Summary of the invention
In order to solve the problems, such as not considering in existing traffic analysis that other modes of transportation influence.The invention proposes one kind
The association analysis method of large aerospace hinge operation and urban traffic blocking makees the airport obtained by queueing theory delay situation
It is introduced into for a principal element in traffic flow forecasting, so that evaluating airport is delayed influence to periphery road traffic.
Specific step is as follows:
Step 1: being directed to the flight schedules of certain time period, the number that should be reached in the period is counted, record should
In period actual conditions get off the plane take off output number and aircraft landing input number.
The division of period are as follows: be divided into 48 periods for 24 hours one day, i.e., per half an hour is a period.
Record the method for taking off output number are as follows: within each period, record the flight number that each moment takes off
Mesh calculates its transportable total number of persons m, and records the time interval t for the flight that it takes off with subsequent time, records t/m.
Mean value and variance are calculated to t/m all in each period.
Step 2: establishing queue theory model M/E according to the statistical result of step 1k/ 1 model calculated in the period
Troop length LqWaiting time W per capitaq。
Formula are as follows:
WhereinK is EkOrder, λ is demand factor, and μ is service rate.
Step 3: utilizing troop length LqWaiting time W per capitaq, establish measure airport in delay situation congestion sense refer to
Number α and Anxiety index β;
Congestion sense index: α=φ1ln(Lq-C)+φ2
Anxiety index: β=φ3ln(Wq-K)+φ4
Whereinφ1, φ2, φ3And φ4For normalization coefficient, C is congestion sense threshold value;K is coke
Consider sense threshold value;
Step 4: calculating in the road network in the period in airport radius M kilometer range, the free flow velocity of every a road section
Degree;
Free stream velocity is the travel speed of vehicle in the completely unimpeded situation of road.
Step 5: calculating the hourage index of every a road section in the period using the free stream velocity in each section
TTI;
Hourage index is that vehicle is travelled with present speed and free stream velocity traveling is equally apart from the ratio of time used
Value.
Step 6: being arranged from small to large each hourage index TTI, and truncation position is chosen, deletes to be greater than and cut
The unreasonable data of disconnected position retain from the zero TTI data to truncation position and do subsequent processing.
Position is truncated to choose from by 99% quantile of hourage index.
Step 7: according to each hourage index in the entire road network of step 5 calculating, when calculating the travelling of the period
Between Mean value of index, as evaluation road network jam situation index.
Step 8: being distributed according to the discrete probability density of each hourage index retained in step 6, when calculating travelling
Between index entropy H (X), as evaluation road network complexity index.
Formula is as follows:
Wherein P (TTIi) be TTI discrete probability density distribution in the corresponding probability value of i-th kind of TTI value, N be from zero to
The TTI of truncation position divides sum at equal intervals.
Step 9: the characteristic quantity for comprehensively considering airport and road index carries out time prediction point using LSTM neural network
Analysis, predicts the mean value of the hourage index of subsequent time period and the entropy of hourage index.
Firstly, the input feature vector of neural network includes: the characteristic quantity and road index by the airdrome beacon of articulamentum
Characteristic quantity;
Specifically: value and congestion sense index α after the number that should be reached in each period is normalized and
Anxiety index β, these three values are detained the index of situation as airport, pass through a full feature of the articulamentum as airdrome beacon
Amount;
Value after the mean value of the hourage index of each period is normalized, the trip for each period
The entropy H (X) of row time index be normalized after value, the value that the period of division is normalized and the same day
Weather condition quantized value, by this four value be used as congestion in road situation index, by a full articulamentum as road index
Characteristic quantity.
The mean value of the hourage index of each period is normalized specifically: travel in each period history
The maximum value of time index mean value is historical data maximum value, and minimum value is historical data minimum value, and place is normalized with this
Reason.
The entropy H (X) of the hourage index of each period is normalized specifically: hourage Exponential Entropy
Maximum entropy HmaxIt is obtained by following formula, Hmax=logN, minimum value are the minimum value obtained in Historical Monitoring data;With
Hourage Exponential Entropy is normalized in this.
Weather condition quantization to the same day specifically: weather index is quantified as 7 etc., corresponding numerical value is
The period of division is normalized and is referred to: the variation of place period is
Then, it will be inputted together in LSTM layers by the two of full articulamentum groups of indexs;
Finally, LSTM layers of output is trained by three full articulamentums, the hourage of subsequent time period is obtained
The mean value of index and the entropy of hourage index.
Step 10: the hourage Mean value of index and hourage Exponential Entropy of the subsequent time period using prediction, calculate
Influence of the delay on current slot airport to peripheral path;
Firstly, calculating airport is detained index k and road complexity effect index l that road pavement congestion in road influences;
Congestion in road Intrusion Index calculation formula is as follows:
Road complexity effect formula of index is as follows:
Wherein, m (t+1) is the subsequent time period hourage Mean value of index of prediction, when m (t) is that current slot is travelled
Between Mean value of index.H (t+1) is subsequent time period hourage Exponential Entropy, and H (t) is current slot hourage Exponential Entropy.
