. Existing mainstream technologies typically rely on collecting historical time data from transit processes, extracting features, and inputting this data into a black-box model to predict the final delay minutes of a target flight. However, with the increasing demands for refined airport operation and management, existing technologies have gradually revealed the following significant defects and limitations: First, the over-reliance on data-driven and black-box models lacks interpretability of the physical state evolution: Most existing predictive models treat transit support as an end-to-end fitted black box, focusing only on the absolute duration of the final delay, without clearly mapping the evolution of the underlying support operation status. Once a delay warning occurs, managers find it difficult to reverse-engineer and accurately trace which micro-level support node's resource shortage or execution anomaly caused the delay propagation. This approach, lacking physical interpretability, cannot meet the stringent traceability requirements of civil aviation operations for safety control and liability determination.

[0004] Second, neglecting the buffering and cascading effects under parallel topologies leads to delayed or distorted early warning mechanisms: Station transit assurance is highly concurrent, and some non-critical path links have inherent "time margins." Minor time delays are buffered by the system and do not directly affect the final departure. However, once this margin is exhausted, it instantly becomes a bottleneck hindering the entire system. Existing technologies often use indiscriminate static weight allocation or simply compare the actual time consumption of each node with the standard time difference. This static logic cannot dynamically identify critical path transitions based on the congestion of the underlying physical services, easily leading to frequent but ineffective alarms or failing to react promptly when truly fatal critical bottlenecks form.

[0005] Third, prediction and scheduling intervention are disconnected, lacking a dynamic closed-loop cybernetics mechanism: current technical solutions generally stop at a one-way output of "providing a prediction result," lacking automated linkage between early warning signals and the dispatch of special vehicles and support personnel at the airport frontline. When an early warning occurs, the system cannot automatically trigger targeted resource allocation rules based on the specific status of bottleneck nodes, nor can it quantify the negative feedback mitigation effect of injecting physical support resources on the time deviation of the current node. This makes it difficult for existing delay predictions to form automated closed-loop control during the event phase, limiting their practical application value.

[0006] Therefore, there is an urgent need in this field for a flight delay early warning and control framework that can accurately characterize the evolution of micro-level support status, dynamically quantify the weight of key bottlenecks, and integrate physical resource scheduling into closed-loop control, thereby improving the airport's ground operation's resistance to disturbances and support efficiency. Summary of the Invention

[0007] The purpose of this invention is to provide a closed-loop management method and system for flight support based on Petri nets and dynamic weights, aiming to solve the technical problems of over-reliance on black-box prediction models, inability of static weights to identify concurrent bottlenecks, and separation of early warning and scheduling in the prior art.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: A closed-loop management method for flight support based on Petri nets and dynamic weights, characterized by the following steps: Step S1: Construct a topology model for the state evolution of flight transit support: The micro-physical support links for flight transit are mapped to a Petri net structure, defining a place to represent the physical state of the support node, a transition to represent the task of state change, and a token to represent the real-time availability of support resources such as personnel and vehicles.

[0009] The difference between the actual time taken and the theoretical standard time taken at each transition stage is extracted and defined as the time deviation propagation variable in the directed transmission of the network. This propagation variable is used to characterize the margin absorption effect of time lag on non-critical paths and the cascading spread effect of time lag on critical paths.

[0010] Step S2: Dynamically quantify the critical path weights in the assurance process. Real-time monitoring of the current token distribution status in the Petri net and the real-time time deviation of each transition stage.

[0011] Establish a dynamic weight update mechanism that is jointly determined by real-time time deviation, token congestion level, and the degree to which the deviation approaches the margin.

[0012] When a non-critical parallel process exhausts its inherent time margin due to resource delays, the weight of that process is non-linearly increased according to the above mechanism, thereby automatically identifying the dynamic critical bottleneck node in the network that has undergone a shift.

[0013] Step S3: Calculate the comprehensive delay risk index and perform adaptive judgment: The real-time time deviation propagation variables of each support link are weighted and summed with the dynamically updated critical path weights to obtain a comprehensive delay risk index that represents the overall risk of flight operation collapse.

