WO2021082393A1 - 基于大数据深度学习的机场场面可变滑出时间预测方法 - Google Patents

基于大数据深度学习的机场场面可变滑出时间预测方法 Download PDF

Info

Publication number
WO2021082393A1
WO2021082393A1 PCT/CN2020/089908 CN2020089908W WO2021082393A1 WO 2021082393 A1 WO2021082393 A1 WO 2021082393A1 CN 2020089908 W CN2020089908 W CN 2020089908W WO 2021082393 A1 WO2021082393 A1 WO 2021082393A1
Authority
WO
WIPO (PCT)
Prior art keywords
scene
out time
time
airport
data
Prior art date
Application number
PCT/CN2020/089908
Other languages
English (en)
French (fr)
Inventor
周龙
Original Assignee
南京智慧航空研究院有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 南京智慧航空研究院有限公司 filed Critical 南京智慧航空研究院有限公司
Publication of WO2021082393A1 publication Critical patent/WO2021082393A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/04Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00Machine learning
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063Operations research, analysis or management
    • G06Q10/0639Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393Score-carding, benchmarking or key performance indicator [KPI] analysis
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/40Business processes related to the transportation industry
    • GPHYSICS
    • G08SIGNALLING
    • G08GTRAFFIC CONTROL SYSTEMS
    • G08G5/00Traffic control systems for aircraft, e.g. air-traffic control [ATC]
    • G08G5/06Traffic control systems for aircraft, e.g. air-traffic control [ATC] for control when on the ground
    • G08G5/065Navigation or guidance aids, e.g. for taxiing or rolling

