CN109460871A - Airport passenger amount prediction technique based on the identification of typical day - Google Patents
Airport passenger amount prediction technique based on the identification of typical day Download PDFInfo
- Publication number
- CN109460871A CN109460871A CN201811327149.5A CN201811327149A CN109460871A CN 109460871 A CN109460871 A CN 109460871A CN 201811327149 A CN201811327149 A CN 201811327149A CN 109460871 A CN109460871 A CN 109460871A
- Authority
- CN
- China
- Prior art keywords
- day
- history
- history day
- period
- passenger
- 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
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000013139 quantization Methods 0.000 claims abstract description 5
- 230000002123 temporal effect Effects 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 9
- 239000000284 extract Substances 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000013480 data collection Methods 0.000 description 3
- 230000010006 flight Effects 0.000 description 3
- 241001269238 Data Species 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013439 planning Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 239000000843 powder Substances 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 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
- 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
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Tourism & Hospitality (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of airport passenger amount prediction technique based on the identification of typical day, the following steps are included: acquisition air station flight history day operation data acquisition system P, for each history day operation data, history Time of Day feature vector and history day flight feature vector are extracted；Quantization characteristic data set similarity, obtains the optimal weights factor；Typical day identification model is established, passenger measures at times for prediction.Airport passenger amount prediction technique provided by the invention based on the identification of typical day has the advantage that the present invention realizes accurately and effectively predicting at times for airport passenger amount, to effectively improve Airport Operation pressure pre-alerting ability.
Description
Technical field
The invention belongs to passenger flow electric powder predictions, and in particular to a kind of airport passenger amount based on the identification of typical day is pre-
Survey method.
Background technique
In recent years, with the rapid growth of China's civil aviation passenger freight volume, as air net hub node airport by
To severe challenge.Contradiction between ever-increasing passenger's amount and airport limited resources is increasingly prominent.Especially leave the port in passenger
Phenomena such as peak period, sales counter resource saturation, passenger are detained frequent occurrence in airport building, the strong influence travelling of passenger
Experience causes the economic loss of airline, and brings negative social influence for airport and Civil Aviation Industry.How airport is realized
Fine-grained management maximally utilizes existing terminal Support Resource, has become the hot issue of Airport Operation.
For airport passenger amount forecasting problem, existing research, which is broadly divided into, both at home and abroad is modeled and is based on based on passenger's behavior
Airport Operation state models two class methods.Passenger's behavior modeling refers to: carrying out Multi simulation running to single passenger's behavior model, generates
Analogue data, it is for statistical analysis to analogue data later.Common method is Computer Simulation and queueing theory, is typically employed in
Predict time, luggage delivery time, boarding waiting time etc. of the passenger spent by queuing process.Airport Operation state modeling master
If the assessment of prediction and airport whole volume for entire airport whole year handling capacity, it is therefore an objective to reflect machine from strategic level
Capacity and development scale, main method include: time series method, trend extrapolation, Econometric method, grey method,
Neural network and forecasting by regression analysis etc..
Above-mentioned two classes method can meet to a certain degree practical application request, however, the distribution of terminal Support Resource needs
The passenger of tactical level is wanted to measure Accurate Prediction, i.e., passenger's amount of seven days in advance (or next day) each periods.It is specific for this
Demand, first kind method need a large amount of repetitions to emulate single passenger's behavior, cause the superposition of slight error accumulative, so that prediction knot
Fruit deviate it is practical, and in order to establish respectively it is various under the conditions of passenger's behavior model (reflect the things such as special weather, festivals or holidays
Part), need a large amount of historical datas and priori knowledge；Although second class method is capable of providing the trend statistics of Long time scale, but
It can not reflect the influence of the dynamic factors such as flights distribution.
Therefore, above two method is unable to satisfy the demand predicted passenger's amount accurately and fast.
Summary of the invention
In view of the defects existing in the prior art, the present invention provides a kind of airport passenger amount prediction side based on the identification of typical day
Method can effectively solve the above problems.