[0014] Meanwhile, the adaptive warning red line is calculated in real time based on the target flight's aircraft size, terminal facility level, and the current peak and off-peak status of airport operations.

[0015] When the comprehensive delay risk index exceeds the currently calculated adaptive warning red line, an over-limit warning signal is triggered to the system.

[0016] Step S4: Execute physical scheduling closed-loop control with negative feedback mechanism: After receiving the early warning signal, the system locks the transition node in the Petri net that has the most severe token congestion and the highest dynamic weight, and generates a targeted resource allocation instruction based on the locking result.

[0017] When physical intervention personnel or special vehicles intervene, a token representing new resources is injected into the corresponding warehouse in the network, triggering the reduction feedback rule for the time deviation of that node, recalculating and reducing the comprehensive delay risk index, thereby achieving closed-loop management of adaptive optimization and automatic intervention.

[0018] In this invention, the delay propagation mechanism, dynamic weighting mechanism, and negative feedback control mechanism are strongly coupled. The absence of any one of these mechanisms will prevent the real-time identification and closed-loop convergence control of dynamic critical path mutations. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating a closed-loop management method and system for flight support based on Petri nets and dynamic weights, provided by the present invention.

[0020] Figure 2 This is a schematic diagram of a Petri net provided for an embodiment of the present invention. Detailed Implementation

[0021] In one specific implementation, step S1, constructing a flight transit support state evolution topology model, includes the following sub-steps: Sub-step S101: Establish the Petri net mapping for the underlying physical guarantee process: The system extracts the standard operating procedures for ground transit support of the target flight and maps the actual physical support process as a Petri net with a directed bipartite graph structure.

[0022] The network consists of a set of places, a set of transitions, a set of directed arcs, and a set of tokens.

[0023] Specifically, a warehouse represents the physical environment state before or after a certain support operation begins or ends, such as when the cabin door is open or the jet bridge is docked.

[0024] Changes represent physical support tasks that cause a change in status, such as cabin cleaning, fuel refueling, or cargo loading and unloading.

[0025] The token represents the physical support resources required to perform the above tasks, including cleaning staff, tanker trucks, and baggage conveyors.

[0026] A transition can be triggered if and only if all upstream input libraries of a transition contain tokens representing resource availability.

[0027] Sub-step S102: Define and extract the time deviation propagation variable: During the operation of the above-mentioned network topology, the system connects to the airport's existing operational database or IoT sensing devices to collect the actual operation time of each transition link in real time.

[0028] The system extracts historical flight operation data and removes outliers to calculate the theoretical standard flight time for the corresponding aircraft type under the current operating environment.

[0029] For any transition step that is in progress or has been completed, the system calculates the difference between its actual operation time and the theoretical standard operation time, and defines this difference as the time deviation propagation variable. .

[0030] Sub-step S103: Quantifying margin absorption and cascade spread effects in topology: The system utilizes the critical path forward and backward bidirectional traversal algorithm in graph theory to dynamically deduce the topological state of the global Petri net.

[0031] The system calculates the earliest possible time and the latest necessary time for each transition in the network, and calculates the structural time margin of the i-th transition in real time using the following mathematical expression:

[0032] In the above mathematical expression, This represents the structural time margin of the i-th transition stage.

[0033] This represents the latest time that the node must occur, as derived by the system through a backward traversal algorithm, without affecting the final flight wheel chock removal time.

[0034] This represents the earliest possible time for this node based on the preceding physical conditions, derived by the system through a forward traversal algorithm.

[0035] Subsequently, the system constructs an effective delay load output equation for the time deviation propagation variable to quantify the network's local fault tolerance and delay propagation behavior. Its specific mathematical expression is as follows:

[0036] In the above mathematical expression, This represents the effective delay load actually output by the i-th transition node to its directly associated subsequent transition node.

[0037] This represents the time deviation propagation variable generated by the i-th transition step itself or inherited from the upstream.

[0038] The system executes the state determination based on the above effective delay load output equation.

[0039] When the first term of the equation is satisfied, i.e. the time deviation propagation variable is less than or equal to the structural time margin, the system determines that the time deviation is completely absorbed by the local network, and the effective delay load of the actual output is zero.

