Time optimization method based on flight normality target

Abstract

The invention relates to the technical field of aviation operation management, and discloses a time optimization method based on a flight normality target. The method comprises the following steps: constructing an operation situation topology network corresponding to an initial flight plan, wherein the operation situation topology network comprises operation association edges between flight nodes; collecting a real-time operation data flow, dynamically mapping the real-time operation data flow to a corresponding flight node in the operation situation topology network, and updating the attribute state of the flight node; extracting network structure features according to the updated network, and identifying potential conflict areas in the network based on the network structure features; for the potential conflict area, generating a plurality of moment adjustment plan sets with different adjustment granularities; inputting the real-time operation environment parameters into a pre-arranged plan set for screening, and determining a target adjustment pre-arranged plan; and performing time sequence adjustment on the initial flight plan according to the target adjustment plan to generate an optimized flight plan. According to the invention, systematic characterization and intelligent and precise time optimization of the flight operation situation are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of air operation management, in particular to a time optimization method based on flight normality target. BACKGROUND

[0002] The current flight time optimization method mainly relies on the analysis of historical statistical data and the adjustment based on fixed rules. These technologies usually regard the flight plan as a static time sequence table, and through the analysis of historical punctuality rate, airport throughput and other macro data, the flight time is pre-planned and coarsely planned. In the real-time operation stage, it mainly depends on the personal experience of air traffic controllers and flight dispatchers to make local manual adjustments to individual flights or a limited number of conflicting flights. The existing method regards the flight as an independent individual, and lacks a systematic description of the dynamic operation correlation between flights.

[0003] The existing technical solutions have defects. The static tabular flight plan is difficult to effectively express and quantify the complex mutual influence relationship between flights in the aspects of take-off and landing, airspace use, transfer connection and the like. This leads to that when a disturbance occurs in real-time operation, the system cannot quickly and accurately evaluate the propagation range and influence degree of local delay in the entire flight network. The adjustment decision-making process based on artificial experience is low in efficiency, and is seriously dependent on the personal ability of the dispatch personnel, and it is difficult to ensure the scientificity and global optimality of the decision-making. For the identified operation conflicts, the granularity of the existing adjustment means is single, and lacks the ability to flexibly select different adjustment strategies according to the conflict severity and real-time environment, which is easy to cause insufficient adjustment or excessive adjustment.

[0004] The current technology cannot meet the demand for fine, intelligent and forward-looking optimization of large-scale flight plans. The core of the problem is how to dynamically and systematically represent the operation correlation between flights, and realize the automatic and accurate decision-making from conflict identification to scheme generation and screening based on this. SUMMARY

[0005] The purpose of the present application is to provide a time optimization method based on flight normality target to solve the problems raised in the background art.

[0006] To achieve the above purpose, the present application provides a time optimization method based on flight normality target, which comprises: An initial flight plan corresponding operation situation topology network is constructed, and the operation situation topology network comprises flight nodes and operation correlation edges between flight nodes; Real-time operation data stream is collected, and the real-time operation data stream is dynamically mapped to the corresponding flight node in the operation situation topology network, and the attribute state of the flight node is updated; According to the running situation topology network updated according to the attribute state, network structure features are extracted, and potential conflict areas in the running situation topology network are identified based on the network structure features; For the identified potential conflict areas, a corresponding time adjustment plan set is generated, and the time adjustment plan set includes multiple plans with different adjustment granularities; The real-time running environment parameters are input into the time adjustment plan set for plan screening to determine a target adjustment plan; According to the target adjustment plan, the initial flight plan is adjusted in time sequence to generate an optimized flight plan.

[0007] Preferably, the running situation topology network corresponding to the initial flight plan is constructed, including: The initial flight plan is parsed, all flight tasks in the plan are extracted, and each flight task is abstracted as a flight node; The time sequence dependency relationship and resource competition relationship between the flight tasks are analyzed, the forward association edges between the flight nodes are constructed based on the time sequence dependency relationship, and the conflict association edges between the flight nodes are constructed based on the resource competition relationship; All flight nodes, forward association edges and conflict association edges are integrated to form a running situation topology network.

[0008] Preferably, the real-time running data stream is collected, and the real-time running data stream is dynamically mapped to the corresponding flight node in the running situation topology network, including: The real-time running data stream is continuously received from multiple heterogeneous data sources, and the real-time running data stream at least includes flight identification, timestamp and running state parameter; The real-time running data stream is distributed to the corresponding flight node in the running situation topology network according to the flight identification; The attribute state of the corresponding flight node is updated using the received real-time running data stream, and the attribute state includes planned time, estimated time, actual time and resource occupation state.

[0009] Preferably, the network structure features are extracted from the running situation topology network updated according to the attribute state, including: The running situation topology network updated according to the attribute state is traversed, and the local features of each flight node are calculated, including degree centrality and adjacent flight state consistency; The subgraph structure in the running situation topology network is analyzed, and the subgraph containing dense conflict association edges is identified as a candidate conflict area; The global influence weight of each candidate conflict area is calculated, and the global influence weight is determined based on the position of the candidate conflict area in the running situation topology network and the importance of the flight nodes contained therein.

