CN108805347B - Passenger flow pool-based method for estimating passenger flow of associated area outside subway station - Google Patents
Passenger flow pool-based method for estimating passenger flow of associated area outside subway station Download PDFInfo
- Publication number
- CN108805347B CN108805347B CN201810567313.3A CN201810567313A CN108805347B CN 108805347 B CN108805347 B CN 108805347B CN 201810567313 A CN201810567313 A CN 201810567313A CN 108805347 B CN108805347 B CN 108805347B
- Authority
- CN
- China
- Prior art keywords
- passenger flow
- passenger
- pool
- model
- area
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 230000008569 process Effects 0.000 claims abstract description 26
- 238000005070 sampling Methods 0.000 claims abstract description 7
- 238000004364 calculation method Methods 0.000 claims description 21
- 238000007689 inspection Methods 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 20
- 238000007781 pre-processing Methods 0.000 claims description 7
- 230000014509 gene expression Effects 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 3
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 238000012804 iterative process Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 2
- 238000005311 autocorrelation function Methods 0.000 description 19
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 18
- 238000011160 research Methods 0.000 description 13
- 238000002955 isolation Methods 0.000 description 9
- 238000012545 processing Methods 0.000 description 9
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000009826 distribution Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- 238000012795 verification Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 238000009795 derivation Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000004069 differentiation Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 101100438426 Drosophila melanogaster Art4 gene Proteins 0.000 description 1
- 240000007232 Illicium verum Species 0.000 description 1
- 235000008227 Illicium verum Nutrition 0.000 description 1
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000004308 accommodation Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000003631 expected effect Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 238000012821 model calculation Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 238000012731 temporal analysis Methods 0.000 description 1
- 238000000700 time series analysis Methods 0.000 description 1
Images
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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Mathematical Analysis (AREA)
- Operations Research (AREA)
- General Business, Economics & Management (AREA)
- Pure & Applied Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Mathematical Physics (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Development Economics (AREA)
- General Engineering & Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Quality & Reliability (AREA)
- Algebra (AREA)
- Probability & Statistics with Applications (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 invention provides a passenger flow pool-based method for estimating passenger flow of a subway station outside association area. Secondly, calculating a passenger flow threshold of a related area outside the subway station. Then, a passenger flow estimation model of the related area outside the subway station is determined. And finally, iterating the input and the output corresponding to each sampling period, and adopting residual error learning for each group of network layers in the iteration process, thereby reducing errors, improving the precision and obtaining more accurate passenger flow estimation quantity. The passenger flow estimation model for the off-site related area can provide a theoretical basis for future site flow limiting measures and site early warning management.
Description
Technical Field
The invention relates to the field of passenger flow estimation of subway station areas, in particular to a passenger flow estimation method of a subway station outside associated area based on a passenger flow pool.
Background
The determination of the range of the associated area outside the subway station is the key for estimating the passenger flow of the associated area, and the range of the associated area of the station is influenced by the position of the subway station, the development degree of the land around the station, the distribution of supporting facilities and other factors in the early stage of the rail transit construction planning. The research on the range of the station associated area is also the analysis on the spatial distribution rule of the passenger flow, and the research on the range of the off-station associated area provides a precondition for establishing a passenger flow estimation model of the off-station associated area. In the prior art, the empirical attraction of the station is obtained by an empirical value initially in the determination of the range of the related area outside the subway station, and due to strong subjectivity and lack of persuasion, a large number of scholars gradually turn the research method to statistical analysis, the accuracy of the research result is improved by utilizing a field investigation mode, and the most important research method is to estimate the attraction of the subway station by analyzing a connection mode and a connection total amount.
Accurate estimation of passenger flow is a key to improving the operation safety of rail transit, wherein the estimation of the number of passenger flow belongs to an important component of traffic volume prediction, such as a four-stage method. The following methods are common passenger flow prediction methods: prediction according to the passenger travel OD, a non-ensemble model method, and the like are performed, but these methods are all based on a four-stage method, and a prediction method according to the classification of the passenger flow OD is most widely used.
Regarding the range of the off-site related area and the passenger flow estimation method, the conventional systematic research method summarizes the following problems in the existing off-site related area passenger flow estimation method:
(1) the research on the off-site related areas is relatively less, and the off-site related areas are not assigned as key areas. And in the research on the passenger flow estimation of the counterweight zone, the quantitative relation between the commute time when the passenger arrives at the subway station and the probability of selecting the vehicle is not shown.
(2) Most of the passenger flow estimation methods estimate the cross-section passenger flow, and the estimation research of the regional passenger flow is less. And the regional passenger flow estimation is usually expressed by passenger flow density, and the regional passenger flow quantity cannot be quantized.
