CN108364091A - Probability en-route sector Traffic Demand Forecasting flow management system - Google Patents
Probability en-route sector Traffic Demand Forecasting flow management system Download PDFInfo
- Publication number
- CN108364091A CN108364091A CN201810082230.5A CN201810082230A CN108364091A CN 108364091 A CN108364091 A CN 108364091A CN 201810082230 A CN201810082230 A CN 201810082230A CN 108364091 A CN108364091 A CN 108364091A
- Authority
- CN
- China
- Prior art keywords
- sector
- aircraft
- time
- prediction error
- probability
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
- 238000009826 distribution Methods 0.000 claims abstract description 52
- 241001269238 Data Species 0.000 claims abstract description 7
- 238000011156 evaluation Methods 0.000 claims abstract description 6
- 238000000034 method Methods 0.000 claims description 40
- 238000005315 distribution function Methods 0.000 claims description 10
- 230000008569 process Effects 0.000 claims description 10
- 230000001939 inductive effect Effects 0.000 claims description 3
- 230000033228 biological regulation Effects 0.000 claims description 2
- 239000004744 fabric Substances 0.000 claims description 2
- 238000007619 statistical method Methods 0.000 abstract description 12
- 230000032258 transport Effects 0.000 description 20
- 238000007726 management method Methods 0.000 description 17
- 238000012360 testing method Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 5
- 238000007689 inspection Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000007812 deficiency Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000013277 forecasting method Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000011002 quantification Methods 0.000 description 1
- 230000001568 sexual effect Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G5/00—Traffic control systems for aircraft, e.g. air-traffic control [ATC]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Tourism & Hospitality (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Operations Research (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Marketing (AREA)
- Mathematical Physics (AREA)
- General Business, Economics & Management (AREA)
- Software Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Databases & Information Systems (AREA)
- Algebra (AREA)
- Development Economics (AREA)
- Probability & Statistics with Applications (AREA)
- Game Theory and Decision Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Aviation & Aerospace Engineering (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention relates to a kind of probability en-route sector Traffic Demand Forecasting flow management systems comprising：It predicts that error sorts out statistical module, classification statistics is carried out by the prediction error of sector borders point time within the corresponding period to aircraft；And probability evaluation entity, probabilistic forecasting en-route sector transport need value；The present invention is based on historical datas and prediction data that it's the time pasts existing aircraft, time prediction error distribution characteristics and its an influence factor are crossed by analyzing aircraft, it establishes aircraft and crosses a statistical method for time prediction error distribution character, and probability en-route sector flow management system is proposed on this basis, finally combine actual operating data, obtain the probability distribution and changing rule of sector transport need in certain time, and it finds that the accuracy rate of the probability Traffic Demand Forecasting result of gained has than the accuracy rate of traditional certainty Traffic Demand Forecasting result and greatly improves.
Description
Technical field
The present invention relates to aviation fields, and in particular to a kind of probability en-route sector Traffic Demand Forecasting traffic management system
System.
Background technology
In recent years, with the rapid development of Chinese aviation industries, air traffic congestion is increasingly prominent, and constantly from terminal
It is spread to route grid in area.In order to alleviate the air route congestion increasingly to take place frequently, the congestion management means of implementation science, premise are needed
One of be exactly accurate, objectively predict transport need.Reality is run according to current spatial domain congestion management, mainly by being based on flight path
The needing forecasting method of supposition is realized, i.e., is according to the running orbit for determining every frame aircraft, prediction with the aircraft equation of motion
The position of every frame aircraft in following a period of time, and then extrapolate the aircraft quantity in day part by certain spatial domain.This side
Final prediction result under formula shows usually in the form of certainty, i.e., traffic corresponding in spatial domain under certain predicted time scale
Requirement forecasting is the result is that a determining numerical value.Although this deterministic prediction result can meet spatial domain to a certain extent
Congestion management demand, but there are several deficiencies：First, although aircraft operational process may be considered during prediction
In many uncertain factors to prediction result influence (for example, flight cancellation in unplanned, change etc. into leaving time it is random
Event deviation caused by aircraft run time, schedule flight path caused by weather reason or the unplanned interior change of height
Deng), the representation of this deterministic forecast result is but unable to fully embody the reality of uncertain factor to a certain extent
Border influences and its degree；Further, since the odjective causes such as prediction model, input data, system inherent shortcoming, certainty result
Accuracy will decline therewith, then the extent of damage of this accuracy can not be also embodied in prediction result.
Invention content
The object of the present invention is to provide a kind of probability en-route sector Traffic Demand Forecasting flow management systems, to be based on missing
En-route sector transport need is predicted in poor distribution character statistical analysis.
In order to solve the above technical problem, the present invention provides a kind of flow management systems, including：
It predicts that error sorts out statistical module, aircraft is missed within the corresponding period by the prediction of sector borders point time
Difference carries out classification statistics；And
Probability evaluation entity, probabilistic forecasting en-route sector transport need value.
Further, the prediction error to aircraft by the sector borders point time within the corresponding period carries out classification statistics
Method include：
The route grid model of aircraft is created, determines the influence factor of prediction error.
