WO2016045195A1

WO2016045195A1 - Passenger flow estimation method for urban rail network

Info

Publication number: WO2016045195A1
Application number: PCT/CN2014/093084
Authority: WO
Inventors: 贾利民; 秦勇; 于鸿飞; 王子洋; 赵忠信; 曾璐; 杜渺; 梁平; 孙方
Original assignee: 北京交通大学
Priority date: 2014-09-22
Filing date: 2014-12-05
Publication date: 2016-03-31
Also published as: CN104217129B; CN104217129A

Abstract

Provided is a passenger flow estimation method for an urban rail network, the method comprising: configuring a multi-factor-influenced resistance calculation method and an effective path calculation method to obtain a multi-path distribution ratio among rail network ODs, and loading a passenger flow according to the ratio; obtaining an overall state of a passenger flow by estimating a passenger flow in the rail network in combination with a passenger flow distribution regularity of a corresponding period in the past. The present invention ensures the safety of urban rail transit through the accurate estimation of the rail network passenger flow, substantially improving the service level and quality of urban public transportation across the country.

Description

Method for estimating passenger flow of urban rail road network

Technical field

The invention relates to a method for estimating passenger flow of a city rail network.

Background technique

With the growth of urban rail transit mileage in China, the importance of its safe operation and emergency dispatch has become increasingly prominent. As the main bearer of urban public transportation, the urban rail network system will cause a large area of passenger flow in the event of an unexpected interruption such as interval interruption, which will have a great impact on passenger transportation safety and service level. Road network emergencies frequently occur. In order to properly allocate transportation capacity and implement effective passenger transportation organization measures, it is necessary to grasp the quantitative data such as the scope and impact degree of emergencies more quickly and accurately.

At present, there is still a lack of a perfect and systematic urban rail network passenger flow estimation technology. The existing systems and methods are difficult to accurately estimate the passenger flow to ensure that the operational decision-making department conducts correct statistics.

Summary of the invention

The invention realizes the derivation of the rail transit passenger flow for the whole road network. The present invention specifically adopts the following technical solutions: the following steps are included:

Step 1: Obtain historical historical passenger flow OD data, and transfer and store the acquired data in a database;

Step 2: Define the path impedance and the effective path in the road network in the path ratio allocation module, establish a path search algorithm to obtain an effective path between the ODs, and use the utility theory to calculate the multipath selection between the OD pairs by using the impedance theory. The probability of inputting the historical synchronous passenger flow OD data obtained in step 1 to the path ratio distribution module loads the passenger flow in the daily situation and outputs the derivation result of the daily passenger flow.

Preferably, the path impedance calculation method in the path ratio allocation module in step 2 is as follows:

(1) Calculate the time impedance

1 Calculate train running time

Train running time refers to the time passengers spend on urban rail transit trains. It consists of two parts: interval running time and stop time:

In the formula:

The train running time of the OD kth path from station a to station b; T _i,j is the running time of station i to station j on path k; T _s is the stop time of the train passing through the intermediate station;

2 Calculate the transfer time

The transfer time refers to the time spent by the passenger outside the train of the transfer station. It consists of two parts: transfer time and transfer time. Car time:

In the formula:

The total transfer time for the ok kth path from a station to b;

The transfer time from the m line to the n line on the path k;

The travel time for the transfer from the m line to the n line;

The waiting time for the transfer from the m line to the n line; α ₁ is the penalty coefficient of the transfer time;

The calculation formula for the transfer travel time is:

In the formula:

The distance traveled from the m line to the n line at the transfer station k,

The average walking speed of the transfer from the m line to the n line at the transfer station k during the t period;

The calculation formula for the transfer waiting time is:

Where: H ⁿ is the departure interval of the transfer line n;

3 Calculate the inbound and outbound time

The entry and exit time is:

In the formula:

The travel time from the stop gate to the start station a platform, and the departure time from the stop to the terminal b;

(2) Calculate the congestion impedance

The formula for calculating the full load rate is as follows:

Where: δ is the train full load rate; P is the section passenger flow per unit time; D is the section transport capacity per unit time; n is the number of trains per unit time; Y is the vehicle capacity; B is the number of trains, Y *B is the entire train capacity;

The congestion degree impedance is:

In Q(δ), the congestion coefficient on a section of the rail transit network; 0, A, B correspond to three levels of congestion coefficient, A is the extra time overhead coefficient in general congestion; B is overcrowded Extra time overhead factor; δ ₀ is the full load rate when the number of passengers in the car is equal to the number of seats; when the number of passengers in the car is equal to the capacity, the full load rate is 1;

(3) Calculate the transfer penalty

Formulated as follows:

In the formula,

Is the number of transfers of the kth path between the wth OD pairs, and α ₂ is the transfer penalty coefficient in the above;

(4) Calculate the retention time

The additional congestion factor is expressed in exponential form:

Where: η and

For the parameter; x _i is the passenger flow to the i station; c is the maximum passenger capacity of the train;

The time of the I station is:

Where: T _i ^γ is the occupancy time of the i station; γ _i is the additional congestion factor of the i station;

The departure interval for the i-station to the n-line direction;

The total time of all stations is:

T ^γ =ΣT _i ^γ

(5) Calculate the passenger's comprehensive travel impedance

The passenger integrated impedance function is expressed as:

Preferably, the effective path screening method in the path ratio allocation module in step 2 is as follows:

(1) K short circuit search implementation

The algorithm is described as follows:

1 Using Dijkstra algorithm to obtain the shortest path tree with the starting node s as the root in the directed graph (N, A), the shortest path between the starting node s and the ending node t is pk, k=1;

2 If k is less than the maximum number K of the required shortest path, and there is still a candidate path present, let the current path p=pk, turn to ○33; otherwise, the program ends;

3 Find the first node of the current path p from the first node with the indegree greater than 1, recorded as nh; if the extended node n'h of nh is not in the node set N, then turn to ○44, otherwise find out Among all the nodes behind nh in path p, the corresponding extended node is not the first node in N, denoted as ni, turn to ○55;