Disclosure of Invention
In order to solve the technical problem, the invention provides a passenger flow pool-based method for estimating the passenger flow of the associated area outside the subway station. In order to reduce the accumulated error possibly occurring in the iteration process, a residual error network algorithm in image processing is introduced, and the accumulated error occurring in the iteration step is reduced by training the error in each iteration process, so that the accuracy of the passenger flow estimation model in the associated region is improved. The technical scheme is as follows:
the method comprises the following steps:
(1) data preprocessing and determining passenger flow pool input according to passenger flow characteristics
Firstly, carrying out data normalization and wavelet denoising on passenger flow inflow rate f (t) of a passenger flow pool, and then carrying out white noise test on the processed data, wherein when f (t) belongs to a white noise sequence, no time rule needing to be extracted exists in the sequence, and f (t) is not adopted as the input of the passenger flow pool; if f (t) does not belong to a white noise sequence, adopting f (t) as the input of the passenger flow pool;
(2) calculating passenger flow threshold of associated area outside subway station
Maximum number of accommodated passenger flows Q in a site-specific areamaxThe calculation method is
Qmax=a·Se/Sp
a=a1×a2
a2=1/1+x
Sp=(bs+bx)×(ds+dx)
Wherein a represents a passenger capacity coefficient, SeRepresenting the effective area, S, of the off-site direct association zonepRepresents the individual floor area of the passenger, a1Representing the formation factor coefficient; a is2Representing the coefficient of transportation, x representing the proportion of passengers carrying large pieces of luggage, SpRepresenting the footprint of a single passenger, bsIndicates the width of the shoulder of the passenger, bxIndicates the left-right psychological distance of the passenger, dsIndicating the thickness of the passenger's body, dsRepresenting the psychological distance of the passenger;
(3) passenger flow estimation model for determining related areas outside subway station
In the model, the model is divided into a plurality of models,
f(t)=f(t)walk+f(t)bike+f(t)bus
g(t)=n(g(t)l+gp)+m(g(t)nl+gp)
wherein Q (T) represents the number of passenger flows in the passenger flow pool at the current T moment, f (T) represents the passenger flow inflow rate of the passenger flow pool, g (T) represents the passenger flow outflow rate of the passenger flow pool, and QmaxIndicating the maximum number of passengers accommodated in the site-associated area, f (t)walkIndicating the rate of flow of pedestrian traffic in the traffic pool, f (t)bikeIndicating the flow rate of bicycle in the flow cell, f (t)busIndicating traffic flow in the traffic pool, g (t)lTraffic rate for passenger security inspection passage with package, g (t)nlThe passing rate of the security check channels for passengers without packages, n is the number of the security check channels for passengers with packages, m is the number of the security check channels for passengers without packages, and gpRepresenting the rate at which subway staff expend safety checks on passengers;
(4) and iterating the input and the output corresponding to each sampling period, establishing a deep residual error network, learning the residual error of each group of network layers in the iteration process, and finally obtaining the passenger flow number of the passenger flow pool.
Preferably, the specific way of establishing the depth residual error network in step (4) is as follows:
residual error learning is adopted for each group of network layers in the iterative process:
yT=h(qT)+Q(T)+qT
qT=f(yT)
wherein, yTFitting of the mapping relation for one through several stacked network layers, h (q)T) Is an identity map, Q (T) represents the estimated value of the passenger flow pool at time T, qTRepresents the estimated error input of the passenger flow pool at time T, f (y)T) Is an activation function;
and obtaining the feature expression of any deep level unit T through recursion:
drawings
FIG. 1 is a flow chart of the method of the present invention.
FIG. 2 is a schematic diagram of a passenger flow pool.
Fig. 3 is a schematic diagram of a site association area.
Fig. 4 is a schematic diagram of a residual network algorithm flow.
Detailed Description
The method estimates the passenger flow of the off-station related area by establishing a passenger flow pool-based passenger flow estimation model and adopting a deep residual error network algorithm in video processing. The specific flow is shown in fig. 1.
(1) Passenger flow estimation model for associated region outside subway station
(1.1) associated region passenger flow estimation model
The maximum aggregation effect benefit of transferring subway passenger flows from various transportation modes to stations is used as an associated area of a subway station, the associated area is a passenger flow pool, and the passenger flow change condition in the passenger flow pool follows the conservation law.
By "traffic pool" is simply understood a pool without physical isolation that stores traffic, similar to the pool concept. The water storage capacity of a pool depends on the water inlet and outlet of the pool, and similarly, in a certain area, the flow of passenger flow depends on the passenger flow entering the area and the passenger flow leaving the area. The principle diagram 2 is shown as the following diagram:
assuming that the number of people in the passenger flow of the relevant area at the time T is Q (T), the passenger flow arrival function of various traffic modes in the relevant area at the time T is f (T), and the passenger flow arrival rate at the security check, namely the outflow rate of the passenger flow in the passenger flow pool is g (T). The passenger flow in the passenger flow pool at the time T meets the following mode:
the derivation process is as follows:
Q(T)=f(T)-g(T)+Q(T-1)
Q(T-1)=f(T-1)-g(T-1)+Q(T-2)
Q(T-2)=f(T-2)-g(T-2)+Q(T-3)
...
in the formula, q (T) represents the number of the passenger flows in the passenger flow pool at the current time T, f (T) represents the passenger flow inflow rate of the passenger flow pool, and g (T) represents the passenger flow outflow rate of the passenger flow pool. Wherein f (t) is a function of arrival of passenger flows in various traffic modes, and can be obtained by analyzing the rule of arrival of passenger flows; g (t) represents a traffic function at the security inspection site and the function has an upper limit, namely, the maximum traffic capacity g at the security inspection sitemaxIn the case of sudden large traffic or morning and evening rush hour, the traffic capacity at the security check is usually maintained at gmax。
Although the defined traffic pool has no physical boundary, there is a maximum traffic capacity Qmax. When the passenger flow pool capacity Q (T) at a certain moment exceeds Q by calculationmaxAt this time, it means that the passenger flow pool has an overflow phenomenon, the number of passengers in the passenger flow pool reaches the peak value, and the capacity Q (T) of the passenger flow pool is not exceededCan continue to be represented by the calculated value, when Q (t) ═ Qmax。
To sum up, the passenger flow estimation model of the passenger flow pool is as follows:
(1.2) associated region passenger flow estimation model description
1. Regional passenger flow input
In the associated regional traffic estimation model, traffic input refers to the rate at which traffic enters the regional scope of a known stop. Under the influence of different traffic modes, the rules of the passenger flows of different traffic modes reaching the station are different, so the passenger flows of different traffic modes are distinguished when the passenger flow input amount is calculated.