Then, k and l is normalized, influence of the airport delay to peripheral path congestion is evaluated as quantizating index.
Choosing the two history maximum value is maximum value, and history minimum value is minimum value, is normalized:
Institute's value illustrates that influence of the airport delay to peripheral path is bigger closer to 1.
Compared with prior art, the beneficial effects of the present invention are:
(1) association analysis method of a kind of large aerospace hinge operation and urban traffic blocking, building to airport periphery
The prediction of short-term traffic congestion situation, not only the traffic congestion index before use, it is also contemplated that the congestion inside airport
Situation.
(2) a kind of association analysis method of large aerospace hinge operation and urban traffic blocking, is inside a kind of pair of airport
The method evaluated of delay situation, and provide the evaluation number of jam situation in airport, i.e. congestion sense index and anxiety
Feel index.By the two indexes, the jam situation in airport is quantified, is introduced into the influence to airport peripheral path congestion.
(3) a kind of association analysis method of large aerospace hinge operation and urban traffic blocking, is detained for evaluating airport
Peripheral path is had an impact, simple possible.
Specific embodiment
Below in conjunction with attached drawing, the present invention is described in further detail.
The invention discloses a kind of large aerospace hinges to run the association analysis method with urban traffic blocking, including as follows
Step: 1) calculating the number of arriving at the airport using basic Airlines Timetable, takes off number, aircraft landing number, building row
Team theory model.2) team leader obtained by queue theory model measures the degree of crowding on airport with the waiting time.3) airport is detected
The free flow speed of peripheral path calculates the hourage index of every road.4) hourage index is arranged, is chosen
Position is truncated, and calculates hourage Mean value of index.5) hourage index discrete probability density is calculated, hourage is asked to refer to
Number entropy.6) required each index is normalized.7) existing airport is comprehensively considered using LSTM neural network
Time prediction analysis is carried out with the feature of road, after the completion of training, is delayed situation when flight occurs, feature needed for calculating network refers to
Number, predicts the hourage mean value and hourage entropy of subsequent time;8) prolong on the statement airport for constructing evaluation criterion to quantify
The accidentally influence to congestion in road situation.
As shown in Figure 1, the specific steps are as follows:
Step 1: being directed to the flight schedules of certain time period, the number that should be reached in the period is counted, record should
In period actual conditions get off the plane take off output number and aircraft landing input number.
The division of period are as follows: delay situation and road the delay situation on airport were with the variation for the period in one day;In order to protect
Real-time is demonstrate,proved, was divided into 48 periods for 24 hours one day, i.e., per half an hour is a period, is denoted as [h1,…,h47,
h48].It is for statistical analysis.
The number that should be reached in certain time period, which is calculated, by flight schedules is denoted as λ (t);
Record the method for taking off output number are as follows: within each period, record the flight number that each moment takes off
Mesh calculates its transportable total number of persons m, and records the time interval t for the flight that it takes off with subsequent time, records t/m.
Mean value and variance, i.e. E (E are calculated to t/m all in each periodKAnd Var (E (t))k(t))。
The number of aircraft landing input, record change the total number of persons to land in the period and are denoted as g (t).
Step 2: establishing queue theory model M/E according to the statistical result of step 1k/ 1 model calculated in the period
Troop length LqWaiting time W per capitaq。
Data used in queue theory model are established to be safety check number and took off number;
Use M/EK/ 1 queue theory model.Then in hiIn section, demand factor are as follows: λ (t)=x (t)/0.5 people/hour
Service rate are as follows: be distributed for k rank Erlang, distribution density is
Therefore its mean value and variance are as follows:
So having
Calculation formula are as follows:
WhereinK is EkOrder, λ is demand factor, and μ is service rate.
Step 3: utilizing troop length LqWaiting time W per capitaq, establish measure airport in delay situation congestion sense refer to
Number α and Anxiety index β;
Congestion sense index: α=φ1ln(Lq-C)+φ2
Anxiety index: β=φ3ln(Wq-K)+φ4
Whereinφ1, φ2, φ3And φ4For normalization coefficient, C is congestion sense threshold value;K is coke
Consider sense threshold value;
Step 4: calculating in the road network in the period in airport radius M kilometer range, the free flow velocity of every a road section
Degree;
Free stream velocity is the travel speed of vehicle in the completely unimpeded situation of the smaller road of the volume of traffic.
Step 5: calculating the hourage index TTI of every a road section in the period using the free stream velocity in each section;
Hourage index is that vehicle is travelled with present speed and free stream velocity traveling is equally apart from the ratio of time used
Value.
The calculation formula of hourage index are as follows:
Step 6: being arranged from small to large each hourage index TTI, and truncation position is chosen, deletes to be greater than and cut
The unreasonable data of disconnected position retain from the zero TTI data to truncation position and do subsequent processing.
Position is truncated to choose from by 99% quantile of hourage index.
Step 7: calculating the mean value of the period according to each hourage index in the entire road network of step 5 calculating, make
For the index for evaluating road network jam situation.
Step 8: being distributed according to the discrete probability density of each hourage index retained in step 6, when calculating travelling
Between index entropy H (X), as evaluation road network complexity index.
Formula is as follows:
Wherein P (TTIi) be TTI discrete probability density distribution in the corresponding probability value of i-th kind of TTI value, N be from zero to
The TTI of truncation position divides sum at equal intervals.
Step 9: the characteristic quantity for comprehensively considering airport and road index carries out time prediction point using LSTM neural network
Analysis, predicts the mean value of the hourage index of subsequent time period and the entropy of hourage index.
Firstly, design LSTM neural network carries out time series analysis training to the traffic conditions on airport periphery;Nerve net
The model of network is as shown in Figure 2: input feature vector includes: the feature of the characteristic quantity and road index by the airdrome beacon of articulamentum
Amount;
Specifically:
First part is airport correlated characteristic amount: after the number that should be reached in each period is normalized
Value and congestion sense index α and Anxiety index β, these three values are detained the index of situation as airport, pass through a full connection
Characteristic quantity of the layer as airdrome beacon;
Second part is road correlated characteristic amount: the mean value of the hourage index of each period is normalized
Value afterwards, the value after the entropy H (X) of the hourage index of each period is normalized, for the time of division
The weather condition quantized value of value and the same day that section is normalized regard this four values as congestion in road situation index, passes through
One full characteristic quantity of the articulamentum as road index.
The mean value of the hourage index of each period is normalized specifically:
TTI is denoted as in each period historymean=mean (TTIi) and mean value is normalized: hourage refers to
The maximum value of number mean value is data with existing maximum value max (TTImean), the minimum value of mean value is the minimum value min of data with existing
(TTImean).Normalizing formula is
The entropy H (X) of the hourage index of each period is normalized specifically:
The maximum entropy H of hourage Exponential EntropymaxIt is obtained by following formula, the upper bound can obtain E by Jensen's inequality
(logY) < log (E (Y)), obtains:
The lower bound of hourage entropy is defined as the minimum value min (H (X)) of the hourage entropy of available data;That is minimum value
For the minimum value obtained in Historical Monitoring data;Formula is normalized to hourage Exponential Entropy with this are as follows:
Weather condition quantization to the same day specifically: weather index is quantified as 7 etc., corresponding numerical value is
The period of division is normalized and is referred to: the variation of place period is
Then, it will be inputted together by the two of full articulamentum groups of feature figureofmerits and carry out time series analysis in LSTM layers;
Finally, LSTM layers of output is trained by three full articulamentums, the hourage of subsequent time period is obtained
The mean value of index and the entropy of hourage index.
Specifically: LSTM network entitled shot and long term memory network entirely, there are memory units and something lost in each LSTM cellular
Forget unit, can be excavated in operation and with hiding Markov property.And traffic data has the characteristic of markov, fits
Conjunction is trained with LSTM network.
By the characteristic quantity of the airport of [t-n ..., t-1, t] period and road index as input, by the t+1 period
Average hourage index and hourage Exponential Entropy are as label.
Using previous airport book flight table, the data on flows training nerve of practical flight-table and airport peripheral path
Network, neural network carry out supervised learning using historical data by backpropagation, and obtaining after successive ignition operation can be quasi-
Really prediction subsequent time period network weight table, thus establish airdrome beacon characteristic quantity and road index characteristic quantity with
Connection between label.After the completion of training, real-time detector data calculates the characteristic quantity of airdrome beacon and the feature of road index
Amount, by the average hourage index and hourage Exponential Entropy that can predict subsequent time period after the operation of network;
Further relate to the influence of the delay to peripheral path on this moment airport.
Step 10: the hourage Mean value of index and hourage Exponential Entropy of the subsequent time period using prediction, calculate
Influence of the delay on current slot airport to peripheral path;
Firstly, calculating airport is detained index k and road complexity effect index l that road pavement congestion in road influences;
Congestion in road Intrusion Index calculation formula is as follows:
Road complexity effect formula of index is as follows:
Wherein, m (t+1) is the subsequent time period hourage Mean value of index of prediction, when m (t) is that current slot is travelled
Between Mean value of index.H (t+1) is subsequent time period hourage Exponential Entropy, and H (t) is current slot hourage Exponential Entropy.
Then, k and l is normalized, influence of the airport delay to peripheral path congestion is evaluated as quantizating index.
Choosing the two history maximum value is maximum value, and history minimum value is minimum value, is normalized:Institute's value illustrates that influence of the airport delay to peripheral path is bigger closer to 1.
Above embodiments are only the embodiment of the present invention, are not used in the limitation present invention, protection scope of the present invention is by right
Claim limits.Those skilled in the art can make various modifications to the present invention within the spirit and scope of the present invention
Or equivalent replacement, this modification or equivalent replacement also should be regarded as being within the scope of the present invention.