An optimization method for random probability distribution model of urban rail transit passenger flow

Abstract

The present invention discloses a method for optimizing a random probability distribution model of urban rail transit passenger flow, including: optimizing the generalized travel impedance in the random probability distribution model of urban rail transit passenger flow by analyzing the travel behavior, travel process and influencing factors of passengers in the urban rail transit system; optimizing the effective path of passenger travel according to the optimized generalized travel impedance; optimizing the structure of the random probability distribution model of urban rail transit passenger flow based on the fully observable control theory to reduce the deviation of the distribution ratio of passenger flow in the section. Based on the theory that the spatiotemporal distribution of passenger flow in the urban rail transit network is fully observable and controllable, the present invention analyzes the main factors affecting the random probability distribution of passenger flow, optimizes the random probability distribution model of urban rail transit passenger flow, and lays a foundation for realizing a standardized, scientific and systematic urban rail transit clearing method.

Description

Technical Field

[0001] The present invention relates to the technical field of urban rail transit passenger flow sorting, and in particular to an optimization method for a random probability distribution model of urban rail transit passenger flow based on a fully observable control theory. Background Art

[0002] Urban rail transit is characterized by being fast, punctual, safe, and large in capacity, and plays an increasingly important role in the comprehensive transportation system. With the increase in urban rail transit lines, the network scale has gradually increased, and the connectivity between stations has increased. The network passenger flow has shown characteristics that are completely different from non-networked passenger flow. Passengers have more and more path choices between OD and travel. How to fully grasp the passenger travel preferences and travel patterns of the urban rail transit network, and then conduct scientific and accurate network passenger flow allocation is an important basis for implementing passenger flow statistical analysis, passenger flow forecasting, transportation planning, passenger transportation organization, and other work.

[0003] At present, many scholars have studied the passenger flow allocation methods of urban rail transit. Mao Baohua et al. summarized that the heuristic methods for passenger flow allocation in urban rail transit networks mainly include three methods: all-or-nothing, proportional allocation method and tolerance allocation method. All-or-nothing allocates all the demand between ODs to the shortest path between OD pairs, while other paths cannot be allocated passenger flow; the proportional allocation method considers the impact of passenger flow on costs and is a flow allocation method that adjusts the passenger flow of the line network according to the change of generalized travel impedance. On the basis of the above two methods, the tolerance allocation method adds a comparison process of the newly allocated and originally allocated interval flow, and the newly calculated and originally calculated interval costs. These methods are based on the shortest path, and perform passenger flow allocation from the perspective of the supply and demand balance of rail transit, without considering the passengers' travel willingness.

[0004] In recent years, many scholars have begun to study the psychological activities and intentions of passengers in the selection of effective paths between OD pairs. By determining the generalized travel impedance for the effective paths of any OD pair in the urban rail transit network, a Logit model is established to determine the probability of each effective path being selected, which is generally called a random probability allocation model. Lin Zhan et al. used the relative cost of the path instead of the absolute cost of the path to improve the traditional Logit model and proposed a passenger flow allocation method for the urban rail transit network based on the improved Logit model. Cai Xiaochun revised the Logit model based on the two problems that the traditional Logit model did not consider the characteristics of regular passengers in terms of parameters and did not adjust the parameter size according to real-time dynamic conditions, making the model more consistent with the actual situation. Sun Luanying et al. proposed the concept and related algorithms of the network random loading model, and used examples to verify the rationality and feasibility of the improved Logit model in the application of urban rail transit passenger flow allocation. Zhou Weiteng et al. constructed a simulation model for the dynamic distribution of urban rail transit passenger flow, used the improved Logit model to select the path of the main body of passengers, and made short-term predictions of the passenger flow distribution of the urban rail transit network. Jing Wuxia studied the subway clearing model, proposed the theoretical basis for the probability of effective path selection and the calculation method of the probability of effective path selection, and gave examples for verification. Yao Xiangming et al. constructed a dynamic simulation model of network passenger flow distribution based on Multi-Agent modeling technology. In the simulation model, it is assumed that the random error of the estimated path impedance value by travelers follows the Gumbel distribution, and constructed a passenger travel path selection model based on the Logit model to simulate the passenger's network travel process. Based on the algorithm of undirected connected graphs in graph theory, Lu Chunjiang used the "shortest time method" to determine the path that most passengers are willing to choose, and calculated the path allocation ratio according to the probability of passengers choosing the path.

[0005] In addition to the random probability allocation model, according to the passenger flow allocation theory and principles, rail transit passenger flow allocation also includes a user equilibrium allocation model. Chen Boyang et al. constructed a path generalized cost model based on the subway space-time expansion network, proposed a subway network passenger flow equilibrium allocation model based on the timetable, and gave a solution algorithm. Based on the stability of the timetable, Szeto proposed a linear complementary equation to solve the random allocation problem considering capacity and additional browsing of the section, and proposed a random stability equilibrium condition for travel time.