Further, the method for creating the route grid model of aircraft includes：
Route grid is reduced to spatial domain, four class element of air route, en-route sector and sector borders point, while by entire spatial domain
It is divided into two class of target and non-targeted spatial domain；Wherein
Target spatial domain refers to the spatial domain that the en-route sector belonged in prediction spatial dimension is formed；And
The en-route sector for being not belonging to this prediction spatial dimension constitutes non-targeted spatial domain；
It is located in prediction target time section T, if quarreling aircraft passes through a certain en-route sector, wherein the i-th frame aircraft
The departure time be set asBy the predicted time of sector borders point, i.e. aircraft is crossed the predicted value of time and is set asThrough
The real time of sector borders point is spent, i.e. aircraft is crossed the actual value of time and is set asAnd aircraft passes through sector borders
The time prediction error of point, i.e. aircraft cross the prediction error delta t of timeiIt is defined as
The prediction error sample that aircraft is crossed to the time is divided into two subsets, i.e.,
First m days are subset I, and the m+1 days to the M days are subset II；Wherein
Subset I is used for the distribution character of statistical forecast error, and subset II is used to verify the validity of statistical law.
Further, determine that the method for predicting the influence factor of error includes：
The influence factor includes：Typical case runs the in a few days busy extent of different periods and predicted time scale；And
Using dimensional probability distribution f (Tk, Pj) describe to predict the distribution character of error with the changing rule of period, wherein
TkIt indicates period subregion, with the busy extent of the typical operation day different periods of reflection, i.e., transports the typical case of aircraft
Row day, whole day with a fixed step size was divided into several period subregions；K indicates error statistics period period, k=1,2 ..., K, K tables
Show carry out error statistics it is total when hop count；And
PjIt is the characteristic quantity for describing aircraft and spending the time.
Further, classification statistics is carried out to the prediction error, i.e., distribution character statistics, method is carried out to prediction error
Including：
By TkPrediction error sample in period by prediction error size and its number be uniformly distributed carry out it is double
It divides.
Further, by TkPrediction error sample in period by prediction error size and its number be uniformly distributed into
The method of the double division of row includes：
First, by TkThe prediction error sample of period carries out first time division by the size of prediction error value, i.e.,
Establish the coordinate system divided for the first time, abscissa Tk+ i and i=0,2 ..., 59, indicate prediction error value institute
Corresponding period points, ordinate is prediction error value, sets classificatory scale, to cross the prediction error of sector borders point time into
Row divides at equal intervals；
Setting prediction error sample size reference interval [l- δ, l+ δ], l indicate that suitable number of samples, δ are fluctuation model
It encloses, is merged from two side of predicted value to third side to the section at equal intervals to Preliminary division；
After aforesaid operations, correspond to TkPeriod obtains total W layers of subregion, if each subregion is LK, w(w=1,
2 ..., W), and LK, wContain R in a subregionK, wA sample；And it is expanded respectively from transverse direction, regulation of longitudinal angle to going through along time shaft
History sample data carries out inducing classification.
Further, further include to the method for prediction error progress distribution character statistics：
L is calculated one by oneK, w, and k=1,2 ..., K；W=1,2 ..., W layers of prediction error sample vr, and r=1,
2 ..., RK, w, to predict percentage error as abscissa, suitable error step-length is chosen as region interval width, according to institute
The v acquiredrValue focuses on sample point in corresponding separate；
To obtain the number of samples in each prediction error band interval it is respectively m through countings, wherein s=1,2 ...,
S, andAnd f is obtained according to the number of samples in prediction error band intervals=ms/RK, w；
Work as RK, wWhen fully big, you can by fsIt is considered as TkThe L of periodK, wThe prediction error of layer subregion it is discrete it is definite generally
Rate is distributed；
All historical datas are traversed, to obtain the discrete exact method distribution function of MW prediction error, i.e. error is united
Count table.
Further, include according to the method that classification statistical probability predicts en-route sector transport need value：
If in prediction object time section [tk, tk+ 14] in, if shared aircraft of quarreling passes through an en-route sector, currently
Aircraft passes through the sector borders point and enters the predicted time error of the sectorProbability densityCorresponding probability
Distribution function isAnd aircraft passes through the sector borders point and leaves the predicted time error of the sectorIt is general
Rate densityCorresponding to probability-distribution function is
Since the predicted time that the i-th frame aircraft enters the en-route sector isWhereinFor according to aviation
The accurate point time excessively that device departure time, flight path length and flying quality obtain, andWithThe regularity of distribution one
It causes, then in [tk, tk+ 14] probability of the aircraft into the en-route sector is in：
And in [tk, tk+ 14] probability that the aircraft leaves en-route sector in the period is：
Have formula (1)(2) it is found that in [tk, tk+ 14] probability that the aircraft is located in en-route sector in the period is：
And
If following [t one dayk, tk+ 14] certain en-route sector deterministic demand prediction result is M sorties in the period, then
There may be M frame aircrafts process, then there are the probability of m frame aircrafts in sector in the period is：PM[m](0≤m
≤ M), by pseudo-program representation, obtain its result.