f(t)=f(t)walk+f(t)bike+f(t)bus+···
2. Regional passenger flow output
The passenger flow output quantity refers to the speed of passenger flow dissipation in the associated area, and the invention mainly researches the passenger flow in the associated area outside the station, so that a security check place is selected as a station passenger flow dissipation place when the passenger flow dissipation speed is determined. The passenger flow dissipation rate depends on the traffic capacity of a security check place, in the early peak period, the station is divided into a security check place with a bag passenger and a security check place without a bag passenger for improving traffic efficiency, the traffic capacities of different types of security check places are different, and the security check places are specifically distinguished in practical application.
g(t)=n(g(t)l+gp)+m(g(t)nl+gp)
Wherein, g (t)lTraffic rate for passenger security inspection passage with package, g (t)nlThe passing rate of the security check channels for passengers without packages is shown, n is the number of the security check channels for passengers with packages, and m is the number of the security check channels for passengers without packages. In the process of security check of passengers, the passengers usually need to be subjected to security check of subway workers, so that the passing rate of a security check passage is further reduced, and a penalty item g is introducedpTo indicate the consumption of safety inspection of passengers by subway staffThe rate of the fee.
3. Maximum volume of passenger flow
The maximum passenger flow output quantity refers to the maximum traffic rate of the security inspection passage. The maximum traffic capacity of different types of security inspection channels is different,the maximum traffic capacity of the security inspection passage of the package to be inspected is represented, namely the passenger flow outflow rate of the passenger flow pool of the passage of the package to be inspected,the maximum traffic capacity of the security inspection passage without inspecting the packages is represented, namely the passenger flow outflow rate of the passenger flow pool without inspecting the package passage.
4. Maximum passenger flow capacity
The maximum passenger flow carrying capacity of the passenger flow pool means that in a station associated area, the maximum passenger flow containing quantity of the area, namely an area passenger flow threshold value, is marked as Qmax。
After determining each parameter in the model, the passenger flow estimation model of the subway station association area is as follows:
in the model:
f(t)=f(t)walk+f(t)bike+f(t)bus+···
g(t)=n(g(t)l+gp)+m(g(t)nl+gp)
(1.3) arrival passenger flow estimation method
In the passenger flow estimation model of the off-station related area, the passenger flow input amount refers to the number of arriving passenger flows in different transportation modes within the time T. The law of passenger flow to the station is different for different traffic modes. And arranging the passenger flow data according to the time sequence of observation statistics to form a time sequence. Time series is commonly used in the fields of air, passenger flow and the like, and is a process for predicting the trend of a certain index by researching the time series of the index. In the field of rail transit, the time series analysis can fully reflect the change rule of passenger flow along with time, and the ARIMA model in the time series can effectively reflect the change condition of the passenger flow in a non-steady state.
When determining the input f (t) of the passenger flow of the associated area, modeling should be performed for the non-stationary time series corresponding to the passenger flow of different transportation modes. Before performing ARIMA model fitting on f (t), f (t) should be judged first. Performing stationarity check, namely white noise check, on a given sequence f (t), wherein when the sequence f (t) belongs to the white noise sequence, it is indicated that there is no dependency between passenger flow data corresponding to each sampling period in the sequence, and at this time, there is no time rule needing to be extracted in the sequence, and the ARIMA model is not suitable for the given time sequence; if f (t) does not belong to a white noise sequence, it is proved that the passenger flow data at each moment in a given time sequence can influence the subsequent data, and at the moment, an ARIMA model can be used for carrying out regular analysis on a certain traffic mode.
The non-stationary time series presented for a certain mode of transportation, i.e. the given time series f (t), can be written as follows:
f(t)=μt+εt
wherein muTA time-varying mean function, epsilon, corresponding to the non-stationary time series of the traffic patterntIs a white noise sequence. The original f (t) sequence is subjected to d-order difference and then converted into a stable sequence, and the essence of the process is the d-order derivation process. The Carmer decomposition theorem theoretically guarantees the feasibility of extracting all effective information in the original sequence after d-order difference, and information contained in the original sequence can be completely extracted through continuous difference. The process of differentiating the original sequence can be written as:
in the formula, Φ (B) represents an autoregressive coefficient polynomial, Θ (B) represents a moving average coefficient polynomial, and the calculation formulas are:
Φ(B)=1-φ1B-φ2B2-···-φpBp
Θ(B)=1-θ1B-θ2B2-···-θqBq
the remaining parameters p, q in the ARIMA model are determined as follows:
step 1: and (4) preprocessing data. For passenger flow data collected on site, data preprocessing is needed. If the passenger flow data is in a non-stationary state, zero-averaging processing and differential stationary processing are required to be carried out on the data.
Step 2: and identifying a model structure. The original passenger flow data is preprocessed to obtain the structure of an autocorrelation function and a partial autocorrelation function. When the objective function is continuously trained, the accuracy of the objective function model is continuously improved, and meanwhile, the defect of overfitting is also caused. By adding the penalty term, overfitting can be effectively avoided, and the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC) are commonly used. The expressions corresponding to the two methods are respectively as follows:
AIC=-2ln(L)+2K
BIC=-2ln(L)+ln(n)*k
both methods can perform parameter determination, but BIC is only used to solve simple models with few parameters. According to the characteristics of the passenger flow model, the AIC determines parameters. The meaning of each parameter in the expression of AIC is: l represents the maximum likelihood estimation of the sequence, n is the data quantity, and k is the variable quantity of the model. In order to prevent the over-fitting phenomenon, the corresponding parameter value when the AIC value is minimum is selected when the model parameters are determined.