The technical solution adopted by the invention is as follows:
The present invention provides a kind of airport passenger amount prediction technique based on the identification of typical day, comprising the following steps:
Step 1, air station flight history day operation data acquisition system P is acquired, by the air station flight history day operation data set
It closes P and is divided into history day sample data sets P1 and history day test data；
Step 2, the history day operation data acquisition system P is made of multiple history day operation data；It described is gone through for each
History day operation data extract history Time of Day feature vector and history day flight feature vector；
The history Time of Day feature vector indicates are as follows: < DATEi,fea_month,fea_day,fea_week,fea_
Holiday >
Wherein, DATEiThe date is represented as i；Fea_month represents the date as the month information of i；Fea_day represents the date as i
Day information；Fea_week represents the date as the week information of i；Fea_holiday represents the date as the holiday information of i；
The history day flight feature vector indicates are as follows:
Wherein: DATEiThe date is represented as i；The 1st period that the date is i is respectively represented ..., at n-th
The plan flight takeoff amount of section；N represents the date as the period total quantity of i；The date is represented as the whole day flight takeoff amount of i
Sum；
The history day flight feature vector is normalized；It is as follows to normalize formula:
Wherein: Di,jFor the amount of the taking off trend for j-th of period that the date after normalization is i, DAi,jFor the jth that the date is i
The plan flight takeoff amount of a period；
Step 3, quantization characteristic data set similarity obtains the optimal weights factor；This step specifically includes:
Step 301, temporal characteristics weight factor ω is assigned1Initial value, and assign flight feature weight factor ω2Initially
Value；
Step 302, successively calculate in history day test data and history day sample data sets P1 each sample history day
The overall similarity of notebook data selects the highest history day sample data of overall similarity, it is assumed that overall similarity is highest to be gone through
History day corresponding to history day sample data is history day r3；History day corresponding to history day test data is history day r2；Then
History day r3With history day r2Typical day each other:
Where it is assumed that history day corresponding to any one sample data is history day in history day sample data sets P1
History day r1, then history day r is calculated using following methods1With history day r2Overall similarity
Step 3021: calculating history day r1With history day r2Temporal characteristics similarity
Wherein, M1For history day r1With history day r2Between different temporal characteristics quantity, M2For history day r1With go through
History day r2Between identical temporal characteristics quantity；WhenWhen being 0, history day r is indicated1With history day r2Temporal characteristics height
It is similar；WhenWhen being 1, history day r is indicated1With history day r2Temporal characteristics it is completely dissimilar；
Step 3022: calculating not normalized history day r1With history day r2Flights distribution characteristic similarity
Wherein:Represent history day r1J-th of period the amount of taking off trend；Represent history day r2J-th when
The amount of the taking off trend of section；
Step 3023:Middle selection maximum value is denoted as max (Δ D)；
Step 3024: using following formula pairIt is normalized, obtains
Wherein, whenWhen being 0, history day r is indicated1With history day r2Flights distribution feature height it is similar, that is, leave the port
Flight planning is similar；WhenWhen being 1, history day r is indicated1With history day r2Flights distribution feature it is completely dissimilar；
Step 3025: calculating history day r1With history day r2Overall similarity
Wherein, whenWhen being 0, history day r is indicated1With history day r2It is highly similar, that is, think that the two history days are mutual
For typical day；WhenWhen being 1, history day r is indicated1With history day r2It is completely dissimilar；
Step 303, total deviation DIF is calculated；
History day r is being calculated2Typical day be history day r3Afterwards, history day r2For the day to be predicted of current iteration, with
Its typical day, that is, history day r3History day r is used as in the practical passenger amount of each period2Benchmark is measured in the prediction passenger of corresponding period
Value, with history day r3History day r is used as in the practical passenger's amount of the maximum of each period2Prediction on dividing value, with history day r3Each
The practical passenger's amount of the minimum of a period is used as history day r2Prediction floor value；
1) day, that is, history day r to be predicted is calculated using following formula2Per period practical passenger's amount measures a reference value with prediction passenger
Average deviation DIF1:
Wherein: n is history day r2Period total quantity, xjIt is day to be predicted i.e. history day r2In jth period practical passenger
Amount, αjIt is history day r2A reference value is measured in the prediction passenger of jth period；
2) day, that is, history day r to be predicted is calculated using following formula2Per period practical passenger's amount is beyond the average inclined of predicted boundary
Poor DIF2:
Wherein: βjIt is history day r2In the predicted boundary value of jth period, that is, work as xjGreater than dividing value in the prediction of corresponding period