The invention has the advantages that the present invention is based on historical datas and prediction number that it's the time pasts existing aircraft
According to crossing time prediction error distribution characteristics and its an influence factor by analyzing aircraft, establishing aircraft, to spend the time pre-
The statistical method of error distribution character is surveyed, and proposes probability en-route sector flow management system on this basis, is finally tied
Actual operating data is closed, the probability distribution and changing rule of sector transport need in certain time are obtained, and finds that gained is general
The accuracy rate of forthright Traffic Demand Forecasting result has than the accuracy rate of traditional certainty Traffic Demand Forecasting result to be greatly improved,
Illustrate that the present invention can provide the Traffic Demand Forecasting foundation of more science for air traffic flow management.
Description of the drawings
Present invention will be further explained below with reference to the attached drawings and examples.
Fig. 1 is the functional block diagram of this flow management system；
Fig. 2 is the ALFISOL IN CENTRAL sectors AR05 day part in May aircraft distributed number；
Fig. 3 is the ALFISOL IN CENTRAL sectors AR01-AR08 aircraft in May distributed number；
Fig. 4 is route grid schematic diagram；
Fig. 5 is subregion L1,1Predict error distribution；
Fig. 6 is subregion L1,2Predict error distribution；
Fig. 7 is that prediction error band divides signal；
Fig. 8 a are the predicted time error Normal distribution test Q-Q figures that flight enters an en-route sector；
Fig. 8 b are the predicted time error Normal distribution test Q-Q figures that flight leaves an en-route sector；
Fig. 9 is the comparison of AR05 day parts certainty and probability requirement forecasting result and actual traffic flow value and capacity
Figure.
Specific implementation mode
In conjunction with the accompanying drawings, the present invention is further explained in detail.These attached drawings are simplified schematic diagram, only with
Illustration illustrates the basic structure of the present invention, therefore it only shows the composition relevant to the invention.
As shown in Figure 1, present embodiments providing a kind of flow management system, this flow management system is mainly for aircraft
Flight route the historical data and prediction data of time are crossed based on a large amount of aircrafts using en-route sector boundary point as object,
Establish aircraft and cross a statistical method for time prediction error distribution character, so to the probability transport need value of en-route sector into
Row prediction comprising：
It predicts that error sorts out statistical module, aircraft is missed within the corresponding period by the prediction of sector borders point time
Difference carries out classification statistics；And
Probability evaluation entity, probabilistic forecasting en-route sector transport need value.
Above-mentioned prediction error sorts out statistical module and the realization of embeded processor module may be used in probability evaluation entity.
The realization process of the present embodiment is discussed in detail below.
Data preparation
En-route sector boundary point this chance event is leapt in order to excavate really reflection aircraft from historical data
Random nature, the present embodiment use direct statistical method, to a large amount of aircrafts in special time period pass through en-route sector
The prediction error of boundary point time carries out classification statistics, so obtain it is more true, objectively respond chance event regularity from
Dissipate distribution function.
For predicting the signature analysis of error, in the present embodiment, when aircraft passes through the prediction of en-route sector boundary point
Between relative error be by historical forecast time and corresponding history it is practical cross point the time between comparison obtain.To big
It measures predicted time and its actual value is for statistical analysis, it can be deduced that it is with certain regularity, and when regularity is because of prediction
The busy extent of section, the location of en-route sector be different and difference.Concrete condition is as follows：
The busy extent of prediction period.As shown in Fig. 2, usually aircraft operation is main concentrates 09:00-24:00 period, sample
Notebook data is more, is easier to find the error law of predicted time, and 01:00-08:00 period belonged to relative free period, sample
Data are less, it is possible to be not easy to find the error law of predicted time.
En-route sector position is different.As shown in figure 3, different en-route sectors are due to location difference, the boat of process
Pocket quantity differs, and in contrast passes through the sector more than aircraft quantity since sample data is more, when being easier to find prediction
Between error law.
It is in the present embodiment, convenient for problem description about the analysis of Influential Factors of descriptive model and prediction error, it will navigate
Road network is reduced to the network model being made of spatial domain, air route, en-route sector and four class element of sector borders point, and by aircraft
The space-time (one-dimensional time and three dimensions) of flight is reduced to one-dimensional time and two-dimensional space (omitting height), entire spatial domain
It is divided into two class of target and non-targeted spatial domain, as shown in Figure 4：Target spatial domain refers to the air route fan belonged in prediction spatial dimension
The spatial domain that area is formed；Other en-route sectors for being not belonging to this estimation range constitute non-targeted spatial domain.Aircraft is by taking off
Airport departing enters target empty domain through sector borders point, flies down an airway and constantly leaps several sector borders points, until leaving
Target spatial domain.