Step 3: and (6) calibrating parameters. And calibrating parameters in the model according to the characteristics of the actual data.
Step 4: and (6) checking the model. In the process of carrying out model test, statistical quantity relative error and absolute percentage error are introduced. And if the prediction precision reaches the expected effect, successfully debugging the model parameters.
Step 5: and (5) determining a model. And exporting the qualified model for actual passenger flow estimation.
(1.4) passenger flow threshold of associated area outside subway station
When estimating the passenger flow of the off-site related area, not only how to estimate needs to be solved, but also the estimation result needs to be reasonably utilized when the estimation result is obtained. Passenger flow threshold Q of off-station related areamaxThe calculation of (2) is not only a judgment condition of an iterative process in the passenger flow estimation model of the associated region, but also can be used as a basis for passenger flow early warning judgment of subway stations. And quantifying the passenger flow value of the off-station related area, wherein the maximum threshold value of the passenger flow in the area can be used as the passenger flow early warning basis. The determination of the passenger flow threshold of the off-station related area directly influences the early warning mechanism of the subway station. At the present stage, passenger flow early warning for subway stations is limited to passenger flow distribution conditions of a station platform waiting area, a stair holding area and a ticket gate area in the station, and passenger flow estimation and early warning research for an off-station related area is relatively few.
In order to ensure the orderly security inspection of passengers, a barrier is usually arranged outside the station. The directly associated area outside the subway station is usually affected by the arrangement of the infrastructure such as the station isolation fence, and the area may show an irregular figure as shown in fig. 3.
In the process of queuing and entering passengers, the passengers can wait for inspection in the direct association area outside the station, and when the early warning threshold value of the direct association area outside the station is calculated, the passenger flow capacity of the direct association area outside the station is the most intuitive index for measuring the passenger flow threshold value. The maximum capacity of passenger flow in the direct association region can be calculated by the individual area of the passenger and the effective area of the direct association region outside the station, theoretically, the passenger follows the guidance of the isolated fence outside the station, queues for entering the station for security inspection, and the maximum capacity of the direct association region can be obtained:
Qmax=a·Se/Sp
wherein Q ismaxRepresenting the maximum capacity of the off-site direct association area, a representing the passenger capacity coefficient, SeRepresenting the effective area, S, of the off-site direct association zonepRepresenting the individual passenger footprint.
Since the passenger queue in the real situation can be different from the theory, it is necessary to introduce a to correct the maximum capacity calculated by the theory. a form factor of passenger flow queuingCoefficient of element a1And coefficient a of passenger's large luggage2The influence is calculated by the formula:
a=a1×a2
if the directly related area outside the subway station covers the area outside the isolation fence, passengers can not queue for waiting for detection strictly according to the sequence of the isolation fence. In the waiting and detecting process, the distance between familiar people is closer than that between strangers, and certain psychological distance exists between the strangers, so that the maximum threshold value of the area directly related to the outside of the station deviates from theoretical data, and the formation factor coefficient a can be utilized1Corrected to a value range of 1.1-a1≤1.25。
The passenger make-up may differ for different types of stops. When carrying large luggage or shopping trolleys, passengers can occupy the passenger space in the directly associated area outside the station, and the transportation coefficient a can be used2To correct the area it occupies. a is2The value of (a) is usually determined according to passenger flow characteristics of different stations, and a is calculated according to the proportion x of large luggage carried by passengers2The calculation formula of (2) is as follows:
a2=1/1+x
in the calculation formula, SpRepresenting a single passenger footprint. In the process of waiting for detection of passengers in a queuing way, certain psychological distance can be kept among the passengers, and the psychological distance needs to be added when the individual floor area of the passengers is calculated. The calculation formula is as follows:
Sp=(bs+bx)×(ds+dx)
in the formula, bsIndicates the width of the shoulder of the passenger, bxIndicates the left-right psychological distance of the passenger, dsIndicating the thickness of the passenger's body, dsRepresenting the passenger's mental distance from front to back.
According to the actual layout of the stations, the threshold calculation method of the directly associated region outside the subway station comprises the following steps:
Qmax=a·Se/Sp
a=a1×a2
a2=1/1+x
Sp=(bs+bx)×(ds+dx)
(2) passenger flow estimation algorithm for associated areas outside subway station
(2.1) deep residual network concept
Ideally, when the input and the output of the passenger flow pool are known, the number of the passenger flows in the passenger flow pool, that is, the number of the passenger flows in the off-station related area can be obtained only by iterating the input and the output corresponding to each sampling period. In practical application, because input and output need to be estimated by means of an ARIMA model, accumulated errors are easily generated in an iteration process, and passenger flow estimation errors of an off-site direct association area are continuously increased in subsequent calculation, so that the accuracy of the model is seriously influenced.
In order to counteract the influence of accumulated errors in the iteration process on subsequent passenger flow estimation, the invention introduces the idea of a Deep Residual error Network (Deep Residual Network) which is established for ensuring that the model precision is not influenced by Network iteration in the image processing process. The residual error network (ResNet) allows all models to be deeply characterized layer by layer, so that the feedforward/back propagation algorithm is smoothly carried out, and the method has strong characterization capability, optimization capability and induction capability.
The deep residual network assumes that a Shallow network (Shallow Net) reaches saturation accuracy, and an Identity Mapping layer (Identity Mapping) is added behind the Shallow network, so that not only is the network depth increased, but also errors are not increased. A Residual Network structure (Residual Network) is introduced into the deep Residual Network, each Residual learning corresponds to one building unit, the Network training level can be deepened through the Residual Network structure, and the error can be guaranteed to be within a confidence interval.