4 Construct an extension node n'h for the node nh and add it to the set N. At the same time, connect all the nh predecessors in the graph (N, A) to an arc of n'h, and the weight corresponding to the arc is unchanged. , add these arcs to the arc set A, except that nh is the previous node nh-1 in p; calculate the shortest path from the start node s to n'h, and record ni=nh+1;

5 For all subsequent nodes in p starting from ni, it may be denoted as nj, and the following operations are performed: adding the extended node n'j of nj to the node set N; except for the previous node nj-1 of nj in the path p, Connect an arc from the nj precursor node to its extension node n'j, the weights on the arc remain unchanged, and add these arcs to the arc set A; in addition, if the previous node nj-1 of n in p With the extended node n'j-1, it is also necessary to connect an arc from n'j-1 to n'j, and the weights and arcs (nj-1, nj) have equal weights; the calculation starts from node s to n The shortest path of 'j; update the current shortest path tree, and find that the shortest path between the current extended node t(k)' from the starting node s to the ending node is the kth shortest path, so k=k+1, turn ○22 continue;

(2) Determination of the effective path set

The unreasonable path in the K-small path obtained by the path search algorithm does not participate in the distribution of the passenger flow, and generates an effective path set according to the limitation of the operation time of different rail transit lines;

1 operation time judgment

During a certain period of time, if one of the K optional trailing path sets is outside the operating time, it is not included in the effective path set; the running time of the path is represented by the effective running time of the starting station of the path. The effective operation time of the starting station is the intersection of the first and last shift time of the starting station and the first and last shift time of each transfer station in the path, and the starting time of the starting station is reversed;

2 travel impedance threshold judgment

The unreasonable path in the K-straighed path obtained by the path search algorithm does not participate in the distribution of the passenger flow; the validity test of the path is judged by the travel impedance threshold; assuming that the K-selectable progressive path sets between the two stations are The impedance value of the shortest path

If the impedance value of the secondary short path or other shorter short path exceeds a certain range of the travel impedance value of the shortest path, the short path or the second short path is considered unreasonable; it can be reasonably assumed that when

When it is small,

versus

In proportion to

When it is large enough, the upper bound of the allowable area of the travel impedance value is fixed; it can be expressed as:

In the formula:

The upper bound of the travel impedance value for the effective path;

The maximum allowable value of the effective path exceeds the shortest path travel resistance value; ξ is a proportional coefficient; U is a constant; their values can be determined by passenger travel survey.

Preferably, the multipath allocation method in the path ratio allocation module in step 2 is as follows:

Let the set of k valid paths between the two stations of the OD be

Select path

The probability is P _i (i=l,...,k); obviously, P _i is a function of the integrated travel impedance of the path; the integrated travel impedance of each effective path is T _i ^f (i=l,...,k) And satisfy

Then for P _i has the following characteristics:

a.

That is, the sum of the proportions of all valid route passenger flow assignments between a pair of stations is equal to 1;

b.

Then P _i =P _j , that is, the paths with equal impedance values are selected with equal probability;

C.1≥P ₁ ≥P ₂ ≥...≥P _k ≥0, that is, the smaller the probability that the path with larger impedance value is selected, the probability that the path of the smallest impedance value is selected is the largest;

d. If T _i ^f is very close to T ₁ ^f (ie the shortest path impedance value)

), I should be very close to the P _i T ₁ ^f, when the impedance in the vicinity of T ₁ ^f, the rate of decrease of the P _i is small;

e. As the impedance value increases, the deceleration rate of P _i will increase rapidly, that is, the probability that the path is selected will decrease rapidly;

The normal distribution is used to describe the passenger's travel path selection behavior. The formula of the normal distribution function is as follows:

Where: μ gives the x value of the maximum expected probability, here is 0; σ is a constant whose value will determine the steepness of the normal curve;

The passenger flow distribution ratio of the path is calculated by the following formula:

Preferably, the method for deriving the road network passenger flow in the path ratio allocation module in step 2 is as follows:

Where t _i refers to the time point when the passenger departs from the O station to reach the i station;

t _O refers to the time when the passenger swipes the card from the O station;

It is the waiting time of passengers at O station;

T _ab is the running time of the train in the interval ab, and M is the interval set;

T _s is the stop time of the train at station S, and N is the station set;

It is the transfer time for passengers to change from line m to n at k station.

The invention has the following beneficial effects:

(1) The present invention constructs a rapid quantitative estimation method for the number of stations and passengers in the entire network, and fills the gap in the industry.

(2) It is conducive to improving the safety level of road network operation, reducing the economic losses caused by security accidents and emergencies, and indirectly improving the operational efficiency of enterprises.

(3) This technology conforms to the demand for rail transit in megacities. Through the accurate estimation of passenger flow, it can improve the operational safety guarantee capability and ensure the safety of urban rail transit, thus greatly improving the service level and quality of urban public transport in China.

DRAWINGS

Figure 1 is a flow chart of the method for estimating the passenger flow of the whole road network.

Figure 2 is a flow chart of the passenger flow impact analysis method based on the historical passenger flow law.

detailed description

Figure 1 is a flow chart of the method for estimating the passenger flow of the whole road network. By designing the multi-factor impedance calculation method and the effective path calculation method, the multi-path distribution ratio between the road network OD is obtained, and the passenger flow is loaded according to the ratio. Then, combined with the distribution law of passenger flow in the same period of history, the full state of passenger flow is obtained through the deduction of passenger flow in the road network.

Figure 2 is a flow chart of the passenger flow impact analysis method based on the historical passenger flow law. As shown in FIG. 2, the passenger flow state derived from FIG. 1 is combined with the information of the emergency event to filter out the affected passenger flow. After the affected passenger flow is processed according to the redistribution rule, the passenger flow index is counted to obtain the final calculation result.

The method includes the following steps:

Step 1: Obtain historical historical passenger flow OD data, and transfer and store the acquired data in a database.