Definitions

  • the invention relates to the field of airport traffic control, in particular to a method for predicting the variable sliding-out time of an airport scene based on deep learning of big data.
  • Taxi time is one of the main coordination contents of the Airport Cooperative Decision-Making (A-CDM) specification, and it is also one of the characteristics of studying the airport surface operation status.
  • A-CDM Airport Cooperative Decision-Making
  • the taxi time of the airport is mostly estimated based on the controller's work experience, and the result is usually the default constant value of the corresponding runway.
  • Surface taxiing is an important part of the flight life cycle, and the accurate prediction of its time is of great significance for improving the efficiency of airport surface operations, optimizing the utilization of time slots, and improving flight regularity.
  • the simulation model uses the existing airport topology model, conflict detection and resolution as factors, and obtains the taxi-out time by simulating the operation of all incoming and outgoing aircraft on the ground.
  • the simulation model has strong pertinence and is not universally applicable to different airports.
  • Previous research on analytical models mainly focused on models such as linear regression, and there were also some studies that tried to use machine learning techniques.
  • determining the main factors affecting the taxi time is a focus of the research.
  • the past analysis models usually have shortcomings such as incomplete influencing factors, and the actual reference value is relatively weak, which cannot meet the actual application requirements.
  • the purpose of the present invention is to provide a method for predicting the variable sliding-out time of an airport scene based on deep learning of big data.
  • the present invention provides a method for predicting variable sliding-out time of an airport scene based on deep learning of big data, including:
  • the prediction model for the time to slide out of the airport is used to predict the time to slide out of the airport.
  • the method of obtaining historical operating data and performing data cleaning to obtain a data set includes:
  • the data set is divided into training set and test set.
  • the method for defining and quantifying the traffic condition indicators of the traffic characteristics of the scene includes:
  • the method for analyzing and extracting a feature set that affects the sliding-out time of the scene based on the data set and the traffic condition index includes:
  • the method of performing feature analysis on the features in the original feature set includes:
  • Correlation measurement The correlation coefficient reflects the statistic of the degree of linear correlation between two variables. Its value is [-1,1]. The larger the absolute value, the stronger the degree of linear correlation. A positive value indicates a positive correlation, and a negative value indicates a negative correlation.
  • X and Y are used to refer to any two variables, and the correlation coefficient P X, Y is defined as:
  • Cov(X, Y) is the covariance of X and Y
  • ⁇ X and ⁇ Y are the standard deviations of X and Y
  • ⁇ X and ⁇ Y are the mean values of X and Y
  • Standardized mutual information is a commonly used correlation measure. Its value range is [0, 1]. The larger the value, the greater the degree of correlation between variables.
  • the standardized mutual information U X, Y is defined as:
  • I X, Y are the mutual information of X and Y
  • H X and H Y are the respective entropy of X and Y
  • p(x,y) is the joint probability distribution of X and Y
  • p(x), p( y) is the probability distribution of X and Y;
  • the method of establishing a prediction model of the time of scene slippage through an integrated machine learning method according to the feature set includes:
  • the method for training the initial model and adjusting the parameter values to complete the establishment of the scene sliding-out time prediction model includes:
  • the method for establishing a prediction model of the time of sliding out of the scene through an integrated machine learning method according to the feature set further includes
  • test set to verify the prediction model of the sliding-out time of the scene and perform the performance evaluation adopts the mean square error in the performance evaluation, and the calculation formula is:
  • N is the number of samples in the test set
  • o i is the actual taxi time of the i-th sample
  • p i is the predicted taxi time of the model.
  • the beneficial effect of the present invention is that the present invention provides an airport scene variable sliding-out time prediction method based on big data deep learning.
  • the airport scene variable sliding-out time prediction method based on deep learning of big data includes: acquiring historical operating data and performing data cleaning to obtain a data set; defining and quantifying the traffic condition indicators of the traffic characteristics of the scene; analyzing and analyzing the traffic conditions based on the data set and the traffic condition. Extract the feature set that affects the time of the scene sliding out; based on the feature set, establish the prediction model of the time of the scene sliding out through the integrated machine learning method, and complete the prediction of the time of the airport scene through the scene sliding time prediction model. Process the original recorded data of the airport, model the traffic conditions of the airport scene, analyze and extract the factors affecting taxi time, train the GBRT integrated learning model, and then obtain the taxi time prediction model, which provides a data basis for the management and optimization of airport operations.
  • Fig. 1 is a flowchart of a method for predicting variable sliding-out time of an airport scene based on big data deep learning provided by the present invention.
  • Fig. 2 is the macroscopic spatio-temporal network topology structure of the coasting process provided by the present invention.
  • Fig. 3 is the correlation coefficient of the correlation measurement between the candidate influencing factors and the sliding-out time provided by the present invention.
  • Fig. 4 is the standardized mutual information relationship between candidate influencing factors and slip-out time provided by the present invention.
  • Figure 5 is a factor analysis result diagram of candidate influencing factors provided by the present invention.
  • Figure 6 is a diagram of the performance change process of the model training and testing phases provided by the present invention.
  • this embodiment 1 provides a method for predicting variable taxi-out time of airport scenes based on big data deep learning, processing the original recorded data of the airport, modeling the airport scene traffic conditions, analyzing and extracting taxi time Influencing factors, the GBRT integrated learning model is trained, and then the sliding-out time prediction model is obtained, which provides a data basis for the management and optimization of airport operations.
  • the airport scene variable sliding-out time prediction method based on deep learning of big data includes:
  • S120 Define and quantify the traffic condition indicators of the traffic characteristics of the scene
  • S140 Establish a scene sliding-out time prediction model through an integrated machine learning method based on the feature set
  • step S110 includes:
  • outlier processing firstly, the data type of all attributes and the basic check whether it is out of bounds are performed, and then the delimited detection method is used to further check outliers for some attributes. Based on the actual situation of the airport surface operation, the attribute value range is defined, and the data whose value is not in the corresponding range is regarded as an abnormal value. Finally, data entries containing outliers are deleted from the data set.
  • the attribute value range is shown in the following table.
  • S113 Perform data integration on the original data set to obtain the data set.
  • this step includes redundant attribute identification, data type conversion, and logic error checking. Identify and delete redundant attributes, identify redundant attributes that carry less information by calculating the information entropy of each attribute, and identify redundant attributes contained in other attributes by calculating mutual information between attributes.
  • the redundant attributes "departure airport" and "execution date" have been deleted. Convert the data type to convert the information that is only used for identification in the non-numeric attribute into an integer value type that is easy to follow-up processing and use.
  • the information contained in the "restricted content” attribute is difficult to quantify and is deleted after comprehensive consideration.
  • Check for logical errors consider the physical meaning of each feature, establish constraint relationships between features, and eliminate logical errors. Check the correspondence between the model and the number of engines, check the sequence of time nodes in the scene operation, and directly delete the information items with logical errors.
  • the data set is divided into two parts, namely the training set and the test set.
  • 90% of the data is the training set used in the training phase of the model, and 10% of the data is used as the test set to verify the effectiveness and robustness of the model. "That is to say, the training set and the test set are of the same origin.
  • 10% of the data set is reserved for testing before the machine learning model training, and the remaining data set is used for testing. The next 90% is used as the training set to train the machine learning model.
  • step S120 includes:
  • S121 Use the macroscopic spatiotemporal network topology model to model the traffic situation of the airport surface to obtain the macroscopic spatiotemporal network topology;
  • the macroscopic spatio-temporal network topology model is used to model the traffic situation of the airport surface.
  • Figure 2 visualizes the general situation of the network topology during the taxiing process of departure and arrival in any time and space. In the actual operation of the airport scene, the processes of sliding in and sliding out are mutually coupled and interdependent. Therefore, the influence of the port arrival on the port departure process is also considered in the model.