[0010] Preferably, the network structure feature-based operation situation topology network potential conflict area identification comprises: A conflict threshold is set, and the global influence weight of the candidate conflict area is compared with the conflict threshold; The candidate conflict area with a global influence weight greater than the conflict threshold is screened out and marked as a potential conflict area; The flight node set contained in each potential conflict area and its associated conflict type are recorded.

[0011] Preferably, for the identified potential conflict area, a corresponding set of time adjustment plans is generated, comprising: For each potential conflict area, analyze the time sequence constraints and resource constraints of its flight node set; A plan for multiple different adjustment strategies is generated to minimize the total delay time, and the adjustment strategies include time translation, flight sequence exchange, and ground waiting; An adjustment granularity parameter is set for each plan, which defines the minimum time unit of time adjustment and the maximum number of flights allowed to adjust.

[0012] Preferably, the real-time running environment parameters are input into the time adjustment plan set for plan screening, comprising: Obtain the current real-time running environment parameters, including airspace traffic density, airport receiving rate, and weather influence coefficient; Establish a plan evaluation function, which takes real-time running environment parameters and plan adjustment parameters as input; Use the plan evaluation function to calculate the fitness evaluation value of each plan in the time adjustment plan set; Select the plan with the optimal fitness evaluation value as the target adjustment plan.

[0013] Preferably, the initial flight plan is time-adjusted according to the target adjustment plan, comprising: Parse the target adjustment plan to obtain the set of flight nodes to be adjusted and the target time set for each flight node to be adjusted in the target adjustment plan; Locate the flight tasks corresponding to the set of flight nodes to be adjusted in the initial flight plan; Modify the planned takeoff time or planned arrival time of the corresponding flight task according to the target time, and check whether the adjusted flight plan meets all the operation constraints.

[0014] Preferably, after generating the optimized flight plan, it further comprises: Continuously monitor the execution of the optimized flight plan and collect plan execution data; Compare the plan execution data with the optimized flight plan to calculate a plan deviation degree; When the plan deviation degree exceeds a preset tolerance, trigger a new round of optimization process, reconstruct the operational situation topology network and perform subsequent steps.

[0015] Preferably, the calculation of the plan deviation degree comprises: Obtain the planned time and the actual execution time of a key flight in the optimized flight plan; Calculate the time deviation of each key flight, and perform weighted fusion based on the weight of the key flight in the operational situation topology network; Compare the weighted and fused time deviation value with a plurality of preset deviation threshold values to determine the final plan deviation degree level.

[0016] Compared with the prior art, the beneficial effects of the present application are: The operational situation topology network is constructed and real-time data is dynamically mapped and updated, and discrete flight information is converted into a network structure containing nodes and edges. This expression method can accurately depict the internal correlation between flights in terms of time sequence, airspace resources and process connection. Based on the network structure characteristics, potential conflicts can be identified, and the cascading effect of local disturbance through the running correlation edge can be understood from the overall system perspective. Network centrality and other indicators can quantify the key degree of nodes, and edge density and other indicators can reveal the running tension of the region, which enables the identification process to change from experience-based fuzzy judgment to accurate calculation based on network topology data, thereby discovering systemic bottlenecks and vulnerable links that are difficult to detect by traditional methods.

[0017] A multi-granularity adjustment plan set is generated for the identified conflict area, so that the decision system has the flexibility to respond to different sizes and different nature of operational conflicts. Fine-grained plans are suitable for fine-tuning local timing, and coarse-grained plans can be used to solve structural resource competition. Real-time running environment parameters are used as input conditions for plan screening, ensuring that the final selected target adjustment plan is highly matched with the current specific airspace capacity, weather conditions and other constraint conditions. This method upgrades the time adjustment from fixed, single rules or decisions relying on personal experience to a dynamic, multi-option, real-time environment feedback-based optimization process, enhancing the adaptability and robustness of flight plans to complex changes in the running environment. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 A working principle diagram of the flight normality target-based time optimization method described in the present application; Figure 2 A flowchart for constructing the operational situation topology network; Figure 3 A flowchart for real-time data stream acquisition and mapping; Figure 4 a global impact weight comparison chart for candidate conflict areas; Figure 5 a flight plan time versus actual time deviation comparison chart for critical flights. DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0020] Please refer to Figure 1 The present application provides a flight schedule optimization method based on flight normality target, which comprises: constructing a running situation topology network corresponding to an initial flight plan, the network being composed of flight nodes and running associated edges, the flight node representing a single flight task, and the running associated edge reflecting the time sequence or resource interaction relationship between flights; then the system collects real-time running data streams from multiple data sources, and dynamically maps these data to the corresponding flight nodes in the topology network to update the attribute state of the nodes, such as the planned time or resource occupation. Based on the updated network, the system extracts network structure features such as node degree centrality or subgraph density, and identifies potential conflict areas, which are usually composed of dense conflict associated edges. For each potential conflict area, a flight schedule adjustment plan set is generated, the plan containing different adjustment strategies such as time translation or flight sequence exchange, and having variable adjustment granularity. Next, real-time running environment parameters such as airspace traffic density are input into the plan set, and the optimal target adjustment plan is selected through an evaluation function. According to the plan, the time sequence of the initial flight plan is adjusted to generate an optimized flight plan, thereby improving the flight normality.