It is located in prediction target time section T, shares N frame aircrafts and pass through a certain en-route sector, wherein i-th (1≤i≤N)
Frame aircraft fiThe departure time beBy the predicted time of en-route sector boundary point, (i.e. aircraft crosses the prediction of time
Value) beIt is by the real time (i.e. aircraft crosses the actual value of time) of en-route sector boundary pointTherefore, it is easy
Obtain the time prediction error delta t that aircraft passes through sector borders pointi(i.e. aircraft crosses the prediction error of time) is：
The forecast sample that aircraft is crossed to the time is divided into two subsets, and first m days are subset I, are within the m+1 days to the M days
Subset II, subset I are used for the distribution character of statistical forecast error, and subset II is used to verify the validity of statistical law.It is verified
Afterwards, you can the time for entering sector to the aircraft of the non-future carries out probabilistic forecasting.The analytic process of the present embodiment is to be directed to
What the case where typical en-route sector and typical normal operation day, carried out.Day emergency situations occur when en-route sector is changed or run
When (such as bad weather, occasion), needs the method using the present embodiment to change sector or run the historical data of day
Statistics and analysis is re-started, just can guarantee the accuracy and validity of finally obtained back propagation net.According to aircraft
Include by the changing rule of sector borders point time and the characteristic distributions of prediction error：Typical case's operation in a few days different periods it is numerous
Busy degree and predicted time scale are not both to influence aircraft to cross a principal element for time prediction error.In this implementation
In example, predicted time scale is specially entire predicted time section, can be, but not limited to 5h, 7h, 10h, for 24 hours etc., and use
Dimensional probability distribution f (Tk, Pj) changing rule of the prediction error characteristics with the period is described：
TkRepresent period subregion.The busy extent of typical case's operation day different periods is different, and the influence to predicting error is past
Toward presentation different characteristics.Therefore, the variation tendency of classic predictive time graph can be analyzed, and according to transport air flow
Typical case's operation day whole day is divided into several period subregions with a fixed step size, and carries out error system by the usage of trade of buret reason
Meter；Then k indicate error statistics period period, k=1,2 ..., K, K indicate carry out error statistics it is total when hop count, and it is each when
Segment length is selected as 15 minutes, can also be 10 minutes.
PjIt is the characteristic quantity for describing aircraft and spending the time.
In the present embodiment, it will a set of two-dimensional probability-distribution function is obtained by analysis, with careful description aircraft
Cross the predicted value of some time.
Predict the statistical method of error distribution character
Statistical method about prediction error distribution character is as follows：
Step S1 selectes a certain en-route sector and typical operation day compared with peak hours/period, analyzes aircraft mistake in the period
The predicted value of point time is to divide interval with 15min time windows, and every section is denoted as Tk(k=1,2 ..., K).
Step S2, by TkPrediction error sample in period by prediction error size and its number be uniformly distributed into
The double division of row.It is as follows：
Step S21, first, by TkThe prediction error sample of period by the size of prediction error value draw for the first time
Point.Establish the coordinate system divided for the first time, abscissa Tk+ i (i=0,2 ..., 59) indicate prediction error value corresponding to when
Section points, ordinate is prediction error value.Using a certain appropriate step-length λ as classificatory scale (such as 2min), to crossing en-route sector
The prediction error of boundary point time is divided at equal intervals.
Then step S22 leads to not in order to avoid sample is very few near the maximum value and minimum value due to prediction error
The problem of reflecting its change of error trend carries out secondary division to prediction error sample set.Sample size reference interval is set
[l- δ, l+ δ], l indicate suitable number of samples (such as l=8), and δ is fluctuation range, can use ± 12% or so of l.From predicted value
Two lateral third sides merge to the section at equal intervals to Preliminary division, make the prediction error sample size in each section as possible
It meets the requirements, while recording the dividing value of each segment.If occur increasing in merging process/not increase subregion sample number equal
The case where not reaching requirement then selects with reference interval at a distance of scheme of the smaller situation as interval division.
After aforesaid operations, correspond to TkPeriod can obtain total W layers of subregion, if each subregion is LK, w(w=1,
2 ..., W), and LK, wContain R in a subregionK, wA sample.Using this method, expanded respectively from lateral, longitudinal along time shaft
Angle carries out inducing classification to historical sample data, and obtained statistics classification can more preferably reflect that different periods descended en-route sector
The prediction error distribution situation of boundary point time, concrete operations are as shown in Figure 5 and Figure 6.
Step S3, calculates L one by oneK, w(k=1,2 ..., K；W=1,2 ..., W) layer prediction error sample vr(r=
1,2 ..., RK, w).To predict percentage error as abscissa, suitable error step-length is chosen as region interval width, according to
Obtained vrValue focuses on sample point in corresponding separate, as shown in Figure 7.
Step S4, can obtain the number of samples in each prediction error band interval through statistics is respectively
ms(s=1,2 ..., S), andIt thus can be according to the number of samples in prediction error band interval
Obtain fs=ms/RK, w.Work as RK, wWhen fully big, you can by fsIt is considered as TkThe L of periodK, wLayer subregion prediction error it is discrete
Exact method is distributed.
Step S5 traverses all historical datas, you can obtains the discrete exact method distribution letter of M*W prediction error
Number, that is, error statistics table.Statistical method in order to ensure carried prediction error distribution character is effective, need to be to its statistics
As a result validity check is carried out.Assuming that in 1 day after historical statistics sample with the prediction data of period be in 1 day following simultaneously
The forecast sample data of section cross the discrete probabilistic minute of the prediction error of time using the statistical method statistics aircraft carried
Cloth, and carry out distribution inspection.
If it's the time is past the aircraft in the above process to be divided into aircraft and cross the time for entering en-route sector, boat
Pocket crosses this two class of time of entrance en-route sector, and obtains the back propagation net of this two classes time by the above process, just
It can carry out probabilistic en-route sector Traffic Demand Forecasting research.
En-route sector transport need is predicted according to statistical probability is sorted out
In the present embodiment, using a certain en-route sector as object, probability Traffic Demand Forecasting is conventional determining sexual intercourse
Logical spreading out for requirement forecasting work stretches.In the prediction error delta t for demonstrating aircraft and crossing the timeiDiscrete probability distribution after,
The error statistics law-analysing aircraft can be utilized to enter and leave the possibility distribution situation of en-route sector time, Jin Ertong
Count the following en-route sector aircraft quantity that may be present, the i.e. probability of the en-route sector transport need value in a certain amount of time
Property result.Which reflects the risk factors implied in Traffic Demand Forecasting, are the potential crowded risk problem encountered of en-route sector
Research provides precondition and foundation.
If in prediction object time section [tk, tk+ 14] in, N is sharedkFrame aircraft is (i.e. in the sections k-th 15min
Aircraft quantity) pass through the en-route sector, can obtain these aircrafts according to preceding method passes through the en-route sector boundary point
And enter the predicted time error of the sectorThe probability density of (i.e. aircraft enters the prediction error of sector time)Corresponding to probability-distribution function isAnd aircraft by the en-route sector boundary point and leaves the sector
Predicted time errorThe probability density of (i.e. aircraft enters the prediction error of sector time)Corresponding probability distribution
Function is
Due to i-th (1≤i≤Nk) frame aircraft fiPredicted time into the sector isWhereinFor
The accurate point time excessively obtained according to departure time of aircraft, flight path length and flying quality, what this time was to determine.
Therefore,The regularity of distribution withUnanimously, then in [tk, tk+ 14] interior aircraft fiProbability into the sector is：
Similarly it is found that then in [tk, tk+ 14] aircraft f in the periodiThe probability for leaving the sector is：
There is formula (1) (2) it is found that in [tk, tk+ 14] aircraft f in the periodiProbability in en-route sector is：
Then for
If following [t one dayk, tk+ 14] certain en-route sector deterministic demand prediction result is M sorties in the period, then
There may be M frame aircrafts process, then there are the probability of m frame aircrafts in sector in the period is：PM[m](0≤m
≤ M), then utilize pseudo-program representation, it is known that its value is：
In order to further verify the validity of institute's extracting method, on the basis of above-mentioned distribution inspection, then from accumulated probability
Angle carries out validity check.Here, utilizing P (i) characterization aircrafts fiIn [tk, tk+ 14] it is located in en-route sector in the period
Probability, to be fully described by the statistical regularity of error distribution.
Assuming that the probability transport need value that prediction obtains following certain day the i-th period is xi, corresponding actual traffic need to
Evaluation counts to obtain as y according to historical dataiIf the same day shares n period, related coefficient is：
Wherein,WithX is indicated respectivelyiAnd yiMean value, then haveIf correlation coefficient ρ is bigger, say
Bright statistical law is better to the simulation effect of value fluctuation in following 1 day, has preferable practical value, while the prediction error is advised
Rule is equally effective with the prediction of period to the following odd-numbered day.
The present invention is in order to make up the potential deficiency of traditional deterministic forecast method, from another angle, it is believed that differ
Surely the accuracy of requirement forecasting must be improved, can also the probabilistic size of accurate quantification requirement forecasting, provide probabilistic
Traffic Demand Forecasting is as a result, Traffic Demand Forecasting result corresponding in en-route sector under i.e. certain predicted time scale includes number
Value and its corresponding probability.Factor due to influencing aircraft operation in actual motion is filled with unpredictability, uncertainty
With dynamic, it there is no method to fully rely on mathematical model and be transformed into deterministic prediction result, and it is uncertain to quantify these
The consequence that factor influences transport need holds the randomness of entire traffic circulation environment, is gathered around for what raising was implemented thereafter
It squeezes for management strategy, is a kind of more rational selection.
Instance analysis
About data statistics, middle southern region of China AR01-AR08 en-route sectors 1 day 00 Mays in 2014:00:00 to 2014
On May 24,23 in:59:59 based on totally 25 days operation datas, in this time daily 07:00 (more idle) is to 11:59
All time datas by above-mentioned en-route sector boundary point are historical data sample in (more busy), obtain 19624 altogether
Time samples, and it is for statistical analysis according to institute's construction method.According to central-limit theorem：When the same distribution of obedience and there is number
Hope the stochastic variable with variance, when sample total is more than or equal to 30, the variable approximation Normal Distribution term[15], then may be used
Assuming that the sample space Normal Distribution, and sample result of calculation is as shown in table 1, table 2.