When, βjIt indicates the prediction upper bound, and works as xjWhen less than the prediction lower bound for corresponding to the period, βjIndicate prediction lower bound；
Envelope refers to the envelope surrounded by predicting upper dividing value and prediction floor value；
3) day, that is, history day r to be predicted is calculated using following formula2The total deviation DIF of practical passenger's amount and prediction passenger's amount:
DIF=DIF1+DIF2；
Step 3.4, continuous adjustment time feature weight factor ω1With flight feature weight factor ω2Assignment, by step
The method of 3.2- step 3.4 calculates total deviation DIF, and selection is so that the smallest weight factor of total deviation DIF combines, as optimal power
Repeated factor；
Step 4, typical day identification model is established, passenger measures at times for prediction, comprising the following steps:
Step 4.1, using the following one day as day to be predicted, day r to be predicted will be denoted as day to be predicted4；
Step 4.2, using all history day operation data as historical sample data, day r to be predicted is extracted4Time it is special
Vector sum flight feature vector is levied, matching primitives is carried out with historical sample data, respectively obtains day r to be predicted4With each history
The overall similarity of sample data；Then overall similarity is ranked up, the highest preceding m sample of overall similarity is selected to make
Gather for similar typical day；Using the highest sample of overall similarity as typical day；
Step 4.2, the practical check-in passenger of the different periods of typical day is measured and is used as day r to be predicted4Corresponding to the pre- of period
Survey a reference value；
The practical passenger of each period in similar typical day set is measured into minimum value as day r to be predicted4When to corresponding
The predicted value upper bound of section；The practical passenger of each period in similar typical day set is measured into maximum value as day r to be predicted4?
The predicted value lower bound of corresponding period；
Thus day r to be predicted is obtained4Passenger in each period measures under prediction a reference value, the predicted value upper bound and predicted value
Boundary.
Preferably, in step 3.5, using the continuous adjustment time feature weight factor ω of following methods1It is weighed with flight feature
Repeated factor ω2Assignment:
Weight factor is enabled to meet ω1+ω2=1, by temporal characteristics weight factor ω1Assignment is gradually increased since 0.01,
Until 0.99；By flight feature weight factor ω2Assignment is gradually reduced since 0.99, until 0.01；When by this rule traversal
Between feature weight factor ω1With flight feature weight factor ω2。
Airport passenger amount prediction technique provided by the invention based on the identification of typical day has the advantage that
The present invention realizes accurately and effectively predicting at times for airport passenger amount, to effectively improve Airport Operation pressure
Pre-alerting ability.
Detailed description of the invention
Fig. 1 is the flow diagram of the airport passenger amount prediction technique provided by the invention based on the identification of typical day.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without creative efforts, shall fall within the protection scope of the present invention.
Traditional passenger measures own limitations present in prediction algorithm and shows the following aspects:
For in single passenger's modeling method, be easy to causeing the superposition of slight error accumulative, and need a large amount of historical datas
And priori knowledge.
For in whole passenger colony modeling methods, it is difficult to reflect the influence of the dynamic factors such as flights distribution.
For the deficiency for solving the above method, prediction side is measured based on the passenger for running typical day identification the invention proposes a kind of
Method: firstly, extracting runing time feature (month, week, festivals or holidays etc.) and (the unit interval plan of flights distribution feature
Leave the port flight amount) on the basis of, manhatton distance is defined respectively and Jaccard distance calculates characteristic similarity, and then identifies history
Run day typical day set similar with day feature to be predicted；Then, select the highest typical day passenger amount of similarity as pre-
A reference value is surveyed, and analyzes the fluctuation of day part passenger amount in typical day set, provides the bound of predicted value.Therefore, this hair
Bright is designed to provide a kind of airport passenger amount prediction technique based on typical day recognizer, considers flight feature and time
Feature measures the following passenger and carries out short interval prediction, improves Airport Operation pressure pre-alerting ability.
Airport passenger amount prediction technique provided by the invention based on the identification of typical day, main thought include:
1) air station flight operation data is acquired, characteristic vector data collection is extracted；
Wherein, the acquisition air station flight operation data, extracting characteristic vector data set method includes: flight feature vector
Extracting method, temporal characteristics vector extracting method.
2) quantization characteristic data set similarity, obtains weight factor；
It is calculated according to characteristic data set similarity relationship, obtaining weight factor method includes: that flight vector similarity calculates
Method, time arrow similarity calculating method, overall similarity calculation method, error evaluation method between any operation day.
3) typical day identification model is established, passenger measures at times for prediction.