[0040] This physical phenomenon is defined as the margin absorption effect, in which the system does not issue delay superposition instructions to the downstream global network.

[0041] When the second term of the equation is satisfied, i.e. the time deviation propagation variable is greater than the structural time margin, the system determines that the physical fault tolerance space of the node has been exhausted, triggering a cascading spread effect.

[0042] The system extracts the overflow difference as an effective delay load based on the equation, and directly superimposes it onto the start-up time constraint of the downstream immediate associated node through a directed arc. In essence, the delay of upstream entity guarantee resources begins to physically squeeze the available time window of subsequent guarantee tasks.

[0043] Step S2, dynamically quantifying the critical path weights of the assurance process, includes the following sub-steps: Sub-step S201: Real-time extraction of state feature parameters of Petri net: The system monitors the operational progress of each physical support link in the Petri net and extracts the network state feature parameters at the current time t according to a preset sampling frequency.

[0044] For the k-th transition that is in progress or waiting in the queue, the system extracts the number of physical resource tokens currently backlogged in its upstream repository and records it as the real-time backlog. .

[0045] Simultaneously, the system synchronously calls the real-time time deviation propagation variable for this stage calculated in step S1. and structural time margin .

[0046] Sub-step S202: Construct state-driven nonlinear dynamic weight calculus equations: The system abandons the static constant weighting strategy and constructs a dynamic critical path weight function with the real-time physical state of the network as the independent variable.

[0047] The function is a nonlinear product determined by three features: the infrastructure dependency of the process, the real-time token congestion coefficient representing resource competition, and the bias exponential sensitivity representing the consumption of the fault tolerance space.

[0048] The system calculates the dynamic weight of the k-th transition in real time using the following mathematical expression. :

[0049] In the above mathematical expression, This represents the dynamic critical path weight of the k-th transition at time t.

[0050] The degree of connectivity of the infrastructure is represented by the ratio of the total number of downstream paths counted by performing a depth-first traversal of the Petri net reachability matrix starting from the k-th transition to the sum of the total number of downstream paths of all transitions in the network.

[0051] This indicates the real-time backlog, representing the number of flights or support vehicles currently waiting to perform the support task.

[0052] This is the congestion adjustment coefficient set by the system based on the current airport resource scheduling capacity.

[0053] This represents the real-time time deviation propagation variable.

[0054] This represents the structural time margin, which is recalculated in each sampling period.

[0055] The deviation sensitivity adjustment coefficient characterizes the safety tolerance limit.

[0056] This represents an exponential function with the natural constant as its base. Sub-step S203: Perform dynamic nonlinear assignment and critical path transition marking: The system performs concurrent calculations on the transition links in the global network based on the above calculation equations.

[0057] When a non-critical parallel process is in its normal operating range, i.e., the time deviation propagation variable... Much smaller than structural time margin At this point, the exponential term in the above equation approaches a constant, and the dynamic weights of the system... It remains in the low range dominated by the degree of correlation of basic infrastructure.

[0058] When this step causes significant delays in actual operations due to the absence of support personnel, the real-time time deviation propagation variable... Constantly approaching or even exceeding the structural time margin At this point, the exponential term in the above equation... This exhibits a non-linear surge, thereby driving the dynamic weights of this stage. It amplifies exponentially.

[0059] The system will dynamically assign weights in real time. The transition link that rises to the top of the overall picture is automatically marked as a dynamic key bottleneck node that is shifting at the current moment, thereby completing the mathematical mapping from physical congestion state to warning weight.

[0060] Step S3, Calculate the comprehensive delay risk index and perform adaptive judgment. The specific implementation process includes the following sub-steps: Step S301: Calculate the overall global delay risk index. Based on a preset sampling clock cycle, the system synchronously traverses all transition stages of support tasks that are being executed or are waiting in the queue in the global Petri net.

[0061] The system extracts the dynamic critical path weights of each transition stage obtained in real-time calculation in step S2, as well as the real-time time deviation propagation variables determined in step S1.