1 flight of table enters the predicted time error of sector AR01-AR08Mean μ and variances sigma (min)
2 flight of table leaves the predicted time error of sector AR01-AR08Mean μ and variances sigma (min)
Using quantile quantile plot (the Quantile-Quantile Figure, Q-Q in normal probability paper method of inspection
Figure) it tests to actual error sample distribution.Q-Q figures using quantile and the specified distribution of sample data quantile it
Between relation curve come inspection data whether Normal Distribution.If relation curve is straight line, data to be verified are obeyed just
It is distributed very much, it is on the contrary then disobey normal distribution.320 hypothesis testings need to be carried out altogether to above-mentioned sample, it is found that result meets just
State distributional assumption.The inspection result 15 minute sector period AR01 and AR05 is enumerated in the present embodiment, such as Fig. 8 a and Fig. 8 b
It is shown.
About interpretation of result, according to prediction error distribution character statistical method, probability en-route sector flow management system,
Predicted time error parameter can get Central-South region AR01-AR08 en-route sector typical cases and run day 07:00-11:Traffic in 59
Demand and its probability distribution.With en-route sector AR05 07:00-07:For the prediction result of 14 periods, first according to the sector
In on May 24,07 1 day to 2014 May in 2014:00-07:The maximum aircraft sortie that 14 periods occurred is M values, then may
The probability value for m framves occur is as shown in table 3.
Table 3 07:00-07:14 en-route sector AR01 transport needs and its probability distribution
As shown in table 3, wherein abscissa is indicated by predicting that en-route sector transport need obtains according to classification statistical probability
To the en-route sector in the following same period by the aircraft quantity M of appearance, what ordinate represented is the boat being likely to occur
Pocket quantity m, if the numerical value as en-route sector AR01 in table is 07:00-07:The prediction of 14 periods has M frame aircrafts to fly
, then there is the probability value of m framves in row.It can obtain similar to the en-route sector transport need of table 2 and its probability distribution table total 160
.
The mean time of flight for counting to obtain each sector in Central-South region based on sample data is 12 minutes.According to this as a result,
With 25 days 07 May in 2014:00-11:It is front and back respectively respectively to expand 12 minutes durations in 59 centered on day part, and then obtain each
Sector, day part predict corresponding transport need and its corresponding probability, and it is most to choose the maximum transport need of probability value
Final value, then the results are shown in Table 4.
4 2014 on May 25,07 of table:00-11:The probability requirement forecasting results of 59 en-route sector AR01-AR08
By above-mentioned prediction result and 25 days 07 May in 2014:00-11:The aircraft sortie of 59 day parts actually flown over
It is compared, and carries out correlation test, the results are shown in Table 5.It can be found that the more busy period (such as 09:00-11:
59) or the related coefficient of the larger sector of sector bulk flow (such as AR01-AR05) can reach 70% or more, highest or even reach
To 99%, it was demonstrated that its statistical law or relatively effective；In contrast, more idle or the smaller sector of bulk flow
Related coefficient is relatively very ideal, and it is less to be primarily due to sample size, so not as good as the former is easier to reflect changing rule,
But in general or acceptable.
5 2014 on May 25,07 of table:00-11:The correlation test of 59 each sector Traffic Demand Forecasting results
According to the demand prediction result, according to the capability value for investigating each en-route sector.By taking the sectors AR05 as an example,
Capability value be 15 framves beat/min, 07:00-11:59 day part certainty and probability Traffic Demand Forecasting result, actual traffic
Flow value as shown in figure 9, it can be found that：
Since deterministic forecast method is limited by predicted time scale, the order of accuarcy of result is in predicted time scale
It is more accurate when smaller；And probabilistic forecasting method is when sample data volume is larger, therefore peak period its result more subject to
Really.
08:45-08:59 periods, probability requirement forecasting result more accurately reflect friendship compared with certainty result
The phenomenon that logical demand is more than capacity, makes controller be more easy to predict sector crowding phenomenon；09:15-09:It is 29 periods, probability
As a result it shows flow and is less than capacity, so as to avoid false-alarm problem caused by certainty result
In general, compared with deterministic demand prediction result, the accuracy rate of probability Traffic Demand Forecasting result is 80%,
Accuracy rate 65% than certainty Traffic Demand Forecasting result greatly improves.
The present embodiment can not be suitable for more ripe probability transport need for existing flow management system data basis
The realistic problem of method crosses the historical data and prediction data of time based on existing aircraft, by analyzing aircraft mistake
Point time prediction error distribution characteristics and its influence factor, establish aircraft and cross a statistics for time prediction error distribution character
Method, and probability en-route sector flow management system is proposed on this basis, actual operating data is finally combined, is obtained
The probability distribution and changing rule of sector transport need in certain time, and the standard of the probability Traffic Demand Forecasting result of gained
True rate is greatly improved than the accuracy rate of traditional certainty Traffic Demand Forecasting result, illustrates that the method can be air traffic
Management provides the Traffic Demand Forecasting foundation of more science.
It is enlightenment with above-mentioned desirable embodiment according to the present invention, through the above description, relevant staff is complete
Various changes and amendments can be carried out without departing from the scope of the technological thought of the present invention' entirely.The technology of this invention
Property range is not limited to the contents of the specification, it is necessary to determine its technical scope according to right.
