Method and system for dynamically refreshing airport service data based on priority scheduling

Abstract

The invention relates to the technical field of airport informatization, in particular to an airport business data dynamic refreshing method and system based on priority scheduling, and the method comprises the steps: obtaining a preset service method list and a to-be-calculated instruction event list, obtaining a mapped service method according to the preset service method list, and obtaining a to-be-calculated instruction event list; and performing flow branch selection based on the updated to-be-calculated instruction event list, performing resource consumption attribute assignment in a non-low load scene, and calculating the final completion time offset of each instruction event based on the to-be-calculated instruction event list after the resource consumption attribute is updated. According to the method, the delay query opportunity and the sequence mode are dynamically adjusted, the resource consumption is distributed to each time interval in a balanced mode, the system overload risk is reduced, and the system performance is improved. And stable operation of the server in a high-business-volume scene is ensured.

Description

Technical Field

[0001] This invention relates to the field of airport information technology, and in particular to a method and system for dynamically refreshing airport business data based on priority scheduling. Background Technology

[0002] In the process of digitalization and informatization of civil airports in the civil aviation industry, the real-time and consistency of airport business data plays a key supporting role in core scenarios such as flight production scheduling, passenger service guarantee, and operational safety management. At present, the dynamic refresh of airport business data mainly adopts two technical solutions: fixed polling and business event notification refresh. However, both of these solutions have significant defects in practical applications and are difficult to balance the needs of data real-time performance, system performance, and load balancing.

[0003] On the one hand, when using the fixed polling method, user terminals need to periodically send data query requests to the backend server at preset time intervals to obtain the latest business data. Its core drawback lies in the irreconcilable conflict between real-time performance and system performance. Setting the polling interval too short in pursuit of high real-time performance will lead to a massive number of duplicate query requests on the server during peak flight periods. Conversely, setting the polling interval too long to reduce server load will significantly increase data refresh latency. On the other hand, when using the business event notification refresh method, the backend server only actively pushes data update notifications to relevant user terminals when specific business events occur. Upon receiving the notification... While this approach avoids invalid requests from fixed polling, its drawback lies in uneven server load distribution, which can easily lead to access peaks. During peak flight arrival and departure times, a large number of business events will trigger notification pushes, causing numerous user terminals to simultaneously send query requests to the server in a short period of time, forming a "request flood." This not only causes delays in data query response but may also lead to server overload and decreased service availability. Furthermore, neither of the above two approaches considers the differences in business priorities among user terminals and the need for dynamic resource scheduling. At present, a method and system for dynamically refreshing airport business data based on priority scheduling is needed. Summary of the Invention

[0004] To address the issues of difficulty in balancing real-time performance and performance, as well as poor dynamic scheduling capabilities in traditional airport business data refresh methods, this invention provides a method and system for dynamic refresh of airport business data based on priority scheduling.

[0005] In a first aspect, the present invention provides a method for dynamically refreshing airport business data based on priority scheduling, which adopts the following technical solution: A method for dynamically refreshing airport business data based on priority scheduling, comprising: S1. In response to the service method update event, obtain the preset service method list and the instruction event list to be calculated, construct the instruction event, obtain the mapped service method according to the preset service method list, and update the initial instruction event through the mapped service method. S2. Select process branches based on the updated list of instructions and events to be calculated, including obtaining the server load record set according to the list of instructions and events to be calculated, and performing low-load processing and non-low-load processing respectively according to the proportion of records in the server load record set that meet the preset load conditions. S3. Assigning resource consumption attributes in non-low load scenarios, including traversing the list of instruction events to be calculated and retrieving corresponding resource consumption data based on the attribute of the service method to be called for each instruction event; S4. Based on the list of instruction events to be calculated after updating the resource consumption attributes, load the pattern list and pattern number list of the delayed query sequence of the service method, and calculate the final completion time offset of each instruction event. S5. Update the flight dynamic event set according to the time offset, including building a time prediction model to predict the time of instruction events in the event set list; S6. Calculate and select the delayed query sequence using the updated flight dynamic event set list, and complete the dynamic refresh based on the selection result.

[0006] Furthermore, updating the initial instruction event via the mapping service method includes: S11. Based on the content of the service method update event, construct an initial instruction event, wherein the initial instruction event includes at least the content corresponding to the service method update event. Attribute information; S12. Locate the specific service method in the service method list by matching the method name with the service method name corresponding to the service method update event. ,according to Get the update tag Service methods mapped in the middle; S13. Assign the acquired mapping service method to the initial instruction event. The attribute, along with a globally unique ID generated by the UUID generator, is assigned to the initial command event. The attribute, which maps the service method's preset initial execution time, is assigned to the initial instruction event. The attribute completes the update of the initial instruction event and adds it to the list of instruction events to be calculated; Where 'o' represents the total number of original instruction events in the instruction event list to be calculated. As a unique identifier for the instruction event, Update the method parameter data corresponding to the event for the service method. Update the event-related flight number to the service method. The service method that triggers this instruction event, To determine the start time of the event service method that triggers this instruction. The service method to be invoked. The initial execution time of the service method to be called. List of preset service methods The first service method.

[0007] Furthermore, the assignment of resource consumption attributes under non-low-load scenarios includes: S31. Load the service method execution average resource consumption preset list H, where each service method in the average resource consumption preset list... The resource consumption is: ; in, For the first One method Resource consumption varies by time period. For the first Service method for segmented time intervals Resource consumption; S32. Traverse the list of instruction events to be calculated, and obtain each instruction event to be calculated in turn. ; S33, Extraction The service method to be called and with To perform a matching search for keywords in H, locate and... The corresponding service method is associated with ; S34, retrieve S34 The resource consumption data is divided into time periods and assigned to the instruction event to be calculated. Resource consumption attributes ; in, For the instruction event to be calculated, The service method to be invoked. For a single Resource consumption data for the service method, where k is the k-th specific instruction event in the list of instruction events to be calculated. Let be the i-th specific resource consumption in H.

[0008] Furthermore, the calculation of the final completion time offset of each instruction event includes the following steps: S41. Using operations research optimization methods to generate a list of delayed query sequence patterns. Optimize computation by constructing an operations research optimization objective function using computational resource overload penalty terms, query latency penalty terms, and load balancing standard deviation terms; S42. Use a mixed integer programming solver to solve the objective function. When the solution time exceeds the set threshold, switch to the NSGA-II heuristic algorithm to solve the objective function, and update the optimization results obtained to the list of delayed query sequence pattern numbers N. S43, Event pending calculation instruction of Based on the matching criteria, the corresponding sequence pattern list P and pattern number list N are retrieved. and ,Will The Chinese number is The delayed query sequence pattern is assigned to Delayed query sequence attribute ,Finish Delay the update of the query sequence; The objective function expression is: ; in, for The preset selected delayed query sequence pattern number, For service methods List of delayed query sequence patterns for Delayed query sequence, Penalty for resource overload Penalty for query delay The standard deviation of load balancing is formed. Weighting for resource overload penalties. For query delay penalty weight, The standard deviation weights are used for load balancing.

