JP7417476B2 - Layout plan creation support system and layout plan creation support method - Google Patents

Description

The present invention relates to a layout plan creation support system and the like that supports the creation of a layout plan for measurement sensors for calculating the OD (Origin-Destination) traffic volume of pedestrians inside a station.

Measuring pedestrian OD traffic volume (origin-destination traffic volume) is considered an important survey not only for urban planning but also for the renovation and management of buildings such as stations. Therefore, various techniques related to OD traffic measurement have been devised (see, for example, Patent Document 1).

Japanese Patent Application Publication No. 2014-106879

In order to accurately estimate the pedestrian OD traffic volume (origin/destination traffic volume) within the station premises, it is necessary to install sensors (e.g., people flow sensors, CCD cameras, passing person counters) at appropriate locations to measure the flow of pedestrians. , etc.) must be installed.

Ideally, sensors should be installed at all locations where pedestrians pass, but in the case of existing buildings, there are locations where sensors can be installed and locations where sensors cannot be installed. For example, there are locations that are unsuitable for installing the sensor due to reasons such as not being able to secure a space to install the sensor, creating a blind spot in the shooting range, not being able to secure a power source, or having difficulty installing signal lines.

Additionally, installing the sensor is costly. The cost varies depending on the type of sensor that can be installed as well as the situation at the installation location. For example, there are costs such as the need for several supporting metal fittings to install the sensor in an appropriate location, and the need for additional work to wire and extend signal lines.

Conventionally, skilled engineers have inspected the site, considered the budget, and determined the appropriate installation location based on their accumulated experience.
However, in recent years, due to the increase in the size of buildings and the renovation of building structures to improve convenience, the number of entrances and exits for the flow of people and the flow of people has increased, making it difficult to determine the installation location based on experience as in the past. It has increased. Furthermore, training of skilled engineers has also become an issue.

The problem to be solved by the present invention is to provide a technology that supports creation of a measurement sensor arrangement plan for calculating pedestrian OD traffic volume.

A first invention for solving the above-mentioned problems is a placement plan creation support system that supports the creation of a placement plan for measurement sensors for calculating pedestrian OD (Origin-Destination) traffic volume, comprising:
A network creation means (for example, a network creation means (for example, FIG. 8 control board 1150, network creation unit 202, pedestrian traffic network setting data 510, step S2 in FIG.
Candidate position setting means (for example, the control board 1150 in FIG. 8, the candidate position setting unit 204, the candidate position setting data 512, Step S4) in FIG.
Solution candidate creation means for creating a solution candidate that determines a combination of candidate positions of the measurement sensors from among the candidate positions (for example, the control board 1150 in FIG. 8, the solution candidate creation unit 210, the solution candidate setting data 530, 9 step S10),
Verification traffic setting means (for example, the control shown in FIG. 8 A board 1150, a verification traffic setting section 216, verification traffic setting data 540, step S8 in FIG. 9),
a solution candidate application network consisting of the total departure traffic volume for verification and the total arrival traffic volume for verification for each of the OD nodes, and a partial network of the pedestrian traffic network in which traffic volume can be measured based on the solution candidate; an estimation means (for example, the control board 1150, the estimation unit 218, the estimated traffic data 550 in FIG. 8, and step S24 in FIG. 9) that estimates the estimated traffic volume between the OD nodes based on the OD nodes;
A first evaluation means that evaluates the solution candidate by comparing the verification traffic volume and the estimated traffic volume between the OD nodes (for example, the control board 1150 in FIG. 8, the measurement accuracy evaluation unit 220, 9 step S26, step S42),
This is a layout plan creation support system that includes:

According to the first invention, the layout plan creation support system sets a network model of people flow (pedestrian traffic network) in which nodes are pedestrian passage points including OD nodes that are departure points/destinations of pedestrians. Then, solution candidates for various combination patterns of "installing" and "not installing" the measurement sensor are created from among the candidate positions for installing the measurement sensor.

Then, the layout planning support system sets the verification traffic volume (total cross-sectional traffic volume between nodes) for each OD node based on the virtual human flow in the set network, and sets the verification traffic volume for each OD node. In the solution candidate application network, assuming that the traffic volume and total arrival traffic volume for verification are set and measurement sensors are installed in the network model according to the created solution candidates, the total departure traffic volume for verification between OD nodes and verification By estimating the estimated traffic volume that realizes the total arrival traffic volume and comparing the estimated traffic volume with the verification traffic volume, the solution candidate can be evaluated.

Therefore, by using the placement plan creation support system of the first invention, the user can know where to place measurement sensors to accurately measure OD traffic volume, and can make a measurement sensor placement plan. can be created smoothly.

A second invention is characterized in that the solution candidate creation means changes a part of the solution candidate to create a new solution candidate, and the estimation means estimates the new solution candidate and the first evaluation means evaluates the new solution candidate. The arrangement plan creation support system according to the first invention further includes a repetition control means (for example, the control board 1150 in FIG. 8, the repetition control unit 224, NO in step S40 → step S12).

According to the second aspect of the invention, the placement plan creation support system can repeat the creation of solution candidates, estimation, and evaluation in a process approximating a so-called genetic algorithm. Therefore, an appropriate solution candidate can be found with a smaller amount of calculation compared to a process in which all possible solution candidates are created and estimated and evaluated using a brute force method.