Claims (8)
1. a kind of flow management system, which is characterized in that including：
Predict error sort out statistical module, to aircraft within the corresponding period by the sector borders point time prediction error into
Row sorts out statistics；And
Probability evaluation entity, probabilistic forecasting en-route sector transport need value.
2. flow management system according to claim 1, which is characterized in that
Include by the method that the prediction error of sector borders point time sort out statistics within the corresponding period to aircraft：
The route grid model of aircraft is created, determines the influence factor of prediction error.
3. flow management system according to claim 2, which is characterized in that
The method of route grid model for creating aircraft includes：
Route grid is reduced to spatial domain, four class element of air route, en-route sector and sector borders point, while entire spatial domain being drawn
It is divided into two class of target and non-targeted spatial domain；Wherein
Target spatial domain refers to the spatial domain that the en-route sector belonged in prediction spatial dimension is formed；And
The en-route sector for being not belonging to this prediction spatial dimension constitutes non-targeted spatial domain；
It is located in prediction target time section T, if quarreling aircraft passes through a certain en-route sector, wherein the i-th frame aircraft rises
The winged time is set asBy the predicted time of sector borders point, i.e. aircraft is crossed the predicted value of time and is set asBy fan
The actual value that the real time of area's boundary point, i.e. aircraft spend the time is set asAnd aircraft is by sector borders point
Time prediction error, i.e. aircraft cross the prediction error delta t of timeiIt is defined as
The prediction error sample that aircraft is crossed to the time is divided into two subsets, i.e.,
First m days are subset I, and the m+1 days to the M days are subset II；Wherein
Subset I is used for the distribution character of statistical forecast error, and subset II is used to verify the validity of statistical law.
4. flow management system according to claim 3, which is characterized in that
Determine that the method for predicting the influence factor of error includes：
The influence factor includes：Typical case runs the in a few days busy extent of different periods and predicted time scale；And
Using dimensional probability distribution f (Tk, Pj) describe to predict the distribution character of error with the changing rule of period, wherein
TkIndicate period subregion, it is with the busy extent of the typical operation day different periods of reflection, i.e., the typical operation day of aircraft is complete
It is divided into several period subregions with a fixed step size；K indicates that error statistics period period, k=1,2 ..., K, K indicate to carry out
Error statistics it is total when hop count；And
PjIt is the characteristic quantity for describing aircraft and spending the time.
5. flow management system according to claim 4, which is characterized in that
Classification statistics is carried out to the prediction error, i.e., distribution character statistics is carried out to prediction error, method includes：
By TkPrediction error sample in period carries out double division by the size and its being uniformly distributed for number of prediction error.
6. flow management system according to claim 5, which is characterized in that
By TkPrediction error sample in period carries out double division by the size and its being uniformly distributed for number of prediction error
Method includes：
First, by TkThe prediction error sample of period carries out first time division by the size of prediction error value, i.e.,
Establish the coordinate system divided for the first time, abscissa Tk+ i and i=0,2 ..., 59, it indicates corresponding to prediction error value
Period points, ordinate is prediction error value, sets classificatory scale, is carried out etc. to the prediction error for spending the sector borders point time
Interval divides；
Setting prediction error sample size reference interval [l- δ, l+ δ], l indicate suitable number of samples, and δ is fluctuation range, from
Two lateral third side of predicted value merges to the section at equal intervals to Preliminary division；
After aforesaid operations, correspond to TkPeriod obtains total W layers of subregion, if each subregion is LK, w(w=1,2 ..., W),
And LK, wContain R in a subregionK, wA sample；And it is expanded respectively from transverse direction, regulation of longitudinal angle to historical sample number along time shaft
According to progress inducing classification.
7. flow management system according to claim 6, which is characterized in that
Further include to the method for predicting error progress distribution character statistics：
L is calculated one by oneK, w, and k=1,2 ..., K；W=1,2 ..., W layers of prediction error sample vr, and r=1,2 ...,
RK, w, to predict percentage error as abscissa, suitable error step-length is chosen as region interval width, according to obtained
vrValue focuses on sample point in corresponding separate；
To obtain the number of samples in each prediction error band interval it is respectively m through countings, wherein s=1,2 ..., S, andAnd f is obtained according to the number of samples in prediction error band intervals=ms/RK, w；
Work as RK, wWhen fully big, you can by fsIt is considered as TkThe L of periodK, wThe discrete exact method point of the prediction error of layer subregion
Cloth；
All historical datas are traversed, to obtain the discrete exact method distribution function of MW prediction error, i.e. error statistics table.