[0021] Embodiment 1: refer to Figure 2, when constructing the initial flight plan corresponding to the running situation topology network, the system performs a parsing operation on the initial flight plan file, which is usually in a structured data format such as Extensible Markup Language or JavaScript Object Notation. The parsing process involves reading the file content and extracting key field information according to a predefined data pattern. In specific implementation, the system processes the initial flight plan file line by line or in blocks, identifies each independent flight task record, and extracts basic attributes such as flight number, planned departure time, planned arrival time, departure airport, arrival airport, aircraft registration number, and aircraft model code. Each valid flight task record is instantiated as a flight node in the running situation topology network, and the flight node is assigned a globally unique node identifier and stores all extracted flight attributes as node metadata. Analyzing the time sequence dependency relationship and resource competition relationship between flight tasks is the core logic of constructing network association edges. The time sequence dependency relationship mainly refers to the connection constraints between consecutive flight tasks performed by the same aircraft.

[0022] In specific implementation, the system groups and sorts all flight tasks according to the aircraft registration number. For each aircraft group, the system arranges the flight tasks in ascending order of planned departure time to form a flight chain for that aircraft. For adjacent flight tasks in the flight chain, if the arrival airport of the previous flight is the same as the departure airport of the next flight or if the minimum transit time requirement is met, the system establishes a forward association edge between the flight nodes corresponding to the two flight tasks. The forward association edge has directionality, with the direction pointing from the previous flight node to the subsequent flight node. The forward association edge itself carries attributes, mainly used to record the calculated minimum transit time requirement or theoretical turnaround time interval. The analysis of resource competition relationship focuses on the potential competition for shared limited resources such as runways, parking spaces, and terminal bridges. In some embodiments, the system loads an airport resource usage strategy template that defines time window allocation rules for different types of resources. The system matches the planned takeoff and landing times of each flight task with the resource available time window. When it finds that two or more flight tasks overlap in time and apply for the same resource, the system establishes a conflict association edge between the flight nodes corresponding to these flight tasks. The conflict association edge is usually undirected, but can have a weight attribute to quantify the severity of the conflict. The weight can be calculated based on factors such as the length of the time overlap and the criticality of the resource.

[0023] Building forward association edges between flight nodes based on time dependency is a systematic process. After the aircraft grouping and flight chain sequencing are completed, the system will traverse each flight chain. For each pair of adjacent flight tasks in the chain, the system will calculate the time interval, which is the difference between the planned departure time of the latter flight task and the planned arrival time of the former flight task. It can be understood that the system will compare this time interval with a preset safety threshold, which takes into account the minimum transit time, taxi time and buffer time. Only when the actual time interval is greater than or equal to the safety threshold, the system considers that the time dependency is loose and may not need special attention. However, when the time interval is less than the safety threshold, it indicates that the connection between the previous and the latter flights is tight, and there is a risk of delay propagation. At this time, the system will force to build a forward association edge between the corresponding flight nodes, and the calculated time interval deficit will be an important attribute of the forward association edge. The establishment of the forward association edge enables the running situation topology network to clearly express the time sequence conduction path between flights.

[0024] Building conflict association edges between flight nodes based on resource competition relationship requires more detailed resource-time matching analysis. The system maintains an airport resource configuration file, which lists the number, available time period and allocation rules of each resource in detail. Optionally, the system simulates the resource allocation process for each flight task, allocates appropriate resource instances to it according to the planned take-off and landing time, aircraft model, etc. For example, the system will record the occupied time interval of each resource instance when it allocates a parking space for the flight. When the system detects that the resource instances allocated by two or more flight tasks have intersection in the time interval, it determines that there is a resource competition relationship between these flight tasks. The system will build a conflict association edge between the flight nodes corresponding to these flight tasks. The conflict association edge can record the specific resource type involved in the conflict and the time range of the conflict. The construction of the conflict association edge enables the running situation topology network to explicitly depict the resource bottleneck area.

[0025] All flight nodes, forward association edges and conflict association edges are integrated to form the final operational situation topology network, which involves the assembly and persistence of data structures. In specific implementation, the system usually adopts a graph data model to represent the operational situation topology network, with flight nodes as vertices of the graph, forward association edges and conflict association edges as edges of the graph, and the system adds all vertices and edges created in the previous steps to a graph structure and sets global attributes for this graph structure, such as the creation timestamp of the graph, the initial flight plan version number based on which the graph is created, etc. It can be understood that after the graph structure is established, the system will perform a series of consistency checks, such as checking whether there are isolated flight nodes, verifying whether the forward association edges constitute a directed acyclic graph to ensure the timing logic is reasonable, checking whether the conflict association edges are consistent with the resource allocation results, and after the checks, the system saves the complete operational situation topology network to a graph database or an in-memory database for subsequent module queries and updates. The construction of the operational situation topology network marks that the flight plan data has been converted into a network model that can be used for complex relationship analysis.

[0026] Embodiment 2: see Figure 3 When collecting real-time operational data streams, the system needs to establish stable connections with multiple heterogeneous data sources, including air traffic control automation systems, airport collaborative decision-making systems, aircraft communication addressing and reporting systems, and ground support systems, etc. In specific implementation, the system deploys a dedicated data collection interface adapter for each type of data source. For example, for air traffic control automation systems, the adapter receives real-time messages containing flight radar trajectories, calculated estimated arrival times, etc. through a dedicated data subscription service; for airport collaborative decision-making systems, the adapter obtains flight support node states such as wheel block removal times, jetway docking completion times, etc. by calling application programming interfaces. These collection interface adapters perform preliminary analysis on the received raw data and convert it into an internal unified standardized data format, which at least contains a flight identifier, a data generation timestamp, a data source system identifier, and one or more running state parameter key-value pairs.