(2.2) correlation area passenger flow depth residual error network
For the passenger flow of the associated area outside the subway station, assume yTThe method is characterized in that a mapping relation is fitted through a plurality of stacked network layers, Q (T) represents a passenger flow pool estimated value at T moment, and q is usedTAnd representing the estimation error input of the passenger flow pool at the time T. The multilayer non-linear hierarchy can gradually approach a certain levelThe assumption of a complex function is equivalent to approximating the residual function that it can asymptotically approach, i.e., Q (T) + qT(assuming that the inputs and outputs belong to the same dimension), the new approximation function can be expressed as Q (T) + qT。
Residual learning is adopted for each group of network layers in the iteration process, and the following forms can be presented:
yT=h(qT)+Q(qT)
qT=f(yT)
wherein h (q)T) Is an identity map, f (y)T) Is an activation function. Since h (q) and f (y) are both identity maps, i.e., h (q)T)=qT,f(yT)=yTIn the early and backward propagation stages in training the deep residual network, signals can be passed from one unit to another, making training simpler and undistorted. At this time, the above formula can be expressed as:
qT+1=qT+Q(T)+qT
through recursion, the feature expression of any deep level unit T can be obtained:
the schematic flow chart of the algorithm for reducing the model degradation degree by constructing the depth residual error network is shown in fig. 4:
the excellent properties of this expression are shown in: features q for arbitrarily deep cells TTCan be expressed as a feature plus shape of the shallow cell tThe residual function shows that any unit T and T has residual characteristics, so that the model precision can be improved in the residual network training process, and the influence of accumulated errors on subsequent passenger flow estimation is reduced.
(3) Example analysis
(3.1) estimation of arrival passenger flow of subway station in octagonal amusement park
1. Pedestrian traffic estimation model
Step 1: data pre-processing
Monitoring the traffic flow of the station entrance of the subway station B in the octagonal amusement park in real time, wherein the time is 6: 50-9: 00 earlier, the time interval is 5 minutes, sampling the traffic flow of ten weeks and one week continuously, and dividing according to different traffic modes to obtain 1040 sample data. Modeling was performed based on the data from the first nine days and an estimation test was performed using the last set of data. In order to eliminate the interference of random passenger flow except data to the sampled data, the wavelet technology is adopted to perform denoising processing on the original sampled data.
Step 2: model structure identification
According to observation, the time sequence formed by the pedestrian flow is in a non-steady state, in order to identify the model structure, the sequence is firstly subjected to first-order difference, and the autocorrelation function and the partial autocorrelation function obtained by difference are shown in table 1. When the original data is differentiated, the objective function starts to converge after the number of difference terms reaches 10 in general, so that the number of difference terms is 10 at most in the differentiation in general.
TABLE 1 Difference results
And after the autocorrelation function value and the partial autocorrelation function value of the original data are obtained by calculation, determining the parameter value range of the passenger flow estimation model by observing the function values of the corresponding item numbers.
Step 3: parameter estimation
From the calculation results, it is clear that the first three terms of the partial autocorrelation function are not 0, and thus p-3 in the ARIMA model can be determined. And the first term of the autocorrelation function in the graph is not 0, and the second term and the third term are within the confidence interval. In order to make the value of q more comprehensive, q can take [1,3 ]. For different combination values of p and q, the AIC values of the combinations were calculated according to the AIC judgment criteria, see table 2.
TABLE 2 calculated AIC values
And fitting the data of the first nine days by using the established model, and bringing the parameters determined by the AIC into the model to establish the ARIMA model of the pedestrian flow.
Step 4: model inspection
The data from the last day was examined based on the established model, and table 3 lists the estimated values and the measured values, and their relative errors, and it can be seen that the accuracy has been relatively high. In the whole passenger flow estimation process, the accuracy of individual time points is reduced, the model can be continuously perfected through more mass historical data, and the estimation accuracy of the model is improved.
TABLE 3 relative error results
Step 5: verifying wavelet de-noising results
In the process of estimating passenger flow by applying an ARIMA model, in order to verify the influence of the wavelet technology on the model, the raw data which is not denoised is brought into the ARIMA model for parameter verification, and then the influence of the wavelet technology on the model output is compared, and the result is shown in the following table 4:
table 4 demonstrates the denoising effect
And comparing the relative error data of the predicted value after data processing with the predicted value without data processing.
The result shows that the estimated value without wavelet de-noising has more deviation of individual points from the measured value and larger deviation degree, and the relative error of each time node after wavelet de-noising is almost within 15% of the confidence interval, thereby obviously improving the passenger flow estimation precision.
2. Bicycle passenger flow estimation model
Step 1: data pre-processing
Step 2: model structure identification
As can be seen from the change trend of the bicycle pedestrian volume, the time sequence formed by the bicycle passenger volume presents a non-steady state. To identify the model structure, first order difference should be performed on the passenger flow data, and the autocorrelation function and the partial autocorrelation function obtained by the difference are shown in table 5. When the original data is differentiated, the objective function starts to converge after the number of difference terms reaches 10 in general, so that the number of difference terms is 10 at most in the differentiation in general.
TABLE 5 model calculation
And (3) calculating an autocorrelation function value and a partial autocorrelation function value of the original passenger flow data, observing a function value corresponding to the number of decomposition items of the autocorrelation function value, and calibrating a parameter value range of the passenger flow estimation model.
Step 3: parameter estimation
The calculation results from the table show that the first three terms of the partial autocorrelation function are obviously not 0, and therefore, the ARIMA model parameter p is judged to be 3. Whereas the first term of the autocorrelation function in the graph is obviously not 0 and the second third term is within the confidence interval. In order to make the value of q more comprehensive, q can take [1,3 ]. For different combination values of p, q, the AIC values for each combination were calculated according to the AIC criteria, see table 6. A set with smaller AIC values is also selected as a model parameter.