The path impedance calculation method of multi-factor influence in the path ratio allocation module is as follows:

Based on the existing passenger integrated travel impedance calculation method, combined with the psychological process and judgment basis of the actual passenger travel choice, the present invention innovatively conducts passenger classification discussion, and respectively gives the corresponding comprehensive travel impedance calculation method, the corresponding parameters. The acquisition was explained.

(1) Time impedance

4 train running time

Train running time refers to the time passengers spend on urban rail transit trains. It consists of two parts: interval running time and stop time.

In the formula:

The train running time of the OD kth path from station a to station b; T _i,j is the running time of station i to station j on path k; T _s is the stopping time of the train passing through the intermediate station.

5 transfer time

The transfer time refers to the time spent by the passenger outside the transfer station train. It consists of two parts: transfer travel time and transfer waiting time.

Where: T _k ^tr is the total transfer time of the OD kth path from station a to b;

The transfer time from the m line to the n line on the path k;

The travel time for the transfer from the m line to the n line;

The waiting time for the transfer from the m line to the n line; α ₁ is the penalty factor of the transfer time.

The calculation formula for the transfer travel time is:

In the formula:

The distance traveled from the m line to the n line at the transfer station k,

The average walking speed of the transfer from the m line to the n line at the transfer station k during the t period.

The calculation formula for the transfer waiting time is:

Where: H ⁿ is the departure interval of the transfer line n. A large amount of statistical data shows that passengers arrive at a train interval independent of the train schedule, showing a random normal distribution. For the average waiting time of the overall passenger flow, the value will approach half of the driving interval. .

6 inbound and outbound time

The entry and exit time is:

In the formula:

They are the travel time from the stop gate to the starting station a, and the departure time from the stop to the terminal b.

(2) Congestion impedance

The degree of congestion is an important indicator of travel comfort, reflecting the sensitivity of passengers to congestion and the amplification of passengers' perception of travel time. There are generally two ways to indicate the degree of congestion: one is to obtain a corresponding amplification factor through questionnaires or evaluation methods; the other is to calculate the congestion level according to the train full load rate and seat situation.

First, the congestion degree of the compartment is divided into three levels according to the full load rate. Class 1 is the number of passengers in the train is less than the number of seats. There is no feeling of discomfort at this time; the second level is the number of people in the train between the number of seats and the passengers in the car. At this time, there will be a certain degree of congestion; the third level is the number of people in the train. More than the passenger compartment, the car is extremely crowded and the passengers feel very uncomfortable.

The formula for calculating the full load rate is as follows:

Where: δ is the train full load rate; P is the passenger flow, usually refers to the section passenger flow per unit time; D is the transport capacity, generally refers to the section transport capacity per unit time; n is the number of trains per unit time; Y is Vehicle capacity; B is the number of trains, Y*B is the entire train.

Congestion impedance:

In Q(δ), the congestion coefficient on a section of the rail transit network; 0, A, B correspond to three levels of congestion coefficient, A is the extra time overhead coefficient in general congestion; B is overcrowded Additional time overhead factor. δ ₀ is the full load rate when the number of passengers in the car is equal to the number of seats; when the number of passengers in the car is equal to the number of seats, the full load rate is 1.

(3) Transfer penalty

In urban rail transit, there are few cases where passengers travel OD just two stations on one line of rail transit. Generally, they need to transfer through two or more lines in the subway, or take the track section as the middle of the travel path. On the road, you will need other public transportation, bicycle transportation or transfer of walking traffic after you leave the station. Due to the increasing size of the city, people's average travel time is getting longer and longer, the number of transfer times is increasing, and the long transfer passage and long waiting time are also factors that people are not willing to transfer. Therefore, in addition to the time impedance described above and the comfort impedance of the passenger in the train, the number of passenger transfers on the road segment is also an important factor for the passenger to select the travel route.

Because passengers not only have to spend time on the transfer, but also need to spend the physical cost of the transfer passage, the cost of congestion, the spiritual cost of finding the road in the station, and the cost of the second waiting, so the passengers are in the same time. Passengers are more inclined to choose a path with fewer transfers. Especially for some elderly people or other special groups, they even have less time utility requirements than the number of transfer times. When some routes do not need to be transferred, passengers tend to choose this path; when they need one or more times When transferring, the probability that the passenger tends to choose the path becomes smaller, which is expressed as follows:

In the formula,

Is the number of transfers of the kth path between the wth OD pairs, and α ₂ is the transfer penalty coefficient in the above.

(4) Sharing problem

In the case of large passenger flow, over-peak or unexpected events in rail transit operations, some passengers may not be able to board the train because of overcrowding, but only consider the congestion degree impedance is that the train has sufficient transportation capacity, no matter how much Crowded passengers can board the train. Obviously, this assumption is not true in reality. In the rail transit of some big cities at home and abroad, congestion is more common due to the limitation of train capacity. To describe this problem, it is introduced. With the additional congestion factor γ, the additional congestion factor can be expressed in exponential form:

Where: η and

For the parameter; x _i is the passenger flow to the i station; c is the maximum passenger capacity of the train.

In addition to the additional congestion factor, the retention time is also related to the departure interval. It is a function of the additional congestion factor and the departure interval. The form is as follows:

In the formula:

The elapsed time for the i station; γ _i is the additional congestion factor of the i station;

The departure interval for the i-station to the n-line direction.

The total time of all stations is:

T ^γ =ΣT _i ^γ (10)

(5) Passenger integrated travel impedance

Considering comprehensively, the passenger integrated impedance function can be obtained, expressed as:

The effective path screening method in the path ratio allocation module is as follows:

An effective path screening method based on K short circuit algorithm.

(1) K short circuit search implementation

The algorithm is described as follows:

6 Using the Dijkstra algorithm to find the shortest path tree with the starting node s as the root in the directed graph (N, A), the shortest path between the starting node s and the ending node t is pk, k=1.

7 If k is less than the maximum number K of required shortest paths, and there are still candidate paths present, let the current path p = pk, turn 3. Otherwise, the program ends.

8 Find the first node in the current path p starting from the first node with an indegree greater than 1, denoted as nh. If the extended node n'h of nh is not in the node set N, then go to 4. Otherwise, find all the nodes behind nh in the path p, and the corresponding extended node is not the first node in N, denoted as ni, turn 5 .

9 Construct an extension node n'h for the node nh and add it to the set N. At the same time, connect all the nh predecessors in the graph (N, A) to an arc of n'h, and the weight corresponding to the arc is unchanged. , add these arcs to arc set A, except that nh is the previous node nh-1 in p. Calculate the shortest path from the start node s to n'h and record ni=nh+1.

10 For all subsequent nodes starting from ni in p, it may be denoted as nj, and the following operations are sequentially performed: adding the extended node n'j of nj to the node set N. Except for the previous node nj-1 of nj in path p, respectively, an arc from the nj precursor node to its extension node n'j is connected, the weights on the arc remain unchanged, and these arcs are added to arc set A. . In addition, if the previous node nj-1 of nj in p has an extended node n'j-1, it is also necessary to connect an arc from n'j-1 to n'j, the weight and the arc (nj-1, nj). The weights are equal. Calculate the shortest path from the start node s to n'j. The current shortest path tree is updated, and the shortest path between the current extended node t(k)' from the start node s to the end node is the kth shortest path, so k=k+1, and 2 continues.

(2) Determination of the effective path set

Among the K-small paths obtained by the path search algorithm, some unreasonable paths can be considered as passengers will not choose, do not participate in the distribution of passenger flow, and take into account the limitation of the operation time of different rail transit lines, which requires reasonable K-paths. The judgment is made to generate a valid path set.

3 operation time judgment

During a certain period of time, if one of the K optional trailing path sets is outside the running time, the path does not participate in the sharing of the passenger flow as a valid path and cannot be included in the effective path set. The operation time of the route can be expressed by the effective operation time of the starting station of the route. The effective operation time of the starting station is the first and last shift time of the starting station and the first and last shift times of the transfer stations in the path are reversed. Intersection.

4 travel impedance threshold judgment

Among the K-smooth paths obtained by the path search algorithm, some unreasonable paths can be considered that passengers do not choose and do not participate in the distribution of passenger flows. The validity test of the path is mainly judged by the travel impedance threshold. Suppose the K-optional set of progressive paths between the two stations, the impedance value of the shortest path

If the impedance value of the secondary short path or other shorter short path exceeds the certain range of the shortest path, the value exceeds a certain range (ie, is greater than

When it is considered that the short path or the second short path is unreasonable. Can reasonably assume that when

When it is small,

versus

In proportion to

When it is large enough, the upper boundary of the allowable area of the travel resistance value is fixed. It can be expressed as:

In the formula:

The upper bound of the travel impedance value for the effective path;

The maximum allowable value of the effective path exceeds the shortest path travel resistance value; ξ is a proportional coefficient; U is a constant. Their values can be determined by passenger travel surveys.

The multipath allocation method in the path ratio allocation module is as follows:

Multipath probability calculation is performed using a time impedance based multipath allocation method. The passenger flow distribution ratio of the path is determined based on the deterministic impedance of each path, that is, the time impedance, according to a certain statistical law and probability distribution model. When the elements of the effective path set are unique, the effective path assumes 100% of the passenger flow; when the elements of the effective path set are not unique, the problem of how the passenger flow is allocated in each path is generated.

Let the set of k valid paths between the two stations of the OD be

Select path

The probability is P _i (i=l,...,k). Obviously, P _i is a function of the integrated travel impedance of the path. Let the comprehensive travel impedance of each effective path be T _i ^f (i=l,...,k) and satisfy

Then for P _i has the following characteristics:

f.

g.

H.1≥P ₁ ≥P ₂ ≥...≥P _k ≥0, that is, the smaller the probability that the path with the larger impedance value is selected, the probability that the path of the smallest impedance value is selected is the largest.

i. If T _i ^f is very close to T ₁ ^f (ie the shortest path impedance value)

), I should be very close to the P _i T ₁ ^f, when the impedance in the vicinity of T ₁ ^f, the rate of decrease P _i is small. In other words, passengers are less sensitive to changes in ride time around T ₁ ^f .

j. As the impedance value increases, the deceleration rate of P _i will increase rapidly, that is, the probability that the path is selected will decrease rapidly. In fact, passengers are more sensitive to a larger extension of the ride time.

Through the passenger travel survey and analysis, it is also basically possible to verify the above characteristics.

Since the number of effective paths between ODs is uncertain, the proportion of passenger flow assignments of each path can be determined by calculating a utility value (S) in which each path participates in passenger flow sharing. The larger the utility value of the passenger flow allocation of the route, the larger the proportion of passenger flow distribution. It is assumed that the shortest path participates in the passenger flow sharing with the largest utility value and S=1. In general, the path passenger flow allocation utility is related to the extent x of the integrated travel impedance that exceeds the shortest path integrated travel impedance. The integrated travel impedance of the path exceeds the shortest path. The more the integrated travel impedance, the smaller the utility value of the path passenger flow distribution, and the smaller the proportion of the shared OD passenger flow. The path passenger distribution utility value (S) distribution pattern is similar to the normal distribution pattern. Considering that the normal distribution can well meet the above five requirements, and has been widely used in the statistical study of group behavior characteristics, a normal distribution is used to describe the passenger's travel path selection behavior. The formula of the normal distribution function is as follows:

Where: μ gives the x value of the maximum expected probability, here is 0; σ is a constant whose value will determine the steepness of the normal curve. Since it is impossible, the weight value T _i ^{f is} less than the minimum impedance value

The path, therefore, only needs to take the positive half of the normal distribution curve x ≥ μ. It can be considered that the parameter σ is a constant for all ODs. Its mathematical significance is very clear, and the fit can be analyzed by the results of the passenger travel survey. In general, the smaller the σ, the stronger the sensitivity of the passenger to the impedance. The passenger flow distribution ratio of the path is calculated by the following formula.

The road network passenger flow derivation method in the path ratio allocation module is as follows:

Based on the travel time road network passenger flow derivation model, through the previous multipath probability calculation, the probability that several paths between each specific OD pair in the road network are selected by passengers can be obtained. However, the travel process of passengers in the road network is a dynamic process that changes with time. It is only possible to completely grasp the full state of the passengers in the road network by relying on the static route selection method. Therefore, a road network passenger flow derivation model based on travel time is designed here.

Assume a full passenger ride behavior: from the passengers starting from the starting station to the station, until the passenger arrives at the terminal to swipe out, the passenger's travel status in the road network includes the travel time from the gate to the platform, waiting time, multiplication Car travel time, transfer time, transfer time, transfer time, bus time and travel time from the station to the gate.

Therefore, the derivation of passenger flow in the road network is as follows:

t _O refers to the time when the passenger swipes the card from the O station;

It is the waiting time of passengers at O station;

T _s is the stop time of the train at station S, and N is the station set;

It is the transfer time for passengers to change from line m to n at k station.

Using the above formula, combined with the passenger route and impedance matrix, the passengers can be inferred in the road network.

Step 3: Input the deduction result of step 2 into the affected passenger flow screening and redistribution module, and introduce the path information in the road network in the case where the line section of the road network is interrupted, and perform screening and redistribution calculation on the affected passenger flow. .

The classification and screening method of the sudden passenger flow is:

The invention divides the passenger flow in the road network into an unreachable passenger flow when the emergency occurs, the passenger flow that needs to be bypassed, the passenger flow whose service level is reduced, and the passenger flow that is not affected by the interruption interval.

(1) Passenger flow that cannot reach the destination

Interval interruption will make some passengers lose accessibility. When the road network is not interrupted, the passenger flow that cannot reach the destination only includes the passenger flow with the starting point or the defect inside the interruption interval; when the road network is interrupted into multiple sections When the sub-networks are not connected to each other, the passenger flow that cannot reach the destination includes the passenger flow of the different sub-networks in addition to the aforementioned passenger flow.

If the starting point of the trip is in the interruption zone, since the departure station is often in a closed state, this part of the passenger flow can only choose the ground bus travel, so this part of the passenger flow belongs to the passenger flow with sudden loss of time and needs to be removed from the total passenger flow;

If the trip is within the interruption interval, it needs to be discussed separately: if the accident is already on the initial path, the passengers will choose to continue to take the subway to the nearest station to the destination; if the accident occurs, the passenger will still If you have not set off, there are two options for this part of the passengers. One will directly abandon the rail transit mode, and the other will continue to take the subway to the nearest station to the destination. This situation should have been distributed in the stations within the interruption zone. Passengers at the station will choose to leave the station at the nearest station from the interruption zone, causing greater pressure on the station and requiring operational work.

When the trip point is located in a disconnected sub-network, if the smallest sub-network after splitting has a certain scale, then for most passengers, the nearest station that rail transit can reach is still far away from the destination. It is not conducive to short-distance transfer connections, so they will be more inclined to give up the choice of rail transit; if the smallest sub-network after splitting has only a few stations, then passengers will choose to take the subway to the nearest operating station to the destination. Then choose another mode of transportation. Since most of the existing urban rail transit networks are mature and have a large number of transfer stations and loops, the latter case should be chosen unless there is a large-scale road network failure.

(2) Passenger flow that needs to be detoured

There are often multiple reachable paths between each OD pair. Most of the passenger flow between the OD pairs is concentrated on the shortest path. If the shortest path passes through the interrupted interval, then this part of the passenger flow needs to select the second short path, the second time is short. Other alternative routes such as trails. This will result in a sharp increase in passenger traffic in some sections of the alternative route, resulting in an inability to adapt to the opening plan and an increase in the number of stations. Passengers also have a certain psychological limit when choosing a path. When the travel time of the alternative route is unacceptable, the passenger will choose other modes of transportation. This limit is related to the passenger's travel purpose, age, gender, and consumer preferences. The relationship between the increase in travel costs and the proportion of passenger flow losses needs to be obtained through detailed passenger flow surveys.

For the passenger flow that needs to be detoured, it needs to be divided into the passenger flow that has not entered the road network at the time of the accident; the passenger flow has been located in the road network after the accident, but has not yet reached the passenger flow in the interruption zone; and has been located in the road network after the accident has passed The passenger flow in the interruption interval. Without considering the loss of passenger flow caused by the increase in travel costs, in the first case, passengers will directly choose alternative routes to travel; in the second case, in the case of complex road networks, most passengers cannot remember the road network structure. Heart, they know that after the accident, the passenger will often get off at the next transfer station with the help of the station staff or the road network diagram to re-select the path, then the next transfer station becomes the new O station; the third Passengers in the situation are not affected by unexpected events.

(3) Increase in the congestion of the road and the decline of the service level

Due to the interruption of some sections, the passenger flow that needs to be interrupted by the path needs to find an alternative path to bypass and arrive. The purpose of the station, this will result in a substantial increase in the passenger flow of the alternative route, thus affecting the waiting time and ride comfort of passengers traveling normally. In the urban rail transit system, passengers can only know the passenger flow of the platform and the comfort of the ride after buying the ticket. Studies have shown that few passengers change their initial travel routes because of changes in congestion, and passengers will pay more time and physical cost when they change their travel routes. Therefore, the tendency of this part of the passenger flow in the path selection will not be affected.

(4) Passenger flow that is not affected by the interruption zone

The intensity and duration of an emergency determine the extent of the impact of the accident on the road network. After the accident, the scope of the impact is gradually decreasing over time. The passenger flow outside the initial impact range and the station is not affected by the emergency, and the passenger flow characteristics of this part of the passenger flow do not change compared with the daily situation.

The passenger flow redistribution calculation method in the case of the emergency event is:

From the point of view of passenger travel, considering the actual impact of interval interruption on passenger travel, the affected passenger flow can be divided into two categories: OD passenger flow that cannot reach the destination, and OD passenger flow whose travel path changes. They have different treatment methods. .