Typical day identification model is established, passenger's amount method includes: typical day recognition methods, a reference value prediction at times for prediction
Calculation method, Volatility forecasting calculation method.
Specifically, the present invention provides a kind of airport passenger amount prediction technique based on the identification of typical day, in detail with reference to Fig. 1
The following steps are included:
Step 1, air station flight history day operation data acquisition system P is acquired, by the air station flight history day operation data set
It closes P and is divided into history day sample data sets P1 and history day test data；
Step 2, the history day operation data acquisition system P is made of multiple history day operation data；Know typical day
Before not, characteristic vector pickup first is carried out to airport historical data.This step extracts typical day by acquisition airport historical data
Feature needed for identification model.
For each history day operation data, extract history Time of Day feature vector and history day flight feature to
Amount；
The history Time of Day feature vector indicates are as follows: < DATEi,fea_month,fea_day,fea_week,fea_
Holiday >
Wherein, DATEiThe date is represented as i；Fea_month represents the date as the month information of i, for example, annual January is denoted as 1,
2 months were denoted as 2, and so on；；Fea_day represents the date as the day information of i, for example, monthly No. 1 be denoted as 1, monthly No. 2 are denoted as
2, and so on；；Fea_week represents the date as the week information of i, for example, Sunday is denoted as 1, Monday is denoted as 2, Saturday
7 are denoted as, is recycled with this；；Fea_holiday represents the date as the holiday information of i, for example, festivals or holidays are denoted as 1, non-festivals or holidays
It is denoted as 0；
The history day flight feature vector indicates are as follows:
Wherein: DATEiThe date is represented as i；The 1st period that the date is i is respectively represented ..., at n-th
The plan flight takeoff amount of section；N represents the date as the period total quantity of i, can be using every 1 hour as one in practical application
Therefore period shares 24 periods on the 1st, n 24, it is of course also possible to use other Time segments division methods, the present invention to this simultaneously
It does not limit；The date is represented as the whole day flight takeoff amount sum of i；
To eliminate influence of total flight amount to flight amount trend, the history day flight feature vector is normalized；
It is as follows to normalize formula:
Wherein: Di,jFor the amount of the taking off trend for j-th of period that the date after normalization is i, DAi,jFor the jth that the date is i
The plan flight takeoff amount of a period；
Step 3, quantization characteristic data set similarity obtains the optimal weights factor；
After respectively obtaining airport time and flight history feature vector data collection, certain any day has two groups of spies simultaneously
Vector is levied, needs to merge to indicate certain day general characteristic the two, therefore one group of weight factor need to be calculated to indicate two groups of spies
Levy weight shared by vector.In order to calculate the optimal weights factor, need to set evaluation criteria.Specifically can by traversal weight because
Sub- value, while predicted value and actual value Accumulated deviation are calculated to every group of weight factor, the weight factor of minimum deflection will be generated
Group is used as the optimal weights factor.
This step specifically includes:
Step 301, temporal characteristics weight factor ω is assigned1Initial value, and assign flight feature weight factor ω2Initially
Value；
Step 302, successively calculate in history day test data and history day sample data sets P1 each sample history day
The overall similarity of notebook data selects the highest history day sample data of overall similarity, it is assumed that overall similarity is highest to be gone through
History day corresponding to history day sample data is history day r3；History day corresponding to history day test data is history day r2；Then
History day r3With history day r2Typical day each other:
Where it is assumed that history day corresponding to any one sample data is history day in history day sample data sets P1
History day r1, then history day r is calculated using following methods1With history day r2Overall similarity
Step 3021: calculating history day r1With history day r2Temporal characteristics similarityThe definition of temporal characteristics similarity
For Jaccard distance:
Wherein, M1For history day r1With history day r2Between different temporal characteristics quantity, M2For history day r1With go through
History day r2Between identical temporal characteristics quantity；WhenWhen being 0, history day r is indicated1With history day r2Temporal characteristics height
It is similar；WhenWhen being 1, history day r is indicated1With history day r2Temporal characteristics it is completely dissimilar；
For example, history day r1Temporal characteristics vector in include month information, day information, week information and holiday information；
History day r2Temporal characteristics vector in include month information, day information, week information and holiday information；Assuming that history day r1With
History day r2Month information it is identical, day information, week information and holiday information be all different, then M1It is 3, M2It is 1.