8. flow management system according to claim 7, which is characterized in that
Include according to the method that statistical probability predicts en-route sector transport need value is sorted out：
If in prediction object time section [tk, tk+ 14] in, if shared aircraft of quarreling passes through an en-route sector, current aerospace
Device passes through the sector borders point and enters the predicted time error of the sectorProbability densityCorresponding probability distribution
Function isAnd aircraft passes through the sector borders point and leaves the predicted time error of the sectorProbability it is close
DegreeCorresponding to probability-distribution function is
Since the predicted time that the i-th frame aircraft enters the en-route sector isWhereinTo be risen according to aircraft
Fly the accurate point time excessively that moment, flight path length and flying quality obtain, andWithThe regularity of distribution it is consistent, then
In [tk, tk+ 14] probability of the aircraft into the en-route sector is in：
And in [tk, tk+ 14] probability that the aircraft leaves en-route sector in the period is：
There is formula (1) (2) it is found that in [tk, tk+ 14] probability that the aircraft is located in en-route sector in the period is：
And
If following [t one dayk, tk+ 14] certain en-route sector deterministic demand prediction result is M sorties in the period, then
There may be M frame aircrafts process, then there are the probability of m frame aircrafts in sector in the period is：PM[m] (0≤m≤M),
By pseudo-program representation, its result is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810082230.5A CN108364091A (en) | 2018-01-26 | 2018-01-26 | Probability en-route sector Traffic Demand Forecasting flow management system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810082230.5A CN108364091A (en) | 2018-01-26 | 2018-01-26 | Probability en-route sector Traffic Demand Forecasting flow management system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108364091A true CN108364091A (en) | 2018-08-03 |
Family
ID=63007190
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810082230.5A Pending CN108364091A (en) | 2018-01-26 | 2018-01-26 | Probability en-route sector Traffic Demand Forecasting flow management system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108364091A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109740818A (en) * | 2019-01-10 | 2019-05-10 | 南京航空航天大学 | A kind of probability density forecasting system applied to en-route sector traffic |
| CN112862197A (en) * | 2021-02-19 | 2021-05-28 | 招商银行股份有限公司 | Intelligent network point number allocation method, device, equipment and storage medium |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105528647A (en) * | 2015-11-25 | 2016-04-27 | 南京航空航天大学 | Airport traffic demand possibility prediction method based on big data analysis |
-
2018
- 2018-01-26 CN CN201810082230.5A patent/CN108364091A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105528647A (en) * | 2015-11-25 | 2016-04-27 | 南京航空航天大学 | Airport traffic demand possibility prediction method based on big data analysis |
Non-Patent Citations (3)
| Title |
|---|
| 杨文佳等: "基于预测误差分布特性统计分析的概率性短期负荷预测", 《电力系统自动化》 * |
| 田文等: "一种航路扇区概率性交通需求预测方法", 《交通运输系统工程与信息》 * |
| 田文等: "空域扇区概率交通需求预测模型", 《西南交通大学学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109740818A (en) * | 2019-01-10 | 2019-05-10 | 南京航空航天大学 | A kind of probability density forecasting system applied to en-route sector traffic |
| CN112862197A (en) * | 2021-02-19 | 2021-05-28 | 招商银行股份有限公司 | Intelligent network point number allocation method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109615163B (en) | Air-ground joint capacity calculation method based on time sequence simulation | |
| CN110176141B (en) | Traffic cell division method and system based on POI and traffic characteristics | |
| CN106023655B (en) | Traffic congestion state monitoring method in a kind of sector null | |
| Marsigli et al. | A spatial verification method applied to the evaluation of high‐resolution ensemble forecasts | |
| CN106651088B (en) | Flight quality monitoring method based on temporal GIS | |
| CN106845768A (en) | Bus hourage model building method based on survival analysis parameter distribution | |
| Shao et al. | Heterogeneous saltation: Theory, observation and comparison | |
| WO2021082394A1 (en) | Layout-variable taxiing-out time prediction system based on big data deep learning | |
| CN105528647B (en) | Airport traffic demand possibility prediction method based on big data analysis | |
| CN108364091A (en) | Probability en-route sector Traffic Demand Forecasting flow management system | |
| CN113362604B (en) | Controller workload assessment method based on sector dynamic traffic characteristics | |
| Du et al. | Traffic events oriented dynamic traffic assignment model for expressway network: a network flow approach | |
| CN109460871A (en) | Airport passenger amount prediction technique based on the identification of typical day | |
| CN109740818A (en) | A kind of probability density forecasting system applied to en-route sector traffic | |
| Gecchele et al. | Freeway rear-end collision risk estimation with extreme value theory approach. A case study | |
| CN108172027A (en) | Probability en-route sector Forecast of Traffic Demand | |
| CN107316503A (en) | A kind of congestion focus space domain sector recognition methods based on multistage matter-element entropy weight | |
| CN113284369B (en) | Prediction method for actually measured airway data based on ADS-B | |
| CN107092988A (en) | Method for predicting station-parking time of bus on special lane | |
| CN107316096A (en) | A kind of track traffic one-ticket pass passenger amount of entering the station Forecasting Methodology | |
| Gazzarri et al. | Estimation of gap acceptance parameters for HCM 2010 roundabout capacity model applications | |
| CN103778782A (en) | Traffic state partitioning method based on semi-supervised machine learning | |
| CN110263962A (en) | Termination environment is marched into the arena Tendency Prediction method | |
| CN106844948B (en) | Landslide prediction method and device | |
| Dong et al. | An identification model of urban critical links with macroscopic fundamental diagram theory |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180803 |