Aircraft multivariate flight information association method and storage medium

Abstract

The invention belongs to the technical field of aviation automatic control, and relates to an aircraft multi-element flight information association method and a storage medium, and the method comprises the steps: obtaining multi-element flight information; associating the multivariate flight information based on a first method to obtain first associated information; if the association is not successful, associating the multivariate flight information based on a second method to obtain second association information; if the association is not successful, judging the flight state of the aircraft; if the flight state shows that the aircraft does not land, continuing to acquire the multi-element flight information; and if the association succeeds or the flight state shows that the aircraft lands, acquiring flight dynamic data, and performing data analysis in combination with the first association information or the second association information to obtain optimized association algorithm parameters. According to the method, a conventional method is firstly used for association, and if the association is not successful, a higher-level second method is switched, so that a complicated and changeable flight information environment can be effectively handled, and the success rate of information association is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of aviation automatic control, and particularly relates to a method for associating multi-source flight information of an aircraft and a storage medium. Background Art

[0002] The dynamic tracking of an aircraft relies on the support of multi-source flight information. The types of information include planned information, surveillance information, and real-time information. In actual operation, it is necessary to associate these three types of information in order to intuitively grasp the detailed flight status of the aircraft, which is of great significance for monitoring the flight consistency of the aircraft and ensuring the safe and stable operation of the aircraft.

[0003] In the related art, the multi-source flight information association technology of an aircraft is generally divided into two categories, which are conventional methods. The first category is the method of associating aircraft identification information; the second category is to establish a planned flight track based on flight plan data and complete information association by using relevant factors and the matching principle of aircraft identification information. Among them, the relevant factors include area factor, yaw factor, altitude factor, time factor, direction factor, speed factor, etc. Through the conventional method, in the case of a large number of air targets, abnormal aircraft status information, and the absence of actual takeoff / landing time, mis-association is likely to occur, reducing the success rate of flight information association. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, the present invention proposes a method for associating multi-source flight information of an aircraft and a storage medium.

[0005] In a first aspect, a method for associating multi-source flight information of an aircraft is provided, including: obtaining multi-source flight information, where the multi-source flight information includes flight plan data before the aircraft takes off, flight dynamic data and flight surveillance data after the aircraft takes off; associating the multi-source flight information based on a first method to obtain first association information; if the association is not successful, then associating the multi-source flight information based on a second method to obtain second association information; if the association is not successful, then judging the flight status of the aircraft; if the flight status indicates that the aircraft has not landed, then continue to execute the step of obtaining the multi-source flight information; if the association is successful or the flight status indicates that the aircraft has landed, then obtain the flight dynamic data and perform data analysis in combination with the first association information or the second association information to obtain optimized association algorithm parameters.

[0006] In an alternative embodiment, if the association is not successful, the multi-source flight information is associated based on a second method to obtain second association data, which specifically includes: obtaining a prediction model based on the multi-source flight information, where the second method is based on the prediction model; calculating an information association decision threshold based on the prediction model; if the information association decision threshold meets an optimization condition, optimizing the information association decision threshold to obtain an optimized threshold; and associating the multi-source flight information according to the optimized threshold to obtain the second association data.

[0007] In an alternative embodiment, obtaining a prediction model based on the multi-source flight information specifically includes: performing data preprocessing on the multi-source flight information to obtain a numerical dataset; dividing the numerical dataset into training set sample data and validation set sample data; screening the training set sample data to obtain screened data, where the screened data includes takeoff anomaly training set sample data and landing anomaly training set sample data; obtaining the prediction model based on the screened data, where the prediction model includes a takeoff delay prediction model and a landing delay prediction model; and inputting the validation set sample data into the prediction model to obtain the optimized association algorithm parameters.

[0008] In an alternative embodiment, performing data preprocessing on the multi-source flight information to obtain a numerical dataset specifically includes: deleting null data and outlier data according to the sample data text format and numerical distribution characteristics to obtain normal text data; converting the normal text data into categorical data and numerical data to obtain the numerical dataset, where the numerical dataset includes three numerical data items: planned flight time, planned takeoff delay, and planned landing delay.