Step 3022: calculating not normalized history day r1With history day r2Flights distribution characteristic similarityBoat
Class's distribution characteristics similarity is defined as manhatton distance:
Wherein:Represent history day r1J-th of period the amount of taking off trend；Represent history day r2J-th when
The amount of the taking off trend of section；
Step 3023:Middle selection maximum value is denoted as max (Δ D)；
Step 3024: using following formula pairIt is normalized, obtains
Wherein, whenWhen being 0, history day r is indicated1With history day r2Flights distribution feature height it is similar, that is, leave the port
Flight planning is similar；WhenWhen being 1, history day r is indicated1With history day r2Flights distribution feature it is completely dissimilar；
Step 3025: calculating history day r1With history day r2Overall similarity
Wherein, whenWhen being 0, history day r is indicated1With history day r2It is highly similar, that is, think that the two history days are mutual
For typical day；WhenWhen being 1, history day r is indicated1With history day r2It is completely dissimilar；
Step 303, total deviation DIF is calculated；
History day r is being calculated2Typical day be history day r3Afterwards, history day r2For the day to be predicted of current iteration, with
Its typical day, that is, history day r3History day r is used as in the practical passenger amount of each period2Benchmark is measured in the prediction passenger of corresponding period
Value, with history day r3History day r is used as in the practical passenger's amount of the maximum of each period2Prediction on dividing value, with history day r3Each
The practical passenger's amount of the minimum of a period is used as history day r2Prediction floor value；
1) day, that is, history day r to be predicted is calculated using following formula2Per period practical passenger's amount measures a reference value with prediction passenger
Average deviation DIF1:
Wherein: n is history day r2Period total quantity, xjIt is day to be predicted i.e. history day r2In jth period practical passenger
Amount, αjIt is history day r2A reference value is measured in the prediction passenger of jth period；
2) day, that is, history day r to be predicted is calculated using following formula2Per period practical passenger's amount is beyond the average inclined of predicted boundary
Poor DIF2:
Wherein: βjIt is history day r2In the predicted boundary value of jth period, that is, work as xjGreater than dividing value in the prediction of corresponding period
When, βjIt indicates the prediction upper bound, and works as xjWhen less than the prediction lower bound for corresponding to the period, βjIndicate prediction lower bound；
Envelope refers to the envelope surrounded by predicting upper dividing value and prediction floor value；
3) day, that is, history day r to be predicted is calculated using following formula2The total deviation DIF of practical passenger's amount and prediction passenger's amount:
DIF=DIF1+DIF2；
Step 3.4, continuous adjustment time feature weight factor ω1With flight feature weight factor ω2Assignment, by step
The method of 3.2- step 3.4 calculates total deviation DIF, and selection is so that the smallest weight factor of total deviation DIF combines, as optimal power
Repeated factor；
In this step, the continuous adjustment time feature weight factor ω of following methods is specifically used1With flight feature weight because
Sub- ω2Assignment:
Weight factor is enabled to meet ω1+ω2=1, by temporal characteristics weight factor ω1Assignment is gradually increased since 0.01,
Until 0.99；By flight feature weight factor ω2Assignment is gradually reduced since 0.99, until 0.01；When by this rule traversal
Between feature weight factor ω1With flight feature weight factor ω2。
Step 4, typical day identification model is established, passenger measures at times for prediction, comprising the following steps:
Step 4.1, using the following one day as day to be predicted, day r to be predicted will be denoted as day to be predicted4；
Step 4.2, using all history day operation data as historical sample data, day r to be predicted is extracted4Time it is special
Vector sum flight feature vector is levied, matching primitives is carried out with historical sample data, respectively obtains day r to be predicted4With each history
The overall similarity of sample data；Then overall similarity is ranked up, the highest preceding m sample of overall similarity is selected to make
For similar typical day set (being chosen using similarity threshold θ)；Using the highest sample of overall similarity as typical day；
Step 4.2, the practical check-in passenger of the different periods of typical day is measured and is used as day r to be predicted4Corresponding to the pre- of period
Survey a reference value；
The practical passenger of each period in similar typical day set is measured into minimum value as day r to be predicted4When to corresponding
The predicted value upper bound of section；The practical passenger of each period in similar typical day set is measured into maximum value as day r to be predicted4?
The predicted value lower bound of corresponding period；
Thus day r to be predicted is obtained4Passenger in each period measures under prediction a reference value, the predicted value upper bound and predicted value
Boundary.