[0009] Furthermore, the calculation of the final completion time offset of each instruction event also includes, for each Matched Extract the next delayed query sequence pattern Simultaneously, in conjunction with the service method execution average resource consumption preset list H, and... corresponding , obtain Calculate the amount of resource consumption in different time periods. The final completion time offset is calculated and the result is assigned to... The final completion time offset attribute, the formula for calculating the final completion time offset is: ; in, For the instruction event to be calculated The final completion time offset, For service methods The maximum offset time of the last delayed query sequence pattern. for The last delayed query sequence pattern in the middle, for Total number of time allocation points for delayed queries. For service methods Resource consumption by time period.

[0010] Furthermore, updating the flight dynamic event set based on the time offset includes loading the flight dynamic data list S and the flight dynamic event set list K, and extracting... Triggering service method properties Related flight number attributes ,based on Lock the corresponding flight dynamic event set list K And then according to locking The corresponding instruction event ,when Final sending time attribute When empty, of , , Assigned to ,when of If it is not empty, then in middle Added later ,Finish Update K from the instruction event; in, for The associated flight number attribute, for The start time of the event service method that triggered this instruction. For dynamic flight instruction event sets, for The e-th instruction event, To update the event corresponding method parameter data for the obtained response service method, To determine the start time of the event service method that triggers this instruction. The initial execution time of the service method. for The final sending time attribute.

[0011] Furthermore, the step of updating the flight dynamic event set based on the time offset also includes constructing a time prediction model consisting of an input layer, a bidirectional LSTM hidden layer, an attention layer, a fusion layer, and an output layer; calculating the maximum predicted interval offset of the instruction event list I to be calculated; combining the current time and the maximum interval offset to filter the instruction event set that needs to be time predicted in the flight dynamic event set list K; inputting the filtered instruction event set into the time prediction model to obtain the instruction event update method start time prediction value; and updating the time attribute of the corresponding instruction event after standardizing the prediction value.

[0012] Furthermore, the dynamic refresh based on the selection result includes: filtering and constructing a set of already occurred instruction events that overlap with the predicted time interval based on the updated flight dynamic event set list K; creating a timing interval; assigning a value to the initial execution time attribute of the service method to be called for the unsent instruction events in K in conjunction with the average resource consumption preset list H; configuring a delayed query sequence for the unsent instruction events in K; and updating the delayed query sequence by combining the real-time user terminal list to count the number of user terminals at each query allocation point.

[0013] Furthermore, the dynamic refresh based on the selection result also includes constructing a cumulative variable of resource consumption over a time interval, accumulating the resource consumption of already occurred instruction events and instruction events not sent within the prediction period, constructing a set of calculation instruction events sorted by priority, traversing the set and judging based on resource thresholds, selecting an appropriate delayed query sequence, assigning a sending time to the instruction events in the instruction event list I to be calculated based on the selection result and distributing them to the user terminal, thereby completing the dynamic refresh of business data.

[0014] Secondly, a priority-based scheduling-based dynamic refresh system for airport business data includes: The data acquisition module is configured to: acquire a list of preset service methods and a list of instruction events to be calculated; acquire the mapped service methods according to the list of preset service methods; and update the initial instruction events through the mapped service methods. The event handling module is configured to: select process branches based on the updated list of instruction events to be calculated, including obtaining the server load record set according to the list of instruction events to be calculated, and performing low-load processing and non-low-load processing respectively according to the proportion of records in the server load record set that meet the preset load conditions; The resource calculation module is configured to: assign resource consumption attributes in non-low load scenarios, including traversing the list of instruction events to be calculated and retrieving corresponding resource consumption data based on the attributes of the service methods to be called for each instruction event; The offset module is configured to: load the list of modes and mode numbers of the delayed query sequence of the service method based on the list of instruction events to be calculated after updating the resource consumption attributes, and calculate the final completion time offset of each instruction event; The event prediction module is configured to update the flight dynamic event set based on the time offset, including building a time prediction model to predict the time of instruction events in the event set list; The update module is configured to: calculate and select the delayed query sequence using the updated flight dynamic event set list, and perform dynamic refresh based on the selection results.

[0015] In summary, the present invention has the following beneficial technical effects: 1. This invention utilizes service method list mapping and instruction event attribute completion techniques. After constructing initial instruction events based on service method update events, it accurately filters valid update-type service methods and obtains corresponding mapped service methods by leveraging the update markers and mapping relationships of service methods in a preset service method list. Simultaneously, it generates globally unique identifiers using UUIDs and supplements attributes such as the initial execution time of the service methods to be called, ensuring that each instruction event has complete attributes and a clear attribution. This method avoids non-update-type service methods triggering invalid instruction events, reduces redundant data occupation, and provides a unified data foundation for subsequent load judgment and resource calculation through standardized attribute configuration, thereby improving the accuracy of instruction event processing.

[0016] 2. This invention is based on statistical analysis of server load record sets. It dynamically selects two processing flows—low load and non-low load—based on the proportion of records meeting preset load conditions. In non-low load scenarios, resource consumption attributes are precisely assigned to instruction events using a service method execution average resource consumption list. Combined with a delayed query sequence pattern list and an optimized pattern number list, a delayed query sequence is configured, and the final completion time offset is calculated. In low load scenarios, the process is simplified to reduce computational consumption. This scenario-based processing method can flexibly adjust the processing logic according to the actual server load, avoiding service lag caused by resource overload in non-low load scenarios and resource waste caused by excessive computation in low load scenarios, achieving a dynamic balance between system performance and business real-time performance.

[0017] 3. This invention constructs a time prediction model, combining flight dynamic data with the preceding temporal features of instruction events, to select instruction event sets within an effective time window for time prediction, and then standardizes the prediction results. Compared to traditional fixed-time allocation methods, this model uses a bidirectional LSTM to learn the forward and reverse temporal dependencies of instruction events, and combines an attention layer to focus on key influencing features, significantly improving the prediction accuracy of the instruction event update start time. This ensures that the temporal attributes of the flight dynamic event set match the actual business scenario, providing an accurate time benchmark for subsequent delayed query sequence selection, and further guaranteeing the real-time nature of business data updates.

[0018] 4. This invention constructs a cumulative variable for resource consumption over a time interval by using cumulative resource consumption calculation and dynamic selection of delayed query sequences. It accumulates the resource consumption of already occurred instruction events and those within the prediction period, and then combines this with a service method priority-based calculation of the instruction event set. Based on resource thresholds, it dynamically selects suitable delayed query sequences. Simultaneously, it combines real-time user terminal lists to count the number of terminals at each query allocation point, ensuring that the delayed query sequence matches the terminal distribution. This mechanism effectively avoids server resource spikes caused by a large number of instruction events being triggered simultaneously. By dynamically adjusting the timing and sequence mode of delayed queries, resource consumption is evenly distributed across time intervals, reducing the risk of system overload and ensuring stable server operation under high-volume scenarios. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall process of a method for dynamically refreshing airport business data based on priority scheduling, according to Embodiment 1 of the present invention. Detailed Implementation

[0020] The present invention will be further described in detail below with reference to the accompanying drawings.

[0021] Example 1 Reference Figure 1 This embodiment of a method for dynamically refreshing airport business data based on priority scheduling includes: Before S100, there is an initialization operation, as detailed below: C100: In response to the initialization event, construct a list of preset service methods based on preset configuration data. Fill in the specific service methods according to the configuration data. of Construct a list of instruction events to be computed. Build a server load record set ,in, The method name. For updating method markers, For method priority, for Mapped service methods.

[0022] C200: Based on preset configuration data, construct a preset list of average resource consumption for service method execution. Simultaneously, a list of service method delayed query sequence patterns is constructed based on the preset service method delayed query sequence configuration data. Based on the preset delayed query sequence pattern for each service method, select the configuration and build... Create an abnormal delayed query sequence Abnormally delayed query sequences will filter out low-priority user terminals.

[0023] C300: Query and retrieve flight status data, and construct a flight status data list based on this data. At the same time, a list of flight dynamic events is constructed. traversal To obtain specific Based on the relationship between flights and update methods, the order is as follows: Create instruction event and combined and ,for renew , , Attributes, traversal complete The relationship between flights and update methods is fixed; specific business events will trigger the update method.

[0024] The relationship between flights and update methods is fixed; specific business events will trigger the update method. An example update method is shown below: Table 1: Example of Update Methods

[0025] C400: Create a real-time user terminal inventory Query and obtain the user terminal's preset configuration information, and combine it with the user terminal's current login information to... Fill in the specific user information, including the user terminal ID and priority.

[0026] C500: Create a scheduled calculation task for instruction event handling, with a set time. Configure the settings according to the presets and start the scheduled task.

[0027] like Figure 1 As shown, S100: In response to the service method update event, obtain the list of preset service methods. Get the list of instruction events to be calculated Construct instruction events, and update the events and their corresponding service methods based on the responses. Get the mapping service method, update the command event, and put the command event into .

[0028] Specifically, in response to a service method update event, a list of preset service methods is obtained. in For the total number of service methods, for The first service method, ,in, The method name. For updating method markers, For method priority, for The mapped service method, and the service method for determining the response. If the method is not an update method, the process will exit directly without further operations. If it is an update method, the process will retrieve the method parameter data corresponding to the update event of the response service method and obtain the list of instruction events to be calculated. in The total number of instruction events to be calculated. For the first One instruction event to be calculated, ,in For the instruction event ID, To update the method parameter data corresponding to the event in the obtained response service method, Update the event-related flight number to the service method. The service method that triggers this instruction event, To determine the start time of the event service method that triggers this instruction. for The initial execution time of the service method The service method to be invoked. For a single Resource consumption of service methods This is a delayed query sequence used to guide different user terminals on when to query backend data. This is obtained based on the longest delayed query sequence pattern. Service method final completion time offset, Construct the instruction event based on the final sending time of this instruction event. Update events and responses based on the service method provided. Get the mapping service method and update this data. of ,Will Put in Correspondingly Increment by one, where 'o' represents the number of events already in the list.

[0029] Following S100, the following steps may also be included: M100: Continuously records abnormal server operation events. Here, abnormal events refer to non-crash exceptions, that is, the occurrence of abnormal events will not cause service unavailability. The abnormal event is the service method running timeout. If the proportion of service method timeouts within 5 minutes exceeds 15% of the total number of accessed service methods, then continue the subsequent operations. If it does not exceed 15%, then execute S200-S700 normally.

[0030] M200: Traversal To obtain specific ,according to middle ,for Set abnormal delayed query sequence ,for Assign the current time and complete the iteration. .

[0031] M300: Loading Flight Dynamic Data List Load the list of flight dynamic events traversal To obtain specific traversal To obtain specific To judge its Is it empty? If not, skip this step. Determine the condition and continue iterating. If its If empty, then iterate through. To obtain specific ,according to of and locking ,if of If empty, then of , , , , Assigned to Conversely, in middle Added later Traversal complete Traversal complete Traversal complete .

[0032] M400: will The command event is sent to each user terminal in the form of a message. After sending, the message is cleared. User terminal Upon receiving the instruction event Then, obtain its delayed query sequence. Based on its own preset priority Find the offset of the time point allocated in the delayed query sequence. Optional, with Centered on a point, the value can be randomly changed within 500ms to 1000ms before and after the change, and the changed value will be reassigned. The user terminal is Create a query task, and the delay time is equal to ,in, Assign time offsets to the delayed query sequence.

[0033] S200: Obtain server load record set Check the load conditions sequentially; if the conditions are met... occupy If 95% of the total number of cases is a low-load scenario, then execute B100-B300 below; otherwise, if the number of cases is not a low-load scenario, then execute S300-S700 below.

[0034] Specifically, obtain the server load record set. in The total number of server load records. This dataset, consisting of system load data collected at 500ms intervals, employs a rolling update strategy, retaining data for approximately 30 minutes. For the first Each server load record ,in, For the monitored CPU utilization, For the monitored memory usage, For the monitored IO utilization, To monitor network bandwidth utilization, traverse To obtain specific Judge in sequence ,in, , , , These are user-adjustable thresholds for CPU utilization, memory utilization, I / O utilization, and network bandwidth utilization, which, for example, could be 20%, 30%, 15%, and 10%, respectively. If the judgment conditions are met... occupy If the total is 95%, then execute B100-B300 below; otherwise, execute S300-S700 below.

[0035] Following S200, the following steps are also included: B100: List of Delayed Query Sequence Patterns for Loading Service Methods Load service method to set delayed query sequence pattern number list ,in, For the i-th service method in the predefined service method list F The corresponding delayed query sequence pattern number, For the i-th service method in the predefined service method list F Iterate through the corresponding set of delayed query sequence patterns. To obtain specific ,according to middle , combined ,for Set a preset delayed query sequence for Assign the current time and complete the iteration. .

[0036] B200: Loading Flight Dynamic Data List ,in, For the complete set of business information corresponding to the l-th dynamic flight in the flight dynamic data list S, load the flight dynamic event set list. , For the flight dynamic event set list K, the corresponding dynamic flight l A dedicated set of command events, traversed To obtain specific traversal To obtain specific To judge its Is it empty? If not, skip this step. Determine the condition and continue iterating. If its If empty, then iterate through. To obtain specific ,according to of and locking ,if of If empty, then of , , , , Assigned to Conversely, in middle Added later Traversal complete Traversal complete Traversal complete .

[0037] B300: will The command event is sent to each user terminal in the form of a message. After sending, the message is cleared. User terminal Upon receiving the instruction event Then, obtain its delayed query sequence. Based on its own preset priority Find the offset of the time point allocated in the delayed query sequence. Optional, with Centered on a point, the value can be randomly changed within 500ms to 1000ms before and after the change, and the changed value will be reassigned. The user terminal is Create a query task with a delay time of [time value missing]. .

[0038] S300: Load the default list of average resource consumption for service method execution. traversal To obtain specific ,according to of Attribute Search Get the resource consumption of the service method and assign it to the appropriate value. of ,Finish Traversal.

[0039] Specifically, the average resource consumption preset list for loading service methods is executed. ,in, For service methods Resource consumption situation, in, For the first One method Resource consumption varies by time period. For the first Service method for segmented time intervals Resource consumption situation, Service methods are evaluated in 500ms increments. The execution time is divided into multiple time intervals. To enable fast subsequent calculations, all time intervals are 500ms. Any shortfall at the end of the service method execution is also converted to 500ms. ,in, , , , The first The usage and consumption of CPU, memory, I / O, and network bandwidth in each segmented time interval; traversal To obtain specific ,according to of Attribute Search Get the resource consumption of the service method and assign it to the appropriate value. of ,Finish Traversal.

[0040] The average resource consumption for service method execution in S300 can be calculated through the following steps: F100: Bytecode enhancement techniques are used to inject timing logic before and after the method to be monitored. A globally unique method execution context ID is created at the method entry point using a UUID generator. During method execution, the context ID is bound to a ThreadLocal variable. For existing threads, a new thread is created, passing the parent thread's context ID, and a global ConcurrentHashMap is created.<Long,String> The structure mapping table records the mapping from thread ID to method execution context ID. At the method execution entry point, the current thread ID and method execution context ID are stored in the mapping table. When the method returns, the context ID is popped from the stack, the corresponding record is removed from the mapping table, and the method execution time is recorded.

[0041] F200: Obtain thread CPU time through the thread management interface, calculate CPU utilization between adjacent sampling points, with a sampling frequency of 500ms, calculate thread CPU utilization, find the corresponding method through the mapping table, and record the method execution status.

[0042] F300: Through the memory management interface, obtain the usage of heap memory and non-heap memory, record the memory usage at each sampling time, with a sampling frequency of 500ms, calculate the memory usage rate, find the corresponding method through the mapping table, and record the method execution status.

[0043] F400: Wraps a standard IO class using the decorator pattern, counts the number of IO operation bytes during method execution, and converts it into IO device utilization. The sampling frequency is 500ms. For synchronous IO, the utilization is directly assigned to the method corresponding to the current method context ID. For asynchronous IO, the submitter method context ID is recorded when the IO task is submitted, and the IO statistics of the corresponding method are updated in the IO completion callback, recording the method execution status.

[0044] F500: By intercepting network socket operations, it counts the number of bytes sent and received over the network during method execution and converts them into a percentage of network bandwidth usage. The sampling frequency is 500ms. It obtains the method context ID of the current thread, finds the corresponding method through the mapping table, and records the method execution status.

[0045] F600: For the recorded method execution status, remove exception records, calculate the average execution time of the method and the consumption of specific resources in time slices (500ms), and then create... The content is then filled in. Since each sampling calculation will affect service performance, the relevant sampling calculation logic is only triggered manually when needed.

[0046] In S300, the average resource consumption of service method execution can also be analyzed and calculated by collecting server operation logs, monitoring data, and performance data. First, the resource usage of a single, non-overlapping service method is analyzed. Then, based on this, the resource consumption of multiple overlapping service methods is analyzed, ultimately calculating the average resource consumption and runtime of each service method. Optionally, if there are very few single, non-overlapping service methods in the collected data, on-site testing can be conducted during system updates and maintenance outside of flight operation hours to generate relevant data, and finally, the on-site environment can be restored.

[0047] S400: Loading Service Method Delayed Query Sequence Pattern List Load service method to set delayed query sequence pattern number list traversal To obtain specific ,according to of Attribute Search Based on the search value and ,renew of , obtain The last delayed query sequence pattern of the retrieved value , combined Calculate the final completion time offset and assign it to of Attributes, traversal complete .

[0048] Specifically, the loading service method delays the query sequence pattern list. ,in, For service methods A list of delayed query sequence patterns, only when A specific list of delayed query sequence patterns is only available when querying service methods. in For the first One method The total number of delayed query sequence patterns, For service methods The A delayed query sequence pattern, In the delayed query sequence pattern, the query time increases as you move further down the sequence. in For the first Delayed query sequence pattern The total number of delayed query time allocation points. for The Each delayed query time allocation point , To delay the query offset time and ensure fast subsequent calculations, the offset time is set with a minimum unit of 500ms. For priority filtering conditions that can be queried at this time allocation point, for example, a specific time allocation point requires that the executable user terminal priority be greater than or equal to 8. According to The actual number of user terminals per delayed query allocation point is calculated based on the conditions; the service method is loaded to set the delayed query sequence pattern number list. ,in For service methods Set the delayed query sequence pattern number; iterate through... To obtain specific ,according to of Attribute Search Find the corresponding At the same time, ,according to and ,Will The corresponding number of the delayed query sequence pattern is assigned to of , obtain The last delayed query sequence pattern , obtain The corresponding The corresponding resource consumption amounts for different time periods are obtained. ,Will Assigned to of Attributes, traversal complete .

[0049] An example of a delayed query sequence is as follows: Table 2 is the delayed query sequence list.

[0050] S400 Service Method Setting Delayed Query Sequence Pattern Number List Setting the value in the middle is achieved through the following steps: H100: Responding to the rolling time domain Calculate the event, load the preset event time prediction model based on LSTM technology, and obtain... traversal To obtain specific ,according to Get the corresponding ,judge The data includes the actual departure time from the previous station, the changed departure time from the previous station, the planned departure time from the previous station, and the previous station CTOT attribute. If the actual departure time from the previous station is not empty, then... This is the actual departure time from the previous station. Otherwise, it checks if the previous station's CTOT (Critical Time of Occurrence) is empty. If it is not empty, then... If the previous station's departure time is CTOT, then check if the previous station's departure time is empty. If it is not empty, then... To change the departure time from the previous station, or vice versa. To determine the departure time from the planned previous station The actual departure time, changed departure time, planned departure time, and CTOT attribute are displayed. If the actual departure time is not empty, then... This is the actual takeoff time; otherwise, it checks if CTOT is empty. If it is not empty, then... If it's CTOT, then check if the change of departure time is empty; if not, then... To change the departure time, or vice versa. To determine the planned takeoff time, when... and When there is an overlap with the prediction period, iterate through... To obtain specific ,if of If not empty, skip this step. Continue traversing Conversely, if it is empty, the event time prediction model is used. , The preceding event time and flight characteristic data are input into the model. The start time of the predicted instruction event update method is then processed to be divisible by 500 milliseconds and entered into the system. until Traversal complete, traversal complete .

[0051] H200: List of delayed query sequence pattern numbers using operations research optimization techniques The optimization calculation is specifically performed using the following objective function: ; Among them, the overall goal Penalty for resource overload Query delay penalty and load balancing standard deviation constitute , Weighting for resource overload penalties. For query delay penalty weight, For load balancing standard deviation weights, Exemplary It is 0.4. It is 0.3. It is 0.3.

[0052] Resource overload penalty The calculation logic is to divide the prediction period into time slices according to fixed time units, and accumulate the portion of the total resource consumption of the update method and query method in each time slice that exceeds the resource threshold. The calculation method is as follows: ; in, The prediction period is divided into time slices with a minimum interval of 500ms, representing the total number of prediction time slices. For example, if the prediction period is 1 hour, then... , Indicates time slice Resource consumption of internal update methods For time slices Resource consumption of inner query method This is the resource threshold vector.

[0053] Query delay penalty The calculation logic is to calculate the average query latency by weighting it according to the service method priority. The calculation method is as follows: ; in, As decision variables, Representative query method Select the Various delay modes, ,and , The total number of query methods. For method The number of preset modes. For decision variables, the representation method is as follows. Select the In the first delay mode, the first Time offset of each query. ,in For pattern The number of delay time points, For method priority, For method Select the In the delay mode, at a certain time point The number of terminals at any given time.

[0054] Load balancing standard deviation Used to measure the degree of fluctuation in resource utilization across different time slices. The calculation method is as follows: ; Meanwhile, the solution also satisfies the constraint that the method can only select one delayed query mode within a solution interval. This constraint is a key boundary rule set for the operations optimization calculation process of the delayed query sequence mode number list N. The core is to clarify that within a complete solution interval (i.e., the dynamic refresh task cycle for a certain batch of instruction events to be calculated, which matches the prediction cycle of flight dynamic events; in this embodiment, it is divided into 500ms as the minimum time unit, for example, a 1-hour prediction cycle corresponds to 7200 time slices), any service method (including the query method and update method to be called, with a focus on the query method) can only select from its corresponding delayed query sequence mode list. Choose a single delayed query sequence mode to guide the allocation of delayed query timings for all instruction events associated with this method. Two or more modes cannot be selected simultaneously. For service methods The set of delayed query sequence patterns means that if the same method selects multiple delayed patterns within the solution interval, the resource consumption (such as CPU usage and memory usage) of its associated instruction events will correspond to multiple time allocation dimensions. This will cause the calculation results of the cumulative resource consumption variable D (used to count the total resource consumption within each 500ms time slice) to overlap or contradict each other. For example, if a method selects both pattern A (the query is triggered in the 3rd time slice) and pattern B (the query is triggered in the 5th time slice) at the same time, the resource consumption statistics of the 3rd and 5th time slices in D will be repeatedly included in the consumption data of this method, which will lead to the resource threshold judgment.

[0055] B300: A mixed-integer programming solver (such as CPLEX) is used to solve the problem. If the solution time is too long, all time-dimensional calculations in this embodiment are performed in 500ms increments. For example, the prediction period is divided into 500ms time slices (e.g., a 1-hour prediction period corresponds to 7200 time slices), and service method resource consumption is statistically divided into 500ms time periods. If the solution time of the mixed-integer programming solver (such as CPLEX) exceeds 500ms, i.e., the set threshold, it will cause a single N optimization calculation to occupy more than one system base time slice, thus affecting subsequent steps. In this case, the NSGA-II heuristic algorithm is used. After the solution is completed, The optimization results have been updated to... middle.

[0056] When solving the objective function using a mixed-integer programming solver or the NSGA-II algorithm, the above equality constraints are embedded as hard constraints into the solution logic: if using the CPLEX solver, this is done by calling the solver.addConstraint() interface. This is transformed into a linear constraint expression, ensuring that the solver searches for the optimal solution only within the solution space that satisfies this constraint. Given the total number of delay modes corresponding to method i, if the NSGA-II algorithm is used, during population initialization, for each chromosome segment (encoded) corresponding to method i... The values ​​of the chromosomes are validated, and only chromosomes that satisfy the activation of a single pattern are retained. After crossover and mutation operations, the newly generated chromosomes are validated again, and individuals that violate the constraints are removed to ensure that all individuals in the population meet the constraints.

[0057] S500: Loading Flight Dynamic Data List Load the list of flight dynamic events ,renew From the command event In the middle, calculate the maximum interval offset of this prediction. Load the pre-defined event timing prediction model based on LSTM technology, predict the timing of instruction events, and update the data. middle.

[0058] Specifically, load the flight dynamic data list. ,in The total number of flights is dynamic. For the first Each dynamic flight loads a list of flight dynamic events. ,in For dynamic flights The set of instructions and events, in For dynamic flights The total number of instruction events, For dynamic flights The One instruction event; traversal To obtain specific ,according to of and locking ,if of If empty, then of , , Assigned to Conversely, in middle Added later Traversal complete Let the maximum interval offset be the one predicted in this case. The default value is 0, and the iteration is performed. To obtain specific ,judge If true, then Otherwise, continue iterating. Traversal complete Load the pre-defined event time prediction model based on LSTM technology and iterate through it. To obtain specific ,according to Get the corresponding ,judge The data includes the actual departure time from the previous station, the changed departure time from the previous station, the planned departure time from the previous station, and the previous station CTOT attribute. If the actual departure time from the previous station is not empty, then... This is the actual departure time from the previous station. Otherwise, it checks if the previous station's CTOT (Critical Time of Occurrence) is empty. If it is not empty, then... If the previous station's departure time is CTOT, then check if the previous station's departure time is empty. If it is not empty, then... To change the departure time from the previous station, or vice versa. To determine the departure time from the planned previous station The actual departure time, changed departure time, planned departure time, and CTOT attribute are displayed. If the actual departure time is not empty, then... This is the actual takeoff time; otherwise, it checks if CTOT is empty. If it is not empty, then... If it's CTOT, then check if the change of departure time is empty; if not, then... To change the departure time, or vice versa. The current time is set to the planned departure time. ,if or Then skip the current one. Continue traversing Conversely, traversal To obtain specific Using an event time prediction model, , The preceding event time and flight characteristic data are input into the model. The start time of the predicted instruction event update method is then processed to be divisible by 500 milliseconds and entered into the system. At the same time, according to , and ,for Assignment to delayed query sequence pattern until Traversal complete, traversal complete .