[0009] In an alternative embodiment, screening the training set sample data to obtain screened data, where the screened data includes takeoff anomaly training set sample data and landing anomaly training set sample data, specifically includes: eliminating outlier data based on the planned takeoff delay and screening out data where the aircraft takes off before the planned takeoff time to obtain the takeoff anomaly training set sample data; eliminating outlier data based on the planned landing delay and screening out data where the aircraft lands after the planned landing time to obtain the landing anomaly training set sample data.

[0010] In an alternative embodiment, obtaining the prediction model based on the screened data specifically includes: based on the screened data, dividing the variable features of the training set sample data into input features of independent variables and target features of dependent variables, where the independent variable features include categorical data and the dependent variable features include numerical data; constructing an initial model based on the input features and the target features; and training the initial model to obtain the prediction model.

[0011] In an alternative embodiment, the initial model is trained to obtain the prediction model, which specifically includes: converting categorical features into numerical features, and the calculation formula is as follows:

[0012] In the formula, D is the entire data set for model training, D k is a subset of D, is the i-th feature of sample k, y j is the feature value of sample j, a is the prior weight, p is the prior distribution, [] is the indicator function, which has a value of 1 if the internal condition is satisfied, otherwise 0; in the first round of iteration, a sample data set in a preset sorting state is selected to train n models M i (i ∈ [1, n]), where n is the number of samples, and M i is the model trained using the first i samples; in the training process, a symmetric tree structure is obtained and the leaf node values are calculated; the loss function and gradient estimation of sample i are calculated using the model M i-1 to establish a residual tree and generate a base learner; in the second and subsequent iterations, the tree structures of the n models M i trained in the first round are reused, and the samples are divided into the corresponding leaf nodes; all the generated base learners are weighted to obtain the prediction model.

[0013] In an alternative embodiment, based on the prediction model, an information association decision threshold is calculated, which specifically includes: inputting categorical data and numerical data into the takeoff delay prediction model to obtain predicted takeoff delay data; inputting categorical data and numerical data into the landing delay prediction model to obtain predicted landing delay data; using the planned takeoff time and the predicted takeoff delay data as the lower boundary of the information association decision threshold, and using the planned landing time and the predicted landing delay data as the upper boundary of the information association decision threshold, and calculating to obtain the information association decision threshold, and the calculation formula is as follows: In the formula, t AL is the lower boundary of the association threshold, t AU is the upper boundary of the association threshold, t PD is the planned takeoff time, t PL is the planned landing time, Δt ED is the predicted takeoff delay data, Δt EL is the predicted landing delay data.

[0014] In an alternative embodiment, if the information association decision threshold meets the optimization condition, the information association decision threshold is optimized to obtain an optimized threshold, which specifically includes: based on the information association decision threshold and the planned flight time, calculating whether the aircraft meets the optimization condition of the information association decision threshold, and the calculation formula is as follows:

[0015] where t AL is the lower boundary of the association threshold, t AU is the upper boundary of the association threshold, λ is the flight time extension coefficient, and t PF is the planned flight time; if the information association decision threshold meets the optimization condition, then optimize the information association decision threshold according to the planned takeoff time, planned landing time, takeoff delay adjustment parameter, and landing delay adjustment parameter to obtain the optimized threshold, and the calculation formula is as follows: where t' AL is the lower boundary of the optimized association threshold, t' AU is the upper boundary of the optimized association threshold, ξ AD is the takeoff delay adjustment parameter, and ξ AL is the landing delay adjustment parameter.

[0016] In a second aspect, a storage medium is further provided, storing a computer program, and when the computer program is executed by a processor, the multi - aircraft flight information association method described in any one of the above is implemented.

[0017] The beneficial effects of the present invention are as follows: By first using a conventional method for association and switching to a more advanced second method if it fails, it can effectively cope with the complex and changeable flight information environment, improve the success rate of information association. When the aircraft has not landed, continuously obtain and associate multi - aircraft flight information, which can ensure real - time grasp of the latest state of the aircraft to reduce the risks brought by information lag. By continuously optimizing the association algorithm, it can better cope with complex flight environments and data changes, and can improve the overall flight information processing ability and monitoring efficiency. Through multi - level and multi - method information association and dynamic optimization, the association accuracy of aircraft flight information and monitoring reliability are significantly improved, flight safety and management efficiency are enhanced, and at the same time, it has good adaptability and flexibility, and can effectively cope with complex and changeable flight environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] By reading the detailed description of the non - restrictive embodiments with reference to the following drawings, other features, objectives, and advantages of the present application will become more obvious:

[0019] Figure 1 is a flowchart of the multi - aircraft flight information association method provided by an embodiment of the present invention;

[0020] Figure 2 is a flowchart of the multi - aircraft flight information association method provided by another embodiment of the present invention;

[0021] Figure 3 is a flowchart of model training of the multi - aircraft flight information association method provided by an embodiment of the present invention;

[0022] Figure 4 A comparison chart of flight information association rates of a method for associating aircraft multi-element flight information provided in one embodiment of the present invention. DETAILED DESCRIPTION

[0023] The following is combined with Figures 1 to 4 The present application is further described in detail with reference to the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It should also be noted that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0024] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0025] The conventional methods used to associate aircraft multi-dimensional flight information mainly include the first method and the second method. The first method associates the aircraft plan data, monitoring data and real-time data according to the aircraft identification information, which can achieve rapid matching of the three and relax the requirements for aircraft monitoring data. The disadvantage is that under the condition that the aircraft identification information is correct, for the connecting aircraft with abnormal status information and missing actual take-off / landing time, it is impossible to determine which flight plan of the current aircraft and the connecting aircraft to match during the take-off / landing phase.

[0026] The second method is based on the accurate calculation of the planned trajectory and strictly associates information according to the trajectory parameters. The disadvantage is that the planned trajectory calculated by the flight plan has a large error compared with the actual flight situation. The thresholds and weights of various factors are difficult to determine in actual engineering applications. It is easy to cause misassociation when the aircraft is executing the takeoff phase of the second flight plan when there are many aerial targets, the aircraft status information is abnormal, and the actual take-off / landing time is missing.

[0027] In addition, there are also information association improvement methods based on the second type of methods, such as association methods based on fuzzy comprehensive functions; association methods based on point pattern matching.

[0028] The association method of fuzzy comprehensive function is based on the maximum comprehensive similarity and threshold discrimination principles in fuzzy mathematics. Under limited conditions, it is suitable for dense target environments with large errors. The disadvantage is that this method uses a simple three-stage flight model: climb-cruise-descent to derive the planned trajectory. In practice, the aircraft takeoff / landing phase follows the standard approach / departure procedure, which is inconsistent with the actual flight conditions of the aircraft and cannot achieve rapid and accurate association during the aircraft takeoff phase.

[0029] The correlation method for point pattern matching is based on the correlation matching of two point sets, which can reduce the dependence of the matching result on single-point information. The disadvantage is that this method can only process data normally when there are enough track points. In practical engineering applications, there is a lag in information correlation, and it is impossible to achieve fast matching of multiple flight information.

[0030] Through the above analysis, the existing technology can no longer meet the existing needs. Therefore, a new technical solution is needed to solve the above problems.

[0031] Please refer to Figure 1 and Figure 2 , and this embodiment of the present invention provides a method for correlating multiple flight information of an aircraft, including the following steps:

[0032] Step S11: Obtain multiple flight information, where the multiple flight information includes flight plan data before the aircraft takes off, flight dynamic data and flight surveillance data after the aircraft takes off.

[0033] Step S12: Correlate the multiple flight information based on the first method to obtain the first correlation information, and the first method is a conventional method.

[0034] Step S13: If the correlation is not successful, then correlate the multiple flight information based on the second method to obtain the second correlation information.

[0035] Step S14: If the correlation is not successful, then judge the flight state of the aircraft.

[0036] Step S15: If the flight state shows that the aircraft has not landed, then continue to execute the step of obtaining multiple flight information.

[0037] Step S16: If the correlation is successful or the flight state shows that the aircraft has landed, then obtain the flight dynamic data, and perform data analysis in combination with the first correlation information or the second correlation information to obtain optimized correlation algorithm parameters.

[0038] In this embodiment, the flight state of the aircraft is judged. Before the aircraft takes off, flight plan data is obtained. After the aircraft takes off, flight dynamic data and flight surveillance data are obtained. Flight plan data usually includes information such as flight routes, estimated take-off and arrival times, etc.; flight dynamic data includes information such as the real-time position, speed, altitude, etc. of the aircraft; flight surveillance data comes from systems such as radar and ADS-B. Conventional methods are used to initially integrate flight plans and real-time flight data. If the multi-source flight information of the aircraft has been associated, it enters step S16; otherwise, it enters step S12. Based on the conventional method, the multi-source flight information of the aircraft is associated. If the association of the multi-source flight information of the aircraft is successful, it enters step S16; otherwise, it enters step S13. If the association of the multi-source flight information of the aircraft is successful, it enters step S16; otherwise, a non-association prompt message is displayed and it enters step S15. The flight state of the aircraft is judged. If the aircraft has not landed, it enters step S2; otherwise, it enters step S16.

[0039] By first using the conventional method for association and switching to a more advanced second method if it fails, it can effectively cope with the complex and changeable flight information environment, improve the success rate of information association. When the aircraft has not landed, continuously obtain and associate multi-source flight information, which can ensure real-time grasp of the latest state of the aircraft to reduce the risks brought by information lag. Continuously optimize the association algorithm, which can better cope with complex flight environments and data changes, and can improve the overall flight information processing ability and monitoring efficiency. Through multi-level and multi-method information association and dynamic optimization, the association accuracy of aircraft flight information and monitoring reliability are significantly improved, flight safety and management efficiency are enhanced, and at the same time, it has good adaptability and flexibility, and can effectively cope with complex and changeable flight environments.

[0040] Based on step S13, if the association is not successful, the multi-source flight information is associated based on the second method to obtain second association data, which specifically includes the following steps:

[0041] Step S131: Obtain a prediction model based on the multi-source flight information. The second method is based on the prediction model.

[0042] Step S132: Calculate the information association decision threshold based on the prediction model.

[0043] Step S133: If the information association decision threshold meets the optimization condition, optimize the information association decision threshold to obtain the optimized threshold.

[0044] Step S134: Associate the multi-source flight information according to the optimized threshold to obtain the second association data.

[0045] In this embodiment, the prediction model is used to analyze and predict the potential correlation relationships between flight information. By using the prediction model to calculate the decision threshold for information correlation, it is possible to determine whether the information between different data sources can be correlated. Through the analysis of the prediction model, a reasonable threshold value can be determined to distinguish between valid and invalid correlations. Through the optimized threshold value, the flight plan data, flight dynamic data, and flight surveillance data can be more accurately integrated, thereby obtaining more reliable correlation results.

[0046] Step S131: Obtain a prediction model based on multivariate flight information, which specifically includes the following steps:

[0047] Step S1311: Perform data preprocessing on the multivariate flight information to obtain a numerical dataset.

[0048] Step S1312: Divide the numerical dataset into training set sample data and validation set sample data. Among them, the sample data of the preprocessed dataset is divided into training set sample data and validation set sample data. Among them, 20% of the sample data is divided as the validation set sample data, and the remaining 80% of the sample data is divided as the training set sample data, which is convenient to select the model in the way of cross-validation and obtain the optimal model parameter combination.

[0049] Step S1313: Screen the training set sample data to obtain the screened data, and the screened data includes takeoff anomaly training set sample data and landing anomaly training set sample data.

[0050] Step S1314: Obtain a prediction model based on the screened data, and the prediction model includes a takeoff delay prediction model and a landing delay prediction model.

[0051] Among them, input the validation set sample data into the prediction model to obtain optimized correlation algorithm parameters.

[0052] Through data preprocessing, sample division, and screening, it is possible to effectively remove noise data and extract features related to flight delays, thereby improving the accuracy of the prediction model. The takeoff delay prediction model and the landing delay prediction model can provide intelligent decision-making support for airlines and airports, help arrange resources in advance, reduce the impact of delays on operations, and evaluate and optimize the model through the validation set, and the optimal model parameters can be found to further improve the performance of the model

[0053] Among them, based on Step S1311, perform data preprocessing on the multivariate flight information to obtain a numerical dataset, which specifically includes the following steps:

[0054] Step S13111: According to the text format and numerical distribution characteristics of the sample data, delete the null data and outlier data to obtain normal text data.

[0055] Step S13112: converting the normal text data into categorical data and numerical data to obtain a numerical data set, which includes three numerical data items: planned flight time, planned take-off delay, and planned landing delay.

[0056] In this embodiment, the text data "planned take-off time" is split into three categorical data items of "month", "day" and "hour". The text data "planned take-off time", "planned landing time", "actual take-off time" and "actual landing time" are converted into numerical data timestamps, and the difference between the planned landing time and the planned take-off time and the difference between the actual time and the planned time are calculated respectively to obtain three numerical data items of "planned flight time", "planned take-off delay" and "planned landing delay".

[0057] Wherein, based on step S1313, the training set sample data is screened to obtain screened data, wherein the screened data includes takeoff abnormality training set sample data and landing abnormality training set sample data, and specifically includes the following steps:

[0058] Step S13131: Eliminate outlier data based on planned takeoff delays, filter out data on aircraft taking off before the planned takeoff time, and obtain takeoff anomaly training set sample data.

[0059] Step S13132: Eliminate outlier data based on planned landing delays, filter out data of aircraft landing after the planned landing time, and obtain landing anomaly training set sample data.

[0060] In this embodiment, the data of aircraft taking off before the planned take-off time, that is, the data with "planned take-off delay" less than zero, are screened out for training the take-off delay prediction model, and the outlier data based on "planned take-off delay" are eliminated. At the same time, the data of aircraft landing after the planned landing time, that is, the data with "planned landing delay" greater than zero, are also screened out for training the landing delay prediction model, and the outlier data based on "planned landing delay" are eliminated.

[0061] Among them, step S1314, obtaining a prediction model based on the screened data, specifically includes the following steps:

[0062] Step S13141: Based on the filtered data, the variable characteristics of the training set sample data are divided into input characteristics of independent variables and target characteristics of dependent variables, the independent variable characteristics include categorical data, and the dependent variable characteristics include numerical data.

[0063] Step S13142: Construct an initial model based on input features and target features.

[0064] Step S13143: Perform model training on the initial model to obtain a prediction model.

[0065] Furthermore, as Figure 3 shown, based on step S13143, the initial model is trained to obtain a prediction model, which specifically includes the following steps:

[0066] Step S131431: Convert categorical features into numerical features, and the calculation formula is as follows:

[0067]

[0068] In the formula, D is the entire dataset used for model training, D k is a subset of D, is the i-th feature of sample k, y j is the feature value of sample j, a is the prior weight, p is the prior distribution, and [] is the indicator function, which has a value of 1 if the internal condition is met, otherwise 0.

[0069] Step S131432: In the first round of iteration, select the sample dataset in the preset sorting state and train n models M i (i ∈ [1, n]), where n is the number of samples, and M i is the model trained using the first i samples;

[0070] Step S131433: Obtain the symmetric tree structure and calculate the leaf node values during the training process;

[0071] Step S131434: Use the model M i-1 to calculate the loss function and gradient estimation of sample i, establish a residual tree, and generate a base learner;

[0072] Step S131435: In the second and subsequent iterations, reuse the tree structures of the n models M i trained in the first round and divide the samples into the corresponding leaf nodes.

[0073] Step S131436: Perform weighted processing on all the generated base learners to obtain a prediction model.

[0074] In this embodiment, first, feature selection of the data is performed, and the variable features of the filtered dataset are divided into input features of independent variables and target features of dependent variables. The independent variable features include: categorical data "airline", categorical data "aircraft type", categorical data "departure airport", categorical data "arrival airport", categorical data "planned departure month", categorical data "planned departure day", categorical data "planned departure time", and numerical data "planned flight time". The dependent variable features include: numerical data "planned departure delay" and numerical data "planned arrival delay".

[0075] Then model training is carried out. In each round of iteration, a sample data set in a certain sorting state is selected. For a certain sample of categorical features, to avoid label leakage, the average value of the categorical labels before this sample is taken, and priority and weight coefficients are introduced for adjustment. The categorical features are converted into numerical features using a formula. In the first round of iteration, a sample data set in a certain sorting state is selected. To ensure that the prediction results do not deviate, model M is used i-1 Calculate the loss function and gradient estimation of sample i, build a residual tree, and generate a base learner. In the second round and subsequent iterations, the model training speed is accelerated and the model robustness is enhanced.

[0076] According to the prediction model constructed after training, use the method of cross-validation to evaluate the sample data in the validation set, obtain the parameter value with the highest score through a scoring function, and in all parameter combinations of the model, obtain the parameter combination that makes the model performance reach the optimal.

[0077] Furthermore, perform rough processing, step S132. Calculate the information association decision threshold based on the prediction model, which specifically includes the following steps:

[0078] Step S1321: Input the categorical data and numerical data into the takeoff delay prediction model to obtain the predicted takeoff delay data.

[0079] Step S1321: Input the categorical data and numerical data into the landing delay prediction model to obtain the predicted landing delay data.

[0080] Step S1321: Use the planned takeoff time and the predicted takeoff delay data as the lower boundary of the information association decision threshold, and use the planned landing time and the predicted landing delay data as the upper boundary of the information association decision threshold, and calculate the information association decision threshold. The calculation formula is as follows:

[0081]

[0082] In the formula, t AL is the lower boundary of the association threshold, t AU is the upper boundary of the association threshold, t PD is the planned takeoff time, t PL is the planned landing time, Δt ED is the predicted takeoff delay data, Δt EL is the predicted landing delay data.

[0083] In this embodiment, the categorical data "airline company", "aircraft type", "departure airport", "arrival airport", "scheduled departure month", "scheduled departure day", "scheduled departure time", and the numerical data "scheduled flight time" are input into the "departure delay model" and the "arrival delay model" to obtain the "predicted departure delay" data and the "predicted arrival delay" data.

[0084] If the information association decision threshold meets the optimization condition, it indicates that based on the calculation of historical aircraft data, the probability of serious delay of this aircraft is relatively high, and it is necessary to introduce the statistical parameters of the takeoff / landing delay historical sample data of the aircraft to optimize the information association decision threshold.

[0085] Further, perform fine processing, step S133: If the information association decision threshold meets the optimization condition, then optimize the information association decision threshold to obtain the optimized threshold, which specifically includes the following steps:

[0086] Step S1331: Calculate whether the aircraft meets the optimization condition of the information association decision threshold based on the information association decision threshold and the scheduled flight time. The calculation formula is as follows:

[0087]

[0088] In the formula, t AL is the lower boundary of the association threshold, t AU is the upper boundary of the association threshold, λ is the flight time expansion coefficient, and t PF is the scheduled flight time;

[0089] Step S1332: If the information association decision threshold meets the optimization condition, then optimize the information association decision threshold according to the scheduled departure time, scheduled arrival time, takeoff delay adjustment parameter, and landing delay adjustment parameter to obtain the optimized threshold. The calculation formula is as follows:

[0090]

[0091] In the formula, t' AL is the lower boundary of the optimized association threshold, t' AU is the upper boundary of the optimized association threshold, ξ AD is the takeoff delay adjustment parameter, ξ AL is the landing delay adjustment parameter. The takeoff / landing delay adjustment parameter is the fluctuation average value of the standard deviation of the takeoff / landing delay of the historical sample data, and is dynamically updated according to the historical sample data.

[0092] The multi - flight - information association method for aircraft with abnormal status information provided by the present invention shows through experiments that the training device is an RTX 3090 server with 24 GB video memory, and the test device is a Quadro P5000 server with 16 GB video memory; based on the deep - learning development framework Pytorch 1.10 under the Ubuntu system; five - day actual flight data was randomly collected for system testing. Among them, on the first day, there were 12,563 flights, on the second day, there were 12,424 flights, on the third day, there were 12,507 flights, on the fourth day, there were 12,668 flights, and on the fifth day, there were 12,108 flight operations. To ensure the effectiveness of the test, historical flight data with flight dates later than the training set samples and validation set samples was actually collected for testing.

[0093] The flight - information association thresholds are generated respectively by the control method and the method of the present invention, and it is judged whether the time stamp of the received surveillance data meets the flight - information association threshold, and the flight information that meets the conditions is associated. In the case where the conventional association method fails, the control method directly uses the planned take - off / landing time of the flight as the lower and upper boundaries of the flight - information association threshold of the flight; the method of the present invention estimates the delay of the planned take - off / landing time of the flight according to the flight - plan data of the flight, combines the predicted delay to calculate the predicted take - off / landing time of the flight, and uses the predicted take - off / landing time as the lower and upper boundaries of the flight - information association threshold of the flight. As Figure 4 shown, through the association processing of the five - day actually collected data, the value of the information - association rate of the control method fluctuates around 57.9%, and the value of the information - association rate of the method of the present invention fluctuates around 94.5%. The method of the present invention has a significant improvement in the information - association success rate compared with the control method.

[0094] The control method directly uses the planned take - off / landing time of the flight as the lower and upper boundaries of the flight - information association threshold of the flight. During the actual flight process, due to reasons such as air traffic control, the flight cannot strictly execute the flight mission according to the flight plan. When the actual take - off time of the flight is earlier than the planned take - off time or the actual landing time of the flight is later than the planned landing time, the flight information cannot be associated. The method of the present invention learns the law of the historical flight - plan data of the associated anomalies. In the case where the conventional method of associating flight information fails, it predicts the planned take - off / landing delay time of the flight according to the flight - plan data, reduces the proportion of data where the actual take - off time is earlier than the predicted take - off time or the actual landing time of the flight is later than the predicted landing time, and uses the predicted take - off / landing time as the lower and upper boundaries of the flight - information association threshold of the flight, effectively improving the success rate of flight - information association in the case where the conventional association method fails.

[0095] On the other hand, the present invention also provides a computer storage medium storing a computer program, which when executed by a processor implements the aircraft multi-flight information association method according to any one of the above.

[0096] The computer storage medium can be simply referred to as a medium. Those of ordinary skill in the art can understand that all or part of the processes in the above-described embodiment methods can be completed by instructing relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the above-described method embodiments. Among them, any reference to a memory, storage, database, or other medium used in the various embodiments provided in the present application can include non-volatile and / or volatile memories. Non-volatile memories can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memories can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and Rambus dynamic RAM (RDRAM). Each embodiment in this specification is described in a progressive manner. The same or similar parts among the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple. For the relevant parts, refer to the partial description of the method embodiments.

[0097] The above description is only a preferred embodiment of the present application and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the inventive concept. For example, the technical solutions formed by mutually replacing the above features with the (but not limited to) technical features having similar functions disclosed in the present application.

Claims

1. A method for correlating multiple flight information of an aircraft, characterized in that, Including: Obtain multi-source flight information, where the multi-source flight information includes flight plan data before the aircraft takes off, flight dynamic data and flight surveillance data after the aircraft takes off; Associate the multi-source flight information based on a first method to obtain first associated information; If the association is not successful, then associate the multi-source flight information based on a second method to obtain second associated information; If the association is not successful, then judge the flight state of the aircraft; If the flight state indicates that the aircraft has not landed, then continue to execute the obtaining of the multi-source flight information; If the association is successful or the flight state indicates that the aircraft has landed, then obtain the flight dynamic data, and perform data analysis in combination with the first associated information or the second associated information to obtain optimized association algorithm parameters.

2. The method for associating multi-flight information of an aircraft according to claim 1, wherein If the association is not successful, then associate the multi-source flight information based on a second method to obtain second associated data, specifically including: Obtain a prediction model based on the multi-source flight information, and the second method is based on the prediction model; Calculate an information association decision threshold based on the prediction model; If the information association decision threshold meets the optimization condition, then optimize the information association decision threshold to obtain an optimized threshold; Associate the multi-source flight information according to the optimized threshold to obtain the second associated data.

3. The aircraft multi-flight information association method according to claim 2, wherein Obtain a prediction model based on the multi-source flight information, specifically including: Perform data preprocessing on the multi-source flight information to obtain a numerical dataset; Divide the numerical dataset into training set sample data and validation set sample data; Screen the training set sample data to obtain screened data, where the screened data includes takeoff anomaly training set sample data and landing anomaly training set sample data; Obtain the prediction model based on the screened data, and the prediction model includes a takeoff delay prediction model and a landing delay prediction model; Among them, input the validation set sample data into the prediction model to obtain the optimized association algorithm parameters.