[0062] The system calculates a comprehensive delay risk index, characterizing the degree to which the overall operation system of the current target flight is approaching the brink of collapse, by coupling and summing the deviation states of each micro-node with dynamic weights. Its specific mathematical expression is as follows:

[0063] in, This represents the overall delay risk index of the target flight at the current time t.

[0064] N represents the total number of entity-guaranteed transition links in the global network that have not yet completed physical handover.

[0065] Step S302: Construct an adaptive early warning redline equation that is bound to the physical environment: Traditional early warning mechanisms typically use a fixed time difference value as a one-size-fits-all alarm benchmark. The system reconstructs this underlying logic to establish an adaptive early warning red line mechanism that dynamically evolves with the real physical operating environment.

[0066] The system calls the airport geographic information system and flight information system to extract the physical parking stand facility level characteristics of the target flight and the physical passenger and cargo capacity characteristics of the corresponding aircraft type.

[0067] Simultaneously, the system scans in real time the total number of active arriving and departing flights within the current terminal area to characterize the overall operational pressure of the airport. The system calculates the adaptive early warning threshold in real time using the following mathematical expression:

[0068] in, This represents the adaptive warning red line generated at time t.

[0069] This represents the basic risk tolerance constant extracted based on the historical flight punctuality baseline.

[0070] This represents the capacity buffer coefficient, which is determined by the physical size of the target aircraft and its load factor. Wide-body aircraft have a higher coefficient value than narrow-body aircraft.

[0071] This represents the efficiency adjustment coefficient determined by the physical facilities of the parking position, used to distinguish the physical differences in time compensation capabilities between parking positions near the jet bridge and parking positions far from the jet bridge.

[0072] and The specific values ​​are derived by the system through least squares regression fitting based on historical flight operation data of at least three calendar months for the airports under its jurisdiction, and are updated on a rolling basis on a quarterly basis.

[0073] This represents the airport wave operation pressure index calculated in real time, and its value is positively correlated with the number of concurrent flights currently in the peak stage of ground support.

[0074] This represents the system's preset pressure-sensitive attenuation constant.

[0075] Step S303: Execute state change judgment and trigger warning signal: The system constructs a real-time state comparator to synchronously compare the numerical relationship between the comprehensive delay risk index and the adaptive early warning red line according to the clock beat.

[0076] When the comprehensive delay risk index is continuously less than the adaptive warning red line, the system determines that the time lag of the current transit support operation is still within the self-recovery physical range of the physical resources. At this time, the system intercepts any secondary alarms that interfere with the front-line operation and maintains a silent monitoring state.

[0077] When the comprehensive delay risk index reaches or exceeds the adaptive warning red line at the current moment, the system determines that the ground operation network of the target flight has lost its self-recovery capability.

[0078] At this point, the system triggers a state change mechanism, generates an early warning signal containing the physical space coordinates of dynamic key bottleneck nodes and the types of resources required for protection, and issues the instruction to the physical scheduling closed-loop module in step S4.

[0079] Step S4: Execute physical scheduling closed-loop control with negative feedback mechanism. The specific implementation process includes the following sub-steps: Step S401: Identify dynamic critical bottlenecks and issue targeted scheduling instructions: After receiving the warning signal triggered in step S3, the system immediately blocks the passive propagation of time deviation in the global network.

[0080] The system traces the topology of the Petri net in reverse, identifies at least one transition node with the highest weight on the dynamic critical path at the current moment that meets the intervention conditions, and confirms it as the dynamic critical bottleneck node that triggers system collapse.

[0081] The system extracts the types of physical support resources that are missing in the upstream warehouse for this node, and generates low-level control signaling that includes the spatial coordinates of the target parking position and the required incremental adjustment parameters of the resources.

[0082] The system sends the control signal to the field physical support execution unit via the communication bus, directly driving the dispatch response of the corresponding special vehicles or automated support equipment, and synchronously updating the available token status constraints of the upstream warehouse.

[0083] Step S402: Perform physical resource injection and negative feedback control calculations. Once ground physical support personnel or special vehicles receive instructions, arrive at the designated location, and actually intervene in the support operation, the system senses the change in the underlying physical state and injects an intervention token representing the newly added entity resource into the corresponding upstream repository of the Petri net.