[0059] The S500's default LSTM-based event-time model can be implemented using the following steps: An event time prediction model is constructed based on a bidirectional LSTM network. The model includes an input layer, a bidirectional LSTM hidden layer, an attention layer, a fusion layer, and an output layer: Input layer: receives... Time base data (actual departure time from previous station / previous station CTOT / changed departure time from previous station / planned departure time from previous station, actual departure time / CTOT / changed departure time / planned departure time), The system generates a bidirectional LSTM hidden layer containing forward and backward LSTM units. The forward LSTM learns the temporal dependencies of the command events from front to back, while the backward LSTM learns the inverse temporal dependencies from back to front, outputting a bidirectional temporal feature vector. An attention layer calculates weights on the bidirectional temporal feature vector output from the bidirectional LSTM hidden layer, focusing on the historical event features that have the greatest impact on the current time prediction, resulting in an attention-weighted feature vector. A fusion layer fuses the attention-weighted feature vector with the global flight features (encoded data such as aircraft type and gate) to generate a comprehensive feature vector. The output layer outputs the feature vector through a fully connected layer. China The predicted start time value for the instruction event update method; E100: Prepare training data, including historical flight data and command event call history data. Preprocess the data. For some key moments in the historical flight data, the planned, changed and actual attribute forms are merged into a unique key moment of the same kind with the actual attribute as the actual value. Convert all times into seconds from the reference time point to form a unified numerical representation.

[0060] E200: One-hot encoding technology is used for attributes such as aircraft type, aircraft number, aircraft position, and boarding gate in historical flight data. Hash encoding technology is used for flight number. One-hot encoding technology is used for the instruction event update method in the historical data. At the same time, the trigger time of the instruction event update method around the same flight is converted into the time difference between the current event and the previous event.

[0061] E300: For each flight's instruction event sequence, extract the features (type, time, flight attributes, etc.) of the first m events in chronological order as input, and the actual execution time of the (m+1)th event as output to construct training sample pairs. Divide the training data into training set, validation set, and test set in an 8:1:1 ratio, with each batch containing instruction event sequences from multiple flights.

[0062] E400: The training model constructs a bidirectional LSTM network to process event sequences. The forward LSTM learns event dependencies from front to back, while the backward LSTM learns inverse dependencies from back to front. An attention layer is designed to calculate the importance weights of features at each time step, focusing on the most critical historical instruction events for the current prediction. The attention-weighted features are fused with global flight features, and a fully connected layer generates an instruction event update method to execute time prediction. In terms of specific training details, the weighted average absolute error (WMAE) is used as the loss function, batch stochastic gradient descent is used to optimize model parameters, and an early stopping strategy is used to prevent overfitting.

[0063] S600: Create a set of instruction events whose delayed query sequence of occurred instruction events coincides with the predicted time interval. ,filter The overlapping time of the delayed query sequence of already occurred instruction events is calculated. Based on the time rotation frequency, the start time of the execution of the query method for no instruction events occurring within the prediction period is calculated. The number of user terminals at different allocation points of the delayed query queue for instruction events that meet the conditions is calculated, and a cumulative calculation variable for resource consumption over the time interval is constructed. Accumulate the resource consumption of delayed query sequences of occurred instruction events, accumulate the resource consumption of instruction event update methods within the prediction period, and construct a computational instruction event set. Sort by priority traverse sequentially , and perform the calculation and selection of the instruction event delayed query sequence.

[0064] S610: Create a set of instruction events whose delayed query sequence of occurred instruction events coincides with the predicted time interval. in This represents the total number of instruction events whose delayed query sequence and prediction time interval overlap. For the first For each occurrence of a delayed query sequence of instruction events that overlaps with the predicted time interval, iterate through the sequence. To obtain specific traversal To obtain specific ,judge Is it empty? If it is empty, skip this step. Determine the condition and continue iterating. ,if Not empty, combined and ,get The final completion time of the delayed query sequence ,if Then Put in .

[0065] S620: The creation instruction event handling time interval is... in The timer interval is user-adjustable and is an integer multiple of the minimum time of 500ms. Then it is a continuously increasing integer used to represent processing time intervals at different times, for example... At that time, the time interval is traversal To obtain specific traversal To obtain specific To judge its Is it empty? If not, skip this step. Determine the condition and continue iterating. If its If it is empty, then check its... Is it empty? If it is empty, skip this step. Determine the condition and continue iterating. If its If not empty, then according to and Get specific ,judge The instruction event processing time interval to which it belongs, let's assume it is... Then Assigned to Traversal complete Traversal complete .

[0066] S630: Traversal To obtain specific traversal To obtain specific To judge its Is it empty? If not, skip this step. Determine the condition and continue iterating. If its If empty, then according to of and , ,for Assign the corresponding delayed query sequence and load the real-time user terminal list. in This represents the total number of user terminals currently in operation. For the first One user terminal. , for ID, for priority, User terminal types, for example, include client, web, and mobile terminals. The service type of the user terminal includes collaborative decision-making, flight support, operations control, airworthiness assurance, safety management, passenger services, and others. The importance of user terminal positions is categorized into different levels of importance: general, important, and very important. For the management level of the user terminal, including different management levels such as frontline, team, dispatch, supervision, management, and decision-making, obtain... Delayed query sequence Assuming Then traverse To obtain specific ,according to of Conditional traversal ,judge of Does the condition meet? If so, then... of The accumulation technique is used to finally complete the process. Traversal, Reassignment .

[0067] S640: Constructing cumulative resource consumption calculation variables over a time interval in The total number of time intervals. , For the first Cumulative resource consumption over a given time interval The corresponding time interval is , ,in , , , The first The cumulative CPU, memory, I / O, and network bandwidth usage over a given time interval is displayed, all defaulting to 0; iterating through... To obtain specific , obtain The delayed query sequence, assumed to be traversal To obtain specific ,according to of as well as Find the resource consumption situation. The matching time interval set is traversed to obtain the specific... ,Will resource consumption corresponding The resource consumption for each time period is divided into segments, with CPU, memory, I / O, and network bandwidth consumption multiplied by the respective percentages. Then add them in separately Finally, the traversal is complete. ; Traversal To obtain specific traversal To obtain specific To judge its Is it empty? If not, skip this step. Determine the condition and continue iterating. If its If empty, check Is it If it is a command event, skip this step. Determine the condition and continue iterating. If not, continue the evaluation. Is it empty? If it is empty, skip this step. Determine the condition and continue iterating. If it is not empty, then according to of exist Obtain the corresponding resource consumption information. ,according to and ,turn up The matching time interval set is traversed to obtain the specific... ,Will correspond Resource consumption across different time periods is analyzed, with CPU, memory, I / O, and network bandwidth consumption added up sequentially for each period. Finally, the traversal is complete. Traversal complete .