[0006] The above literatures have conducted rigorous and detailed research on the passenger flow allocation method of urban rail transit, and provided theoretical basis and algorithm process. However, in the research process, the sensitivity of passengers to the choice of path and the correlation between the generalized travel cost value of the shortest path were not taken into account, which would result in the same decrease in the probability of selection of other paths except the first path regardless of the generalized travel cost value of the shortest path between OD (travel origin and destination).

[0007] The temporal and spatial distribution of urban rail transit passenger flow in the rail transit network is fully observable and controllable, so by adjusting the control variables such as model parameters, the causal relationship between the control variables and the controlled variables (section passenger flow, section full load rate, transfer volume, etc.) can be mastered. Based on this theory and the problems existing in the current urban rail transit network passenger flow path allocation method, the present invention establishes a network model architecture, analyzes the main factors affecting the random probability distribution of passenger flow, optimizes the passenger flow random probability distribution model, adopts a dynamic passenger flow equilibrium distribution method, and combines the passenger flow loading method based on multi-agent simulation technology to realize the temporal and spatial distribution of passenger flow between different facilities, stations, sections and lines, and then verifies the model optimization effect, providing a scientific method and basis for network operation management. Summary of the invention

[0008] The purpose of the present invention is to provide an optimization method for the random probability distribution model of urban rail transit passenger flow. Based on the theory that the spatiotemporal distribution of passenger flow in the urban rail transit network is fully observable and controllable, the main factors affecting the random probability distribution of passenger flow are analyzed, and the urban rail transit passenger flow classification model framework is established to optimize the micro-path selection behavior and decision-making process of passenger travel.

[0009] To solve the above technical problems, the embodiments of the present invention provide the following solutions:

[0010] An optimization method for a random probability distribution model of urban rail transit passenger flow includes the following steps:

[0011] By analyzing the travel behavior, travel process and influencing factors of passengers in the urban rail transit system, the generalized travel impedance in the random probability distribution model of urban rail transit passenger flow is optimized;

[0012] According to the optimized generalized travel impedance, the effective path of passenger travel is optimized;

[0013] Based on the fully observable control theory, the structure of the random probability distribution model of urban rail transit passenger flow is optimized to reduce the deviation of passenger flow distribution ratio in section.

[0014] Preferably, the travel behavior includes: travel starting and ending points, and travel time; the travel process includes: the process of entering the urban rail transit station, waiting, boarding, transferring, waiting, boarding, and exiting the station; the influencing factors include: interval running time, stop time, departure interval, waiting time, transfer preference, and passenger travel purpose.

[0015] Preferably, the generalized travel impedance refers to the factors affecting the path selection being uniformly converted into travel time, which is called the generalized travel impedance; the optimization of the generalized travel impedance specifically includes:

[0016] The generalized travel impedance includes interval impedance and station impedance; the interval impedance is represented by the running time of the train in the interval; the station impedance is the time spent by the passengers at the station; the impedance of the station is the stop time of the train, and for the transfer station, the time spent includes the transfer travel time and the transfer waiting time. The transfer station impedance is represented by the transfer time multiplied by a transfer magnification factor α, α≥1, that is, the transfer time is converted into the same sense of the ride time through a transfer magnification factor;

[0017] The interval impedance calculation method is as follows:

[0018]

[0019] Where:

[0020] A represents the generalized travel impedance of a certain path in a certain period of time;

[0021] n represents the number of intervals in a path; t i represents the interval running time of the i-th interval;

[0022] The impedance calculation method of the station is as follows:

[0023]

[0024] Where:

[0025] It represents the generalized travel impedance of all stations along a route in the period of d days and p;

[0026] s represents the total number of stations passed by a certain route;

[0027] It represents the train stop time at the jth station during the d-day p-period, where d is weekdays, weekends, holidays, and p is peak or off-peak.

[0028] The transfer station impedance calculation method is as follows:

[0029]

[0030] Where:

[0031] represents the generalized travel impedance of all transfer stations in a certain path in d days and p periods;

[0032] R represents the total number of transfer stations in a certain route;

[0033] α h The transfer magnification factor representing the travel time at the h-th transfer station;

[0034] represents the travel time of the hth transfer station in the d-day p-period;

[0035] It represents the waiting time of the hth transfer station in the d-day p-period, which can be taken as half of the departure interval of the transfer line train;

[0036] The generalized travel impedance of a path is calculated as follows:

[0037]

[0038] Where:

[0039] C d，p It represents the generalized travel impedance of a path in d days and p periods.

[0040] Preferably, the effective path is a physical path between ODs for passengers to choose, determined based on the topological structure of the urban rail transit network and in combination with the time rationality principle of passengers' generalized travel impedance selection between ODs; the optimization of the effective path for passenger travel according to the optimized generalized travel impedance specifically includes:

[0041] Set effective path selection rules:

[0042] Paths that exceed the generalized travel impedance of the shortest path by too much will not be chosen by passengers;

[0043] The probability of a path being selected is inversely proportional to the size of the path's generalized travel impedance;

[0044] The number of valid paths between ODs is uncertain and is greater than or equal to 1;

[0045] Based on the above conclusions, a threshold is set according to the shortest generalized travel impedance path. When the generalized travel impedance of a path is less than the threshold, it is selected as a valid path, otherwise it is an invalid path.

[0046] Add the following constraints to the above valid path selection rules:

[0047] Constraint 1: Backtracking is not allowed, and each node in the path cannot be visited repeatedly, that is, there is no loop, and all paths are elementary paths;

[0048] Constraint 2: Backtracking is not allowed, and each arc in the path cannot be visited repeatedly, that is, the same arc cannot appear repeatedly in each path;

[0049] Constraint 3: The generalized travel impedance of the path between ODs cannot exceed the generalized travel impedance threshold between ODs;

[0050] Among the K optional gradually shortening paths between two stations, it is assumed that the generalized travel impedance value of the shortest path is If the comprehensive impedance value of the second shortest path or other shorter paths exceeds a certain range compared with the comprehensive impedance value of the shortest path, that is, T range , the range can be a relative range m and an absolute range U. In this case, the shortest path or the second shortest path is considered unreasonable, which can be expressed as the following formula:

[0051]

[0052] Where:

[0053] T range Indicates the maximum allowable value of the effective path exceeding the generalized travel impedance value of the shortest path;

[0054] m represents the proportionality coefficient;

[0055] U represents a constant;

[0056] The choice of the values ​​of m and U is Related, that is Within a certain threshold range, the values ​​of m and U are different, so T range The optimization is a piecewise function, and the specific expression is as follows to express the influence of the generalized travel impedance of the shortest path for passengers on the effective path selection:

[0057]

[0058] Where:

[0059] m1, m2 represent the proportionality coefficient;

[0060] U1, U2 represent constants;

[0061] T represents the threshold of the shortest path generalized travel impedance;

[0062] The values ​​of m1, m2, U1, and U2 need to be determined comprehensively after statistical analysis of the survey results in combination with OD passenger flow surveys as the line network structure and operating methods change.

[0063] Preferably, the optimization of the structure of the random probability distribution model of urban rail transit passenger flow based on the fully observable control theory specifically includes:

[0064] The original urban rail transit passenger flow random probability distribution model constructed includes:

[0065] Determine the effective paths involved in the clearing according to the size of the generalized travel impedance of the possible paths between the travel origin and destination OD, and calculate the passenger flow allocation ratio of each effective path according to the normal distribution model;

[0066] The normal distribution model is used to describe the travel path selection behavior of passengers. The formula is as follows:

[0067]

[0068] Among them: The random variable x is taken as

[0069] a is the x value that obtains the maximum expected value of probability, which is 0 here; σ is a constant, and the value of σ determines the steepness of the normal curve; since there is no weight Less than the shortest path generalized travel impedance Therefore, we only need to take the positive half of the normal distribution curve x ≥ a. Here, we assume that the parameter σ is a constant for all the origin and destination points OD. The passenger flow distribution ratio of the path is calculated by the following formula:

[0070]

[0071]

[0072] Where S i is the benefit value of a certain effective path, P i is the passenger flow distribution ratio of the effective path;

[0073] The passenger flow distribution state generated by the current normal distribution model structure and parameter combination causes deviations in the distribution ratio of passenger flow on the cross section. The reasons are:

[0074] The normal distribution model does not distinguish the boundary value of the generalized travel impedance of the shortest path, and uses σ as a constant for all OD pairs, resulting in a problem that the allocation ratio of the second or third path is low for OD pairs with a large generalized travel impedance of the shortest path, thereby causing a deviation in the distribution ratio of passenger flow on the cross section;

[0075] The optimization of the random probability distribution model of urban rail transit passenger flow specifically includes:

[0076] According to the value of σ and That is, the passenger's sensitivity to path selection is related to the generalized travel impedance value of the shortest path. The normal distribution model is optimized into a piecewise function model. The boundary is divided into two sections, as shown in the following formula:

[0077]

[0078]

[0079] Where: S 优化 is the benefit value of a certain effective path after optimization; σ1, σ2 are the sensitivity of passengers to the path selection; T is the limit value of the generalized travel impedance of the shortest path, T range It is the maximum allowable value by which a valid path exceeds the generalized travel impedance of the shortest path.

[0080] Preferably, the smaller the value of σ is, the higher the passenger's sensitivity to route selection is.

[0081] Preferably, the value of the piecewise function model is as follows:

[0082] when σ takes the value of 0.2;

[0083] when σ takes the value of 0.25;

[0084] when σ takes the value of 0.4;

[0085] σ1 and σ2 in the piecewise function model represent different values ​​of σ.

[0086] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0087] In the embodiment of the present invention, relying on the passenger travel data source and guided by business needs, based on the fully observable control theory, a method for optimizing and improving the random probability distribution model of passenger flow at key nodes in the urban rail transit clearing process is proposed; by verifying the effective path passenger flow proportion distribution and the line section passenger flow of the optimization results, it is proved that the optimization method proposed in the present invention has achieved the expected goal, laying a solid foundation for realizing a standardized, scientific and systematic urban rail transit passenger flow clearing method. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0089] Figure 1 It is a flow chart of a method for optimizing a random probability distribution model of urban rail transit passenger flow provided by an embodiment of the present invention;

[0090] Figure 2 1 is a schematic diagram of passenger flow at peak time section between Dongzhimen and Dongsishitiao on Line 2 provided in an embodiment of the present invention;

[0091] Figure 3 Schematic diagram of three routes from Huilongguan East Street to Chaoyangmen provided by an embodiment of the present invention;

[0092] Figure 4 Schematic diagram of two routes from Shunyi to Xi'erqi provided in an embodiment of the present invention;

[0093] Figure 5 It is a schematic diagram of the path probability distribution of OD pairs with different shortest path values ​​provided by an embodiment of the present invention;

[0094] Figure 6 It is a schematic diagram of the magnitude of the decrease in the allocation ratio of each pair of ODs to two paths provided by an embodiment of the present invention;

[0095] Figure 7 is a normal distribution diagram of effective path benefit values ​​provided by an embodiment of the present invention;

[0096] Figure 8 It is a schematic diagram of the effective path allocation ratio from Caofang to Xi'erqi provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0097] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0098] The process of urban rail transit clearing is to allocate passengers to specific paths according to their OD, which is called passenger flow allocation. The essence of passenger flow allocation is the interaction, propagation, influence and evolution process of individual flows formed by individual passengers with autonomous consciousness and behavioral characteristics on the network. It is a method to infer the passenger flow distribution of the line network. The basic principle of the clearing model is to determine the effective paths participating in the clearing according to the size of the generalized travel impedance (or generalized travel cost) of possible paths between ODs, and calculate the passenger flow sharing ratio of each effective path according to the normal distribution model. In the clearing process, how to optimize the random probability distribution model of passenger flow is the focus of this invention.

[0099] like Figure 1 As shown, an embodiment of the present invention provides an optimization method for a random probability distribution model of urban rail transit passenger flow based on the fully observable control theory, and the method comprises the following steps:

[0100] By analyzing the travel behavior, travel process and influencing factors of passengers in the urban rail transit system, the generalized travel impedance in the random probability distribution model of urban rail transit passenger flow is optimized;

[0101] According to the optimized generalized travel impedance, the effective path of passenger travel is optimized;

[0102] Based on the fully observable control theory, the structure of the random probability distribution model of urban rail transit passenger flow is optimized to reduce the deviation of passenger flow distribution ratio in section.

[0103] Among them, the travel behavior includes: travel starting and ending points, travel time, etc.; the travel process includes: urban rail transit entering the station, waiting, boarding, transferring, waiting, boarding, and exiting the station; the influencing factors include: interval operating time, stop time, departure interval, waiting time, transfer preference, passenger travel purpose, etc.

[0104] Furthermore, the generalized travel impedance refers to the factors affecting the path selection being uniformly converted into travel time, which is called the generalized travel impedance; the optimization of the generalized travel impedance specifically includes:

[0105] The generalized travel impedance includes interval impedance and station impedance; the interval impedance is represented by the running time of the train in the interval; the station impedance is the time spent by the passengers at the station; the impedance of the station is the stop time of the train, and for the transfer station, the time spent includes the transfer travel time and the transfer waiting time. The transfer station impedance is represented by the transfer time multiplied by a transfer magnification factor α, α≥1, that is, the transfer time is converted into the same sense of the ride time through a transfer magnification factor;

[0106] The interval impedance calculation method is as follows:

[0107]

[0108] Where:

[0109] A represents the generalized travel impedance of a certain path in a certain period of time;

[0110] n represents the number of intervals in a path; t i represents the interval running time of the i-th interval;

[0111] The impedance calculation method of the station is as follows:

[0112]

[0113] Where:

[0114] It represents the generalized travel impedance of all stations along a route in the period of d days and p;

[0115] s represents the total number of stations passed by a certain route;

[0116] It represents the train stop time at the jth station during the d-day p-period, where d is weekdays, weekends, holidays, and p is peak or off-peak.

[0117] The transfer station impedance calculation method is as follows:

[0118]

[0119] Where:

[0120] represents the generalized travel impedance of all transfer stations in a certain path in d days and p periods;

[0121] R represents the total number of transfer stations in a certain route;

[0122] α h The transfer magnification factor representing the travel time at the h-th transfer station;

[0123] represents the travel time of the hth transfer station in the d-day p-period;

[0124] It represents the waiting time of the hth transfer station in the d-day p-period, which can be taken as half of the departure interval of the transfer line train;

[0125] The generalized travel impedance of a path is calculated as follows:

[0126]

[0127] Where:

[0128] C d，p It represents the generalized travel impedance of a path in d days and p periods.

[0129] Furthermore, the effective path is a physical path between ODs for passengers to choose, determined based on the topological structure of the urban rail transit network and in combination with the time rationality principle of passengers' generalized travel impedance selection between ODs; the optimization of the effective path for passenger travel based on the optimized generalized travel impedance specifically includes:

[0130] Set effective path selection rules:

[0131] Paths that exceed the generalized travel impedance of the shortest path by too much will not be chosen by passengers;

[0132] The probability of a path being selected is inversely proportional to the size of the path's generalized travel impedance;

[0133] The number of valid paths between ODs is uncertain and is greater than or equal to 1;

[0134] Based on the above conclusions, a threshold is set according to the shortest generalized travel impedance path. When the generalized travel impedance of a path is less than the threshold, it is selected as a valid path, otherwise it is an invalid path.

[0135] Add the following constraints to the above valid path selection rules:

[0136] Constraint 1: Backtracking is not allowed, and each node in the path cannot be visited repeatedly, that is, there is no loop, and all paths are elementary paths;

[0137] Constraint 2: Backtracking is not allowed, and each arc in the path cannot be visited repeatedly, that is, the same arc cannot appear repeatedly in each path;

[0138] Constraint 3: The generalized travel impedance of the path between ODs cannot exceed the generalized travel impedance threshold between ODs;

[0139] Among the K optional gradually shortening paths between two stations, it is assumed that the generalized travel impedance value of the shortest path is If the comprehensive impedance value of the second shortest path or other shorter paths exceeds a certain range compared with the comprehensive impedance value of the shortest path, that is, T range , the range can be a relative range m and an absolute range U. In this case, the shortest path or the second shortest path is considered unreasonable, which can be expressed as the following formula:

[0140]

[0141] Where:

[0142] T range Indicates the maximum allowable value of the effective path exceeding the generalized travel impedance value of the shortest path;

[0143] m represents the proportionality coefficient;

[0144] U represents a constant;

[0145] The choice of the values ​​of m and U is Related, that is Within a certain threshold range, the values ​​of m and U are different, so T range The optimization is a piecewise function, and the specific expression is as follows to express the influence of the generalized travel impedance of the shortest path for passengers on the effective path selection:

[0146]

[0147] Where:

[0148] m1, m2 represent the proportionality coefficient;

[0149] U1, U2 represent constants;

[0150] T represents the threshold of the shortest path generalized travel impedance;

[0151] The values ​​of m1, m2, U1, and U2 need to be determined comprehensively after statistical analysis of the survey results in combination with OD passenger flow surveys as the line network structure and operating methods change.

[0152] Furthermore, the optimization of the structure of the random probability distribution model of urban rail transit passenger flow based on the fully observable control theory specifically includes:

[0153] The original urban rail transit passenger flow random probability distribution model constructed includes:

[0154] The effective paths participating in the clearing are determined according to the size of the generalized travel impedance of the possible paths between OD, and the passenger flow allocation ratio of each effective path is calculated according to the normal distribution model.

[0155] Usually, the normal distribution model is used to describe the travel path selection behavior of passengers. The formula is as follows:

[0156]

[0157] Among them: The random variable x is taken as

[0158] a is the x value that obtains the maximum expected value of probability, which is 0 here; σ is a constant, and the value of σ determines the steepness of the normal curve; since there is no weight Less than the shortest path generalized travel impedance Therefore, we only need to take the positive half of the normal distribution curve x ≥ a. Here, we assume that the parameter σ is a constant for all the origin and destination points OD. The passenger flow distribution ratio of the path is calculated by the following formula:

[0159]

[0160]

[0161] Where S i is the benefit value of a certain effective path, P i is the passenger flow distribution ratio of the effective path.

[0162] When verifying the passenger flow of urban rail transit, it is found that the passenger flow distribution state generated by the above normal distribution model will cause deviations in the distribution ratio of passenger flow on the cross section. After research, the reason for the deviation is that the normal distribution model does not distinguish the boundary value of the generalized travel impedance of the shortest path, and uses σ as a constant for all OD pairs, resulting in a problem that the distribution ratio of the second or third path is low for OD pairs with larger generalized travel impedance of the shortest path, thereby causing deviations in the distribution ratio of passenger flow on the cross section.

[0163] Based on the above analysis results, the steps of optimizing the random probability distribution model of urban rail transit passenger flow in the present invention specifically include:

[0164] According to the value of σ and That is, the passenger's sensitivity to path selection is related to the generalized travel impedance value of the shortest path. The normal distribution model is optimized into a piecewise function model. The boundary is divided into two sections, as shown in the following formula:

[0165]

[0166]

[0167] Where: S 优化 is the benefit value of a certain effective path after optimization; σ1, σ2 are the sensitivity of passengers to the path selection; T is the limit value of the generalized travel impedance of the shortest path, T range It is the maximum allowable value by which a valid path exceeds the generalized travel impedance of the shortest path.

[0168] The smaller the value of σ is, the more sensitive the passenger is to choosing a route.

[0169] As a preferred implementation of the present invention, the value of the piecewise function model is as follows:

[0170] when σ takes the value of 0.2;

[0171] when σ takes the value of 0.25;

[0172] when σ takes the value of 0.4;

[0173] σ1 and σ2 in the piecewise function model represent different values ​​of σ.

[0174] The method of the present invention is further described in detail below through a specific embodiment.

[0175] Taking Beijing Subway Line 2 as an example, when the normal distribution model is used to calculate the passenger flow distribution ratio, it is found that the passenger flow in the peak section between Dongzhimen and Dongshishitiao on Line 2 is relatively high (e.g. Figure 2 After analysis, it is found that part of the reason is due to the problem of allocation ratio when calculating the path allocation ratio. For example, there are three paths from Huilongguan East Street to Chaoyangmen (such as Figure 3 As shown in the figure, the path allocation ratio of path 1 is significantly less than that of path 2 and path 3. The reason is that the generalized travel impedance value of the shortest path between OD is greater than 30 minutes, and the generalized travel impedance of these three paths is not much different. The use of normal distribution function for path allocation results in a significantly lower allocation ratio for path 1, that is, the generalized travel impedance value of path 1 is slightly larger than that of the shortest path, but the allocated benefits are much less, so passengers are mainly concentrated in paths 2 and 3, causing a high passenger flow in the peak section of Dongzhimen-Dongsishitiao.

[0176] In order to further analyze the reasons for the deviation in path allocation ratio, OD pairs with relatively large passenger flow between ODs during the morning peak are extracted from the K short-circuit table, and there are two paths between these OD pairs. According to the different generalized travel impedance values ​​of the shortest paths between the OD pairs, the changes in the path allocation probabilities corresponding to OD pairs with different shortest path values ​​are analyzed.

[0177] Example: OD pair: Shunyi to Xi'erqi

[0178] Route 1. Shunyi, "Shimen", "Nanfaxin", "Houshayu", "Hualikan", "Guozhan", "Sunhe", "Maquanying", "Cuigezhuang", "Wangjing East", "Wangjing", "Wangjing West", "Wangjing West", "Beiyuan", "Lishuiqiao", "Huoying", "Huilongguan", "Longze", "Xierqi".

[0179] Route 2. Shunyi, "Shimen", "Nanfaxin", "Houshayu", "Hualikan", "Guozhan", "Sunhe", "Maquanying", "Cuigezhuang", "Wangjing East", "Wangjing", "Wangjing West", "Guanzhuang", "Datun Road East", "Datun Road East", "Beiyuan Road North", "Lishuiqiao South", "Lishuiqiao", "Lishuiqiao", "Huoying", "Huilongguan", "Longze", "Xierqi".

[0180] The two paths are detailed in Figure 4 shown.

[0181] Draw a graph to compare the path probability distribution of OD pairs with different shortest path values, such as Figure 5 As shown, Figure 6 It is a schematic diagram of the decrease in the allocation ratio of each OD pair to the two paths. Figure 5 and Figure 6 It can be seen that under different shortest path generalized travel impedance values, the decrease in the allocation ratio of each pair of OD to the two paths is basically the same, and is relatively fast. This is different from the path selection of ordinary passengers. Generally speaking, if the shortest path generalized travel impedance value is less than a certain limit, the allocation ratio of the two paths decreases relatively quickly, but if it is greater than a certain limit, the allocation ratio of the two paths decreases relatively slowly.

[0182] In view of the above problems, it is found that the current normal distribution function does not distinguish the boundary value of the shortest path generalized travel impedance. For all OD pairs, the effective path benefit value distribution diagram σ = 0.1 is used (such as Figure 7 As shown in Figure 2, this results in a problem that for OD pairs with relatively large generalized travel impedance values ​​of the shortest path, the allocation ratio of the second or third path is relatively low, which causes the problem of deviation in the distribution of passenger flow on the cross section. The purpose of the present invention is to solve this problem by optimizing the random allocation model of urban rail transit passenger flow.

[0183] The real-time monitoring system for urban rail transit passenger flow consists of a series of algorithms, steps, rules, etc. It is a fully observable and controllable system, so the system can be optimized based on the fully observable and controllable experimental method. During the entire optimization process, a set of passenger flow distribution states generated by different model structures and parameter combinations are analyzed, and the main reasons for the state deviation are analyzed, and the next set of model structures and parameter combinations are adjusted in a targeted manner to make the process objectives controllable and improve the optimization efficiency.

[0184] Specifically, the value of σ is related to That is, the passenger's sensitivity to path selection is related to the generalized travel impedance value of the shortest path. Therefore, the above normal distribution model is optimized to a piecewise function model. The boundary is divided into two sections, as shown in the following formula:

[0185]

[0186]

[0187] Where: S 优化 is the benefit value of a certain effective path after optimization; σ1, σ2 are the sensitivity of passengers to the path selection; T is the limit value of the generalized travel impedance of the shortest path, T range It is the maximum allowable value by which a valid path exceeds the generalized travel impedance of the shortest path.

[0188] Generally speaking, the smaller the value of σ, the more sensitive the passenger is to the route selection. After investigation and optimization analysis, the values ​​of the piecewise function are as follows:

[0189] when σ takes the value of 0.2;

[0190] when σ takes the value of 0.25;

[0191] when σ takes the value of 0.4;

[0192] σ1 and σ2 in the piecewise function model represent different values ​​of σ.

[0193] According to the passenger flow random probability allocation model optimization method of the present invention, the K short circuits between the OD of the urban rail transit network are recalculated, and it is found that when the generalized travel impedance of the minimum path exceeds 60 minutes, the allocation ratio between the paths is more reasonable, and the allocation ratio of the second or third path will not be much lower than that of the first path. For details, see Figure 8 .

[0194] Figure 8The allocation ratios of some effective paths from Caofang to Xi'erqi were intercepted. From the allocation ratios of these three paths, it can be seen that the minimum generalized travel impedance value of the path from Caofang to Xi'erqi exceeds 1 hour, and the generalized travel cost values ​​of these three paths are not much different. By applying the passenger flow random probability allocation model optimization method, the allocation ratios of these three paths are also similar, which is consistent with the actual situation.

[0195] Furthermore, by adjusting the OD path allocation ratio, the direct effect is to make the passenger flow of the rail transit line network and the up and down sections closer to the actual situation, as shown in Table 1 for details.

[0196] Table 1 Analysis of passenger flow deviation at each end of rail transit network by line and by up and down direction

[0197]

[0198] After optimization, the up and down deviations of Line 2 have been within 5%, achieving the expected effect.

[0199] In summary, the present invention relies on the current passenger travel data source, is guided by existing business needs, and based on a fully observable and controllable experimental method, proposes a method for optimizing and improving the random probability distribution model of passenger flow at key nodes in the rail transit clearing process. Through the test and analysis of the results, the expected optimization effect is achieved, which lays a solid foundation for realizing a standardized, scientific and systematic urban rail transit clearing method.