TABLE 6AIC calculation results
Step 4: model inspection
The data from the last day was examined based on the established model, and table 7 lists the estimated values and the measured values, and their relative errors, and it can be seen that the accuracy has been relatively high.
Table 7 model test
Step 5: verifying wavelet de-noising results
In the process of estimating passenger flow by using the ARIMA model, in order to verify the influence of the wavelet technology on the model, the raw data which is not denoised is brought into the ARIMA model for parameter verification, and then the influence of the wavelet technology on the model output is compared, and the result is shown in the following table 8.
TABLE 8 relative error contrast
The result shows that the estimated value without wavelet de-noising has more deviation of individual points from the measured value and larger deviation degree, and the relative error of each time node after wavelet de-noising is relatively converged, thereby obviously improving the passenger flow estimation precision.
3. Bus passenger flow estimation model
Step 1: data pre-processing
Step 2: model structure identification
The first order difference is performed on the passenger flow sequence and the calculation results are shown in table 9. And after the autocorrelation function value and the partial autocorrelation function value of the original data are obtained by calculation, determining the parameter value range of the passenger flow estimation model by observing the function values of the corresponding item numbers.
TABLE 9
Step 3: parameter estimation
The calculation results show that the first three terms of the partial autocorrelation function are obviously not 0, so that p can be determined to be 3 in the ARIMA model. Whereas the first term of the autocorrelation function in the graph is obviously not 0 and the second third term is within the confidence interval. In order to make the value of q more comprehensive, q can take [1,3 ]. For different combination values of p and q, the AIC values of the combinations were calculated according to the AIC judgment criteria, see table 10.
TABLE 10AIC results
And substituting the calculated parameter values into an ARIMA model, thereby establishing the ARIMA model of the bus passenger flow, and then predicting the data of the tenth day according to the actual data of the first nine days.
Step 4: model inspection
The data from the last day was examined based on the established model, and table 11 lists the estimated values and the measured values, and their relative errors, and it can be seen that the accuracy has been relatively high.
TABLE 11 prediction of relative error
Step 5: verifying wavelet de-noising results
In the process of estimating passenger flow by applying an ARIMA model, in order to verify the influence of the wavelet technique on the model, the raw data which is not denoised is brought into the ARIMA model for parameter verification, and then the influence of the wavelet technique on the model output is compared, and the result is shown in the following table 12:
TABLE 12 relative error comparison
The result shows that the estimated value without wavelet de-noising has more deviation of individual points from the measured value and larger deviation degree, and the relative error of each time node after data processing is converged a lot, thereby obviously improving the passenger flow estimation precision.
(3.2) octagonal amusement park subway station external association region threshold
The calculation formula for calculating the passenger flow threshold of the off-station direct association area is as follows:
Qmax=a·Se/Sp
a=a1×a2
a2=1/1+x
Sp=(bs+bx)×(ds+dx)
a coefficient is formed by1And a2Jointly determining that the passengers have low degree of recognition due to commuting habits in early peak period according to actual research of subway stations in the octagonal amusement park1A value of 1.15 may be taken. Commuter traffic during early peak hours is predominant, passengers carrying large pieces of luggage are only about 5%, so a2The value was 0.95. The value of a is calculated to be 1.09.
And calculating to obtain the radius of the direct association area of the subway station of the octagonal amusement park, wherein the radius of the direct association area is 109.2 meters, the area which takes the security inspection equipment as the center of a circle and the 109.2 meters as the direct association area of the subway station. Since the off-site associated area covers public facilities such as greening facilities, express ways and the like, the area of the public facilities should be removed when calculating the area of the area directly associated with the site. According to the actual on-site investigation condition, measuring the direct correlation region to obtain the effective area S of the off-site correlation regione=684。
The physical sign parameters of the subway passenger flow in China mainly follow the research data of hong Kong and Japan. When calculating the passenger footprint, the individual thickness averages 0.24 meters and the average shoulder width 0.45 meters. The value of the psychological distance between people is generally between 0.15 and 0.25, and in order to follow the safest principle, the value of the psychological distance of the invention is 0.25 meter. Is calculated to obtain Sp=0.343。
Through calculation, the maximum passenger flow threshold value of the direct association area of the octagonal amusement park subway station is about Qmax2059, the maximum threshold for passenger flow in the off-site association zone is about 2059 people.
(3.3) octagonal amusement park subway station outside association region passenger flow estimation model
When the passenger flow of the associated area outside the subway station of the octagonal amusement park is estimated, the sampling time is 6: 50-9: 00 in the early peak period, the station is used for controlling the arrival time of passengers, and normal current limiting is carried out in the period of 7: 00-8: 30 by prolonging the length of the progress isolation fence. During early peak period, the subway security check place carries out security check with the maximum traffic capacity, and through field check, the maximum traffic capacity of a security check channel of a package to be checked at the station is g (t)l160 persons/5 min, the maximum traffic capacity of the security inspection channel without inspection packages is g (t))nl200 persons/5 min, and only one of two security check channels is opened, namely m-n-1 in the model. And (4) performing security inspection on the security inspection channel with the maximum traffic capacity during the early peak period, namely, fixing the outgoing passenger flow rate of the passenger flow pool.
Because the three transportation modes are all covered in the direct association area outside the station, the time of three passenger flows entering the passenger flow pool does not need to be changed, and the speed of the different transportation modes reaching the station is superposed to be the input speed of the passenger flow pool.