[0068] S650: Constructing a set of computation instruction events in This represents the total number of instruction event sets to be calculated subsequently. For the first One instruction event, The command events are stored in order of priority according to the update service method; command events with higher priority are scheduled to be placed in [the appropriate location]. Before; traversal To obtain specific traversal To obtain specific To judge its Is it empty? If not, skip this step. Determine the condition and continue iterating. If its If it is empty, continue the evaluation. Is it empty? If it is empty, skip this step. Determine the condition and continue iterating. If it is not empty, Put in Traversal complete Traversal complete ,according to Using quicksort to sort The command events are sorted according to the priority of the update method.

[0069] S660: Traversal Receive specific instruction events ,get The delayed query sequence, assumed to be traversal To obtain specific ,according to of as well as Find the resource consumption situation. The matching time interval set is traversed to obtain the specific... ,Will resource consumption corresponding The resource consumption for each time period is divided into segments, with CPU, memory, I / O, and network bandwidth consumption multiplied by the respective percentages. Later obtained , , , ,judge ,in , , , This is a user-adjustable resource consumption threshold, for example, it could be 80%, 85%, 80%, or 75%. When the condition is met, then... , , , Accumulation If the condition is not met, then according to The service method to be called and , to obtain the corresponding ,exist Select The deferred query sequence pattern immediately following the current deferred query sequence is assigned a value. The delayed query attribute is then calculated using the same steps as above to obtain a new result. , , , Continue to evaluate the condition; if it is not met, continue in... Replace the immediately following delayed query sequence pattern until the judgment condition is met or The current sequence no longer contains a immediately following delayed query sequence pattern. , , , Accumulation Traversal complete .

[0070] S700: Traversal , obtain ,for In Assignment Traversal complete ,Will The command event is sent to each user terminal in the form of a message. After sending, the message is cleared. User terminal Upon receiving the instruction event Then, obtain its delayed query sequence. Based on its own preset priority Find the offset of the time point allocated in the delayed query sequence. Optional, with Centered on a point, the value can be randomly changed within 500ms to 1000ms before and after the change, and the changed value will be reassigned. The user terminal is Create a query task with a delay time of [time value missing]. .

[0071] The preset terminal priority is implemented through the following steps: L100: Obtain user terminal preset configuration data and obtain real-time user terminal list. traversal To obtain specific Fill in sequentially according to the user terminal's preset data. of property.

[0072] L200: Loading factor quantification standard. Specifically, in terms of user terminal type, client is 10, web is 2, and mobile is 6. In terms of the business type of the user terminal, collaborative decision-making is 10, flight support is 6, operation control is 8, airworthiness support is 1, safety management is 3, passenger service is 5, and others are 4. In terms of the importance of the user terminal position, general is 2, important is 6, and very important is 10. In terms of the management level of the user terminal, front-line is 1, team is 2, dispatcher is 6, supervisor is 4, manager is 8, and decision-making is 10.

[0073] L300: According to The attributes are converted according to the quantification standard. ,calculate priority The specific calculations are as follows: ; in , , , These are the weighting coefficients for user terminal type, the service type to which the user terminal belongs, the importance of the user terminal's position, and the management level of the user terminal. Exemplary , , , They are 15%, 30%, 30%, and 25% respectively.

[0074] Traversal complete .

[0075] Example 2 The difference between this embodiment and Embodiment 1 is that this embodiment provides a priority-based scheduling-based dynamic refresh system for airport business data, including: The data acquisition module is configured to: acquire a list of preset service methods and a list of instruction events to be calculated; acquire the mapped service methods according to the list of preset service methods; and update the initial instruction events through the mapped service methods. The event handling module is configured to: select process branches based on the updated list of instruction events to be calculated, including obtaining the server load record set according to the list of instruction events to be calculated, and performing low-load processing and non-low-load processing respectively according to the proportion of records in the server load record set that meet the preset load conditions; The resource calculation module is configured to: assign resource consumption attributes in non-low load scenarios, including traversing the list of instruction events to be calculated and retrieving corresponding resource consumption data based on the attributes of the service methods to be called for each instruction event; The offset module is configured to: load the list of modes and mode numbers of the delayed query sequence of the service method based on the list of instruction events to be calculated after updating the resource consumption attributes, and calculate the final completion time offset of each instruction event; The event prediction module is configured to update the flight dynamic event set based on the time offset, including building a time prediction model to predict the time of instruction events in the event set list; The update module is configured to: calculate and select the delayed query sequence using the updated flight dynamic event set list, and perform dynamic refresh based on the selection results.

[0076] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for dynamically refreshing airport business data based on priority scheduling, characterized in that, include: S1. In response to the service method update event, obtain the preset service method list and the instruction event list to be calculated, construct the instruction event, obtain the mapped service method according to the preset service method list, and update the initial instruction event through the mapped service method. S2. Select process branches based on the updated list of instructions and events to be calculated, including obtaining the server load record set according to the list of instructions and events to be calculated, and performing low-load processing and non-low-load processing respectively according to the proportion of records in the server load record set that meet the preset load conditions. S3. Assigning resource consumption attributes in non-low load scenarios, including traversing the list of instruction events to be calculated and retrieving corresponding resource consumption data based on the attributes of the service methods to be called for each instruction event; S4. Based on the list of instruction events to be calculated after updating the resource consumption attributes, load the pattern list and pattern number list of the delayed query sequence of the service method, and calculate the final completion time offset of each instruction event. S5. Update the flight dynamic event set according to the time offset, including building a time prediction model to predict the time of instruction events in the event set list; S6. Calculate and select the delayed query sequence using the updated flight dynamic event set list, and complete the dynamic refresh based on the selection result.

2. The method for dynamically refreshing airport business data based on priority scheduling according to claim 1, characterized in that, The update of the initial instruction event via the mapping service method includes: S11. Based on the content of the service method update event, construct an initial instruction event, wherein the initial instruction event includes at least the content corresponding to the service method update event. Attribute information; S12. Locate the specific service method in the service method list by matching the method name with the service method name corresponding to the service method update event. ,according to Get the update tag Service methods mapped in the middle; S13. Assign the acquired mapping service method to the initial instruction event. The attribute, along with a globally unique ID generated by the UUID generator, is assigned to the initial command event. The attribute, which maps the service method's preset initial execution time, is assigned to the initial instruction event. The attribute completes the update of the initial instruction event and adds it to the list of instruction events to be calculated; Where 'o' represents the total number of original instruction events in the instruction event list to be calculated. As a unique identifier for the instruction event, Update the method parameter data corresponding to the event for the service method. Update the event-related flight number to the service method. The service method that triggers this instruction event, To determine the start time of the event service method that triggers this instruction. The service method to be invoked. The initial execution time of the service method to be called. List of preset service methods The first service method.