[0200] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for optimizing the random probability distribution model of urban rail transit passenger flow, characterized in that: The following steps are involved: By analyzing the travel behavior, travel process and influencing factors of passengers in the urban rail transit system, the generalized travel impedance in the random probability distribution model of urban rail transit passenger flow is optimized; According to the optimized generalized travel impedance, the effective path of passenger travel is optimized; Based on the fully observable control theory, the structure of the random probability distribution model of urban rail transit passenger flow is optimized to reduce the deviation of the passenger flow distribution ratio in the section. The travel behavior includes: travel starting and ending points, travel time; the travel process includes: the process of entering the urban rail transit station, waiting, riding, transferring, and exiting the station; the influencing factors include: interval operation time, stop time, departure interval, waiting time, transfer preference, and passenger travel purpose; The generalized travel impedance refers to the factors that affect the path selection converted into travel time, which is called generalized travel impedance. The optimization of generalized travel impedance specifically includes: The generalized travel impedance includes interval impedance and station impedance; the interval impedance is represented by the running time of the train in the interval; the station impedance is the time spent by passengers at the station; among them, the impedance through the station is the stop time of the train, and for the transfer station, the time spent includes the transfer travel time and the transfer waiting time; The interval impedance calculation method is as follows: Where: A represents the generalized travel impedance of a certain path in a certain period of time; n represents the number of intervals in a path; t i represents the interval running time of the i-th interval; The impedance calculation method of the station is as follows: Where: It represents the generalized travel impedance of all stations along a route in the period of d days and p; s represents the total number of stations passed by a certain route; It represents the train stop time at the jth station during the d-day p-period, where d is weekdays, weekends, and holidays, and p is peak or off-peak. The transfer station impedance calculation method is as follows: Where: represents the generalized travel impedance of all transfer stations in a certain path during d days and p periods; R represents the total number of transfer stations in a certain route; α h The transfer magnification factor representing the travel time at the h-th transfer station; represents the travel time of the hth transfer station in the d-day p-period; It represents the waiting time of the hth transfer station in the d-day p-period, which can be taken as half of the departure interval of the transfer line train; The generalized travel impedance of a path is calculated as follows: Where: C d,p It represents the generalized travel impedance of a certain path in d days and p periods; The effective path is a physical path between ODs for passengers to choose, which is determined based on the topological structure of the urban rail transit network and the principle of time rationality of passengers' generalized travel impedance selection between ODs. The optimization of the effective path for passengers' travel based on the optimized generalized travel impedance specifically includes: Set effective path selection rules: Paths that exceed the generalized travel impedance of the shortest path by too much will not be chosen by passengers; The probability of a path being selected is inversely proportional to the size of the path's generalized travel impedance; The number of valid paths between ODs is uncertain and is greater than or equal to 1; Based on the above effective path selection rules, a threshold is set according to the shortest generalized travel impedance path. When the generalized travel impedance of a path is less than the threshold, it is selected as a valid path, otherwise it is an invalid path. Add the following constraints to the above valid path selection rules: Constraint 1: Backtracking is not allowed, and each node in the path cannot be visited repeatedly, that is, there is no loop, and all paths are elementary paths; Constraint 2: Backtracking is not allowed, and each arc in the path cannot be visited repeatedly, that is, the same arc cannot appear repeatedly in each path; Constraint 3: The generalized travel impedance of the path between ODs cannot exceed the generalized travel impedance threshold between ODs; Among the K optional gradually shortening paths between two stations, it is assumed that the generalized travel impedance value of the shortest path is If the generalized travel impedance value of the second shortest path or the shorter path exceeds a certain range of the generalized travel impedance value of the shortest path, that is, T range , the range includes the relative range m and the absolute range U. At this time, the second shortest path or the second shortest path is considered unreasonable, which is expressed as the following formula: Where: T range Indicates the maximum allowable value of the effective path exceeding the generalized travel impedance value of the shortest path; m represents the proportionality coefficient; U represents a constant; The choice of the values ​​of m and U is Related, that is Within a certain threshold range, the values ​​of m and U are different, so T range The optimization is a piecewise function, and the specific expression is as follows to express the influence of the generalized travel impedance of the shortest path for passengers on the effective path selection: Where: m1, m2 represent the proportionality coefficient; U1, U2 represent constants; T represents the threshold of the shortest path generalized travel impedance; The values ​​of m1, m2, U1, and U2 need to be determined comprehensively after statistical analysis of the survey results in combination with the OD passenger flow survey as the line network structure and operation mode change; The optimization of the structure of the random probability distribution model of urban rail transit passenger flow based on the fully observable control theory specifically includes: The original urban rail transit passenger flow random probability distribution model constructed includes: The effective paths participating in the clearing are determined according to the size of the generalized travel impedance of the paths between OD and OD, and the passenger flow allocation ratio of each effective path is calculated according to the normal distribution model; The normal distribution model is used to describe the travel path selection behavior of passengers. The formula is as follows: Among them: the random variable x is taken as a is the x value that obtains the maximum expected value of probability, which is 0 here; σ is a constant, and the value of σ determines the steepness of the normal curve; since there is no weight T i OD Less than the shortest path generalized travel impedance Therefore, we only need to take the positive half of the normal distribution curve x ≥ a. Here, we assume that the parameter σ is a constant for all the origin and destination points OD. The passenger flow distribution ratio of the path is calculated by the following formula: Where S i is the benefit value of a certain effective path, P i is the passenger flow distribution ratio of the effective path; The passenger flow distribution state generated by the current normal distribution model structure and parameter combination causes deviations in the distribution ratio of passenger flow on the cross section. The reasons are: The normal distribution model does not distinguish the boundary value of the generalized travel impedance of the shortest path, and uses σ as a constant for all OD pairs, resulting in a problem that the allocation ratio of the second or third path is low for OD pairs with a large generalized travel impedance of the shortest path, thereby causing a deviation in the distribution ratio of passenger flow on the cross section; The optimization of the random probability distribution model of urban rail transit passenger flow specifically includes: According to the value of σ and That is, the passenger's sensitivity to path selection is related to the generalized travel impedance value of the shortest path. The normal distribution model is optimized into a piecewise function model. The boundary is divided into two sections, as shown in the following formula: Where: S 优化 is the benefit value of a certain effective path after optimization; σ1, σ2 are the sensitivity of passengers to the path selection; T is the threshold of the generalized travel impedance of the shortest path, T range It is the maximum allowable value by which a valid path exceeds the generalized travel impedance of the shortest path.

2. The method for optimizing the random probability distribution model of urban rail transit passenger flow according to claim 1, characterized in that: The smaller the value of σ, the more sensitive the passenger is to choosing a route.

3. The method for optimizing the random probability distribution model of urban rail transit passenger flow according to claim 1, characterized in that: The values ​​of the piecewise function model are as follows: when σ takes the value of 0.2; when σ takes the value of 0.25; when σ takes the value of 0.4; σ1 and σ2 in the piecewise function model represent different values ​​of σ.

Images