  • the spatio-temporal network topology model is a general framework for describing the macro resource flow of the airport system. As shown in Figure 2, the departure d 1 ,..., d 4 represent all four different relationships with the reference departure flight d 0, which are " Before launch, before takeoff", “before launch, after takeoff”, “after launch, before takeoff” and "after launch, after takeoff”.
  • inbound a 1 ,..., a 4 represent all four different relationships with the reference inbound flight a 0 , namely "before landing, before arrival”, “before landing, after arrival”, and “after landing” , Before it is in place” and "after it is in place, after it is in place”.
  • t on t in represents the landing time and arrival time of the reference inbound flight a 0.
  • t out t off means the launch time and departure time of the reference departure flight.
  • represents the time threshold of arrival and departure.
  • S122 Based on the macroscopic spatio-temporal network topology, define and quantify four types of indicators that reflect the traffic volume on the scene.
  • SIFIs surface instantaneous flow index
  • SCFIs surface cumulative flow index
  • AQLIs aircraft queue length index
  • SRDIs slot resource demand index
  • SIFIs include D-SIFI and A-SIFI, which respectively represent the number of taxi departure and arrival flights when d 0 is launched from the boarding gate.
  • SCFIs include D-SCFI and A-SCFI, which respectively represent the amount of overlap between the taxi period d 0 and the taxi period of the departing and arriving aircraft.
  • AQLIs include D-AQLI and A-AQLI, which respectively represent the number of aircraft taking off and landing on the runway during the entire taxiing process d 0.
  • SRDIs include D-SRDI and A-SRDI, which represent the number of aircraft launched and landed during the departure slot of aircraft d 0 [t 0 - ⁇ , t 0 + ⁇ ].
  • the value of ⁇ can be set between 10 minutes and 30 minutes.
  • step S130 includes:
  • S131 Extract the features that affect the sliding-out time of the scene from the data set and the traffic condition indicators and form an original feature set.
  • the relevant factors influencing the sliding-out time of the scene acquired in step S110 and step S120 are sorted to form an original feature set. Process the original feature set, and extract new features from the original feature set to replace some of the features in the original feature set.
  • the relevant factors that affect the slide-out time of the scene obtained in step S110 are: flight number, flight attributes, destination airport, planned departure time, aircraft type, airline company, launch time, actual departure time, departure runway, departure stand, parking Seat type, engine type, corridor entrance, restricted or not, boarding gate.
  • the time-related factors of the impact scene obtained by S120 are: D-SIFI, D-SCFI, D-AQLI, D-SRDI, Corridor_NO.
  • the final acquired original feature set, that is, the candidate influencing factors are shown in the following table:
  • S133 Construct a feature set based on the feature analysis result.
  • step S132 based on the analysis result of step S132, important features are selected from the original feature set formed in step S131 to form a feature set for the integrated machine learning model.
  • the features that are less correlated with the coasting time of the surface are screened out. Including "engine type”, “slot type”, “month”, “week”, “day”, “minute”.
  • the final feature set, that is, the influencing factors, is shown in the following table:
  • S132 includes:
  • Figure 3 Figure 4, and Figure 5 respectively show the candidate influencing factors and the slip-out Time Pearson correlation coefficient, standardized mutual information between candidate influencing factors and slip-out time, and factor analysis results of candidate influencing factors.
  • Correlation measurement The correlation coefficient reflects the statistic of the degree of linear correlation between two variables. Its value is [-1,1]. The larger the absolute value, the stronger the degree of linear correlation. A positive value indicates a positive correlation, and a negative value indicates a negative correlation.
  • X and Y are used to refer to any two variables, and the correlation coefficient P X, Y is defined as:
  • Cov(X, Y) is the covariance of X and Y
  • ⁇ X and ⁇ Y are the standard deviations of X and Y
  • ⁇ X and ⁇ Y are the mean values of X and Y
  • Standardized mutual information is a commonly used correlation measure, and its value range is [0, 1]. The larger the value, the greater the degree of correlation between variables.
  • the standardized mutual information U X, Y is defined as:
  • I X, Y are the mutual information of X and Y
  • H X and H Y are the respective entropy of X and Y
  • p(x,y) is the joint probability distribution of X and Y
  • p(x) is the probability distribution of X and Y
  • step S140 includes:
  • S142 Train the initial model and adjust the hyperparameter values to complete the establishment of the prediction model for the time to slide out of the scene.
  • S142 includes: selecting "maximum depth” as the control method for controlling the decision tree; selecting “least squares” as the loss function; under the optimal product value, selecting the maximum learning rate and corresponding The smallest number of estimators; according to the overall data distribution of the sliding-out time in the training set, the minimum sample is set to be divided into 200; the initial model training is completed to establish a scene sliding-out time prediction model.
  • the GradientBoostedRegressionTrees (GBRT) model which is a typical representative of integrated learning, is used to complete the prediction operation of the slide-out time of the scene.
  • the feature set obtained in step S133 is used as the input of the model, and the GBRT model is quickly trained by executing the algorithm in the scikit-learn library.
  • the hyperparameters that need to be set are: decision tree size control, loss function type, number of estimators and learning rate, and minimum sample partition. There are two options for controlling the size of the decision tree, which are "max_depth” and "max_leaf_nodes”.
  • the GBRT model F(x) is an additive model of the following form:
  • h m (x) is a basis function, usually called a weak learner under the concept of boosting
  • ⁇ m is the corresponding weight of the weak learner
  • M is the sum of the number of weak learners.
  • GBRT uses a fixed-size decision tree as a weak learner. Similar to other boosting algorithm ideas, GBRT greedily builds an additive model:
  • Fm(x) represents the GBRT model obtained in the mth iteration.
  • hm(x) is determined by inferred.
  • n is the total number of training samples
  • L is the selected loss function
  • yi is the label of the i-th sample
  • Fm-1(xi) is the prediction value of the i-th sample of the GBRT model obtained in the m-1th iteration
  • h (xi) is the predicted value of the i-th sample of the weak learner to be obtained. While ⁇ m is determined by inferred.
  • n is the total number of training samples
  • L is the selected loss function
  • yi is the label of the i-th sample
  • Fm-1(xi) is the prediction value of the i-th sample obtained by the GBRT model obtained in the m-1 iteration
  • the initial model F 0 is related to the problem. For least squares regression, the average value of the target value is usually selected.
  • S140 further includes:
  • S143 Use the test set to verify and evaluate the performance of the surface slide-out time prediction model.
  • test set is used to verify the scene sliding-out time prediction model and the performance evaluation in the performance evaluation adopts the mean square error, and the calculation formula is:
  • N is the number of samples in the test set
  • o i is the actual taxi time of the i-th sample
  • p i is the predicted taxi time of the model.
  • the MSE is used to monitor the performance changes of the model during training and testing, and the result is shown in FIG. 6.
  • MSE reached 2.5 in the training set, and the performance on the test set was 5.5.
  • the MSE performance of the training set and the test set it reflects the generalization ability of the model to a certain extent.
  • the table below compares the prediction accuracy within different error ranges of the test set.
  • 85.7% of the data sets have a taxi time error within 3 minutes; more than 93% of the data have a prediction error of 4 minutes; about 96.5% of the data have an error of less than 5 minutes. From the verification results on the above test set, it can be seen that the designed data mining model and algorithm can better meet the accuracy requirements of the actual scene dynamic sliding-out time prediction task.
  • the present invention provides a method for predicting variable sliding-out time of an airport scene based on deep learning of big data.
  • the airport scene variable sliding-out time prediction method based on deep learning of big data includes: acquiring historical operating data and performing data cleaning to obtain a data set; defining and quantifying the traffic condition indicators of the traffic characteristics of the scene; analyzing and analyzing and quantifying the traffic characteristics of the scene based on the data set. Extract the feature set that affects the time of the scene sliding out; based on the feature set, establish the prediction model of the time of the scene sliding out through the integrated machine learning method, and complete the prediction of the time of the airport scene through the scene sliding time prediction model. Process the original recorded data of the airport, model the traffic conditions of the airport scene, analyze and extract the factors affecting taxi time, train the GBRT integrated learning model, and then obtain the taxi time prediction model, which provides a data basis for the management and optimization of airport operations.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Educational Administration (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Mathematical Physics (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Traffic Control Systems (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Artificial Intelligence (AREA)