Then the passenger flow estimation model of the eight-corner amusement park subway station outside association area at the time T is as follows:
f(t)=f(t)walk+f(t)bike+f(t)bus
g(t)=g(t)l+g(t)nl
and when errors occur in the training iteration process, selecting the residual error trained by the investigation data of the first nine days, and predicting the actual passenger flow of the last day.
(3.4) passenger flow early warning of association area outside subway station of octagonal amusement park
In the process of estimating passenger flow of the associated area outside the subway station and early warning, the passenger flow change condition in the associated area is the only decisive index. The short-time passenger flow early warning is used as an early warning system and has the functions of forecasting, warning and decision-making suggestion. The system and the method have the advantages that early warning is given for the conditions of upcoming rush hour passenger flow and the like, theoretical basis is provided for station passenger flow control, and station operation efficiency is improved. In order to reflect the early warning condition, the warning condition is divided into red, yellow and green[53]The three colors respectively represent the passenger flow distribution conditions of the off-station related areas under different conditions, and can be determined according to the station characteristics when the police degree is divided.
In order to prolong the time of passenger flow entering the station in the rush hour, the subway station of the octagonal amusement park prolongs the length of an isolation fence outside the station.
Because passengers walk in the isolated fence, the passengers need to be isolated according to the requirementsThe fence guide direction is walked, when the threshold value of the isolated fence area is calculated, the left and right psychological distances among passengers do not need to be considered, only the front and rear psychological distances are calculated, and the maximum passenger capacity Q is obtained through calculationmax=1541.08。
Through calculation, the threshold value of the maximum passenger flow accommodation of the isolated column area occupies about seven percent of the passenger flow threshold value of the off-station related area, and the station early warning level needs to be manually divided for determining the station early warning level. When the number of passenger flows in the off-site association area is smaller than the maximum passenger-holding threshold of the isolation fence area, defining the early warning alarm as green; when the passenger flow quantity of the off-station associated area is greater than the maximum threshold value of the isolated fence area and less than the maximum passenger holding threshold value of the off-station associated area, defining the early warning alarm condition at the moment as yellow; and when the number of the passenger flows in the off-site associated area is greater than the maximum passenger containing threshold value of the off-site associated area, defining the early warning alarm to be red at the moment. According to the estimation result of the passenger flow of the off-site related area, the warning situation of the site is judged, as shown in table 13:
TABLE 13 site Pre-warning conditions
The early warning result shows that the passenger flow does not exceed the maximum passenger capacity threshold value at the isolation fence in the time period and does not exceed the maximum passenger capacity threshold value of the associated area outside the station. The arrangement of the extended isolation fence in the star anise amusement park subway station plays a role in guiding passenger flow in early peak periods, and the safety of subway operation is guaranteed.
The early warning of the passenger flow condition of the associated area of the subway station provides a theoretical basis for subway operation management and provides guidance for passenger flow control of the subsequent off-station associated area.
Claims (1)
1. A passenger flow estimation method for a related area outside a subway station is characterized by comprising the following steps:
(1) data preprocessing and determining passenger flow pool input according to passenger flow characteristics
Firstly, carrying out data normalization and wavelet denoising on passenger flow inflow rate f (t) of a passenger flow pool, and then carrying out white noise test on the processed data, wherein when f (t) belongs to a white noise sequence, no time rule needing to be extracted exists in the sequence, and f (t) is not adopted as the input of the passenger flow pool; if f (t) does not belong to a white noise sequence, adopting f (t) as the input of the passenger flow pool;
(2) calculating passenger flow threshold of associated area outside subway station
Maximum number of accommodated passenger flows Q in a site-specific areamaxThe calculation method is
Qmax=a·Se/Sp
a=a1×a2
a2=1/1+x
Sp=(bs+bx)×(ds+dx)
Wherein a represents a passenger capacity coefficient, SeRepresenting the effective area, S, of the off-site direct association zonepRepresents the individual floor area of the passenger, a1Representing the formation factor coefficient; a is2Representing the coefficient of transportation, x representing the proportion of passengers carrying large pieces of luggage, SpRepresenting the footprint of a single passenger, bsIndicates the width of the shoulder of the passenger, bxIndicates the left-right psychological distance of the passenger, dsIndicating the thickness of the passenger's body, dsRepresenting the psychological distance of the passenger;
(3) passenger flow estimation model for determining related areas outside subway station
In the model, the model is divided into a plurality of models,
f(t)=f(t)walk+f(t)bike+f(t)bus
g(t)=n(g(t)l+gp)+m(g(t)nl+gp)
wherein Q (T) represents the number of passenger flows in the passenger flow pool at the current T moment, f (T) represents the passenger flow inflow rate of the passenger flow pool, g (T) represents the passenger flow outflow rate of the passenger flow pool, and QmaxIs shown at the siteMaximum number of passengers in the region of interest, f (t)walkIndicating the rate of flow of pedestrian traffic in the traffic pool, f (t)bikeIndicating the flow rate of bicycle in the flow cell, f (t)busIndicating traffic flow in the traffic pool, g (t)lTraffic rate for passenger security inspection passage with package, g (t)nlThe passing rate of the security check channels for passengers without packages, n is the number of the security check channels for passengers with packages, m is the number of the security check channels for passengers without packages, and gpRepresenting the rate at which subway staff expend safety checks on passengers;