[0027] After receiving real-time running data streams continuously from multiple heterogeneous data sources, the system needs to efficiently distribute the data to the corresponding flight nodes in the running situation topology network, and this process relies on a real-time updated flight identifier and node identifier mapping table. In specific implementation, the mapping table is initialized after the running situation topology network is built, which associates the unique node identifier of each flight node with one or more business keys that can uniquely identify the flight task, usually including the combination of flight number, execution date, aircraft registration number and planned departure and arrival airport codes. It can be understood that when a real-time running data stream arrives at the system, the data distribution engine will first extract the flight identifier in the data packet, such as flight number "CA1234" and date "20240601", and then use this identifier as a query key to search in the flight identifier and node identifier mapping table. If a matching item is found, the corresponding node identifier is obtained, and the complete data packet content is sent to the flight node message queue pointed by the node identifier. In some embodiments, in order to deal with the situation of dynamic increase or cancellation of flights, the system will continuously monitor the change notification of the initial flight plan, and once a plan change is detected, the flight identifier and node identifier mapping table and the structure of the running situation topology network will be updated immediately to ensure the accuracy of data distribution.

[0028] Updating the attribute state of the corresponding flight node with the received real-time running data stream is a state maintenance process, and the attribute state of the flight node includes planned time, estimated time, actual time and resource occupation state. In specific implementation, each flight node maintains a state object in memory, and the state object contains multiple fields, which respectively record various timestamps and resource allocation situations of the flight in different life cycles. When a new real-time running data stream is successfully distributed to the message queue of the flight node, the state processor of the node will consume the message, and the state processor will update the corresponding field in the state object according to the type of the running state parameter in the data packet. For example, when receiving the estimated arrival time update from the air traffic control, the state processor will update the "estimated time" field in the state object; when receiving the actual takeoff time report from the airport system, the state processor will update the actual takeoff time in the "actual time" field, and may trigger a chain assessment of the state of the subsequent associated flight node. The update logic of the resource occupation state is relatively complex, which needs to judge whether the resource has been actually occupied or released according to the real-time guarantee data, for example, when the system receives the signal of the completion of the apron docking, it will mark the "stand" in the resource occupation state of the corresponding flight node as "occupied" and record the occupation start time.

[0029] In order to quantitatively evaluate the immediacy of the influence of real-time running data stream on the state of flight node, the system introduces a data freshness index for internal monitoring, and the calculation method of the index is as follows: wherein: represents the data freshness score of a specific flight node at the current time, the higher the score value, the fresher the data received by the node recently. represents the total number of real-time running data streams successfully processed and applied to the flight node within a fixed time window in the past. represents the current standard time of the system. represents the original timestamp carried in the data packet of the th real-time running data stream. is a decay coefficient used to control the decay speed of the influence of historical data on the freshness score. Optionally, the system can set a freshness threshold, when the value of a flight node is lower than the threshold, the system will generate an alarm to indicate that the state of the flight node may have lagged behind, and attention should be paid to the integrity of the data link or attempts to obtain updates from other data sources.

[0030] The update operation of the attribute state is not only a simple field replacement, but also may trigger the logical conversion of the internal state of the flight node and the adjustment of the attributes of the associated edges in the running situation topology network. In some embodiments, the state object of the flight node is designed as a state machine, when the "actual time" field is filled with the actual take-off time, the state of the flight node may change from "planned" to "take-off"; when the actual arrival time is recorded, the state may change to "arrival". It can be understood that such state transition events will be published to the event bus of the system, and other components in the running situation topology network, such as the conflict detection engine, can subscribe to these events, when receiving the flight state change event, the conflict detection engine will re-evaluate the weight of the conflict associated edge connected with the flight node, for example, the delay of a flight may cause the weight of the conflict associated edge with the subsequent flight to increase, because the possibility of overlapping time window of resource competition becomes larger. Through this dynamic updating mechanism, the running situation topology network can reflect the latest changes of the flight running situation in real time, and provide accurate data basis for the subsequent potential conflict area identification.

[0031] ​In the embodiment, the system randomly selects an unvisited flight node from the running situation topology network as the starting point, and calculates the local features of each flight node during the visiting process. The local features include the degree centrality and the adjacent flight state consistency. The degree centrality is calculated for all associated edges of the flight node. For undirected conflict associated edges, the degree centrality is the number of conflict associated edges directly connected to the flight node. For directed forward associated edges, the in-degree (the number of forward associated edges pointing to the flight node) and the out-degree (the number of forward associated edges pointing from the flight node) are calculated respectively. The calculation of the adjacent flight state consistency involves the state attributes of the flight node itself and its first-order neighbor nodes. The system extracts the numerical values of these nodes on a specific attribute, for example, extracts the current delay time estimate of all related nodes, and then calculates the statistical consistency measure of these numerical values, such as variance or standard deviation. The smaller the variance, the higher the adjacent flight state consistency.