4. The method for correlating multiple flight information of an aircraft according to claim 3, wherein Perform data preprocessing on the multi-source flight information to obtain a numerical dataset, specifically including: Delete null data and outlier data according to the text format and numerical distribution characteristics of the sample data to obtain normal text data; Convert the normal text data into categorical data and numerical data to obtain the numerical dataset, and the numerical dataset includes three numerical data items: planned flight time, planned takeoff delay, and planned landing delay.

5. The method for associating multiple flight information of an aircraft according to claim 3, characterized in that Screen the training set sample data to obtain screened data, where the screened data includes takeoff anomaly training set sample data and landing anomaly training set sample data, specifically including: Eliminate outlier data based on the planned takeoff delay, and screen out data where the aircraft takes off before the planned takeoff time to obtain the takeoff anomaly training set sample data; Eliminate outlier data based on the planned landing delay, and screen out data where the aircraft lands after the planned landing time to obtain the landing anomaly training set sample data.

6. The method for correlating multiple flight information of an aircraft according to claim 3, wherein Obtain the prediction model based on the screened data, specifically including: Based on the filtered data, divide the variable features of the training set sample data into input features of independent variables and target features of dependent variables. The independent variable features include categorical data, and the dependent variable features include numerical data; Construct an initial model based on the input features and the target features; Perform model training on the initial model to obtain the prediction model.

7. The method for associating multiple flight information of an aircraft according to claim 6, wherein Performing model training on the initial model to obtain the prediction model specifically includes: Convert categorical features into numerical features, and the calculation formula is as follows: Where D is the entire dataset for model training, D k is a subset of D, is the i-th feature of sample k, y j is the eigenvalue of sample j, a is the prior weight, p is the prior distribution, [] is the indicator function, which has a value of 1 if the internal condition is met and 0 otherwise; In the first round of iteration, a sample data set in a preset sorting state is selected to train n models M i (i ∈ [1, n]), where n is the number of samples, and M i is the model trained using the first i samples; Obtain a symmetric tree structure and calculate leaf node values during training; Use model M i-1 Calculate the loss function and gradient estimate of sample i, build a residual tree, and generate a base learner; For the second and subsequent iterations, reuse the tree structures of the n models M trained in the first round to partition the samples to the corresponding leaf nodes; i ​ Perform weighted processing on all generated base learners to obtain the prediction model.

8. The method for associating multiple flight information of an aircraft according to any one of claims 3 to 7, characterized in that, Calculate the information correlation decision threshold based on the prediction model, specifically including: Input categorical data and numerical data into the takeoff delay prediction model to obtain predicted takeoff delay data; Input categorical data and numerical data into the landing delay prediction model to obtain predicted landing delay data; Use the planned takeoff time and the predicted takeoff delay data as the lower boundary of the information correlation decision threshold, and use the planned landing time and the predicted landing delay data as the upper boundary of the information correlation decision threshold, and calculate to obtain the information correlation decision threshold. The calculation formula is as follows: where t AL is the lower boundary of the association threshold, t AU is the upper boundary of the association threshold, t PD is the planned takeoff time, t PL is the planned landing time, Δt ED is the predicted takeoff delay data, Δt EL is the predicted landing delay data.

9. The method for correlating multi-flight information of an aircraft according to claim 8, wherein If the information correlation decision threshold meets the optimization condition, optimize the information correlation decision threshold to obtain the optimized threshold, specifically including: Based on the information correlation decision threshold and the planned flight time, calculate whether the aircraft meets the optimization condition of the information correlation decision threshold. The calculation formula is as follows: where t AL is the lower boundary of the correlation threshold, t AU is the upper boundary of the correlation threshold, λ is the flight time extension coefficient, and t PF is the planned flight time; If the information correlation decision threshold meets the optimization condition, optimize the information correlation decision threshold according to the planned takeoff time, planned landing time, takeoff delay adjustment parameter, and landing delay adjustment parameter to obtain the optimized threshold. The calculation formula is as follows: where t′ AL is the lower boundary of the optimized correlation threshold, and t′ AU is the upper boundary of the optimized correlation threshold, ξ AD is the takeoff delay adjustment parameter, and ξ AL is the landing delay adjustment parameter.

10. A storage medium, characterized in that, There is a computer program stored, and when the computer program is executed by a processor, it implements the aircraft multi-flight information correlation method according to any one of claims 1 to 9.

Images