[0084] The system then constructs a negative feedback control equation characterizing the effect of resource intervention. This equation uses the number of newly added intervention tokens as the control input parameter and calculates the reduction in the time deviation propagation variable of the node after physical resource injection through a nonlinear decay mechanism. Its specific mathematical expression is as follows:

[0085] In the above mathematical expression, This represents the time deviation propagation variable that updates the transition node after a very short time period following the implementation of physical resource intervention.

[0086] This represents the propagation variable of the original time deviation before the intervention.

[0087] This indicates the actual number of intervention tokens injected during this targeted dispatch, i.e., the total number of support personnel or special vehicles that have been objectively increased. The incremental number of tokens injected in each control cycle is calculated independently.

[0088] This represents the marginal utility conversion constant determined by the physical properties of the operation guaranteed by this type of entity.

[0089] It represents a logarithmic function with the natural constant as its base, used to characterize the physical law of diminishing marginal utility resulting from the continuous increase of physical resources in reducing working time.

[0090] Step S403: Reassess the closed-loop state and perform dynamic optimization adjustment: The system uses the updated time deviation propagation variable as the new initial state parameter and re-substitutes it into the calculation link of steps S2 and S3.

[0091] The dynamic critical path weights of the global network are re-derived using the updated parameters, and the comprehensive delay risk index after intervention is calculated simultaneously.

[0092] The state change decision is executed again, and the comprehensive delay risk index after intervention is compared with the adaptive warning red line at the current moment.

[0093] If the comparison results show that the comprehensive delay risk index after intervention falls back below the adaptive warning red line, the system determines that the negative feedback closed-loop intervention was successful, automatically removes the warning signal and restores the silent monitoring state.

[0094] If the comparison results show that the overall delay risk index after intervention is still higher than or equal to the adaptive warning red line, the system determines that the current physical intervention is insufficient.

[0095] At this point, the system extracts the maximum resource carrying capacity threshold of the physical workspace corresponding to the target transition stage.

[0096] If the current cumulative number of intervention tokens has not reached the maximum resource carrying capacity threshold, the system will automatically increase the resource requirement level of the directional scheduling command and trigger a new round of control cycle.

[0097] If the current cumulative number of intervention tokens has reached the maximum resource carrying capacity threshold, the system determines that the physical intervention at this single point has reached its limit, automatically blocks the incremental injection of this node, and triggers the resource compensation instruction of the secondary critical path, thereby achieving optimization and convergence of the global flight ground operation status under physical extreme value constraints.

[0098] To more clearly illustrate the evolution of time deviation and the physical closed-loop scheduling mechanism in this invention, the following is combined with... Figure 2 The local topology operation mechanism of the present invention will be described in detail.

[0099] like Figure 2 The diagram shows a schematic of a closed-loop management method and system for flight support based on Petri nets and dynamic weights, constructed according to the present invention. In this network topology, circular nodes represent storage locations that characterize physical states, rectangular nodes represent changes in support tasks, solid directed arcs represent the forward data flow of physical entities and state transitions, and dashed lines with arrows represent the reverse control flow triggered by the issuance of control signaling at the system's lower level.

[0100] This local network demonstrates a typical "split-parallel-convergence" physical guarantee scenario. The network evolves from left to right, and after being triggered by the preceding transition node, it splits into two parallel physical guarantee paths, one above the other.

[0101] Figure 2 The non-critical path branches located at the top represent safeguards with inherent tolerance margins, such as cabin cleaning.

[0102] During the initialization phase, the system has already used a bidirectional traversal algorithm to deduce the transition node above and allocated a structural time margin. .

[0103] In the actual execution of this transition process, the difference between the actual time taken and the theoretical time taken is quantified as a time deviation propagation variable. .

[0104] like Figure 2 As shown, when the path above is running normally, the system performs real-time comparisons. and The numerical relationship is used to quantify the margin absorption effect and cascade spread effect of the path, and based on the aforementioned state-driven nonlinear calculation equation, its dynamic weight is maintained in the basic low range.

[0105] Figure 2 The dynamic critical bottleneck branches located at the bottom represent core processes where resource congestion occurs at any given moment, such as jet fuel refueling.

[0106] When the system detects that the dynamic critical path weight of the lower transition node jumps to the top of the global ranking due to physical resource lag, and causes the system's comprehensive delay risk index to exceed the adaptive warning red line, the system immediately locks the node as a dynamic critical bottleneck.

[0107] like Figure 2 As shown by the dashed control flow, the system triggers a negative feedback scheduling loop, injecting an intervention token representing newly added physical support resources into the upstream storage immediately preceding the downstream transition node. .

[0108] The intervention token The injection directly changes the underlying physical state distribution of the local network, thereby activating the logarithmic negative feedback control equation. The physical law of diminishing marginal utility forces the reduction of the time deviation propagation variable of the bottleneck node, ultimately achieving the resolution of local topological congestion and the convergence of the global flight ground operation status.

Claims

1. A method and system for closed-loop management of flight support based on Petri nets and dynamic weights, characterized in that, include: a. Map the physical support links of flight transit to a Petri net structure, define the location representing the physical state of the support node, the transition representing the task of state change, and the token representing the real-time availability of physical support resources. Extract the difference between the actual operation time and the theoretical standard time of each transition link as the time deviation propagation variable, and construct a delay propagation mechanism in the form of a piecewise function based on the comparison of time deviation and structural time margin. When the time deviation is less than or equal to the margin, the delay is absorbed, and when the time deviation is greater than the margin, the excess part is propagated to downstream nodes as a delay load. The maximum tolerable lag of each non-critical path transition link is calculated as the structural time margin, so that the time deviation propagation variable participates in the network topology evolution process as a Petri net state variable. b. Monitor the token distribution status and time deviation propagation variables in the Petri net in real time, establish a dynamic weight function coupled with the Petri net state variables, the weight function includes at least an adjustment term based on the degree of resource congestion and an exponential nonlinear amplification term based on the ratio of time deviation to structural margin, and calculate the dynamic critical path weight of each guarantee link in real time. c. The time deviation propagation variable and the dynamic critical path weight are weighted and summed to calculate the comprehensive delay risk index. The physical capacity characteristics of the target flight and the real-time wave operation pressure index of the current terminal are extracted to construct an adaptive warning red line with a decaying trend. When the comprehensive delay risk index exceeds the adaptive warning red line, an over-limit warning signal is triggered. d. In response to the over-limit warning signal, the transition node with the highest dynamic critical path weight in the Petri net is locked as the dynamic critical bottleneck node. A low-level control signal containing the target physical space coordinates and resource demand adjustment parameters is generated and the physical execution unit is driven to respond. An intervention token representing the new physical resource is injected into the corresponding storage location. Based on the number of intervention tokens injected, a bias compensation term with diminishing marginal utility is constructed. The bias compensation term is used to perform negative feedback offset calculation on the original time deviation propagation variable to obtain the updated time deviation propagation variable. The comprehensive delay risk index is then re-derived to complete the dynamic closed-loop management. The intervention token directly affects the topology propagation process of the time deviation propagation variable by changing the token distribution state in the Petri net.

2. The method according to claim 1, characterized in that, After the step of extracting the time deviation propagation variables of each transition stage, the method further includes: a. The network is deduced using a bidirectional forward and backward traversal algorithm for the critical path. By calculating the difference between the latest possible time and the earliest possible time for each transition step, the structural time margin of each non-critical path transition step is dynamically determined. When the time deviation propagation variable is less than or equal to the structural time margin, the margin absorption effect is triggered, and the effective delay load actually output to the downstream node is reduced to zero. When the time deviation propagation variable is greater than the structural time margin, the cascading spread effect is triggered. The overflow difference between the time deviation propagation variable and the structural time margin is extracted according to the effective delay load output equation, and it is superimposed on the start-up time constraint of the downstream immediate predecessor node through a directed arc.