[0032] The subgraph structure in the running situation topology network is analyzed to identify candidate conflict areas. The system calls a graph community discovery algorithm to segment the network. In the embodiment, the commonly used algorithm is the Louvain method, which takes the modularity maximization as the optimization objective, and divides the entire running situation topology network into multiple subgraphs with tight internal connections and sparse external connections by iteratively merging nodes into communities. It can be understood that after the system executes the Louvain method, it will analyze each generated subgraph, especially focusing on the density of conflict associated edges within the subgraph. The conflict associated edge density is calculated by the ratio of the number of conflict associated edges contained in the subgraph to the maximum number of conflict associated edges that the subgraph can theoretically contain. The system will set a density threshold, and mark the subgraphs with conflict associated edge density higher than the threshold as candidate conflict areas, because these subgraphs have dense conflict interactions between flight nodes, which are potential running bottlenecks.

[0033] The global impact weight of each candidate conflict region is a comprehensive evaluation process, and the global impact weight is determined based on the topological position of the candidate conflict region in the operational situation topology network and the importance of the flight nodes contained therein. In specific implementations, the evaluation of the topological position can be achieved by calculating the betweenness centrality of the candidate conflict region, that is, how many shortest paths pass through the candidate conflict region, and the more paths passing through, the stronger the hub nature in the network. The importance of the flight node is quantified in multiple dimensions, for example, the importance of the flight node can be measured by the eigenvector centrality, and the flight node with high eigenvector centrality means that the other nodes connected thereto are also important; in addition, the importance of the flight node can also be combined with business rules, for example, international flights, wide-body aircraft flights or flights carrying important passengers can be assigned higher importance base weight. The system aggregates the importance scores of all flight nodes in the candidate conflict region, and combines the topological position index of the region to synthesize the final global impact weight value through a weighted formula , the calculation method of which can be represented as: , wherein: represents the global impact weight of the candidate conflict region. represents the average value of the eigenvector centrality of all flight nodes in the candidate conflict region, reflecting the average influence of the region nodes. represents the betweenness centrality of the candidate conflict region in the entire operational situation topology network, reflecting the key degree of the region in network connectivity. and are preset weighting coefficients for balancing the contribution of node influence and topological position in the evaluation of global impact weight. In some embodiments, the weighting coefficients and can be obtained by training historical data, so that the global impact weight can more accurately predict the actual impact of regional out-of-control on the overall network.

[0034] Based on the network structure features, the system identifies potential conflict regions in the operational situation topology network, and compares the global impact weight of the candidate conflict region with a preset conflict threshold. In specific implementations, the conflict threshold is not fixed but a dynamically adjusted parameter, and the system will set the conflict threshold according to the overall pressure level of the current running stage, for example, during the peak period of flight arrival and departure, the conflict threshold can be adjusted accordingly to avoid marking too many potential conflict regions to cause system overload. The screening process is to sequentially check the global impact weight of each candidate conflict region , once the global impact weight of a candidate conflict region is greater than the currently effective conflict threshold, the system will formally mark this candidate conflict region as a potential conflict region.

[0035] While marking the potential conflict regions, the system records detailed information of each potential conflict region. Optionally, the system creates a descriptor object for each potential conflict region, which contains the following key information: first, the set of flight nodes contained in the potential conflict region, usually stored in the form of a list of node identifiers; second, the conflict type associated with the potential conflict region, which is mainly divided according to the nature of the dominant conflict association edge in the region, for example, it can be divided into "resource competition type conflict", which is mainly caused by the competition for resources such as parking spaces and runways; "time sequence conduction type conflict", which is mainly caused by tight forward association edges and delay propagation risks; or mixed conflict. In some embodiments, the descriptor object also records the timestamp of the first identification of the potential conflict region, the historical stability of the region, and other information, which helps to assess the persistence and urgency of the potential conflict region. All identified potential conflict regions and their descriptor objects are passed to the subsequent pre-plan generation module as target regions for time optimization.

[0036] Referring to Figure 4 , the horizontal coordinate is 5 candidate conflict regions, and the vertical coordinate is the global impact weight. The system first identifies candidate regions containing dense conflict association edges from the updated operational situation topology network, and then calculates the global impact weight thereof; by comparing with the conflict threshold, the weights of regions C, A and E in the figure exceed the threshold, which are marked as potential conflict regions, and regions B and D are temporarily excluded from the priority processing range due to insufficient weight. This figure is a visualization of the transition of conflict identification from experience judgment to data-driven, which intuitively quantifies the operational risk level of each region and clearly defines the core target region for the generation of subsequent time adjustment pre-plans, improving the accuracy and operability of conflict identification.

[0037] Embodiment 4: For the identified potential conflict regions, a corresponding set of time adjustment pre-plans is generated. The system first parses the potential conflict region descriptor object and extracts the set of flight nodes contained therein. In specific implementation, the system obtains the complete attribute information of these flight nodes from the operational situation topology network according to the list of flight node identifiers, including the planned time, current estimated time, aircraft information, departure and arrival airport, and allocated resource information of each flight node. Analyzing the time sequence constraints and resource constraints of the set of flight nodes in the potential conflict region is the basis for pre-plan generation. Time sequence constraint analysis focuses on checking the dependency relationship between flight nodes established through forward association edges, such as the minimum transit time requirement between consecutive flight tasks of the same aircraft; resource constraint analysis reviews whether there is an irreconcilable overlap in the planned occupation of shared resources such as runway time period and parking space.