(1) OD passenger flow that cannot be transferred to the destination by rail transit

Point O is in the passenger flow within the interruption interval, directly deletes the relevant data from the table, does not participate in the passenger flow allocation; passengers who have not boarded the vehicle after the accident and passengers who have already been on the initially selected route at the time of the accident, according to the originally selected path Go to the last stop of the reach and change this station to D station for passenger flow distribution.

(2) OD passenger flow with changed travel path

Passengers who have not yet boarded the vehicle after the accident are assigned according to the new route selection ratio; passengers who have already been on the originally selected route at the time of the accident: if the passenger is at the station, change the station to O, follow the new route The ratio is assigned; if the passenger is in the interval, the next transfer station is changed to O to match the flow according to the new route selection ratio.

There is accessibility between the location and destination station, which means that the original route is destroyed. If the passenger is at the station, the station is used as the new O station. If the passenger is on the range A->B, then As the new O station is adjusted to the alternative line, the B station needs to generate a new distribution ratio. From equation (14) we can obtain the utility value S _i of an OD effective path, assuming that an OD effective path set is N, i=1, 2,...n, the interval interrupt causes the path k to be unavailable, according to the utility The principle of constant value, the remaining route passenger flow distribution ratio is:

According to the actual situation of the interval operation interruption, the interruption interval is set to the unavailable state in the basic road network: all the paths passing through the interruption interval are deleted in the full OD passenger flow distribution path of the normal network, and the interval operation interruption is regenerated according to the above principle. Under the condition, the multi-path passenger flow distribution ratio of the road network OD, and the calculation of the network dynamic distribution passenger flow, the steps are as follows:

1 All the OD pairs in the original OD table are assigned, and the distribution results are stored in the “OD distribution intermediate table”;

2 Screen out the OD pairs affected by the interruption and save the result set;

3 Delete the record associated with the “OD pair of affected passenger flow” in the “OD distribution intermediate table”;

4 Delete the passenger flow record of the O station in the interruption interval in the “OD allocation intermediate table”;

5 Recalculate the effective path of the road network and the distribution ratio, and redistribute the “affected passenger flow OD pair”, and store the result in the “OD distribution intermediate table”;

6 Classification and statistics of the affected passenger flow.

Step 4: Calculate the impact range of the emergency event by using the accident information and the road network structure, input the updated path allocation ratio, reload the affected passenger flow, and finally calculate the relevant passenger flow index. The output result is:

1 The number of stranded people in the interruption interval and their changes over time;

○22 The amount of inbound and outbound stations at each station in the affected area, the transfer amount of the transfer station, and the section passenger flow in the interval and its changes with time.

The calculation method of the network passenger flow index in the case of the emergency event is:

From several levels of road network, line, interval and station, the index system of passenger flow in rail transit network is established according to different time periods:

Passenger flow at the station: 5 minutes of passenger flow, outbound passenger flow, passenger flow (transfer station), and the number of stranded persons;

Section passenger flow: 5 min section passenger flow.

1 pair of affected stations outside the interruption zone

In the microscopic case, considering the actual operation of the train, we can calculate the position of the passengers of any OD in the road network at each moment. In fact, it is possible to simulate the actual passenger flow changes of the road network, so The passenger flow outside the interruption interval, the number of passengers entering the station, the number of outbound stations, the number of people transferred to the transfer station, and the number of exchanges at the transfer station can be directly obtained from the OD distribution intermediate table, where a is the length of the statistical period. , the same as the AFC statistical system granularity can be set to 5min.

In(ta,t) _sta = number of inbound credits,

Out(ta,t) _sta = outbound traffic,

TranIn(ta,t) _sta = station exchanges passenger traffic,

TranOut(ta,t) _sta = The station exchanges passenger traffic.

The affected section Sec located outside the interruption interval can also be directly obtained from the OD distribution intermediate table, where a is the length of the statistical period.

Sec(t) _{sta, sta-1} = cross-sectional traffic.

2 pairs of stations located in the interruption zone

Calculation of the number of stranded persons in the station:

Where: Retention(t) _i refers to the number of stations in station i during t-hours;

Section(t) _ij refers to the section passenger flow of the interval ij of the t period;

In(t) _i is the number of inbound stations at station i during t-hours;

Out(t) _i is the number of outbound stations _i at t time;

TranIn(t) _i is the number of people who have entered the station i during the t period;

TranOut(t) _i is the number of people who exchanged station i during t time.

The following is an example of the Beijing-Guangzhou railway network, and an example is given to further illustrate the analysis method of passenger flow impact under the interruption of part of the road network.

On the basis of the Beijing urban rail transit road network, the OD distribution algorithm of the middle view is used to calculate the specific OD to verify the method. The data and parameters used are divided into three categories.

Road network basic data, including: station table, line table, interval table.

Road network passenger flow data: Select the passenger flow data of the Beijing subway on July 22, 2013 as the research object, the required data includes: OD schedule, entry and exit passenger flow statistics table, transfer amount statistics table, train operation schedule, change Take the travel schedule and the cross-section passenger flow table.

Computational parameters: draw on the parameters of the “Beijing Municipal Rail Transit Automatic Ticketing System Clearing Management Center (ACC) Clearing Method Research” and the document “Beijing Rail Transit Networked Traffic Organization Research Work Report”, where σ is The value is 0.25, the value of U is 10, the value of θ is 60%, and the values of α ₁ and α ₂ are both 1.5.

The model is modified based on time impedance to obtain the multipath assignment probability. The network basic data involved (including interval running time, stop time, line departure interval, transfer travel time, etc.) are taken from the actual operation of Beijing rail transit. The data and other related parameters are set and analyzed by passenger flow survey.

Set the interruption interval: Line 1 Gongzhufen - Xidan; Interruption time: 9:10-9:25 am. The interruption lasts for 15 minutes. The simulation results are as follows: The statistical data of the road network related data under normal operating conditions are shown in Table 1 and Table 2.

Table 1 Daily passenger traffic statistics for each line

The road network transfer coefficient is the ratio of the transfer volume of the entire road network to the passenger traffic volume of the line. It is calculated that the transfer amount in the entire road network accounts for 45% of the total passenger traffic.

Table 2 Maximum section passenger flow of each line and its interval and time period statistics

The historical OD data is distributed under the normal operating road network. The maximum cross-section passenger flow of each line is calculated every 5 minutes, and the corresponding interval and time period are obtained. It can be seen that the maximum cross-section passenger flow of most lines appears during the morning peak period.

Impact analysis in the event of an interruption: impact range calculation.

When the interruption interval is from Xidan to Gongzhufen, the interruption duration is 15 minutes and 30 minutes, and the affected station range pairs are shown in Table 3.

Table 3 Comparison of affected station ranges

It can be seen from Table 3 that the interruption time of the interrupt interval is increased at any time, and the influence range of the emergency event is increased. When the time is interrupted In only 15 minutes, the affected lines include Line 1, Line 2, Line 4, Line 9 and Line 10. When the duration reaches 30 minutes, the line 6 is connected to the rest of the line. The station was also affected. At this time, the corresponding affected stations should be prepared to cope with the large passenger flow in time, and timely release PIS information to passengers to effectively guide the passenger flow.

Passenger flow statistics in the event of an interruption:

The interruption interval is the No. 1 line Gongzhufen-Xidan two-way interruption, the corresponding Apple Orchard-Princess Tomb, Xidan-Sihuidong is operated by small traffic.

During the interruption period, the stations whose inbound and outbound quantities have changed are as follows:

Table 49: Statistics of affected stations and their inbound and outbound stations at 05-9:30

From the point of view of the number of inbounds, Wangfujing and Beijing Station have more than doubled the number of inbounds in the normal situation, so the flow restriction control should be carried out at the station entrance port; from the outbound capacity, Beijing station In the case of the interruption of Tiananmen West and Dawang Road The outbound quantity is more than doubled compared with the normal situation and should be promptly diverted. From the overall analysis of the inbound and outbound volume, the impact level of each station can be obtained. The stations with significant impacts include Gongzhufen, Tiananmen West, Wangfujing, Dawang Road, Beijing Railway Station and Beijing West Railway Station.

Detention in the interruption interval: The number of stranded persons here is the worst estimate. It is based on the actual number of stranded persons in the station plus the demand for inbound passengers in each period. A maximum estimate of the demand for passenger flow that needs to be diverted by other vehicles during the time period. It can be seen from the above statistical results that the propagation of the affected stations in the road network is gradually spread around the interruption interval. The interval farther from the interruption interval is less affected than the relatively close interval, and the time of influence is delayed. In the incident, the interruption interval is located in Gongzhufen-Xidan on Line 1, so the passenger flow on Line 1 is the most affected; the station within the interruption interval, where the Military Museum is the transfer station of Line 1 and Line 9, rejuvenation The gate is the transfer station of Line 1 and Line 2, and the Xidan is the transfer station of Line 1 and Line 4, so the stations near the station in the interruption zone of Lines 2, 4 and 9 are also affected, and Gradually extend to the line; the remaining line sections are less affected. The operation management department needs to timely determine the new traffic scheduling plan, as well as the guidance and current limiting measures according to the interruption interval, the interruption time, and the affected range. The above interrupt time lasts for 15 minutes. It is also possible to change the interrupt start time, interrupt duration, interrupt interval, etc. in the system. If the interrupt start time is 11:35 and the interrupt time lasts for 40 minutes, it can be seen that the interrupt duration is longer. Long, the scope of influence is greater.

The analysis results of passenger flow impact under the partial interruption of the road network can reveal the capacity bottleneck of the road network under the condition of interruption, and provide the decision-making basis for the operational management department to adopt targeted driving organization and passenger transportation organization.

The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the specific details of the embodiments described above, and various modifications may be made to the technical solutions of the present invention within the scope of the technical idea of the present invention. These simple variations are within the scope of the invention.

In addition, the specific technical features described in the above specific embodiments may be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present invention will not be further described in various possible combinations. In addition, any combination of various embodiments of the invention may be made as long as it does not deviate from the idea of the invention, and it should be regarded as the disclosure of the invention.

Claims

A method for estimating a passenger flow of a city rail network, characterized in that the method comprises the following steps:

Step 1: Obtain historical historical passenger flow OD data, and transfer and store the acquired data in a database;

Step 2: Define the path impedance and the effective path in the road network in the path ratio allocation module, establish a path search algorithm to obtain an effective path between the ODs, and use the utility theory to calculate the multipath selection between the OD pairs by using the impedance theory. The probability of inputting the historical synchronous passenger flow OD data obtained in step 1 to the path ratio distribution module loads the passenger flow in the daily situation and outputs the derivation result of the daily passenger flow.
The urban rail network passenger flow estimation method according to claim 1, wherein the path impedance calculation method in the path ratio allocation module in step 2 is as follows:

(1) Calculate the time impedance

Calculate train running time

Train running time refers to the time passengers spend on urban rail transit trains, including interval running time and stop time:

In the formula:
The train running time of the OD kth path from station a to station b; T i,j is the running time of station i to station j on path k; T s is the stop time of the train passing through the intermediate station;

Calculate transfer time

The transfer time refers to the time spent by the passenger outside the train of the transfer station. It consists of two parts: transfer travel time and transfer waiting time:

In the formula:
The total transfer time for the ok kth path from a station to b;
The transfer time from the m line to the n line on the path k;
The travel time for the transfer from the m line to the n line;
The waiting time for the transfer from the m line to the n line; α 1 is the penalty coefficient of the transfer time;

The calculation formula for the transfer travel time is:

In the formula:
The distance traveled from the m line to the n line at the transfer station k,
The average walking speed of the transfer from the m line to the n line at the transfer station k during the t period;

The calculation formula for the transfer waiting time is:

Where: H n is the departure interval of the transfer line n;

Calculate inbound and outbound time

The entry and exit time is:

In the formula:
The travel time from the stop gate to the start station a platform, and the departure time from the stop to the terminal b;

(2) Calculate the congestion impedance

The formula for calculating the full load rate is as follows:

Where: δ is the train full load rate; P is the section passenger flow per unit time; D is the section transport capacity per unit time; n is the number of trains per unit time; Y is the vehicle capacity; B is the number of trains, Y *B is the entire train capacity;

The congestion degree impedance is:

In the Q(δ) formula, the congestion coefficient on a section of the rail transit network; 0, A, B correspond to three levels of congestion systems respectively

Number, A is the extra time overhead factor when it is generally crowded; B is the extra time cost factor when overcrowding; δ 0 is the full load rate when the number of passengers in the car is equal to the number of seats; when the number of passengers in the car is equal to the capacity, the full load rate Is 1;

(3) Calculate the transfer penalty

Formulated as follows:

In the formula,
Is the number of transfers of the kth path between the wth OD pairs, and α 2 is the transfer penalty coefficient in the above;

(4) Calculate the retention time

The additional congestion factor is expressed in exponential form:

Where: η and
For the parameter; x i is the passenger flow to the i station; c is the maximum passenger capacity of the train;

The time of the I station is:

In the formula:
The elapsed time for the i station; γ i is the additional congestion factor of the i station;
The departure interval for the i-station to the n-line direction;

The total time of all stations is:

(5) Calculate the passenger's comprehensive travel impedance

The passenger integrated impedance function is expressed as:
The urban rail network passenger flow estimation method according to the preceding claim, wherein the effective path screening method in the path ratio allocation module in step 2 is as follows:

(1) Using K short circuit search, the algorithm is described as follows:

The Dijkstra algorithm is used to obtain the shortest path tree with the starting node s as the root in the directed graph (N, A). The shortest path between the starting node s and the ending node t is pk, k=1;

If k is less than the maximum number K of the required shortest path, and there is still a candidate path present, let the current path p=pk, turn ○3 3 ; otherwise, the program ends;

Find the first node in the current path p from the first node with the indegree greater than 1, denoted as nh; if the extended node n'h of nh is not in the node set N, then turn to ○4 4 , otherwise find out Among all the nodes behind nh in path p, the corresponding extended node is not the first node in N, which is denoted as ni, turn ○ 5 5 ;

Construct an extension node n'h for the node nh and add it to the set N. At the same time, connect all the nh predecessors in the graph (N, A) to an arc of n'h, and the weight corresponding to the arc does not change. Add these arcs to arc set A, except that nh is the same as the previous node nh-1 in p; calculate the shortest path from the start node s to n'h, and record ni=nh+1; for p from ni All subsequent nodes, may be recorded as nj, in turn perform the following operations: add nj extended node n'j to node set N; except for the previous node nj-1 of nj in path p, respectively connect a nj precursor node To the arc of its extension node n'j, the weights on the arc remain unchanged, and these arcs are added to the arc set A; in addition, if the previous node nj-1 of nj in p has the extension node n'j- 1 , you also need to connect an arc from n'j-1 to n'j, the weights and arcs (nj-1, nj) have the same weight; calculate the shortest path from the starting node s to n'j; update The current shortest path tree, finds that the shortest path between the current extended node t(k)' from the starting node s to the ending node is the kth shortest path, so that k=k+1, and ○2 2 continues;

(2) Determine the effective path set:

Operation time judgment

During a certain period of time, if one of the K optional trailing path sets is outside the operating time, it is not included in the effective path set; the running time of the path is represented by the effective running time of the starting station of the path. The effective operation time of the starting station is the intersection of the first and last shift time of the starting station and the first and last shift time of each transfer station in the path, and the starting time of the starting station is reversed;

Travel impedance threshold judgment

The unreasonable path in the K-straighed path obtained by the path search algorithm does not participate in the distribution of the passenger flow; the validity test of the path is judged by the travel impedance threshold; assuming that the K-selectable progressive path sets between the two stations are The impedance value of the shortest path
If the impedance value of the secondary short path or other shorter short path exceeds a certain range of the travel impedance value of the shortest path, the short path or the second short path is considered unreasonable;
When it is small,
versus
In proportion to
When it is large enough, the upper bound of the allowable area of the travel impedance value is fixed; expressed as:

In the formula:
The upper bound of the travel impedance value for the effective path;
The maximum allowable value of the effective path exceeds the shortest path travel resistance value; ξ is a proportional coefficient; U is a constant.
The urban rail network passenger flow estimation method according to the preceding claim, wherein the multi-path allocation method in the path ratio allocation module in step 2 is as follows:

Let the set of k valid paths between the two stations of the OD be
Select path
The probability is P i (i=l,...,k); obviously, P i is a function of the integrated travel impedance of the path; the comprehensive travel impedance of each effective path is
And satisfy
Then for P i has the following characteristics:

That is, the sum of the proportions of all valid route passenger flow assignments between a pair of stations is equal to 1;

Then P i = P j , that is, paths with equal impedance values are selected with equal probability;

1≥P 1 ≥P 2 ≥...≥P k ≥0, that is, the smaller the probability that the path with the larger the impedance value is selected, the probability that the path of the smallest impedance value is selected is the largest;

If
very close
(ie, the shortest path impedance value
), then P i should be very close
When the impedance is
When nearby, the rate of decline of P i is small;

As the impedance value increases, the deceleration rate of P i will increase rapidly, and the probability that the path is selected will decrease rapidly;

The normal distribution is used to describe the passenger's travel path selection behavior. The formula of the normal distribution function is as follows:

Where: μ gives the x value of the maximum expected value of the probability, here is 0; σ is a constant whose value determines the steepness of the normal curve;

The passenger flow distribution ratio of the path is calculated by the following formula:
The urban rail network passenger flow estimation method according to the preceding claim is characterized in that, in the path ratio distribution module in step 2, the road network passenger flow derivation method is as follows:

Where t i refers to the time point when the passenger departs from the O station to reach the i station;

t O refers to the time when the passenger swipes the card from the O station;

It is the waiting time of passengers at O station;

T ab is the running time of the train in the interval ab, and M is the interval set;

T s is the stop time of the train at station S, and N is the station set;

It is the transfer time for passengers to change from line m to n at k station.