Airport passenger amount prediction technique provided by the invention based on the identification of typical day passes through acquisition air station flight history fortune
Row data extract characteristic vector data collection, the optimal weight factor are calculated according to characteristic data set similarity relationship；The trip of foundation
Volume of passenger traffic prediction model, obtains prediction result.The present invention realizes the accurate and effective prediction of the following short-term passenger amount, helps to navigate
The stand administrative department of building Support Resource accurately judges operating pressure in advance, and carries out early warning to the special circumstances being likely to occur,
And then airport Support Resource is effectively deployed, promote operational efficiency.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
Depending on protection scope of the present invention.
Claims (2)
1. a kind of airport passenger amount prediction technique based on the identification of typical day, which comprises the following steps:
Step 1, air station flight history day operation data acquisition system P is acquired, the air station flight history day operation data acquisition system P is drawn
It is divided into history day sample data sets P1 and history day test data；
Step 2, the history day operation data acquisition system P is made of multiple history day operation data；For each history day
Operation data extracts history Time of Day feature vector and history day flight feature vector；
The history Time of Day feature vector indicates are as follows: < DATEi,fea_month,fea_day,fea_week,fea_
Holiday >
Wherein, DATEiThe date is represented as i；Fea_month represents the date as the month information of i；Fea_day represents the date as the day of i
Information；Fea_week represents the date as the week information of i；Fea_holiday represents the date as the holiday information of i；
The history day flight feature vector indicates are as follows:
Wherein: DATEiThe date is represented as i；The 1st period that the date is i is respectively represented ..., the meter of n-th of period
Draw flight takeoff amount；N represents the date as the period total quantity of i；The date is represented as the whole day flight takeoff amount sum of i；
The history day flight feature vector is normalized；It is as follows to normalize formula:
Wherein: Di,jFor the amount of the taking off trend for j-th of period that the date after normalization is i, DAi,jWhen being j-th of i for the date
The plan flight takeoff amount of section；
Step 3, quantization characteristic data set similarity obtains the optimal weights factor；This step specifically includes:
Step 301, temporal characteristics weight factor ω is assigned1Initial value, and assign flight feature weight factor ω2Initial value；
Step 302, successively calculate in history day test data and history day sample data sets P1 each sample number history day
According to overall similarity, select the highest history day sample data of overall similarity, it is assumed that overall similarity highest history day
History day corresponding to sample data is history day r3；History day corresponding to history day test data is history day r2；Then history
Day r3With history day r2Typical day each other:
Where it is assumed that history day corresponding to any one sample data is history history day in history day sample data sets P1
Day r1, then history day r is calculated using following methods1With history day r2Overall similarity
Step 3021: calculating history day r1With history day r2Temporal characteristics similarity
Wherein, M1For history day r1With history day r2Between different temporal characteristics quantity, M2For history day r1With history day
r2Between identical temporal characteristics quantity；WhenWhen being 0, history day r is indicated1With history day r2Temporal characteristics height phase
Seemingly；WhenWhen being 1, history day r is indicated1With history day r2Temporal characteristics it is completely dissimilar；
Step 3022: calculating not normalized history day r1With history day r2Flights distribution characteristic similarity
Wherein:Represent history day r1J-th of period the amount of taking off trend；Represent history day r2J-th of period rise
Fly amount trend；
Step 3023:Middle selection maximum value is denoted as max (Δ D)；
Step 3024: using following formula pairIt is normalized, obtains
Wherein, whenWhen being 0, history day r is indicated1With history day r2Flights distribution feature height it is similar, that is, leave the port flight meter
It draws similar；WhenWhen being 1, history day r is indicated1With history day r2Flights distribution feature it is completely dissimilar；
Step 3025: calculating history day r1With history day r2Overall similarity
Wherein, whenWhen being 0, history day r is indicated1With history day r2It is highly similar, that is, think that the two history days are typical each other
Day；WhenWhen being 1, history day r is indicated1With history day r2It is completely dissimilar；
Step 303, total deviation DIF is calculated；