3. The method according to claim 2, characterized in that, The step of calculating the dynamic critical path weight of each support link specifically includes: a. Extract the number of physical resource tokens currently backed up in the upstream warehouse of the k-th transition link as the real-time backlog, and extract the real-time time deviation propagation variable and the structural time margin of the transition link; b. Calculate the dynamic critical path weight of the transition stage by multiplying a constant term dominated by the basic structural correlation, a linear term dominated by the real-time backlog, and an exponential term dominated by the proportion of the time deviation propagation variable to the structural time margin.

4. The method according to claim 1, characterized in that, The step of calculating the adaptive early warning red line by combining the physical attributes of the target flight with the real-time operational pressure of the airport specifically includes: a. Extract the physical parking stand facility level characteristics and the physical passenger and cargo capacity characteristics of the corresponding aircraft type associated with the target flight, and scan the total number of inbound and outbound flights currently in active operation within the terminal area to calculate the real-time wave operation pressure index. b. Multiply the basic risk tolerance constant, the capacity buffer coefficient determined by the physical passenger and freight capacity characteristics, and the guarantee efficiency adjustment coefficient determined by the physical parking space facility level characteristics, and then multiply by an exponential decay function with a natural constant as the base and a negative real-time wave operation pressure index as the independent variable to obtain the adaptive early warning red line calculated in real time.

5. The method according to claim 1, characterized in that, The step of calculating the reduction of the time deviation propagation variable after resource injection using the negative feedback control equation specifically includes: a. Use the number of intervention tokens actually injected into the corresponding warehouse according to the resource allocation instruction as the control input parameter; b. Calculate the natural logarithm of the control input parameter after adding one, and multiply it by the marginal efficiency transformation constant characterizing the entity support operation to obtain the bias compensation term; c. Subtract the bias compensation term from the time deviation propagation variable before intervention to obtain the updated time deviation propagation variable after physical resource intervention.

6. The method according to claim 1, characterized in that, The step of injecting the intervention token also includes: a. Perform dynamic optimization based on physical extremum constraints: b. Extract the maximum resource carrying capacity threshold of the physical workspace corresponding to the target transition stage; c. If the current cumulative number of injected intervention tokens has not reached the maximum resource carrying capacity threshold, perform the next round of incremental resource injection; d. When the number of intervention tokens injected cumulatively reaches the maximum resource carrying capacity threshold, it is determined that the single-point physical intervention has reached its limit and the incremental injection of the current node is blocked. Instead, the resource compensation instruction of the secondary critical path is triggered to achieve the convergence of the global risk index under physical space constraints.

7. A method and system for closed-loop management of flight support based on Petri nets and dynamic weights, characterized in that, include: a. Topology evolution modeling module, used to map the physical support links of flight transit to a Petri net structure, defines the place representing the physical state of the support node, the transition representing the task of state change, and the token representing the real-time availability of physical support resources, and extracts the difference between the actual operation time and the theoretical standard time of each transition link as the time deviation propagation variable. b. Dynamic weight calculation module, used to monitor the token distribution status and time deviation propagation variable in the Petri net in real time, establish dynamic weight calculation logic with the basic structure correlation degree as the dominant benchmark, the real-time backlog and congestion degree of the token as the positive linear gain, and the degree to which the time deviation propagation variable approaches the structural time margin as the nonlinear mutation driving condition, and calculate the dynamic critical path weight of each guarantee link in real time. c. A state change decision module is used to calculate a comprehensive delay risk index by weighted summation of the time deviation propagation variable and the dynamic critical path weight, and to calculate an adaptive warning red line by combining the physical attributes of the target flight and the real-time operational pressure of the airport. When the comprehensive delay risk index exceeds the adaptive warning red line, an over-limit warning signal is triggered. d. A closed-loop scheduling feedback module is used to lock the transition node with the highest dynamic critical path weight in the Petri net as the dynamic critical bottleneck node and issue a resource allocation instruction, inject an intervention token representing new physical resources into the corresponding warehouse, calculate the reduction of the time deviation propagation variable after resource injection through the negative feedback control equation, and recalculate the comprehensive delay risk index to complete the dynamic closed-loop management.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 6.

Images