[0038] The system generates multiple adjustment strategies using optimization algorithms. In practice, meta-heuristic algorithms such as genetic algorithm and tabu search are commonly used to search for near-optimal solutions under complex constraints. Adjustment strategies mainly include time shift, sequence swap, and ground waiting. Time shift strategy moves the planned departure or arrival time of one or more flight nodes in the potential conflict area forward or backward. Sequence swap strategy adjusts the usage order of multiple flight nodes sharing the same critical resource. Ground waiting strategy adds a controllable ground delay time to the flight node before departure. The system runs multiple instances of optimization algorithms in parallel, each focusing on different strategy combinations or parameter settings, generating diverse plans. Setting the adjustment granularity parameter for each generated plan is a key step to ensure the feasibility of the solution. The adjustment granularity parameter clearly defines the minimum time unit for time adjustment, such as five minutes or ten minutes, and the maximum number of flights allowed for adjustment, such as only allowing thirty percent of the flights in the potential conflict area to be adjusted. The adjustment granularity parameter makes the generated plans both principled and flexible, avoiding drastic or unrealistic changes to the flight plan.

[0039] Real-time operating environment parameters are input into the schedule adjustment plan set for plan screening. The system needs to continuously obtain the latest real-time operating environment parameters from external data interfaces. In practice, real-time operating environment parameters mainly include airspace traffic density, airport reception rate, and weather influence coefficient. Airspace traffic density is provided by air traffic control systems, representing the number of aircraft in a specific sector or waypoint. Airport reception rate is calculated by the airport operation center, reflecting the number of incoming flights that can be safely received by the airport per unit time. Weather influence coefficient comes from meteorological services and is a quantitative value that integrates the effects of visibility, wind direction and speed, precipitation, and other factors on operational efficiency. These parameters are encapsulated in an environment parameter object with a valid timestamp. Establishing a plan evaluation function is the core of the screening decision. The plan evaluation function is a multivariate function whose input parameters include the real-time operating environment parameter object and the adjustment parameters of the plan to be evaluated, including the total delay reduction estimated by the plan, the number of flights involved, and the complexity score of the adjustment operation. The goal of the plan evaluation function is to comprehensively evaluate the adaptability and expected effect of the plan in a specific environment.

[0040] The plan evaluation function is used to calculate the fitness evaluation value of each plan in the schedule adjustment plan set , which can be expressed as: Where: represents the fitness evaluation value of the plan, and the higher the value, the better the plan. is the total delay time expected to be reduced after applying the plan. is the baseline total delay time of the potential conflict area without adjustment. is the number of flights that need to be adjusted in time for the plan to be implemented. is a coefficient representing the complexity of the plan operation, the higher the complexity the larger the coefficient value. is the weather impact coefficient in the real-time operating environment parameter. , , is the weight coefficient used to balance the importance of delay reduction, impact range, complexity and weather environment under different decision preferences.

[0041] The plan with the best fitness value is selected as the target adjustment plan, and the system ranks all the plans according to their values. It can be understood that the selection process usually directly selects the plan with the highest value, but in some cases, for example, when the difference between the highest score and the second highest score is small and the plan with the second highest score has other advantages, the system can be configured to submit the top few plans for the operator to make the final decision. Referring to Table 1, a simplified evaluation including three plans is shown.

[0042] Table 1: Plan Evaluation Results Table According to the above example, plan P002 is selected as the target adjustment plan because it achieves a good balance in reducing delay, controlling impact range and complexity, and thus obtains the highest fitness value. Finally, the target adjustment plan and its detailed adjustment instruction set are output, providing clear basis for subsequent time adjustment execution.

[0043] Embodiment 5: The system first needs to parse the data structure of the target adjustment plan, which is usually stored in a machine-readable format such as JavaScript Object Notation object or a predefined binary protocol buffer format. In specific implementation, the parsing process includes reading the target adjustment plan file or message stream, extracting the key instruction set, which at least contains two core parts: the set of flight nodes to be adjusted and the adjustment instruction details. The set of flight nodes to be adjusted is a list of unique identifiers of all flight nodes that need to modify the planned time. The adjustment instruction details specify the precise target time for each flight node in the set. The target time can be represented as an absolute time or as an offset relative to the original planned time. The parsing engine checks the integrity and logical reasonableness of the target adjustment plan, such as checking whether the target time is within the reasonable time range of flight operation, and whether the adjusted time still meets the minimum transit time requirement for consecutive flights of the same aircraft.

[0044] Locating the flight tasks corresponding to the set of flight nodes to be adjusted in the initial flight plan, the system uses the mapping relationship established in the running situation topology network construction stage for reverse lookup. In specific implementation, the system maintains a mapping table of flight node identifiers and flight task record primary keys in the initial flight plan database. When receiving the set of flight nodes to be adjusted, the system traverses the list and, for each flight node identifier in the list, finds its corresponding unique flight task record in the initial flight plan database by querying the mapping table. It can be understood that the positioning operation must be performed in a transactional environment to ensure the accuracy and consistency of the data when reading the flight task record, preventing data chaos caused by simultaneous modification of the same flight task record by other concurrent processes. After successful positioning, the system temporarily stores the primary key of the flight task record and the corresponding adjustment instruction in a working list in memory, preparing for subsequent batch updates.

[0045] According to the target time, the plan departure time or the plan arrival time of the corresponding flight task is modified, and whether the adjusted flight plan meets all the operation constraints is checked, which is a key data operation and verification phase. In specific implementation, the system processes the entries in the worklist one by one, and for each flight task record, the system updates its plan departure time field or plan arrival time field according to the adjustment instruction. If the target adjustment plan specifies a relative offset, the system will add or subtract the offset from the original plan time to calculate the new plan time. The update operation is usually performed through a database update statement, such as an UPDATE statement of Structured Query Language, and these update operations are wrapped in a database transaction, which means that either all flight task time adjustments are successfully committed, or if any adjustment fails or subsequent constraint check fails, all executed time adjustment operations will be rolled back, and the initial flight plan remains unchanged, which ensures the integrity of the flight plan data during the adjustment process.

[0046] After all the time update operations are temporarily stored in the database transaction, whether the adjusted flight plan meets all the operation constraints is checked, and the system starts a constraint verification module. The constraint verification module simulates the execution of the adjusted flight plan and checks various types of operation constraints, including but not limited to: aircraft turnaround constraints, i.e. whether the adjusted transit time between consecutive flight tasks performed by the same aircraft is still greater than or equal to the minimum transit time required by the aircraft model and airport regulations; airspace capacity constraints, i.e. whether the number of flights passing through a specific airspace sector in the dense period after adjustment exceeds the maximum capacity announced by the sector; airport resource constraints, i.e. whether the adjustment of flight plan time will cause new conflicts in the occupation of resources such as parking spaces, runways, etc., especially the resource competition with unadjusted flights; and policy constraints such as airport noise restrictions, curfew regulations, etc. The constraint verification module generates a detailed verification report listing all discovered constraint violations and their severity. Only when the verification report confirms that the adjusted flight plan completely meets all operation constraints, or only has acceptable, low-risk minor deviations, will the system finally commit the database transaction, making all time adjustments take effect and generating the optimized flight plan. If the verification finds serious constraint violations, the transaction is rolled back, and the system may need to record the failure reason and trigger the pre-plan selection or generation process.

[0047] After generating the optimized flight schedule, the system enters a continuous monitoring and feedback loop, continuously monitoring the execution of the optimized flight schedule and collecting plan execution data. In specific implementation, the monitoring function is achieved by subscribing to real-time data streams related to flight schedule execution, including flight tracking systems, air traffic control status messages, airport ground support systems, etc. The collected plan execution data mainly includes key time points of actual flight execution, such as actual wheel block removal time, actual takeoff time, actual landing time, actual bridge connection time, etc. The system stores these actual time stamps in association with the corresponding plan time stamps in the optimized flight schedule. By comparing the plan execution data with the optimized flight schedule, the plan deviation degree is calculated, which is a quantitative index for measuring the consistency between the flight plan and the actual execution. When calculating the plan deviation degree, the system first obtains the plan time and actual execution time of the key flights marked in the optimized flight schedule. The selection criteria for key flights can be based on node importance in the running situation topology network (such as high eigenvector centrality), flight type (such as international flights, hub transfer flights), or airline designation. For each key flight, the time deviation at a specific milestone (such as takeoff or arrival) is calculated, which is the difference between the actual execution time and the plan time (usually in minutes, positive for delay, negative for advance). Then, the system weights and integrates based on the weight of the key flight in the running situation topology network to calculate a comprehensive plan deviation degree index , whose calculation formula is as follows: Wherein: represents the final weighted average plan deviation degree, is the total number of key flights used to calculate the deviation degree, is the time deviation value of the th key flight at the main milestone, is the weight of the th key flight in the running situation topology network. This weight can be the node importance measure used in the calculation of the global influence weight, or it can be assigned according to business rules. represents taking the absolute value of the time deviation, because both advance and delay are considered as deviation from the plan.

[0048] The weighted and integrated time deviation value is compared with a plurality of preset deviation degree thresholds to determine the final plan deviation degree level. In specific implementation, the system usually sets at least three deviation degree level thresholds, for example: low deviation degree threshold , medium deviation degree threshold , and high deviation degree threshold . The comparison logic is: if If the plan deviation degree is less than 0.5, the plan deviation degree level is "low", indicating that the optimized flight plan is well executed; if If the plan deviation degree is between 0.5 and 1, the plan deviation degree level is "medium", prompting attention to the operation situation; if If the plan deviation degree is greater than 1, the plan deviation degree level is "high", indicating that there is a significant difference between the actual operation and the plan. When the plan deviation degree level is determined to be "high", or the "medium" level continues for a certain period of time under certain strategies, the system determines that the plan deviation degree exceeds the preset tolerance, at which time a new round of optimization process is triggered. Triggering a new round of optimization process means that the system will start from building the operation situation topology network corresponding to the current latest flight plan, and repeatedly execute all subsequent steps, including dynamically mapping real-time operation data, identifying new potential conflict areas, generating a new set of time adjustment plans and screening and applying, thereby forming a closed-loop and adaptive flight time optimization system that can continuously cope with changes in the operating environment and improve flight normality.