History day r is being calculated2Typical day be history day r3Afterwards, history day r2For the day to be predicted of current iteration, with its allusion quotation
Type day, that is, history day r3History day r is used as in the practical passenger amount of each period2A reference value is measured in the prediction passenger of corresponding period,
With history day r3History day r is used as in the practical passenger's amount of the maximum of each period2Prediction on dividing value, with history day r3Each
The practical passenger's amount of the minimum of period is used as history day r2Prediction floor value；
1) day, that is, history day r to be predicted is calculated using following formula2Per period practical passenger's amount measures the average inclined of a reference value with prediction passenger
Poor DIF1:
Wherein: n is history day r2Period total quantity, xjIt is day to be predicted i.e. history day r2In jth period practical passenger's amount, αjIt is
History day r2A reference value is measured in the prediction passenger of jth period；
2) day, that is, history day r to be predicted is calculated using following formula2Per period practical passenger's amount exceeds the average deviation of predicted boundary
DIF2:
Wherein: βjIt is history day r2In the predicted boundary value of jth period, that is, work as xjGreater than in the prediction of corresponding period when dividing value, βj
It indicates the prediction upper bound, and works as xjWhen less than the prediction lower bound for corresponding to the period, βjIndicate prediction lower bound；
Envelope refers to the envelope surrounded by predicting upper dividing value and prediction floor value；
3) day, that is, history day r to be predicted is calculated using following formula2The total deviation DIF of practical passenger's amount and prediction passenger's amount:
DIF=DIF1+DIF2；
Step 3.4, continuous adjustment time feature weight factor ω1With flight feature weight factor ω2Assignment, by step 3.2-
The method of step 3.4 calculates total deviation DIF, selection so that the smallest weight factor combination of total deviation DIF, as optimal weights because
Son；
Step 4, typical day identification model is established, passenger measures at times for prediction, comprising the following steps:
Step 4.1, using the following one day as day to be predicted, day r to be predicted will be denoted as day to be predicted4；
Step 4.2, using all history day operation data as historical sample data, day r to be predicted is extracted4Temporal characteristics to
Amount and flight feature vector carry out matching primitives with historical sample data, respectively obtain day r to be predicted4With each historical sample
The overall similarity of data；Then overall similarity is ranked up, selects the highest preceding m sample of overall similarity as phase
Gather like typical day；Using the highest sample of overall similarity as typical day；
Step 4.2, the practical check-in passenger of the different periods of typical day is measured and is used as day r to be predicted4In the prediction benchmark of corresponding period
Value；
The practical passenger of each period in similar typical day set is measured into minimum value as day r to be predicted4Corresponding to the pre- of period
The measured value upper bound；The practical passenger of each period in similar typical day set is measured into maximum value as day r to be predicted4When to corresponding
The predicted value lower bound of section；
Thus day r to be predicted is obtained4Prediction a reference value, the predicted value upper bound and predicted value lower bound are measured in the passenger of each period.
2. the airport passenger amount prediction technique according to claim 1 based on the identification of typical day, which is characterized in that step
In 3.5, using the continuous adjustment time feature weight factor ω of following methods1With flight feature weight factor ω2Assignment:
Weight factor is enabled to meet ω1+ω2=1, by temporal characteristics weight factor ω1Assignment is gradually increased since 0.01, until
0.99；By flight feature weight factor ω2Assignment is gradually reduced since 0.99, until 0.01；It is special by this regular traversal time
Levy weight factor ω1With flight feature weight factor ω2。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811327149.5A CN109460871A (en) | 2018-11-08 | 2018-11-08 | Airport passenger amount prediction technique based on the identification of typical day |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811327149.5A CN109460871A (en) | 2018-11-08 | 2018-11-08 | Airport passenger amount prediction technique based on the identification of typical day |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109460871A true CN109460871A (en) | 2019-03-12 |
Family
ID=65609804
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811327149.5A Pending CN109460871A (en) | 2018-11-08 | 2018-11-08 | Airport passenger amount prediction technique based on the identification of typical day |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109460871A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110147749A (en) * | 2019-05-10 | 2019-08-20 | 中国民航大学 | A kind of civil aviaton's missing baggage auxiliary method for retrieving |