(4) iterating the input and the output corresponding to each sampling period, establishing a deep residual error network, learning the residual error of each group of network layers in the iteration process, and finally obtaining the passenger flow number of a passenger flow pool;
the specific way of establishing the depth residual error network in the step (4) is as follows:
residual error learning is adopted for each group of network layers in the iterative process:
yT=h(qT)+Q(T)+qT
qT=f(yT)
wherein, yTFitting of the mapping relation for one through several stacked network layers, h (q)T) Is an identity map, Q (T) represents the estimated value of the passenger flow pool at time T, qTRepresents the estimated error input of the passenger flow pool at time T, f (y)T) Is an activation function;
and obtaining the feature expression of any deep level unit T through recursion:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810567313.3A CN108805347B (en) | 2018-06-05 | 2018-06-05 | Passenger flow pool-based method for estimating passenger flow of associated area outside subway station |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810567313.3A CN108805347B (en) | 2018-06-05 | 2018-06-05 | Passenger flow pool-based method for estimating passenger flow of associated area outside subway station |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108805347A CN108805347A (en) | 2018-11-13 |
CN108805347B true CN108805347B (en) | 2021-11-23 |
Family
ID=64088675
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810567313.3A Active CN108805347B (en) | 2018-06-05 | 2018-06-05 | Passenger flow pool-based method for estimating passenger flow of associated area outside subway station |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108805347B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109695215A (en) * | 2018-12-29 | 2019-04-30 | 大连交通大学 | Safety check automatic current limiting system and current-limiting method |
CN111815937B (en) * | 2019-11-22 | 2022-01-04 | 北京嘀嘀无限科技发展有限公司 | Control method and device for dispatching of super-parking vehicles, storage medium and electronic equipment |
CN112906179B (en) * | 2020-11-27 | 2022-04-19 | 西南交通大学 | Urban rail transit passenger flow control optimization method based on fluid queuing network |
CN113065701B (en) * | 2021-03-30 | 2023-10-31 | 重庆交通职业学院 | Intelligent prediction method and device for rail transit passenger flow |
CN113628432B (en) * | 2021-10-13 | 2022-01-04 | 中通服建设有限公司 | Intelligent early warning system for subway people flow current limiting |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105279572A (en) * | 2015-09-16 | 2016-01-27 | 北京城建设计发展集团股份有限公司 | City track traffic passenger flow density index calculating and releasing system |
CN106295854A (en) * | 2016-07-28 | 2017-01-04 | 中国民用航空总局第二研究所 | For analyzing data processing method and the device that the passenger in airport security district arrives |
CN106875314A (en) * | 2017-01-31 | 2017-06-20 | 东南大学 | A kind of Urban Rail Transit passenger flow OD method for dynamic estimation |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170109764A1 (en) * | 2015-10-19 | 2017-04-20 | Xerox Corporation | System and method for mobility demand modeling using geographical data |
-
2018
- 2018-06-05 CN CN201810567313.3A patent/CN108805347B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105279572A (en) * | 2015-09-16 | 2016-01-27 | 北京城建设计发展集团股份有限公司 | City track traffic passenger flow density index calculating and releasing system |
CN106295854A (en) * | 2016-07-28 | 2017-01-04 | 中国民用航空总局第二研究所 | For analyzing data processing method and the device that the passenger in airport security district arrives |
CN106875314A (en) * | 2017-01-31 | 2017-06-20 | 东南大学 | A kind of Urban Rail Transit passenger flow OD method for dynamic estimation |
Also Published As
Publication number | Publication date |
---|---|
CN108805347A (en) | 2018-11-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108805347B (en) | Passenger flow pool-based method for estimating passenger flow of associated area outside subway station | |
CN105404850B (en) | Station flow monitoring system for passenger, station passenger flow monitoring method and station passenger flow managing and control system, station passenger flow management-control method | |
CN110070713B (en) | Traffic flow prediction method based on bidirectional nested LSTM neural network | |
CN104200286B (en) | A kind of urban track traffic timetable optimisation technique application framework | |
CN114783183B (en) | Traffic situation algorithm-based monitoring method and system | |
CN104778837A (en) | Multi-time scale forecasting method for road traffic running situation | |
CN106779429B (en) | Track transfer station passenger flow congestion risk evaluation method based on AFC card swiping data | |
CN108665191B (en) | Subway station outside association area calculation method based on aggregation effect | |
CN115240431B (en) | Real-time online simulation system and method for traffic flow of highway toll station | |
CN111063204B (en) | Expressway vehicle speed prediction model training method based on toll station flow | |
CN110299011A (en) | A kind of traffic flow forecasting method of the highway arbitrary cross-section based on charge data | |
CN104732286A (en) | Method for sensing safety situation of urban rail transit station | |
CN109190507A (en) | A kind of passenger flow crowding calculation method and device based on rail transit train | |
CN107767011B (en) | Pedestrian characteristic acquisition system and service level dynamic evaluation method for railway station | |
CN107180534B (en) | The express highway section average speed estimation method of support vector regression fusion | |
CN108364464B (en) | Probability model-based public transport vehicle travel time modeling method | |
CN109308343A (en) | A kind of Forecasting of Travel Time and degree of reiability method based on Stochastic Volatility Model | |
CN112767684A (en) | Highway traffic jam detection method based on charging data | |
CN107978148A (en) | A kind of traffic status prediction method based on multi-source traffic data dynamic reliability | |
Haputhanthri et al. | Short-term traffic forecasting using LSTM-based deep learning models | |
CN117238126A (en) | Traffic accident risk assessment method under continuous flow road scene | |
CN116597642A (en) | Traffic jam condition prediction method and system | |
Khetarpaul et al. | Bus arrival time prediction using a modified amalgamation of fuzzy clustering and neural network on spatio-temporal data | |
CN113487872B (en) | Bus transit time prediction method based on big data and artificial intelligence | |
CN103793761B (en) | A kind of recognition methods of the abnormal volume of the flow of passengers factor of hinge |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |