JP3931190B2 - Method and apparatus for generating aggregated data of traffic-dependent quantities in stochastic equilibrium allocation - Google Patents

Description

  The present invention relates to a method and apparatus for generating aggregate data of traffic-dependent quantities in stochastic equilibrium distribution.

In the practice of traffic estimation, it is necessary to indicate the breakdown of traffic (OD breakdown of link traffic, average trip length by link, traffic by node direction, etc.) along with route traffic. Conventionally, the traffic volume is estimated by dividing the OD table that gives the traffic volume between regions in a matrix and updating the link cost while changing the OD traffic volume (in the OD table corresponding to the traffic volume from a certain start point to a certain end point). Divided distribution in which each element) is sequentially loaded on the shortest path has been widely used. However, divisional distribution has a problem that the estimation result varies depending on the division condition (number of divisions, ratio, etc.), and an estimation method using balanced distribution is being studied (for example, Non-Patent Document 1, Non-Patent Document 2). reference).
"Theory and Application of Road Traffic Demand Prediction Part I Toward Application of Equilibrium User Distribution" Japan Society of Civil Engineers Civil Engineering Planning Research Committee, Traffic Demand Prediction Technology Review Subcommittee, Maruzen Co., Ltd. August 2003 31st "Equilibrium Traffic Network Distribution-Latest Theory and Solution-" Japan Society of Civil Engineers Civil Engineering Planning Committee, Maruzen Co., Ltd. March 20, 1998

Among the balanced allocations, deterministic user balanced allocation is that all users always act on the same evaluation criteria of minimizing travel time, and the user gets complete travel time information on the route used. The traffic equilibrium distribution model is based on the assumption that the route traffic volume is not unique according to the definite user equilibrium distribution. Therefore, there is a problem that even if the results of definite user equilibrium distribution can be tabulated, the uniqueness of the route is not guaranteed, so that the legitimacy of the breakdown of traffic cannot be claimed.
On the other hand, stochastic user equilibrium allocation is a traffic equilibrium allocation model that considers the state that no user can further shorten their travel time by changing the route. The travel time of each route recognized by the user is not deterministic, but is a model that is considered to include an error that varies stochastically (randomly). According to this stochastic user equilibrium distribution, Theoretically, route traffic is unique, so route traffic is the only solution. However, in the stochastic user balanced distribution, the traffic volume is probabilistically distributed according to the time required for each route for a plurality of routes between ODs, and the number of routes is enormous in an actual traffic network. Therefore, it is difficult to enumerate routes. For this reason, the aggregation method used in the conventional distribution method is applied as it is, and the breakdown of traffic related to route traffic, for example, OD breakdown of link traffic, average trip length by link, and traffic by direction of node The amount cannot be determined. In other words, it is necessary to develop a new tabulation method for stochastic user equilibrium allocation.

  Accordingly, an object of the present invention is to provide an aggregate data generation method and apparatus that can aggregate various amounts depending on traffic after distribution in a stochastic user equilibrium distribution for a predetermined aggregate object. .

  In order to achieve the above object, the invention of the aggregated data generation method for traffic-dependent quantities in stochastic equilibrium distribution according to claim 1 is characterized in that an OD traffic related to a traffic network is converted from a single origin node to a plurality of endpoint nodes. A total data generation method for stochastically balancing and distributing to a plurality of routes leading to a total, and counting various amounts depending on the traffic volume after distribution for a predetermined total object, wherein the OD traffic volume is input and the input The process of distributing the OD traffic volume stochastically in the order of distance from the end node side toward the start node side in a descending order of the shortest route is repeated a plurality of times, and the allocated traffic volume for each repeated stage. A step of obtaining an approximate solution, a step of obtaining a distribution rate for each iteration stage, and a traffic volume corrected using the distribution rate from a specific point of interest toward the starting node A step of generating and outputting aggregated data of various quantities depending on the allocated traffic volume by executing the process of allocating by the backward process at each repetition stage and accumulating the results according to preset aggregation conditions. It is characterized by including these.

  In addition, the invention of the aggregated data generation method for the traffic-dependent quantities in the stochastic equilibrium distribution according to claim 2 is characterized in that the OD traffic related to the traffic network is probable on a plurality of routes from one origin node to a plurality of end nodes. And a total data generation method for summarizing various amounts depending on the traffic volume after distribution for a predetermined target, wherein the OD traffic volume is input, and the input OD traffic volume is A step of probabilistically allocating a plurality of times by retreating from the end node side toward the starting node side in the order of distance from the shortest route a plurality of times, and obtaining an approximate solution of the allocated traffic volume at each repetition stage; A step of obtaining a distribution rate for each repetition stage, and a traffic volume corrected using the distribution rate is distributed by a backward process from a specific point of interest toward the starting node. After that, using the allocated traffic distributed by the backward process as a pattern, the process of distributing by the forward process from the specific point of interest to the end point node is executed for each repetition stage, and the result is set in advance. A step of generating and outputting aggregated data of various quantities depending on the allocated traffic volume by accumulating according to the aggregation conditions.

  The invention of a total data generation method according to claim 3 is the method according to claim 1 or 2, wherein the point of interest is a specific node or link included in a traffic network.

  According to a fourth aspect of the present invention, there is provided a method for generating aggregated data of traffic-dependent quantities in stochastic equilibrium distribution according to claim 4, wherein the OD traffic related to a traffic network is changed from one origin node to a plurality of end nodes. A method of generating aggregate data that probabilistically distributes to a plurality of routes to reach and aggregates the amount of traffic depending on the traffic volume after distribution for a predetermined aggregate object, wherein the OD traffic volume is input, and one starting point A selection probability of each link connected to each node is obtained based on a shortest route search solution from the node to each of a plurality of end nodes, and the input OD traffic volume is calculated using the selection probability. First step to obtain an approximate solution of link traffic volume, and first step to obtain an approximate solution of the link traffic volume in the order of the distance from the end node side to the starting node side. The second step of repeating the first step while correcting the approximate solution until the first step converges, and obtaining the link selection probability and the approximate solution of the allocated link traffic volume for each iteration step, and the iteration step A third step of obtaining a distribution rate for each and the traffic volume corrected using the distribution rate, using the link selection probability, retreats from the specific point of interest to the start point node in order from the specific point of interest. 4th step of allocating by processing and accumulating the result in the aggregation table according to preset aggregation conditions, and repeating the 4th step for the number of repetitions, and aggregating data depending on the allocated traffic volume And a fifth step of generating and outputting.

  Furthermore, the invention of the aggregated data generation method for the traffic-dependent quantities in the stochastic equilibrium distribution according to claim 5, which is another preferable aspect of the present invention, provides a plurality of OD traffic volumes related to the traffic network from one origin node. A total data generation method that probabilistically distributes a plurality of routes to an end node of a node and totals the amounts that depend on the traffic volume after distribution for a predetermined total object, and inputs the OD traffic volume. The selection probability of each link connected to each node is obtained based on the solution of the shortest route search from one origin node to each of a plurality of end nodes, and the input OD traffic volume is used as the selection probability. A first step for obtaining an approximate solution of the link traffic volume by a backward process from the end node side toward the starting node side in the order of distance of the shortest route, and the first step. A second step of calculating the link selection probability for each iteration stage and an approximate solution of the allocated link traffic volume, repeating the first step multiple times while correcting the approximate solution obtained in step B. A third step of obtaining a distribution rate for each repetition stage, and the traffic volume corrected using the distribution rate, starting from a specific point of interest and using the link selection probabilities, the starting points in order from the starting point 4th step to be allocated by the backward process to the node and the traffic volume obtained in the 4th step as a pattern, and the forward process to the end node in order from the specific focus point to the start node A fifth step, a sixth step of accumulating the processing result of the fifth step in the tabulation table in accordance with a preset tabulation condition, and a fourth step Et sixth processing from step repeated the repetition number of times of generating the aggregated data quantities that depend on the distribution traffic, characterized in that it comprises a and a seventh step of outputting.

  Further, the invention of the aggregate data generation method according to claim 6 is the method according to claim 4 or 5, wherein the distribution rate for each iteration stage is based on the step size for each iteration stage obtained in the second step. It is determined.

  Further, the invention of the aggregate data generation method according to claim 7 is the method according to claim 4 or 5, wherein the point of interest is a specific node or link other than the start node included in a traffic network. Features.

  Further, the invention of the aggregate data generation method according to claim 8 is the method according to claim 4, wherein the fourth step is executed for at least one specific OD pair traffic volume, and the OD breakdown of the link designated in advance is performed. It is characterized by tabulating.

  Furthermore, the invention of the total data generation method according to claim 9 is characterized in that, in the method according to claim 5, one specific link is set as the specific point of interest.

  According to another aspect of the present invention, the invention of the aggregated data generation device for traffic-dependent quantities in the stochastic equilibrium distribution according to claim 10 provides the OD traffic related to the traffic network from one origin node to a plurality of end nodes. An aggregated data generation device that probabilistically distributes to a plurality of routes to reach and aggregates various amounts that depend on the traffic volume after allocation for a predetermined aggregation target, the input means for inputting OD traffic volume; A process of probabilistically allocating the input OD traffic volume in order of distance from the end node side to the starting node side in the order of distance from the shortest route is repeated a plurality of times, and the traffic allocated to each repeated stage is repeated. A distribution processing means for obtaining an approximate amount solution, a distribution rate determining means for obtaining a distribution ratio for each iteration stage, and a traffic volume corrected using the distribution ratio from a specific point of interest The backward processing means for executing processing to be distributed by the backward processing toward the starting node and the backward processing means are controlled so that the backward processing is executed once or a plurality of times at each repetition stage, and the result is set in advance. It includes a totaling means for generating aggregated data of various quantities depending on the allocated traffic volume by accumulating in the aggregate table according to the aggregation condition, and an output means for outputting the aggregated data .

  Furthermore, according to another aspect of the present invention, there is provided an invention of a total data generation device for traffic-dependent quantities in stochastic equilibrium distribution according to claim 11, wherein an OD traffic amount related to a traffic network is converted from a single origin node to a plurality of end points. A tally data generation device for stochastically balancing and distributing to a plurality of routes to a node, and tallying various amounts depending on traffic after distribution for a predetermined tally object, wherein the input means inputs the OD traffic amount And a process of probabilistically allocating the input OD traffic volume in order of distance from the end node side toward the starting node side in the order of distance from the shortest route a plurality of times. A distribution processing means for obtaining an approximate solution of the traffic volume, a distribution ratio determining means for obtaining a distribution ratio for each iteration stage, and a traffic volume corrected using the distribution ratio, in a specific manner Forward processing from a specific point of interest toward the destination node using a backward processing means for performing processing to be distributed by the backward processing from the eye point to the starting node, and the allocated traffic distributed by the backward processing means as a pattern Controlling the forward processing means for executing the processing distributed by the processing, the backward processing means and the forward processing means, and executing the backward processing and the subsequent forward processing once or a plurality of times for each repetition stage; By accumulating the results in a tabulation table according to preset tabulation conditions, tabulation means for generating tabulated data of various amounts depending on the allocated traffic volume, and output means for outputting the tabulated data, It is characterized by including.

  As a result of earnestly examining the method of tabulating the distribution result of the stochastic user equilibrium distribution, the present inventor has obtained a probability by using a predetermined backward process alone or a combination of the predetermined backward process and forward process. The present invention has been completed by discovering that it is possible to generate aggregate data of various traffic-dependent quantities in the balanced distribution of users.

  Here, the reverse process means a process of allocating the link traffic from the end point side of the OD traffic to the start point side. As a specific example of the preferred backward process, the totaling process can be performed using the backward process of Dial algorithm described later.

  On the other hand, the forward process refers to a process of allocating link traffic from the same starting point side to the ending point side as the traffic flow. In the present invention, it is possible to perform a totaling process combining the backward process and the forward process by using the distribution result of the backward process performed in advance as a pattern and performing the forward process according to an appropriate starting point and order described later. It was.

FIG. 1 is a diagram for explaining a basic flow of a Dial algorithm known as a solution for probabilistic user equilibrium distribution. As shown in the figure, the Dial algorithm calculates the minimum traffic cost c (m) from the starting point r to all the nodes m and calculates the link likelihood L [i → j] for all links (Step 0 processing). ), The step of calculating the link weight W [m → j] in ascending order from the starting point r (c (m) ascending order) (processing in Step 1), and the order from the starting point r (descending order of c (m)) The step of calculating the traffic volume x _im flowing into each node m (the process of Step 2) is included. Here, m represents a node of interest in each process, i represents an upstream node thereof, and j represents a downstream node thereof. Further, q _rm in the process of Step 2 is traffic concentrated on the point of _interest m when it is a centroid. The link weight means a value proportional to the probability that each link connected to a certain node is selected under the condition that a certain node is selected.

  According to such a Dial algorithm, it is possible to distribute a flowing traffic volume at a time for a route from a certain node to all other nodes (see the processing in Step 2 above). This is because the shortest route and the sub route are distributed in a tree shape as described later, and the Dial algorithm is configured based on the tree shape. On the other hand, according to the Dial algorithm, it is also possible to distribute the traffic volume for only one OD pair. In one embodiment of the tabulation process according to the present invention, as one of the basic methods, a process of allocating traffic only for one OD pair is used. Thereby, the OD breakdown of a specific link can be totaled.

  The conventional allocation process of split allocation and deterministic user equilibrium allocation is the distribution of traffic volume on one shortest route by reverse processing, and the traffic volume on the shortest route from the end point to the starting point is on a single road. By accumulating, various quantities can be tabulated. On the other hand, in the stochastic user equilibrium distribution, the traffic volume distribution result is represented by the shortest route and the sub route from the start point to the end point distributed in a tree shape as shown in the image diagram of FIG. Therefore, it is necessary to develop a new method for aggregation.

The process of obtaining the allocated traffic volume by using the Stochastic User Balance Allocation, for example, the Dial algorithm, is based on the shortest route and the sub route (slightly longer than the shortest route). The route to which the traffic is to be distributed) includes a process of distributing the traffic at a ratio of the link weight.
FIG. 3 is an image diagram of an example in which the traffic volume is allocated by the backward processing. In FIG. 3, since node A is farther from the starting point (not shown) than node B, first, with respect to node A, 60 traffic volumes (10 + 20 + 30) flowing out from node A are upstream at a predetermined link weight ratio. 25 and 35 are distributed to the two links on the side. Next, with respect to the node B closer to the starting point than the node A, 70 traffic volumes (45 + 25) flowing out from the node B are distributed as 30 and 40 to the two upstream links at a predetermined link weight ratio. In this way, in the dial backward processing algorithm, for all the nodes, the traffic flowing out downstream is summed in the order of distance from the starting point and distributed to the upstream link at the ratio of the link weight. ing. Therefore, in the aggregation process of stochastic user equilibrium distribution, if the backward process used in the distribution process is applied, it is possible to go back from the end point toward the start point, and the traffic volume from the start point to the start point in order from the start point. By accumulating the results of the backward processing to be distributed, various quantities can be aggregated.
Of course, in the tabulation process of the stochastic user equilibrium distribution, the backward process may be applied a plurality of times according to the purpose of the tabulation.

  Next, the forward process used in the counting method according to the present invention is a process of allocating by a forward process from a specific point of interest toward the end point node. In actual traffic flow, this refers to a specific section (for example, a specific link, a node (intersection), a group of a plurality of nodes (intersections), etc.) These are hereinafter referred to as “points of interest”, particularly a specific link. It is natural that the traffic passing through the section reaches the destination because it is only necessary to track the vehicles that have passed through the section. However, the path of the stochastic user equilibrium allocation process is generally defined to go from the end point in the opposite direction to the starting point, and the allocation process by the Dial algorithm is also in the order from the end point to the starting point. Backward processing is performed (see Step 2 in the Dial algorithm).

When the result of the distribution calculation is expressed as “traffic using a certain section”, of course, some distribution prior to that is performed. By using this preceding distribution, a route from the point of interest to the end point can be obtained by the forward process. In general, the traffic using the link of interest is a part of the traffic allocated by the preceding backward processing. Therefore, as shown in the image diagram of FIG. 4, in the forward process, the distribution result of the preceding backward process is used as a pattern.
Of course, in the tabulation process of the stochastic user equilibrium distribution, the backward process and the forward process may be applied a plurality of times depending on the purpose of the tabulation.

  According to the present invention, aggregate data of various traffic-dependent quantities in stochastic user equilibrium distribution is generated by using a predetermined backward process alone or in combination with a predetermined backward process and forward process. It becomes possible. In addition, it is possible to aggregate probabilistic user equilibrium distribution for a practical scale traffic network and complete the processing within a sufficiently acceptable time.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings. In addition, embodiment described below is an illustration and this invention is not limited to these.
FIG. 5 is a block diagram showing a basic configuration of an example of a total data generation apparatus according to the present invention. In FIG. 5, the total data generation device 100 includes an input unit 10, a distribution processing unit 20, a total processing unit 30, and an output unit 40. Further, the totalization processing unit 30 includes a distribution rate determination unit 311, a backward processing unit 312, a forward processing unit 313, and a totaling unit 314.

The input unit 10 performs aggregation processing using data necessary for causing the distribution processing unit 20 to execute stochastic user equilibrium distribution, which will be described later, and aggregation processing using the backward processing (or backward processing and forward processing), which will be described later. Data necessary for execution by the unit 30 is input.
Examples of the former data input from the input unit 10 include: (1) OD table, (2) distribution control data (number of zones, number of vehicle types, convergence determination conditions, etc.), (3) network data (link data) (Nodes at both ends, distance, performance function code, etc.), zone center node number (one node for each zone, corresponding to the network for traffic to / from the zone), performance function (link according to traffic volume) Parameters for determining the time required for passing), direction restriction data (data for specifying a direction in which travel is restricted in a three-node sequence), and the like.

  FIG. 6 shows an image diagram of the OD table input from the input unit 10. The OD table gives a traffic volume from a certain starting point to another ending point in a matrix. In FIG. 6, as an example, the traffic volume between the departure zone (corresponding to the start node) and the arrival zone (corresponding to the end node) is defined.

  Moreover, as an example of the latter data input from the input part 10, it is totaling control data (data specifying a totaling link, data specifying a totaling area, etc.), and the totaling control data is described later. Aggregation conditions in the aggregation processing unit 30 to be set are set in advance.

  Based on the OD table, distribution control data, and network data input from the input unit 10, the distribution processing unit 20 moves the OD traffic from the end node side to the start node side in order of distance from the shortest route. A process of stochastic equilibrium distribution is executed by the backward process. By repeating this process a plurality of times until a predetermined convergence condition is satisfied, an approximate solution of the traffic volume allocated for each iteration stage (hereinafter also referred to as “allocated traffic volume”) is obtained. The distribution processing unit 20 outputs a link selection probability, that is, a link weight, an approximate solution of the allocated traffic, and a step size for each repetition stage described later as a processing result.

  In the aggregation processing unit 30, the distribution rate determination unit 311 obtains the distribution rate for each repeated stage based on the processing result (specifically, the step size) in the distribution processing unit 20, and outputs it to the backward processing unit 312. The retreat processing unit 312 corrects the traffic volume using the distribution rate output by the distribution rate determination unit 311, and executes a process of distributing the corrected traffic volume by a retreat process from a specific point of interest toward the starting node. Further, the forward processing unit 313 acquires the allocated traffic distributed by the reverse processing unit 312 from the reverse processing unit 312 and uses the allocated traffic as a pattern to perform forward processing from a specific point of interest toward the end node. The process of allocating is executed. Further, the totaling unit 314 controls only the backward processing unit 312 or both the backward processing unit 312 and the forward processing unit 313, and performs only the backward processing or both the backward processing and the forward processing once for each repetition stage. Alternatively, it is executed a plurality of times, and the result is accumulated in a totaling table (not shown) according to a preset totaling condition. Note that the operation of the totaling unit 314 here follows the totaling condition input by the input unit 10.

As described above, desired aggregated data (aggregated data) relating to various quantities depending on the allocated traffic volume is generated in the aggregation unit 314 and output to the output unit 40.
The total data generation device 100 includes a CPU (Central Processing Unit) and a RAM (Random Access) interconnected by a bus.
Needless to say, it can be configured by a normal personal computer including a memory, a storage device, an input device, and an output device (for example, a display device such as a display). Since the hardware configuration of such a personal computer is self-explanatory, its illustration is omitted. However, in brief description according to the present embodiment, the input device of the personal computer is connected to the input unit 10 of the total data generation device 100. The CPU, RAM, and storage device of the personal computer correspond to the distribution processing unit 20 and the aggregation processing unit 30 of the total data generation device 100, and the output device of the personal computer corresponds to the output unit 40 of the total data generation device 100. . The storage device of the personal computer stores a main program for causing the personal computer to execute each process in the distribution processing unit 20 and the backward processing unit 30 and a table for aggregation. Then, the CPU of the personal computer reads out the main program stored in the storage device, develops it in a work area provided in the RAM, and executes the program, thereby driving and controlling each unit of the total data generation device 100. The RAM of the personal computer forms a main program executed by the CPU and a work area that temporarily stores data that is read from, updated, or generated from the storage device according to the program.
When the total data generation device 100 is configured by a personal computer, the output result is displayed in a predetermined format on a display (not shown) separately connected to the personal computer using the output unit 40 as an interface, and a printer ( (Not shown) can be printed in a predetermined format or stored in an external storage medium (not shown).

  The totaling unit 314 can include a predetermined totaling subprogram for each quantity depending on various types of allocated traffic for the purpose of counting. As an example, a subprogram that calculates the OD breakdown and trip length distribution of a specific link, a subprogram that calculates an OD breakdown and trip length distribution of a screen, an area-related traffic counting program, an OD table counting program between ramps, and traffic using a specific route A route totaling program, a route pattern totaling program, and the like are created in advance and stored in a storage device (not shown), and a CPU (not shown) stores an appropriate subprogram according to the totaling purpose in the storage device (FIG. (Not shown) may be selected, read, and executed.

  A plurality of embodiments of the tabulation method of the present invention performed using the tabulation data generation apparatus 100 configured as described above will be described in detail below with reference to the drawings. In connection with the operations of the distribution processing unit 20 and the aggregation processing unit 30 in the present embodiment, reference is again made to the Dial algorithm, which is a solution for the stochastic user equilibrium distribution, which is assumed by the method of the present embodiment. A preliminary mention will be made of the Frank-Wolfe method, which is a solution to user equilibrium allocation.

Referring again to FIG. 1, the basic flow of the Dial algorithm is to calculate the minimum traffic cost c (m) from the starting point r to all nodes m, and to calculate the link likelihood L [i → j] for all links. Step of calculating (step 0 processing), calculating the link weight W [m → j] in ascending order from the starting point r (ascending order of c (m)) (step 1 processing), and order of increasing distance from the starting point r (c (in descending order of (m)) includes a step of calculating the traffic amount x _im flowing into each node m (processing of Step 2). Of the basic flow of the Dial algorithm, the backward process of Step 2, that is, the process of distributing the traffic flowing out from the node to the upstream link at the link weight ratio plays an important role in the aggregation process in the aggregation processor 30.

FIG. 7 is an explanatory diagram of the backward processing in Step 2 of the Dial algorithm, and FIG. 8 is a table showing an image of traffic distribution by the Dial algorithm. In both figures, i, m, and j are node numbers, x and X are link traffic, q _rm is OD traffic, and W [] is link weight. In the example of FIG. 7, in the backward process of Step2 of Dial algorithm, node by focusing on m, the downstream link traffic xmj _1, xmj _2, the ratio of the link weights to xmj ₃ upstream link W [i Distribute with ₁ → m] and W [i ₂ → m]. As can be seen from the definition formula of Step 2 shown in FIG. 1, the ratio of distributing traffic flowing downstream to the upstream link (ratio of ii and ii, ro and ro, ha and ha, di and ni) Are the same. When the node of interest m is a centroid of the zone, it is also distributed to the upstream links at link weight ratios W [i ₁ → m] and W [i ₂ → m].

  The balanced distribution of traffic networks can be numerically solved by the Frank-Wolfe method as a mathematical optimization problem. This is introduced in Non-Patent Document 1 as a definite user equilibrium distribution solution. The Frank-Wolfe method can also be used for traffic distribution processing in stochastic user equilibrium allocation. In addition, when the aggregation process is further performed using the result obtained in the distribution process, the corrected link traffic volume at each stage can be used according to the Frank-Wolfe method. In the balanced distribution using the Frank-Wolfe method, a descent direction vector determined to indicate the direction in which the function value is lowered for a predetermined objective function and a step size indicating the distance that can be lowered in the one-dimensional search are determined, and the link Correct the traffic vector. In this embodiment, the distribution processing unit 20 executes this method.

(First embodiment)
Next, a tabulation method according to the first embodiment performed using the tabulation data generation apparatus 100 will be described in order with reference to the drawings. In the present embodiment, an example will be described in which the aggregation processing unit 30 performs the aggregation by executing only the backward processing.

  First, the distribution process executed by the distribution processing unit 20 of the total data generation device 100 will be described below using a flowchart. As a premise for operating the distribution processing unit 20, it is assumed that external data necessary for distribution processing such as an OD table, distribution control data, and network data is input from the input unit 10.

FIG. 9 is a flowchart illustrating an example of distribution processing executed by the distribution processing unit 20.
First, as initial values, the number of iterations n until the equilibrium solution converges is set to 0, and the total link traffic volume is set to 0 (Step S11). Next, the distribution processing unit 20 obtains an initial link cost t _ij ⁽⁰⁾ = t _ij (0) (Step S12).

The allocation processing unit 20 allocates the total OD traffic volume input from the input unit 10 to the initial link cost t _ij ⁽⁰⁾ obtained in Step S12 according to the Dial algorithm, and determines the link traffic volume x _ij ⁽⁰⁾ . Obtain (Step S13). The traffic distribution by the Dial algorithm has already been described with reference to FIG. 1, FIG. 7, and FIG.

Next, the distribution processing unit 20 corrects the link cost t _ij ⁽ⁿ⁾ based on the link traffic x _ij ^{(0) obtained} in Step S13 (Step S14). Here, in the first calculation immediately after obtaining the initial link traffic volume, it means _obtaining t _ij ⁽⁰⁾ = t _ij (x _ij ⁽⁰⁾ ).

Next, the distribution processing unit 20 allocates the total OD traffic volume to the link cost corrected in Step 14 according to the Dial algorithm, and obtains the link traffic volume y _ij ⁽ⁿ⁾ (Step S15). Again, in the first calculation immediately after obtaining the initial link traffic volume, this means _obtaining y _ij ⁽⁰⁾ .

Then the distributing unit 20, the equation of ^{_{d ij (n) = y ij}} (n) -x ij (n) determining the lowering direction vector. The descending direction vector is used in the calculation in Step S18 described later. In Step S17, the step size α ⁽ⁿ⁾ of the descending direction vector is determined (Step S16). Again, in the first calculation immediately after obtaining the initial link traffic volume, this means obtaining α ⁽⁰⁾ .

Next, the distribution processing unit 20 corrects the link traffic volume according to the formula x _ij ^{(n + 1)} = x _ij ⁽ⁿ⁾ + α ⁽ⁿ⁾ d _ij ⁽ⁿ⁾ (Step S18). In this correction process, x _ij ^{(n + 1)} is substituted into a predetermined objective function, and the step size and link traffic volume that minimize the objective function are obtained.

  One stage of processing is completed by a series of processing from Step S11 to Step S18. The step size and the link traffic volume thus obtained are approximate solutions at each stage unless the convergence condition of the next Step S19 is satisfied.

Subsequently, the distribution processing unit 20 determines whether or not a preset convergence condition is satisfied using x _ij ^{(n + 1)} and x _ij ⁽ⁿ⁾ (Step S19). When the convergence condition is not satisfied (Step S19; NO), n = n + 1 is set, and the process returns to Step S14 (Step S20). If the convergence condition is satisfied (Step S19; YES), the process is terminated.

  In this way, the distribution processing unit 20 repeats the distribution processing of Step S14 to Step S18 until a predetermined convergence condition is satisfied. The convergence condition can be set by combining multiple conditions such as satisfying the upper limit of the number of repetitions set in advance in consideration of the operation time of the device, etc., and the amount of change in the approximate solution being below a certain constant. For example, the convergence condition that is also used in the deterministic balanced distribution process using the Frank-Wolfe method may be used in the same manner.

When the distribution processing unit 20 completes the above distribution processing, an approximate solution (equilibrium solution) in which the link traffic has converged is finally obtained. In the process, the link weight at the distribution stage is determined each time it is repeated. Obtained in Step S15 (see Step 1 of the Dial algorithm), and for each iteration, the step size of the allocation stage and an approximate solution of link traffic (hereinafter also referred to as “allocated traffic”) are obtained in Step S18. Yes. Since these are also used for aggregation processing in the aggregation processing unit 30 to be described later, in each iteration, they are collected separately immediately after each step of Step S15 and Step S18, or immediately after Step S18 or after convergence determination of Step S19. And stored in a storage device (not shown).
In the present embodiment, the step size obtained in the distribution process is used to obtain a distribution rate described later in the aggregation process. Moreover, in this embodiment, the allocated traffic volume and link weight obtained in the allocation process can be used to reproduce the state of each stage k for aggregation. Here, the state is specifically a link traffic volume. In particular, the allocated traffic volume can be used to create a state in which the backward process of the stochastic equilibrium distribution can be performed in the aggregation process. For example, in the tabulation process, using the reproduced link traffic volume, the time required for each link (link travel time) is obtained using the performance function of each link, or the time required for the link thus obtained is calculated. Further, it is possible to obtain the travel time of the shortest route (travel time between ODs).
Since the link weight can be calculated separately from the allocated traffic volume, it is not always necessary to store the link weight obtained in the allocation process in the storage device (thus, reading from the storage device in the aggregation process described later). Instead, it may be calculated individually using the allocated traffic volume at the stage of the aggregation process.
Furthermore, in the aggregation process described later, the state of the next stage, that is, the corrected link traffic volume is reproduced from the state of one stage using the descent direction vector and step size of each stage obtained in the distribution process described above. In addition, instead of reproducing the state of each stage using the result of the distribution process, the state of the next stage may be reproduced by performing the same distribution process again as a part of the aggregation process. good.
In this way, in the aggregation process described later, it is necessary to reproduce the state of each stage of the distribution process, but as a reproduction method, various methods can be used as appropriate, and what reproduction method is adopted? It goes without saying that the present invention is not limited by the above.

  Next, in the first embodiment, an example of aggregation processing executed by the distribution processing unit 30 of the aggregation data generating device 100 will be described below using a flowchart. As a premise for operating the aggregation processing unit 30, external data necessary for aggregation processing such as aggregation control data (data specifying aggregation target links, data specifying aggregation target areas, etc.) is input from the input unit 10. It is assumed that

FIG. 10 is a flowchart showing an example of basic counting processing executed by the counting processing unit 30 in the first embodiment.
First, aggregation conditions are set (Step S21). The totaling condition sets what kind of traffic-dependent total data, which region, zone, or link in the network is to be counted.

Next, the totalization processing unit 30 calculates the step size α ^(k) , the allocated traffic volume x _ij ^(k) , and the link weight W ^(k) [i → j] of each stage k obtained in the above allocation process. The data is read from a storage device (not shown) (Step S22).

Then, allocation ratio determining unit 311, using the step size for each step k read alpha ^(k), obtaining the distribution ratio P ^(k) of the traffic to be distributed at each iteration step in the Frank-Wolfe method (StepS23 ). The distribution rate P ^(k) can be expressed as follows, ^where α ^(k) is the step size of the descent vector of each step.

Then, the distribution rate determination unit 311 passes the distribution rate P ^(k) of each stage obtained by the equation (1 ⁾ to the backward processing unit 312.

In the case of equilibrium allocation using the Frank-Wolfe method, after the allocation calculation process is performed in the allocation processing unit 20 as described above until the solution converges (that is, after the equilibrium solution is obtained), the allocation rate of each stage is calculated in the OD traffic volume. against was determined by multiplying the P ^(k) "OD traffic volume _{^{q (k) rm = P (}} k) × q rm for aggregation", do the aggregation process (StepS24). .

Next, the reverse processing unit 312 starts from the aggregation start link (or node) using the OD traffic q ^(k) _rm for aggregation and the link weight W ^(k) [] according to the equation (2). Then, in reverse order from the starting point (in descending order of the minimum traffic cost), backward processing is performed to the starting point and the traffic volume is allocated (Step S25).

Processing here, Step2 for Dial algorithms while allocating traffic by the backward process as for all the nodes, obtained by multiplying the (A) OD traffic volume in distribution rate P of each stage ^(k) A point for allocating “OD traffic for aggregation” (see Step S24 above), (A) For example, for a specific one or a plurality of OD pairs, etc. It differs from the distribution by the normal backward process in the distribution processing unit 20 in that the distribution of traffic volume along the line is the target of aggregation. The latter (A) is also related to the processing of the counting unit 314 described later (see Step S27). The backward processing unit 312 passes the result of the backward processing in Step S25 to the counting unit 314.

  The totaling unit 314 accumulates the result of the backward processing in Step S25 in the totaling table by writing the result in the totaling table (not shown) (Step S26).

  As an example of the specific processing in Steps S25 and S26, for example, in the aggregation of the OD breakdown of the links and the distribution of the trip lengths of the links, the backward processing is executed for each OD pair in the OD table and accumulated in the aggregation table. . Needless to say, such totaling conditions can be appropriately set in advance in Step S21.

  Next, the totaling unit 314 determines whether or not the backward processing of the backward processing unit 312 has been executed for all the totaling targets and the result has been added to the totaling table (Step S27). When the accumulation has been performed for all the aggregation objects (Step S27; YES), the process proceeds to the next Step S28. When that is not right (Step S27; NO), it returns to Step S25 and makes the backward process part 312 perform the process of Step S25 until the remaining aggregation object is lost, and the aggregation part 314 itself also executes the process of Step S26.

  When the accumulation is performed for all the aggregation objects, the aggregation unit 314 looks at the parameter k indicating the number of repetitions, and whether or not the aggregation is performed for n steps, that is, until the distribution process in the distribution processing unit 20 converges. For each obtained step k, it is determined whether or not the number of repetitions n has been completed (Step S28). When the number of repetitions has not been reached (Step S28; NO), k = k + 1 is set, and the process returns to Step S24 (Step S29). If the number of repetitions has been reached (Step S28; YES), the process is terminated.

Note that the OD traffic volume for aggregation is a value corrected by the distribution rate P ^(k) as in Step S24 above, and therefore, by accumulating n steps (where n is the number of repetitions until convergence in the distribution process). The traffic volume is not counted twice, and the total traffic volume flowing into or concentrated on one node does not fluctuate.

As described above, the tabulation table in which the tabulated data of the traffic-dependent quantities is written is completed. As an example, when the backward processing and the accumulation to the tabulation table are executed for each OD pair of the OD table, the tabulation of the OD breakdown of the links is completed.
The aggregated data generated in this way is displayed in a predetermined format on a display (not shown) separately connected to the personal computer using the output unit 40 as an interface, and printed in a predetermined format on a printer (not shown), or externally. It can be stored in a storage medium (not shown).

(Second Embodiment)
Next, an example of a tabulation method according to the second embodiment performed using the tabulation data generation device 100 will be described in order with reference to the drawings. In the present embodiment, the distribution process executed by the distribution processing unit 20 is the same as that in the first embodiment, and detailed description thereof is omitted. In this embodiment, an example in which the aggregation processing unit 30 performs aggregation by executing both the backward processing and the forward processing will be described. As in the first embodiment, as a premise for operating the aggregation processing unit 30, it is necessary for aggregation processing such as aggregation control data (data specifying an aggregation target link, data specifying an aggregation target area, etc.). It is assumed that external data is input from the input unit 10.

FIG. 11 is a flowchart showing an example of basic tabulation processing executed by the tabulation processing unit 30 in the second embodiment.
The difference from the flowchart in the first embodiment shown in FIG. 10 will be described. In the first embodiment, the backward processing unit 312 starts from the aggregation start link (or node) and goes away from the starting point (minimum). In the descending order of the transportation cost, the processing is performed to reverse the starting point and the processing for allocating the traffic (Step S25) is executed, and then the processing proceeds to the counting to the counting unit 314. After the backward processing, the forward processing unit 313 uses the allocated traffic volume x, which is the result of the backward processing by the backward processing unit 312 as a pattern, starting from the aggregation start link (or node), and in order from the starting point (minimum traffic) In this order, the processing is advanced to the end point in ascending order of cost, and the processing for allocating the traffic volume, that is, the forward processing for counting is executed (Step S35).

That is, the forward processing unit 313 acquires the allocated traffic volume x, which is the result of the backward processing by the backward processing unit 312, from the backward processing unit 312, and performs forward processing according to Expression (3) using the allocated traffic volume as a pattern. Allocate traffic.

  Intuitively expressing the significance of Equation (3) is that “the total traffic flowing into a node is equal to the outflow traffic” and “the ratio of link traffic in the pattern of traffic flow To "require traffic flow".

FIG. 12 is a diagram showing an example of traffic distribution by forward processing executed in the forward processing unit 313, and FIG. 13 is a diagram showing the state of forward processing in a table format. In these drawings and equation (4) described below, r, i, m, and j are node numbers, x and X are link traffic volumes, q and Q are OD traffic volumes, and W [] is a link weight.
In the illustrated example, as a result of the backward processing by the backward processing unit 312, the state of the node m is the traffic volume concentrated on the downstream three links traffic 300, 600, 100 and the node m (here, centroid) (that is, centroid). OD traffic volume) 200 is distributed at a link weight ratio, and 960 and 240 flow in the upstream side (see FIGS. 12A and 13A). If this distributed traffic volume is used as a pattern, and there is a traffic volume 576 in the link i ₁ → m of interest, the traffic volume concentrated on the centroid (that is, the OD traffic volume) 96, The traffic volumes 144, 288, and 48 flow through the links on the downstream side (see FIGS. 12B and 13B).
The following formula (4) shows a specific example for obtaining the traffic volume allocated by the forward process using the previous formula (3). However, it is assumed that there is no travel direction restriction at the point of interest.

  In this embodiment, Step S31 to Step S35 in the flowchart of FIG. 11 correspond to Step S21 to Step S25 in the flowchart (FIG. 10) according to the first embodiment, respectively, and Step S37 to Step S40 are the same flowchart (FIG. 10). ) Correspond to Step S26 to Step S29, respectively, and the processing contents are the same. Therefore, detailed description here is omitted.

According to the present embodiment, by using the reverse process and the forward process in combination, the route traffic volume using a specific link can be aggregated as an example. This tabulation is useful for viewing the range of influence of the link because it tabulates “what route the traffic using a specific link will take before and after that”.
FIG. 14 is an image diagram of totaling route traffic using specific links, and FIG. 15 is an image diagram of totaling results of route traffic using specific links. This total is calculated by allocating the flow of traffic flowing through the link of interest from the starting point o to that link by the reverse process, and using the result as a pattern, the route traffic volume from the link of interest to the end point d is totaled by the forward process. (See FIGS. 15A and 15B). In addition, in FIG. 14, the magnitude of route traffic is displayed by the thickness of the line.

ここで、xはパターンとして用いる後退処理実行後のリンク交通量、W^(k)[]は後退処理において用いた各段階のリンクウェイトである。 In addition, regarding the processing by the forward processing unit 313 of the second embodiment, the example in which the forward processing is performed according to the expression (3) and the traffic volume is allocated when there is no traveling direction restriction at the focused location has been described. If there is a travel direction restriction at a location, the traffic volume can be distributed by forward processing according to the following equation (5) generalized from equation (3).

Here, x is the link traffic volume after executing the reverse process used as a pattern, and W ^(k) [] is the link weight of each stage used in the reverse process.

(Application example 1 for other tabulations)
As an example of the first embodiment, the aggregation processing in the aggregation processing unit 30 is performed by using a specific OD as an aggregation target, a destination node as a specific point of interest, and starting with the destination node and moving backward from the source node to the source node. In the above example (image diagram of FIG. 3), an arbitrary link closer to the start node than the end node is set as the target link, and distribution is performed from the target link to the start node in reverse order starting from the target link. You may do it. FIG. 16 is an image diagram of the reverse processing starting from the link of interest, and FIG. 17 is an image diagram of the allocated traffic volume by such reverse processing.

By using the backward processing method starting from a specific link, it is possible to aggregate the OD breakdown of the screen and the trip length distribution of the screen. Here, the OD breakdown of the screen refers to the OD breakdown of traffic passing through a plurality of links (used once or a plurality of times) like traffic crossing a river section.
In this tabulation, for example, a plurality of links that cross a river cross section are specified as a tabulation target. However, the links are not limited to those forming a cross section, and other types of links may be tabulated. For example, if the entire highway network is counted, the OD breakdown of the highway traffic and the trip length can be totaled.

The outline of the OD breakdown calculation process for traffic using the screen is as follows.
(1) First, the reverse processing unit 312 of the totalization processing unit 30 distributes the traffic volume by the reverse processing for one OD pair.
(2) Next, a totaling condition is set so that the traffic flowing in the totaling link (a plurality of links constituting the screen) remains, and the traffic of all other links is replaced with zero.
(3) Next, in the retreat processing unit 312, from the state of (2) above, the distribution by the retreat process is executed starting from the link of interest toward the starting node, and the results are accumulated. Note that the traffic volume is not accumulated when the other aggregation target link is reached during the backward process.
(4) When the reverse process reaches the starting point, the traffic gathered at the starting point becomes the traffic volume that passes through the screen at least once.

(Application example 2 for other tabulations)
Furthermore, in the second embodiment, the example of totaling the route traffic volume of the traffic using the specific link by using the reverse process and the forward process together has been described. The total number of traffic routes that use this link is required. For example, the calculation is to obtain the area affected by the IC on the highway. Therefore, the method of counting the route traffic volume of traffic using a specific link is further expanded to total the route traffic volume of traffic using a plurality of links.

  In totaling the usage route of one specific link (focused link), the aggregated data is obtained by performing forward processing and backward processing, but in totaling the usage route of multiple target links, both forward processing and backward processing are performed. As a result of the distribution, there will be duplicated traffic, so it is necessary to remove this overlap. FIG. 18 is an image diagram of aggregation of route traffic volume using a plurality of links. In the case of FIG. 18, the traffic of a is totaled by the distribution by the backward process of the target link A and the distribution by the forward process of the target link B. In other words, if traffic that is counted in duplicate is obtained (this is referred to as “traffic between links of interest”), traffic that uses a plurality of links can be expressed by the following equation.

[Traffic using multiple links]
= [Distribution result of forward process] + [Distribution result of reverse process]-[Traffic between links of interest] (6)

The outline of the route totalization processing of traffic between a plurality of links of interest is as follows.
(1) First, the backward processing unit 312 of the totalization processing unit 30 allocates the traffic volume by the backward processing from all the target links to other target links, and obtains the traffic that reaches the target link. At the same time, it memorizes the traffic volume of the link that passed through this time.
(2) From the traffic volume in step (1) above, only the traffic that has reached the other link of interest remains, and the other link traffic volume is set to zero. As a result, the traffic volume between the links of interest is obtained (however, the route is not obtained).
(3) Next, in the forward processing unit 313 of the totalization processing unit 30, the traffic volume determined in the above (2) is allocated by the forward processing using the traffic volume determined in the above (1) as a pattern. , Cumulative results.
The traffic determined by the above (1) to (3) can always return to the first link, and this becomes the route traffic between the links of interest. That is, the route traffic volume using a plurality of links can be totaled according to the equation (6).

  In order to confirm the practicality of the present invention, the time required for tabulation was measured. An example will be described below.

  Using the total data generation device of FIG. 5, the total data in the stochastic user equilibrium distribution was obtained, and the required time was measured. Note that the data used for the required time measurement converged to an objective function value of 0.1% or less in about 20 iterations in the stochastic equilibrium distribution by the Dial algorithm, but here the number of iterations is 10 times for convenience. It was cut off and the required time was measured.

The required time measurement conditions are as follows.
Network: 28,270 links, 9,626 nodes, all links are one-way street networks.
OD table: 946 zones (the number of OD pairs with traffic is 94,386 pairs).
Performance function: BPR function.
Execution environment: OS (Operating System) ... Windows XP (Windows is a trademark of Microsoft Corporation), CPU (Central Processing Unit) ... Pentium4, 2.8GHz (Pentium is a trademark of Intel Corporation), Memory ... 1GB.

(Time required for each tabulation)
The distribution calculation took 170 seconds.
Of the various traffic-dependent totals, 256 seconds for the total traffic volume by node direction (10 nodes), 485 seconds for the total OD breakdown of links (10 links, 10 zone blocks), and screen OD 704 seconds for the breakdown (two links screens are two), 901 seconds for the total trip length distribution (10 links, 14 trip length ranks), and the total number of traffic routes using multiple links (4 It took 322 seconds to use the link). Therefore, the time required for aggregation was a little more than five times as long as the allocation calculation time.

  According to the present invention, aggregate data of various traffic-dependent quantities in stochastic user equilibrium distribution is generated by using a predetermined backward process alone or in combination with a predetermined backward process and forward process. It becomes possible. In addition, it is possible to aggregate probabilistic user equilibrium distribution for a practical scale traffic network and complete the processing within a sufficiently acceptable time. Therefore, the significance of the present invention that solves the difficulty of tabulation for the practical use of stochastic user equilibrium distribution is extremely great.

It is a figure explaining the basic flow of the algorithm of Dial known as a solution of probabilistic user equilibrium distribution. It is an image figure of the distribution result of traffic volume. It is an image figure of the example which allocates the traffic volume by a reverse process. It is an image figure of the totalization process which uses a backward process and a forward process together. It is a block diagram which shows the basic composition of an example of the total data generation device which concerns on this invention. 4 is an image diagram of an OD table input from an input unit 10. FIG. It is explanatory drawing of the backward process of Step2 of the Dial algorithm. It is an image figure of traffic distribution by Step 2 (retreat process) of the Dial algorithm. It is a flowchart which shows an example of the distribution process which the distribution process part 20 performs. In the first embodiment, it is a flowchart showing an example of a basic counting process executed by a counting processing unit 30. In a 2nd embodiment, it is a flowchart which shows an example of the basic total process which the total process part 30 performs. It is a figure which shows the example of allocation of the traffic volume by the advance process performed in the advance process part 313. FIG. It is a figure showing the state of forward processing in a tabular format It is an image figure of totalization of the route traffic which uses a specific link. It is an image figure of the total result of the route traffic which uses a specific link. It is an image figure of the retreat process which starts from the attention link. It is an image figure of the distribution traffic by a reverse process. It is an image figure of totalization of route traffic using a plurality of links.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Input part 20 Distribution process part 30 Total process part 40 Output part 100 Total data production | generation apparatus 311 Distribution rate determination part 312 Backward process part 313 Advance process part 314 Total part

Claims (11)

OD traffic volume related to the traffic network is stochastically balanced and distributed over multiple routes from one origin node to multiple end nodes, and various quantities that depend on the traffic volume after distribution are aggregated for a given target A data generation method,
The OD traffic volume is input, and the input OD traffic volume is probabilistically balanced and distributed multiple times in reverse order from the end node side to the start node side in the order of distance from the shortest route. Obtaining an approximate solution of the allocated traffic for each iteration stage;
Obtaining a distribution rate for each iteration stage;
A process of allocating the traffic volume corrected using the distribution rate by a backward process from a specific point of interest toward the starting node is executed at each repetition stage, and the results are accumulated according to preset aggregation conditions. To generate and output aggregated data of various quantities depending on the allocated traffic volume, and
A method for generating aggregated data of traffic-dependent quantities in stochastic equilibrium distribution, characterized by comprising: OD traffic volume related to the traffic network is stochastically balanced and distributed over multiple routes from one origin node to multiple end nodes, and various quantities that depend on the traffic volume after distribution are aggregated for a given target A data generation method,
The OD traffic volume is input, and the input OD traffic volume is probabilistically balanced and distributed multiple times in reverse order from the end node side to the start node side in the order of distance from the shortest route. Obtaining an approximate solution of the allocated traffic for each iteration stage;
Obtaining a distribution rate for each iteration stage;
After allocating the traffic volume corrected using the allocation rate by the backward process from the specific point of interest toward the starting node, using the allocated traffic volume allocated by the reverse process as a pattern, from the specific point of interest Generates aggregated data of various quantities depending on the allocated traffic volume by executing the process of allocation by the forward process toward the end node at each iteration stage and accumulating the results according to preset aggregation conditions. And output step,
A method for generating aggregated data of traffic-dependent quantities in stochastic equilibrium distribution, characterized by comprising:   The total data generation method according to claim 1, wherein the point of interest is a specific node or link included in a traffic network. OD traffic volume related to the traffic network is stochastically balanced and distributed over multiple routes from one origin node to multiple end nodes, and various quantities that depend on the traffic volume after distribution are aggregated for a given target A data generation method,
The OD traffic volume is input, a selection probability of each link connected to each node is obtained based on a shortest path search solution from one origin node to each of a plurality of end nodes, and the input OD traffic volume Using the selection probability, the first step of obtaining an approximate solution of link traffic volume by allocating in reverse order of the shortest route and by retreat processing from all the end node sides to the start node side;
The second step of finding the link selection probability and the approximate solution of the allocated link traffic volume at each iteration stage is repeated a plurality of times until the first step converges while correcting the approximate solution obtained in the first step. And the steps
A third step for obtaining a distribution rate for each of the repetition stages;
Using the link selection probability, the traffic volume corrected using the allocation rate is allocated by reverse processing starting from a specific point of interest to the starting point node in order of distance from the starting point node, and the results are set in advance. A fourth step of accumulating in the tabulation table according to conditions;
A fifth step of repeating the fourth step by the number of repetitions, generating and outputting data in which various amounts depending on the allocated traffic volume are output;
A method for generating aggregated data of traffic-dependent quantities in stochastic equilibrium distribution, characterized by comprising: OD traffic volume related to the traffic network is stochastically balanced and distributed over multiple routes from one origin node to multiple end nodes, and various quantities that depend on the traffic volume after distribution are aggregated for a given target A data generation method,
The OD traffic volume is input, a selection probability of each link connected to each node is obtained based on a shortest path search solution from one origin node to each of a plurality of end nodes, and the input OD traffic volume Using the selection probability, the first step of obtaining an approximate solution of link traffic volume by allocating in reverse order of the shortest route and by retreat processing from all the end node sides to the start node side;
The second step of finding the link selection probability and the approximate solution of the allocated link traffic volume at each iteration stage is repeated a plurality of times until the first step converges while correcting the approximate solution obtained in the first step. And the steps
A third step for obtaining a distribution rate for each of the repetition stages;
A fourth step of allocating the traffic volume corrected using the allocation rate by using the link selection probability, starting from a specific point of interest, and moving backward from the starting point to the starting point node;
A fifth step of using the allocated traffic volume obtained in the fourth step as a pattern and allocating by a forward process from the specific focus point to the end node in order from the start point node;
A sixth step of accumulating the processing result of the fifth step in the tabulation table according to preset tabulation conditions;
A seventh step of repeating the processes from the fourth step to the sixth step for the number of repetitions, generating aggregated data of various amounts depending on the allocated traffic volume, and outputting the data,
A method for generating aggregated data of traffic-dependent quantities in stochastic equilibrium distribution, characterized by comprising:   The total data generation method according to claim 4 or 5, wherein the distribution rate for each repetition stage is determined based on a step size for each repetition stage obtained in the second step.   The total data generation method according to claim 4, wherein the point of interest is a specific node or link other than a start node included in a traffic network.   The total data generation method according to claim 4, wherein the fourth step is executed for at least one specific OD pair traffic volume, and the OD breakdown of links specified in advance is totaled.   6. The total data generation method according to claim 5, wherein one specific link is set as the specific point of interest. OD traffic volume related to the traffic network is stochastically balanced and distributed over multiple routes from one origin node to multiple end nodes, and various quantities that depend on the traffic volume after distribution are aggregated for a given target A data generator,
Input means for inputting OD traffic volume;
The input OD traffic volume is distributed at each repetition stage by repeating the process of probabilistically allocating the input OD traffic volume in order of distance from the shortest route to all the destination nodes from the destination node side by a plurality of times. A distribution processing means for obtaining an approximate solution of traffic,
A distribution rate determining means for determining a distribution rate for each of the repetition stages;
Retreat processing means for executing a process of allocating the traffic volume corrected using the distribution rate by a retreat process from a specific point of interest toward the starting node;
By controlling the backward processing means, the backward processing is executed once or a plurality of times at each repetition stage, and the result is accumulated in the totaling table according to the preset totaling conditions, thereby depending on the distribution traffic. An aggregation means for generating the aggregated data of the quantity;
Output means for outputting the aggregated data;
A total data generation device for traffic-dependent quantities in stochastic equilibrium distribution, characterized by comprising: OD traffic volume related to the traffic network is stochastically balanced and distributed over multiple routes from one origin node to multiple end nodes, and various quantities that depend on the traffic volume after distribution are aggregated for a given target A data generator,
Input means for inputting the OD traffic volume;
The input OD traffic volume is distributed at each repetition stage by repeating the process of probabilistically allocating the input OD traffic volume in order of distance from the shortest route to all the destination nodes from the destination node side by a plurality of times. A distribution processing means for obtaining an approximate solution of traffic,
A distribution rate determining means for determining a distribution rate for each of the repetition stages;
Retreat processing means for executing a process of allocating the traffic volume corrected using the distribution rate by a retreat process from a specific point of interest toward the starting node;
Forward processing means for performing processing to distribute by forward processing from a specific point of interest toward the end point node, using the allocated traffic distributed by the backward processing means as a pattern;
The backward processing means and the forward processing means are controlled so that the backward processing and the subsequent forward processing are executed once or a plurality of times for each repetition stage, and the results are accumulated in a totaling table according to preset totaling conditions. A means for generating aggregated data of various quantities depending on the allocated traffic volume,
Output means for outputting the aggregated data;
A total data generation device for traffic-dependent quantities in stochastic equilibrium distribution, characterized by comprising:

Images