3. The method for dynamically refreshing airport business data based on priority scheduling according to claim 1, characterized in that, Assigning resource consumption attributes in non-low-load scenarios includes: S31. Load the service method execution average resource consumption preset list H, where each service method in the average resource consumption preset list... The resource consumption is: ； in, For the first One method Resource consumption varies by time period. For the first Service method for segmented time intervals Resource consumption; S32. Traverse the list of instruction events to be calculated, and obtain each instruction event to be calculated in turn. ; S33, Extraction The service method to be called and with To perform a matching search for keywords in H, locate and... The corresponding service method is associated with ; S34, retrieve S34 The resource consumption data is divided into time periods and assigned to the instruction event to be calculated. Resource consumption attributes ; in, For the instruction event to be calculated, The service method to be invoked. For a single Resource consumption data for the service method, where k is the k-th specific instruction event in the list of instruction events to be calculated. Let be the i-th specific resource consumption in H.

4. The method for dynamically refreshing airport business data based on priority scheduling according to claim 1, characterized in that, The calculation of the final completion time offset of each instruction event includes the following steps: S41. Using operations research optimization methods to generate a list of delayed query sequence patterns. Optimize computation by constructing an operations research optimization objective function using computational resource overload penalty terms, query latency penalty terms, and load balancing standard deviation terms; S42. Use a mixed integer programming solver to solve the objective function. When the solver's solution time exceeds the set threshold, switch to the NSGA-II heuristic algorithm to solve the objective function, and update the optimization results obtained to the list of delayed query sequence pattern numbers N. S43, Event pending calculation instruction of Based on the matching criteria, the corresponding sequence pattern list P and pattern number list N are retrieved. and ,Will The Chinese number is The delayed query sequence pattern is assigned to Delayed query sequence attribute ,Finish Delay the update of the query sequence; The objective function expression is: ； in, for The preset selected delayed query sequence pattern number, For service methods List of delayed query sequence patterns for Delayed query sequence, Penalty for resource overload Penalty for query delay The standard deviation of load balancing is formed. Weighting for resource overload penalties. For query delay penalty weight, The standard deviation weights are used for load balancing.

5. The method for dynamically refreshing airport business data based on priority scheduling according to claim 1, characterized in that, The calculation of the final completion time offset of each instruction event also includes, for each Matched Extract the next delayed query sequence pattern Simultaneously, in conjunction with the service method execution average resource consumption preset list H, and... corresponding , obtain Calculate the amount of resource consumption in different time periods. The final completion time offset is calculated and the result is assigned to... The final completion time offset attribute, the formula for calculating the final completion time offset is: ； in, For the instruction event to be calculated The final completion time offset, For service methods The maximum offset time of the last delayed query sequence pattern. for The last delayed query sequence pattern in the middle, for Total number of time allocation points for delayed queries. For service methods Resource consumption by time period.

6. The method for dynamically refreshing airport business data based on priority scheduling according to claim 1, characterized in that, The step of updating the flight dynamic event set based on the time offset includes loading the flight dynamic data list S and the flight dynamic event set list K, and extracting... Triggering service method properties Related flight number attributes ,based on Lock the corresponding flight dynamic event set list K And then according to locking The corresponding instruction event ,when Final sending time attribute When empty, of , , Assigned to ,when of If it is not empty, then in middle Added later ,Finish Update K from the instruction event; in, for The associated flight number attribute, for The start time of the event service method that triggered this instruction. For dynamic flight instruction event sets, for The e-th instruction event, To update the event corresponding method parameter data for the obtained response service method, To determine the start time of the event service method that triggers this instruction. The initial execution time of the service method. for The final sending time attribute.

7. The method for dynamically refreshing airport business data based on priority scheduling according to claim 1, characterized in that, The step of updating the flight dynamic event set based on the time offset also includes constructing a time prediction model consisting of an input layer, a bidirectional LSTM hidden layer, an attention layer, a fusion layer, and an output layer; calculating the maximum predicted interval offset of the instruction event list I to be calculated; combining the current time and the maximum interval offset to filter the instruction event set that needs to be time predicted in the flight dynamic event set list K; inputting the filtered instruction event set into the time prediction model to obtain the start time prediction value of the instruction event update method; and updating the time attribute of the corresponding instruction event after standardizing the prediction value.

8. The method for dynamically refreshing airport business data based on priority scheduling according to claim 1, characterized in that, The dynamic refresh based on the selection result includes: filtering and constructing a set of already occurred instruction events that overlap with the predicted time interval based on the updated flight dynamic event set list K; creating a time interval; assigning a value to the initial execution time attribute of the service method to be called for the unsent instruction events in K in conjunction with the average resource consumption preset list H; configuring a delayed query sequence for the unsent instruction events in K; and updating the delayed query sequence by combining the real-time user terminal list to count the number of user terminals at each query allocation point.

9. The method for dynamically refreshing airport business data based on priority scheduling according to claim 1, characterized in that, The dynamic refresh based on the selection result also includes constructing a cumulative variable of resource consumption over a time interval, accumulating the resource consumption of the command events that have occurred and the command events that have not been sent within the prediction period, constructing a set of command events to be calculated in order of priority, traversing the set and judging based on the resource threshold, selecting an appropriate delayed query sequence, assigning the sending time to the command events in the command event list I to be calculated based on the selection result and distributing them to the user terminal, thereby completing the dynamic refresh of business data.

10. A priority-based scheduling-based dynamic refresh system for airport business data, executed according to claim 1, characterized in that, include: The data acquisition module is configured to: acquire a list of preset service methods and a list of instruction events to be calculated; acquire the mapped service methods according to the list of preset service methods; and update the initial instruction events through the mapped service methods. The event handling module is configured to: select process branches based on the updated list of instruction events to be calculated, including obtaining the server load record set according to the list of instruction events to be calculated, and performing low-load processing and non-low-load processing respectively according to the proportion of records in the server load record set that meet the preset load conditions; The resource calculation module is configured to: assign resource consumption attributes in non-low load scenarios, including traversing the list of instruction events to be calculated and retrieving corresponding resource consumption data based on the attributes of the service methods to be called for each instruction event; The offset module is configured to: load the pattern list and pattern number list of the delayed query sequence of the service method based on the list of instruction events to be calculated after updating the resource consumption attributes, and calculate the final completion time offset of each instruction event; The event prediction module is configured to update the flight dynamic event set based on the time offset, including building a time prediction model to predict the time of instruction events in the event set list; The update module is configured to: calculate and select the delayed query sequence using the updated flight dynamic event set list, and perform dynamic refresh based on the selection results.

Images