| CN110288177A (en) * | 2019-05-09 | 2019-09-27 | 南京航空航天大学 | A method of capacity is announced using history data envelope assessment airport |
| CN112216151A (en) * | 2020-10-15 | 2021-01-12 | 北京航空航天大学 | Air traffic four-dimensional track regulation and control decision method |
| CN114239325A (en) * | 2022-02-28 | 2022-03-25 | 民航成都物流技术有限公司 | Airport check-in consignment counter configuration planning method, device, equipment and storage medium |
| CN115545996A (en) * | 2022-12-02 | 2022-12-30 | 成都智元汇信息技术股份有限公司 | Subway abnormal history passenger flow identification method and device based on similarity matrix |
-
2018
- 2018-11-08 CN CN201811327149.5A patent/CN109460871A/en active Pending
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110288177A (en) * | 2019-05-09 | 2019-09-27 | 南京航空航天大学 | A method of capacity is announced using history data envelope assessment airport |
| CN110288177B (en) * | 2019-05-09 | 2023-03-28 | 南京航空航天大学 | Method for evaluating airport publishing capacity by adopting historical operating data envelope curve |
| CN110147749A (en) * | 2019-05-10 | 2019-08-20 | 中国民航大学 | A kind of civil aviaton's missing baggage auxiliary method for retrieving |
| CN110147749B (en) * | 2019-05-10 | 2023-05-05 | 中国民航大学 | Auxiliary retrieving method for lost luggage in civil aviation |
| CN112216151A (en) * | 2020-10-15 | 2021-01-12 | 北京航空航天大学 | Air traffic four-dimensional track regulation and control decision method |
| CN112216151B (en) * | 2020-10-15 | 2021-12-28 | 北京航空航天大学 | Air traffic four-dimensional track regulation and control decision method |
| US11922817B2 (en) | 2020-10-15 | 2024-03-05 | Beihang University | 4-dimensional trajectory regulatory decision-making method for air traffic |
| CN114239325A (en) * | 2022-02-28 | 2022-03-25 | 民航成都物流技术有限公司 | Airport check-in consignment counter configuration planning method, device, equipment and storage medium |
| CN115545996A (en) * | 2022-12-02 | 2022-12-30 | 成都智元汇信息技术股份有限公司 | Subway abnormal history passenger flow identification method and device based on similarity matrix |
| CN115545996B (en) * | 2022-12-02 | 2023-03-10 | 成都智元汇信息技术股份有限公司 | Similarity matrix-based subway abnormal historical passenger flow identification method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109460871A (en) | Airport passenger amount prediction technique based on the identification of typical day | |
| Jacquillat et al. | A roadmap toward airport demand and capacity management | |
| Abed et al. | An econometric analysis of international air travel demand in Saudi Arabia | |
| US6580998B2 (en) | System and method for estimating aircraft flight delay | |
| US20180032964A1 (en) | Transportation system and method for allocating frequencies of transit services therein | |
| Kasturi et al. | Airline Route profitability analysis and Optimization using BIG DATA analyticson aviation data sets under heuristic techniques | |
| Wang et al. | Frequency and aircraft size dynamics in a concentrated growth market: The case of the Chinese domestic market | |
| Anwar et al. | Changinow: A mobile application for efficient taxi allocation at airports | |
| CN110363333A (en) | The prediction technique of air transit ability under the influence of a kind of weather based on progressive gradient regression tree | |
| US10789558B2 (en) | Non-linear systems and methods for destination selection | |
| CN111192090A (en) | Seat allocation method and device for flight, storage medium and electronic equipment | |
| CN109214612A (en) | One kind being based on XGBOOST airport traffic spatial and temporal distributions prediction technique | |
| CN110335506A (en) | A kind of termination environment airspace utilization rate calculation method in short-term | |
| Andreatta et al. | A set of approximate and compatible models for airport strategic planning on airside and on landside | |
| Chen et al. | Decentralised cooperative cruising of autonomous ride-sourcing fleets | |
| Al-Sultan | Simulation and Optimization for Modeling the Passengers Check-in System at Airport Terminal | |
| Marcos et al. | A machine learning approach to air traffic route choice modelling | |
| Liang et al. | An improved adaptive parallel genetic algorithm for the airport gate assignment problem | |
| CN110852792A (en) | Big data analysis-based airline value evaluation method and related products | |
| CN113191533A (en) | Warehouse employment prediction method, device, equipment and storage medium | |
| Rodič et al. | Airport ground crew scheduling using heuristics and simulation | |
| Viggiano et al. | Journey-based characterization of multi-modal public transportation networks | |
| Novrisal et al. | Simulation of departure terminal in Soekarno-Hatta International airport | |
| CN114239325B (en) | Airport check-in consignment counter configuration planning method, device, equipment and storage medium | |
| Yang | Entry effect of low-cost carriers on airport-pairs demand model using market concentration approach |
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: 20190312 |