Abstract

一种基于大数据深度学习的机场场面可变滑出时间预测方法,该方法包括:获取历史运行数据并进行数据清洗从而获得数据集(S110);定义并量化场面交通特性的交通状况指标(S120);基于数据集以及交通状况指标分析和提取影响场面滑出时间的特征集(S130);依据特征集通过集成机器学习方法建立场面滑出时间预测模型(S140),通过场面滑出时间预测模型完成对机场场面滑出时间的预测(S150)。处理机场原始记录数据,对机场场面交通状况进行建模,分析和提取滑行时间影响因素,训练GBRT集成学习模型,进而得到滑出时间预测模型,为机场运行的管理和优化提供数据依据。

Description

基于大数据深度学习的机场场面可变滑出时间预测方法 技术领域
本发明涉及机场交通管制领域,具体涉及一种基于大数据深度学习的机场场面可变滑出时间预测方法。
背景技术
滑行时间是机场协同决策(A-CDM)规范的主要协调内容之一,也是研究机场场面运行状态的特征之一。在实际应用中,机场的滑行时间大多根据管制员的工作经验进行估计,其结果通常为对应跑道的默认常量值。当前,既缺乏对航班场面滑行时间的有效预测方法,也没有系统级的工具自动化实现对滑行时间的预测,这样导致在对航班执行离场排序时,不能最大化利用时隙资源,提升航班放行正常率。场面滑行作为航班生命周期的重要组成部分,其时间的精确预测对于提升机场场面运行效率、优化时隙利用以及提高航班正常性具有显著的意义。
以往研究中,飞行器滑出时间预测大多从两方面建立模型:仿真和分析。仿真模型使用了已有的机场拓扑结构模型、冲突探测以及解决方案作为因素,通过仿真地面上所有进离场航空器的运行进而获取滑出时间。仿真模型具有很强的针对性,对不同机场没有很好的普适性。分析模型的以往研究主要聚焦在线性回归等模型上,也有一些尝试使用机器学习技术的研究。对于分析模型而言,确定影响滑行时间的主要因素是研究的一个侧重点。以往的分析模型通常有影响因素不全等缺点,实际参考价值较弱,不能满足实际应用需求。
如何解决上述问题,是目前亟待解决的。
发明内容
本发明的目的是提供一种基于大数据深度学习的机场场面可变滑出时间预测方法。
为了解决上述技术问题,本发明提供了一种基于大数据深度学习的机场场面可变滑出时间预测方法,包括:
获取历史运行数据并进行数据清洗从而获得数据集;
定义并量化场面交通特性的交通状况指标;
基于数据集以及交通状况指标分析和提取影响场面滑出时间的特征集;
依据特征集通过集成机器学习方法建立场面滑出时间预测模型;
通过场面滑出时间预测模型完成对机场场面滑出时间的预测。
进一步的,所述获取历史运行数据并进行数据清洗从而获得数据集的方法包括:
获取历史运行数据构建原始数据集;
对原始数据集进行数据清理;
将原始数据集进行数据集成获取数据集;
将数据集分为训练集以及测试集。
进一步的,所述定义并量化场面交通特性的交通状况指标的方法包括:
采用宏观时空网络拓扑模型,对机场场面运行交通态势进行建模获取宏观时空网络拓扑结构
基于宏观时空网络拓扑结构,定义体现场面交通量的四类指标并进行量化。
进一步的,所述基于数据集以及交通状况指标分析和提取影响场面滑出时间的特征集的方法包括:
从数据集以及交通状况指标提取影响场面滑出时间的特征并构成原始特征集;
对原始特征集中的特征进行特征分析
依据特征分析结果构建特征集。
进一步的,所述对原始特征集中的特征进行特征分析的方法包括:
采用相关性度量相关系数、标准化互信息以及因子分析三者中的一种或多种对原始特征集重的特征进行特征分析;
相关性度量相关系数反映两个变量线性相关程度的统计量,其取值为[-1,1],绝对值越大表示线性相关程度越强,正值表示正相关,负值表示负相关,用X、Y代指任意两个变量,相关性度量相关系数P X,Y的定义为:
Figure PCTCN2020089908-appb-000001
其中Cov(X,Y)为X与Y的协方差,σ X、σ Y为X、Y的标准差,μ X、μ Y为X、Y的均值;
标准化互信息为常用相关度量其取值范围为[0,1],值越大表示变量间的相关程度越大,标准化互信息U X,Y的定义为:
Figure PCTCN2020089908-appb-000002
其中,I X,Y为X、Y的互信息,H X、H Y为X、Y各自的信息熵,p(x,y)为X、Y的联合概率分布,p(x)、p(y)为X、Y各自的概率分布;
因子分析,即,提取到的特征x是完全被潜在影响因子z控制的,表达式为x=Az+ε,其中A为系数矩阵,ε为误差,加以影响因子之间互相独立、影响因子与误差互相独立,最终推导得出:∑ x=AA T+∑ ε,其中∑表示协方差矩阵,从而可以求出A与z。
进一步的,所述依据特征集通过集成机器学习方法建立场面滑出时间预测模型的方法包括:
将特征集作为集成学习模型GBRT的输入获取初始模型;
对初始模型进行训练并调整超参数取值从而完成场面滑出时间预测模型的建立。
进一步的,所述对初始模型进行训练并调整参数取值从而完成场面滑出时间预测模型的建立的方法包括:
选取最大深度作为控制决策树的控制方式;
选取最小二乘作为损失函数;
最优乘积值下,选择能保持性能稳定下最大的学习率和相应最小的估计器数量;
根据训练集中滑出时间的整体数据分布,设置最小样本划分为200;
完成对初始模型的训练从而建立场面滑出时间预测模型。
进一步的,所述依据特征集通过集成机器学习方法建立场面滑出时间预测模型的方法还包括
使用测试集对场面滑出时间预测模型进行验证并进行性能评估。
进一步的,所述使用测试集对场面滑出时间预测模型进行验证并进行性能评估中的性能评估采用均方误差,计算公式为:
Figure PCTCN2020089908-appb-000003
其中N为测试集样本数量,o i为第i个样本的实际滑行时间,p i为模型的预测滑行时间。
本发明的有益效果是,本发明提供了一种基于大数据深度学习的机场场面可变滑出时间预测方法。基于大数据深度学习的机场场面可变滑出时间预测方法包括:获取历史运行数据并进行数据清洗从而获得数据集;定义并量化场面交通特性的交通状况指标;基于数据集以及交通状况指标分析和提取影响场面滑出时间的特征集;依据特征集通过集成机器学习方法建立场面滑出时间预测模型,通过场面滑出时间预测模型完成对机场场面滑出时间的预测。处理机场原始记录数据,对机场场面交通状况进行建模,分析和提取滑行时间影响因素,训练GBRT集成学习模型,进而得到滑出时间预测模型,为机场运行的管理和优化提供数据依据。
附图说明
下面结合附图和实施例对本发明进一步说明。
图1是本发明所提供的基于大数据深度学习的机场场面可变滑出时间预测方法的流程图。
图2是本发明所提供的滑行过程宏观时空网络拓扑结构。
图3是本发明所提供的候选影响因素与滑出时间的相关性度量相关系数。
图4是本发明所提供的候选影响因素与滑出时间的标准化互信息关系。
图5是本发明所提供的候选影响因素的因子分析结果图。
图6是本发明所提供的模型训练与测试阶段性能变化过程图。
具体实施方式
现在结合附图对本发明作进一步详细的说明。这些附图均为简化的示意图,仅以示意方式说明本发明的基本结构,因此其仅显示与本发明有关的构成。
实施例1
如图1所示,本实施例1提供了一种基于大数据深度学习的机场场面可变滑出时间预测方法,处理机场原始记录数据,对机场场面交通状况进行建模,分析和提取滑行时间影响因素,训练GBRT集成学习模型,进而得到滑出时间预测模型,为机场运行的管理和优化提供数据依据。具体的,基于大数据深度学习的机场场面可变滑出时间预测方法包括:
S110:获取历史运行数据并进行数据清洗从而获得数据集;
S120:定义并量化场面交通特性的交通状况指标;
S130:基于数据集以及交通状况指标分析和提取影响场面滑出时间的特征集;
S140:依据特征集通过集成机器学习方法建立场面滑出时间预测模型;
S150:通过场面滑出时间预测模型完成对机场场面滑出时间的预测。
在本实施例中,步骤S110包括:
S111:获取历史运行数据构建原始数据集。
具体的,从机场场面运行数据库中尽可能多的提取数据,构成机场航班离港运行原始数据集。收集滑行轨迹相关信息,包括离港跑道、离港停机位、走廊口编号、滑行长度等;收集航班属性相关信息,包括航班号、航班类型、机型、所属航司、引擎类型等;收集交通管制相关信息,包括是否受限、管制员信息、通话信息、延误情况、本场气象、机场通播等;收集飞行计划相关信息,包括起飞机场、目的机场、计划起飞时间、计划撤轮档时间、航路点信息等;收集滑行过程实录信息,包括撤轮档时间、推出时间、请求/许可开车时间、实际起飞时间、滑行速度、跑道头等待时间等。
S112:对原始数据集进行数据清理。
具体的,考虑机场实获数据集的情况,为实际工作制定具体的处理方案。在缺失值处理方面,采用设置默认缺省值和直接删除两种方法。设置默认缺省值,为“是否受限”设置默认缺省值“否”,为“受限内容”设置默认缺省值“无”。在默认缺省值填充完毕后,直接删除了信息缺失超过半数的属性,包括“请求开车”、“许可开车”、“撤轮挡时间”、“尾流”、“滑行速度”、“离场排队数”。之后,对数据集进行完备性检查,删除缺失信息的数据条目。在异常值处理方面,首先对所有属性的数据进行数据类型和是否越界的基本检查,再采用定界检测法对部分属性进一步检查异常值。基于机场场面运行实际情况为属性划定取值范围,将取值不在对应范围内的数据视作异常值。最后从数据集中删除含有异常值的数据条目。属性取值范围如下表所示。
部分属性的取值范围
Figure PCTCN2020089908-appb-000004
Figure PCTCN2020089908-appb-000005
S113:将原始数据集进行数据集成获取数据集。
具体的,本步骤包含冗余属性识别、数据类型转换以及逻辑错误检验的工作。识别并删除冗余属性,通过计算各属性的信息熵识别携带信息较少的冗余属性,通过计算属性间的互信息识别信息被其他属性包含的冗余属性。删除了冗余属性“起飞机场”、“执行日期”。转换数据类型,将非数值型属性中仅具标识作用的信息转换为易于后续处理和使用的整数值型。“受限内容”属性包含的信息难以量化,综合考虑后予以删除。检查逻辑错误,考虑各特征的物理意义,建立特征间的约束关系,排除逻辑错误。检查机型与引擎数量的对应关系,检查场面运行中各时间节点的先后关系,直接删除存在逻辑错误的信息条目。
S114:将数据集分为训练集以及测试集。
具体的,数据集被划分为两个部分,分别是训练集和测试集。其中90%的数据为训练集用于模型的训练阶段,而10%的数据作为测试集被用于验证模型有效性和鲁棒性。”也就是说,训练集与测试集是同源同性质的。在得到最终处理好的数据集之后,在机器学习模型训练之前,从数据集中预留10%用作测试,将数据集中的剩下90%当做训练集训练机器学习模型。
在本实施例中,步骤S120包括:
S121:采用宏观时空网络拓扑模型,对机场场面运行交通态势进行建模获取宏观时空网络拓扑结构;
具体的,采用宏观时空网络拓扑模型,对机场场面运行交通态势进行建模。图2可视化了在任何时空域下离港和进港的滑行过程中网络拓扑的一般情况。在机场场面的实际运行中,滑入和滑出的过程是相互耦合、相互依存的。因此,在模型中同时考虑到进港对出港过程的影响。时空网络拓扑模型是描述机场系统宏观资源流动的通用框架,如图2所示,离港d 1,...,d 4表示与参考离港航班d 0的所有四种不同关系,分别是“推出前,起飞前”、“推出前,起飞后”、“推出后,起飞前”以及“推出后,起飞后”。相似地,进港a 1,...,a 4表示与参考进港航班a 0的所有四种不同关系,分别是“落地前,到位前”、“落地前,到位后”、“落地后,到位前”以及“落地后,到位后”。t on,t in表示参考进港航班a 0的落地时间和到位时间。t out,t off表示参考离港航班的推出时间和起飞时间。δ表示进港和离港的时间阈值。
S122:基于宏观时空网络拓扑结构,定义体现场面交通量的四类指标并进行量化。
具体的,基于宏观时空网络拓扑结构,定义了体现场面交通量的四类共八个指标。这四类分别是场面瞬时流量指数(SIFIs)、场面累积流量指数(SCFIs)、飞机排队长度指数(AQLIs)和槽资源需求指数(SRDIs)。每个类别中计算两个统计量,分别是离港航空器的数量(前缀为D-)和进港航空器的数量(前缀为A-)。下表显示了以d 0为参考离港航班在图2情况下的各种统计量。
离港航班d 0场面交通态势指标统计结果
Figure PCTCN2020089908-appb-000006
Figure PCTCN2020089908-appb-000007
以图2为例,下面详细介绍了表1中指标的定义和计算方法。对于任何离场航班d 0,SIFIs包括D-SIFI和A-SIFI,分别表示当d 0从登机口推出时,滑行离港和进港的航班数量。SCFIs包括D-SCFI和A-SCFI,分别表示离港和进港航空器的滑行周期与d 0滑行周期重叠的数量。AQLIs包括D-AQLI和A-AQLI,分别表示d 0整个滑行过程中在跑道上的起飞和降落的航空器数量。SRDIs包括D-SRDI和A-SRDI,表示在航空器d 0的离港槽[t 0-δ,t 0+δ]期间推出和降落航空器的数量。一般来说,δ的取值可以设置为10分钟到30分钟之间。
在本实施例中,步骤S130包括:
S131:从数据集以及交通状况指标提取影响场面滑出时间的特征并构成原始特征集。
具体的,对步骤S110以及步骤S120获取的影响场面滑出时间的相关因素进行整理,构成原始特征集。处理原始特征集,从原始特征中提取新特征替换原始特征集中的部分特征。
步骤S110获取的影响场面滑出时间相关因素为:航班号、航班属性、目的机场、计划起飞时间、机型、所属航司、推出时间、实际起飞时间、离港跑道、离港停机位、停机位类型、引擎类型、走廊口、是否受限、登机口。S120获取的影响场面画出时间相关因素为:D-SIFI、D-SCFI、D-AQLI、D-SRDI、Corridor_NO。使用推出时间与实际起飞时间之差作为场面滑行时间,替代原特征。从计划起飞时间中提取月、日、周、小时、分钟新特征替代原特征。对停机位、登机口特征进行进一步划分和分析。提取跑道与机位/登机口的对应关系作为新特征。最终获取的原始特征集即候选影响因素如下表所示:
候选影响因素
Figure PCTCN2020089908-appb-000008
S132:对原始特征集中的特征进行特征分析;
S133:依据特征分析结果构建特征集。
具体的,基于步骤S132的分析结果,从步骤S131形成的原始特征集中选出重要特征,构成用于集成机器学习模型的特征集。筛除部分与场面滑行时间相关性较小的特征。包括“引擎类型”、“停机位类型”、“月”、“周”、“日”、“分钟”。最终获取的特征集即影响因素如下表所示:
最终选取的影响因素
Figure PCTCN2020089908-appb-000009
在本实施例中,S132包括:
采用相关性度量相关系数、标准化互信息以及因子分析三者中的一种或多种对原始特征集重的特征进行特征分析,图3、图4、图5分别展示了候选影响因素与滑出时间的Pearson相关系数、候选影响因素与滑出时间的标准化互信息以及候选影响因素的因子分析结果。
相关性度量相关系数反映两个变量线性相关程度的统计量,其取值为[-1,1],绝对值越大表示线性相关程度越强,正值表示正相关,负值表示负相关,用X、Y代指任意两个变量,相关性度量相关系数P X,Y的定义为:
Figure PCTCN2020089908-appb-000010
其中Cov(X,Y)为X与Y的协方差,σ X、σ Y为X、Y的标准差,μ X、μ Y为X、Y的均值;
标准化互信息是常用相关度量其取值范围为[0,1],值越大表示变量间的相关程度越大,标准化互信息U X,Y的定义为:
Figure PCTCN2020089908-appb-000011
其中,I X,Y为X、Y的互信息,H X、H Y为X、Y各自的信息熵,p(x,y)为X、Y的联合概率分 布,p(x)、p(y)为X、Y各自的概率分布
因子分析,即,提取到的特征x是完全被潜在影响因子z控制的,表达式为x=Az+ε,其中A为系数矩阵,ε为误差,加以影响因子之间互相独立、影响因子与误差互相独立,最终推导得出:∑ x=AA T+∑ ε,其中∑表示协方差矩阵,从而可以求出A与z。
在本实施例中,步骤S140包括:
S141:将特征集作为集成学习模型GBRT的输入获取初始模型;
S142:对初始模型进行训练并调整超参数取值从而完成场面滑出时间预测模型的建立。
在本实施例中,S142包括:选取“最大深度”作为控制决策树的控制方式;选取“最小二乘”作为损失函数;最优乘积值下,选择能保持性能稳定下最大的学习率和相应最小的估计器数量;根据训练集中滑出时间的整体数据分布,设置最小样本划分为200;完成对初始模型的训练从而建立场面滑出时间预测模型。
具体的,采用集成学习的典型代表GradientBoostedRegressionTrees(GBRT)模型来完成场面滑出时间的预测操作。将步骤S133所获的特征集作为模型的输入,通过执行scikit-learn库中的算法快速训练GBRT模型。需要设置的超参数有:决策树大小控制、损失函数类型、估计器个数与学习率以及最小样本划分。在控制决策树大小上,共有两种方式选择,分别是“最大深度(max_depth)”和“最大叶子结点个数(max_leaf_nodes)”。在回归任务中共有四种可选损失函数,分别是“最小二乘(ls)”、“最小绝对偏差(lad)”、“Huber损失(huber)”以及“分位数损失(quantile)”。由于学习率和估计器个数是具有高度的相互作用,二者的乘积大致反映迭代训练情况。因此在设置参数的时候,根据经验设置不同的乘积值,并选择在训练集中获得最好性能的乘积值。最小样本划分用于控制叶子节点中样本个数下限,用于提高模型的鲁棒性。总的来说需要根据应用场景的数据条件合理的调整超参数取值。
具体的,GBRT模型F(x)是以下形式的可加模型:
Figure PCTCN2020089908-appb-000012
其中h m(x)是基函数,通常在boosting的概念下被称为弱学习器,γ m是弱学习器对应的权重,M是弱学习器的数量和。GBRT使用固定大小的决策树作为弱学习器。与其他boosting算法思想类似,GBRT贪婪地构建了可加模型:
F m(x)=F m-1(x)+γ mh m(x)
其中,Fm(x)表示第m次迭代得到的GBRT模型。其中hm(x)由
Figure PCTCN2020089908-appb-000013
得出。n为训练样本总数,L为选定的损失函数,yi为第i个样本的标签,Fm-1(xi)为第m-1次迭代获取的GBRT模型对第i个样本的预测值,h(xi)为要获取的弱学习器对第i个样本的预测值。而γ m
Figure PCTCN2020089908-appb-000014
得出。n为训练样本总数,L为选定的损失函数,yi为第i个样本的标签,Fm-1(xi)为第m-1次迭代获取的GBRT模型对第i个样本的预测值
最初的模型F 0是与问题相关的,对于最小二乘回归,通常选择目标值的平均值。
即,模型未训练状态为
Figure PCTCN2020089908-appb-000015
在本实施例中,S140还包括:
S143:使用测试集对场面滑出时间预测模型进行验证并进行性能评估。
具体的,所述使用测试集对场面滑出时间预测模型进行验证并进行性能评估中的性能评估采用均方误差,计算公式为:
Figure PCTCN2020089908-appb-000016
其中N为测试集样本数量,o i为第i个样本的实际滑行时间,p i为模型的预测滑行时间。
在本实施例中,使用MSE监测模型在训练和测试过程中的性能变化,结果如图6所示。最终,MSE在训练集中达到2.5,而在测试集的性能为5.5。尽管训练集和测试集的MSE性能上有一定的距离,但在一定程度上反应了模型的泛化能力。
另一方面,下表比较了测试集不同误差范围内的预测精度情况。在所有测试集中,85.7%的数据集其滑行时间误差在3分钟之内;超过93%的数据,其预测误差在4分钟之间;大约96.5%的数据,其误差在5分钟之内。由上述测试集上的验证结果可知,所设计的数据挖掘模型及算法能够较好的达到实际场面动态滑出时间预测任务的精度要求。
不同误差范围内的测试集精度
误差范围 [-3,3] [-4,4] [-5,5]
精度 85.7% 93.1% 96.5%
综上所述,本发明提供了一种基于大数据深度学习的机场场面可变滑出时间预测方法。基于大数据深度学习的机场场面可变滑出时间预测方法包括:获取历史运行数据并进行数据清洗从而获得数据集;定义并量化场面交通特性的交通状况指标;基于数据集以及交通状况指标分析和提取影响场面滑出时间的特征集;依据特征集通过集成机器学习方法建立场面滑出时间预测模型,通过场面滑出时间预测模型完成对机场场面滑出时间的预测。处理机场原始记录数据,对机场场面交通状况进行建模,分析和提取滑行时间影响因素,训练GBRT集成学习模型,进而得到滑出时间预测模型,为机场运行的管理和优化提供数据依据。
以上述依据本发明的理想实施例为启示,通过上述的说明内容,相关工作人员完全可以在不偏离本项发明技术思想的范围内,进行多样的变更以及修改。本项发明的技术性范围并不局限于说明书上的内容,必须要根据权利要求范围来确定其技术性范围。

Claims (9)

  1. 一种基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,包括:
    获取历史运行数据并进行数据清洗从而获得数据集;
    定义并量化场面交通特性的交通状况指标;
    基于数据集以及交通状况指标分析和提取影响场面滑出时间的特征集;
    依据特征集通过集成机器学习方法建立场面滑出时间预测模型;
    通过场面滑出时间预测模型完成对机场场面滑出时间的预测。
  2. 如权利要求1所述的基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,
    所述获取历史运行数据并进行数据清洗从而获得数据集的方法包括:
    获取历史运行数据构建原始数据集;
    对原始数据集进行数据清理;
    将原始数据集进行数据集成获取数据集;
    将数据集分为训练集以及测试集。
  3. 如权利要求2所述的基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,
    所述定义并量化场面交通特性的交通状况指标的方法包括:
    采用宏观时空网络拓扑模型,对机场场面运行交通态势进行建模获取宏观时空网络拓扑结构;
    基于宏观时空网络拓扑结构,定义体现场面交通量的四类指标并进行量化。
  4. 如权利要求3所述的基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,
    所述基于数据集以及交通状况指标分析和提取影响场面滑出时间的特征集的方法包括:
    从数据集以及交通状况指标提取影响场面滑出时间的特征并构成原始特征集;
    对原始特征集中的特征进行特征分析;
    依据特征分析结果构建特征集。
  5. 如权利要求4所述的基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,
    所述对原始特征集中的特征进行特征分析的方法包括:
    采用相关性度量相关系数、标准化互信息以及因子分析三者中的一种或多种对原始特征集重的特征进行特征分析;
    相关性度量相关系数反映两个变量线性相关程度的统计量,其取值为[-1,1],绝对值越大表示线性相关程度越强,正值表示正相关,负值表示负相关,用X、Y代指任意两个变量,相关性度量相关系数P X,Y的定义为:
    Figure PCTCN2020089908-appb-100001
    其中Cov(X,Y)为X与Y的协方差,σ X、σ Y为X、Y的标准差,μ X、μ Y为X、Y的均值;
    标准化互信息为常用相关度量其取值范围为[0,1],值越大表示变量间的相关程度越 大,标准化互信息U X,Y的定义为:
    Figure PCTCN2020089908-appb-100002
    其中,I X,Y为X、Y的互信息,H X、H Y为X、Y各自的信息熵,p(x,y)为X、Y的联合概率分布,p(x)、p(y)为X、Y各自的概率分布;
    因子分析,即提取到的特征x是完全被潜在影响因子z控制的,表达式为x=Az+ε,其中A为系数矩阵,ε为误差,加以影响因子之间互相独立、影响因子与误差互相独立,最终推导得出:∑ x=AA T+∑ ε,其中∑表示协方差矩阵,从而可以求出A与z。
  6. 如权利要求5所述的基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,
    所述依据特征集通过集成机器学习方法建立场面滑出时间预测模型的方法包括:
    将特征集作为集成学习模型GBRT的输入获取初始模型;
    对初始模型进行训练并调整超参数取值从而完成场面滑出时间预测模型的建立。
  7. 如权利要求6所述的基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,
    所述对初始模型进行训练并调整参数取值从而完成场面滑出时间预测模型的建立的方法包括:
    选取最大深度作为控制决策树的控制方式;
    选取最小二乘作为损失函数;
    最优乘积值下,选择能保持性能稳定下最大的学习率和相应最小的估计器数量;
    根据训练集中滑出时间的整体数据分布,设置最小样本划分为200;
    完成对初始模型的训练从而建立场面滑出时间预测模型。
  8. 如权利要求7所述的基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,
    所述依据特征集通过集成机器学习方法建立场面滑出时间预测模型的方法还包括
    使用测试集对场面滑出时间预测模型进行验证并进行性能评估。
  9. 如权利要求8所述的基于大数据深度学习的机场场面可变滑出时间预测方法,其特征在于,
    所述使用测试集对场面滑出时间预测模型进行验证并进行性能评估中的性能评估采用均方误差,计算公式为:
    Figure PCTCN2020089908-appb-100003
    其中N为测试集样本数量,o i为第i个样本的实际滑行时间,p i为模型的预测滑行时间。
PCT/CN2020/089908 2019-10-30 2020-05-13 基于大数据深度学习的机场场面可变滑出时间预测方法 WO2021082393A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201911043824.6A CN110826788A (zh) 2019-10-30 2019-10-30 基于大数据深度学习的机场场面可变滑出时间预测方法
CN201911043824.6 2019-10-30

Publications (1)

Publication Number Publication Date
WO2021082393A1 true WO2021082393A1 (zh) 2021-05-06

Family

ID=69551262

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/089908 WO2021082393A1 (zh) 2019-10-30 2020-05-13 基于大数据深度学习的机场场面可变滑出时间预测方法

Country Status (2)

Country Link
CN (1) CN110826788A (zh)
WO (1) WO2021082393A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115712281A (zh) * 2022-11-15 2023-02-24 中国电子科技集团公司第三十八研究所 一种基于数据挖掘的回流焊工艺质量预测方法
CN117668497A (zh) * 2024-01-31 2024-03-08 山西卓昇环保科技有限公司 基于深度学习实现环境保护下的碳排放分析方法及系统
CN118313848A (zh) * 2024-06-11 2024-07-09 贵州省畜牧兽医研究所 一种用于肉牛冻精溯源过程的数据保护方法及系统

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110826788A (zh) * 2019-10-30 2020-02-21 南京智慧航空研究院有限公司 基于大数据深度学习的机场场面可变滑出时间预测方法
CN110852497A (zh) * 2019-10-30 2020-02-28 南京智慧航空研究院有限公司 基于大数据深度学习的场面可变滑出时间预测系统
CN113537506B (zh) * 2020-04-22 2023-08-29 百度在线网络技术(北京)有限公司 针对机器学习效果的测试方法、装置、设备和介质
CN111652427B (zh) * 2020-05-29 2023-12-29 航科院中宇(北京)新技术发展有限公司 一种基于数据挖掘分析的航班到达时刻预测方法及系统
CN111616721B (zh) * 2020-05-31 2022-05-27 天津大学 基于深度学习和脑机接口的情绪识别系统及应用
CN111616682B (zh) * 2020-05-31 2022-05-27 天津大学 基于便携式脑电采集设备的癫痫发作预警系统及应用
US11610497B2 (en) 2020-12-15 2023-03-21 Rockwell Collins, Inc. System and method to display SVS taxi mode exocentric view of aircraft
CN112749754B (zh) * 2021-01-18 2024-05-14 中国民航信息网络股份有限公司 一种对不正常计算撤轮档时间预警的方法及装置
US12025456B2 (en) 2021-03-29 2024-07-02 Rockwell Collins, Inc. System and method to display airport moving map and taxi routing guidance content

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106529734A (zh) * 2016-11-18 2017-03-22 中国民航大学 一种基于knn和svr的航班滑出时间预测方法
CN108846523A (zh) * 2018-07-31 2018-11-20 中国民航大学 一种基于贝叶斯网的航班离港滑行时间动态预测方法
CN110363333A (zh) * 2019-06-21 2019-10-22 南京航空航天大学 一种基于渐进梯度回归树的天气影响下空中通行能力的预测方法
CN110826788A (zh) * 2019-10-30 2020-02-21 南京智慧航空研究院有限公司 基于大数据深度学习的机场场面可变滑出时间预测方法
CN110852497A (zh) * 2019-10-30 2020-02-28 南京智慧航空研究院有限公司 基于大数据深度学习的场面可变滑出时间预测系统

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105023068B (zh) * 2015-08-06 2019-03-01 北京航空航天大学 一种基于规则挖掘的航班进离场协同调度方法
CN106339358B (zh) * 2016-08-16 2018-11-09 南京航空航天大学 基于多元回归分析的航空器场面滑行时间预测方法
CN106709594A (zh) * 2016-11-18 2017-05-24 中国民航大学 一种基于排队论的航班滑出时间预测方法
CN108764560B (zh) * 2018-05-22 2022-03-15 电子科技大学 基于长短期记忆神经网络的航空器场面轨迹预测方法
CN109285346B (zh) * 2018-09-07 2020-05-05 北京航空航天大学 一种基于关键路段的城市路网交通状态预测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106529734A (zh) * 2016-11-18 2017-03-22 中国民航大学 一种基于knn和svr的航班滑出时间预测方法
CN108846523A (zh) * 2018-07-31 2018-11-20 中国民航大学 一种基于贝叶斯网的航班离港滑行时间动态预测方法
CN110363333A (zh) * 2019-06-21 2019-10-22 南京航空航天大学 一种基于渐进梯度回归树的天气影响下空中通行能力的预测方法
CN110826788A (zh) * 2019-10-30 2020-02-21 南京智慧航空研究院有限公司 基于大数据深度学习的机场场面可变滑出时间预测方法
CN110852497A (zh) * 2019-10-30 2020-02-28 南京智慧航空研究院有限公司 基于大数据深度学习的场面可变滑出时间预测系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LI XIA; BAI RUIBIN: "Freight Vehicle Travel Time Prediction Using Gradient Boosting Regression Tree", 2016 15TH IEEE INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND APPLICATIONS (ICMLA), IEEE, 18 December 2016 (2016-12-18), pages 1010 - 1015, XP033055658, DOI: 10.1109/ICMLA.2016.0182 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115712281A (zh) * 2022-11-15 2023-02-24 中国电子科技集团公司第三十八研究所 一种基于数据挖掘的回流焊工艺质量预测方法
CN117668497A (zh) * 2024-01-31 2024-03-08 山西卓昇环保科技有限公司 基于深度学习实现环境保护下的碳排放分析方法及系统
CN117668497B (zh) * 2024-01-31 2024-05-07 山西卓昇环保科技有限公司 基于深度学习实现环境保护下的碳排放分析方法及系统
CN118313848A (zh) * 2024-06-11 2024-07-09 贵州省畜牧兽医研究所 一种用于肉牛冻精溯源过程的数据保护方法及系统

Also Published As

Publication number Publication date
CN110826788A (zh) 2020-02-21

Similar Documents

Publication Publication Date Title
WO2021082393A1 (zh) 基于大数据深度学习的机场场面可变滑出时间预测方法
WO2021082394A1 (zh) 基于大数据深度学习的场面可变滑出时间预测系统
US12026440B1 (en) Optimizing aircraft flows at airports using data driven predicted capabilities
CN111652427B (zh) 一种基于数据挖掘分析的航班到达时刻预测方法及系统
WO2020248228A1 (zh) 一种Hadoop平台计算节点负载预测方法
Choi et al. Artificial neural network models for airport capacity prediction
CN106339358B (zh) 基于多元回归分析的航空器场面滑行时间预测方法
CN110210648B (zh) 基于灰色长短期记忆网络的管制空域战略流量预测方法
CN110348601A (zh) 一种基于双向长短期记忆网络的地铁短期客流量预测方法
Wu et al. An improved svm model for flight delay prediction
CN109637196A (zh) 航路扇区交通概率密度预测方法
Provan et al. A probabilistic airport capacity model for improved ground delay program planning
CN105809280A (zh) 一种机场容量需求预测方法
WO2017071369A1 (zh) 一种预测用户离网的方法和设备
Zhang et al. Data‐driven flight time prediction for arrival aircraft within the terminal area
Truong et al. Applications of business analytics in predicting flight on-time performance in a complex and dynamic system
CN112365037A (zh) 基于长短期数据预测模型的机场空域流量预测方法
CN114781704B (zh) 一种基于过站航班保障流程的航班延误预测方法
CN116911434A (zh) 一种机场运行态势预测方法、装置、系统以及存储介质
CN114611781A (zh) 一种面向航班计划的离港旅客聚集信息预测方法及系统
Lu et al. Flight delay prediction using gradient boosting machine learning classifiers
CN116109212B (zh) 一种机场运行效率评价指标设计及监测方法
CN118211034A (zh) 一种基于knn回归模型的多维度民航客流量预测方法
Baspinar et al. Large scale data-driven delay distribution models of european air traffic flow network
Ehsanifar et al. A simulation approach to evaluate performance indices of fuzzy exponential queuing system (An M/M/C model in a banking case study)

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20880742

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20880742

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 20880742

Country of ref document: EP

Kind code of ref document: A1