[0049] Referring to Figure 5 The figure is the core visualization result of plan execution effect monitoring: the horizontal coordinate is the key flight number, the vertical coordinate is the operation time, and the legend distinguishes between planned time and actual time. As can be seen in the figure, the actual time and the planned time of different numbered key flights have different degrees of deviation. These deviation data are weighted and fused to obtain the plan deviation degree, which is used to determine whether to trigger a new round of optimization process. The figure directly presents the execution fit degree of the optimized flight plan and is the visualization carrier of the closed-loop dynamic optimization mechanism, providing a quantifiable execution effect reference for the continuous guarantee of flight normality.

[0050] It should be noted that, in this text, relational terms such as first and second are used merely to distinguish one entity or action from another, and do not necessarily require or imply that there is any such actual relationship or order between these entities or actions. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or device.

[0051] Although embodiments of the present application have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A time optimization method based on flight regularity objectives, characterized in that, The method comprises: constructing a running situation topology network corresponding to the initial flight plan, the running situation topology network comprising running association edges between flight nodes; collecting real-time running data streams and dynamically mapping the real-time running data streams to corresponding flight nodes in the running situation topology network to update attribute states of the flight nodes; extracting network structure features according to the running situation topology network after the attribute states are updated, and identifying potential conflict areas in the running situation topology network based on the network structure features; generating a corresponding set of time adjustment plans for the identified potential conflict areas, the set of time adjustment plans comprising multiple plans with different adjustment granularities; inputting real-time running environment parameters into the set of time adjustment plans for plan screening to determine a target adjustment plan; performing time sequence adjustment on the initial flight plan according to the target adjustment plan to generate an optimized flight plan.

2. The method of claim 1, wherein, The method comprises: parsing the initial flight plan, extracting all flight tasks in the plan, and abstracting each flight task as a flight node; analyzing the time sequence dependency relationship and resource competition relationship between the flight tasks, constructing forward association edges between the flight nodes based on the time sequence dependency relationship, and constructing conflict association edges between the flight nodes based on the resource competition relationship; integrating all flight nodes, forward association edges and conflict association edges to form a running situation topology network.

3. The method of claim 2, wherein, The method comprises: continuously receiving real-time running data streams from multiple heterogeneous data sources, the real-time running data streams at least comprising flight identifiers, timestamps and running state parameters; distributing the real-time running data streams to corresponding flight nodes in the running situation topology network according to the flight identifiers; updating the attribute states of the corresponding flight nodes using the received real-time running data streams, the attribute states including planned time, estimated time, actual time and resource occupation state.

4. The method of claim 3, wherein, The method comprises: traversing the running situation topology network after the attribute states are updated, calculating the local features of each flight node, the local features including degree centrality and adjacent flight state consistency; analyzing the subgraph structure in the running situation topology network to identify subgraphs containing dense conflict association edges as candidate conflict areas; calculating the global influence weight of each candidate conflict area, the global influence weight being determined based on the position of the candidate conflict area in the running situation topology network and the importance of the flight nodes contained therein.

5. The method of claim 4, wherein, The method comprises: setting a conflict threshold, comparing the global influence weight of the candidate conflict area with the conflict threshold; screening out candidate conflict areas with a global influence weight greater than the conflict threshold and marking them as potential conflict areas; recording the flight node set contained in each potential conflict area and the associated conflict type.

6. The method of claim 5, wherein, The identified potential conflict area, generate a corresponding set of time adjustment plan, including: For each potential conflict area, analyze the timing constraints and resource constraints of its flight node set; With the goal of minimizing the total delay time, generate multiple different adjustment strategies, including time translation, flight order exchange and ground waiting; For each plan, set the adjustment granularity parameter, which defines the minimum time unit of time adjustment and the maximum number of flights allowed to adjust.

7. The method of claim 6, wherein, The real-time running environment parameters are input into the time adjustment plan set for plan screening, including: Get the current real-time running environment parameters, including airspace traffic density, airport receiving rate and weather influence coefficient; Establish a plan evaluation function, which takes real-time running environment parameters and plan adjustment parameters as input; Use the plan evaluation function to calculate the fitness evaluation value of each plan in the time adjustment plan set; Select the plan with the optimal fitness evaluation value as the target adjustment plan.

8. The method of claim 7, wherein, According to the target adjustment plan, the initial flight plan is adjusted in time, including: Parse the target adjustment plan to get the set of flight nodes to be adjusted and the target time set for each flight node to be adjusted; Locate the flight tasks corresponding to the set of flight nodes to be adjusted in the initial flight plan; Modify the planned take-off time or planned arrival time of the corresponding flight task according to the target time, and check whether the adjusted flight plan meets all the running constraints.

9. The method of claim 8, wherein, After generating the optimized flight plan, it also includes: Continuously monitor the execution of the optimized flight plan and collect plan execution data; Compare the plan execution data with the optimized flight plan to calculate the plan deviation; When the plan deviation exceeds the preset tolerance, trigger a new round of optimization process, rebuild the running situation topology network and execute the subsequent steps.

10. The method of claim 9, wherein, The calculation of the plan deviation includes: Get the planned time and actual execution time of the key flights in the optimized flight plan; Calculate the time deviation of each key flight, and weight and fuse based on the weight of the key flight in the running situation topology network; Compare the weighted and fused time deviation value with the preset multiple deviation threshold values to determine the final plan deviation level.

Images