In particular, the iterative control means causes the solution candidate creation means to create a new solution candidate for each of the solution candidates selected based on the result of the first evaluation, and By repeatedly controlling the estimation by the estimation means and the evaluation by the first evaluation means for new solution candidates, it is possible to realize an efficient search for an optimal solution or an approximate solution thereof based on a genetic algorithm.

A third aspect of the invention is a placement cost setting means (for example, the control board 1150 in FIG. 8, the placement cost setting unit 206, the placement cost setting data 516 in FIG. 9, Step S6) and
Based on the placement cost, a total placement cost calculation means (for example, the control board 1150 in FIG. 8, a total placement cost calculation unit) that calculates the total placement cost when the measurement sensor is placed at the candidate position related to the solution candidate. 214, total placement cost 537 in FIG. 5, step S12 in FIG. 9),
a second evaluation means for evaluating the solution candidate based on the total placement cost (for example, the control board 1150 in FIG. 8, the total placement cost evaluation unit 212, step S14 in FIG. 9);
The layout plan creation support system according to the first or second invention further comprises:

According to the third invention, the placement plan creation support system performs the second evaluation and selects appropriate solution candidates that take into consideration the placement cost according to the circumstances of each candidate position and the total placement cost (so to speak, the upper limit of the budget). You can choose.

In a fourth aspect of the present invention, the arrangement plan is an arrangement plan of measurement sensors for calculating the OD traffic volume of pedestrians in the station premises, and the candidate position setting means includes an automatic ticket gate and/or an automatic ticket gate in the station premises. Alternatively, an existing camera (hereinafter collectively referred to as "existing sensor") is regarded as the measurement sensor, and the installation position of the existing sensor is set to be included in the candidate position, and the placement cost setting means This is a placement plan creation support system according to a third aspect of the invention, in which the placement cost is set to zero or a predetermined low cost for candidate positions corresponding to the installation location.

According to the fourth invention, the layout planning support system uses existing sensors that can be used/repurposed for measuring the flow of people, such as automatic ticket gates and CCD (Charge-Coupled Device) cameras, to measure the flow of people at zero or low cost. Therefore, it is possible to select an appropriate candidate solution regarding the arrangement of measurement sensors including existing sensors.

In a fifth invention, the solution candidates whose evaluation results by the first evaluation means and the evaluation results by the second evaluation means satisfy a predetermined high evaluation condition are used as a recommended combination of candidate positions of the measurement sensor. The layout plan creation support system according to the third or fourth invention, further comprising presentation means (for example, the control board 1150 in FIG. 8, the presentation control unit 226, the recommended solution candidate ID list 560, and step S44 in FIG. 9). be.

According to the fifth invention, the placement plan creation support system allows the user to select solution candidates that have good measurement accuracy (highly rated in the first evaluation) and whose total placement cost is within the budget (highly rated in the second evaluation). can be presented.

A sixth invention is a layout plan creation support method for a computer system to support the creation of a layout plan for measuring sensors for calculating pedestrian OD (Origin-Destination) traffic volume, the OD The pedestrian traffic locations including the OD nodes that are the origin/destination of traffic volume are set as nodes, and a pedestrian traffic network connecting each node is created, and the measurement is performed on the pedestrian traffic network. setting candidate positions including at least positions where measurement sensors can be placed; creating solution candidates defining combinations of candidate positions for the measurement sensors from among the candidate positions; By setting the verification traffic volume, the total verification departure traffic volume and the verification total arrival traffic volume for each OD node can be set, and the verification total departure traffic volume and the verification total traffic volume for each OD node can be set. Estimating the estimated traffic volume between each of the OD nodes based on the total arriving traffic volume and a solution candidate application network that is a partial network that can measure traffic volume based on the solution candidate of the pedestrian traffic network. and evaluating the solution candidate by comparing the verification traffic volume and the estimated traffic volume between the OD nodes.

According to the sixth invention, it is possible to realize a layout plan creation support method that exhibits the same effects as the first invention.

A schematic diagram showing an example of a station premises. FIG. 2 is a diagram of a pedestrian traffic network that models the configuration of FIG. 1. The figure which shows the data structure example of candidate position setting data. The figure which shows the example of a data structure of placement cost setting data and a total placement cost upper limit value. The figure which shows the data structure example of solution candidate setting data. The figure which shows the data structure example of the traffic volume setting data for verification. The figure which shows the example of a solution candidate application network. 1 is a diagram illustrating a hardware configuration, a functional configuration, and an example of programs and data stored in the layout planning support system; FIG. 2 is a flowchart for explaining the flow of processing in the layout plan creation support system.

An example of an embodiment of a placement plan creation support system that supports the creation of a placement plan for measurement sensors for the purpose of calculating pedestrian OD traffic volume will be described below. Of course, the present invention is not limited to this embodiment.

FIG. 1 is a schematic diagram showing an example of a station premises 10 used for explanation.
In the station premises 10, a plurality of platforms 11 (11a, 11b,...) and a plurality of station entrances/exits 13 (13a, 13b) are connected by a passageway 15 (including stairs), and an automatic ticket gate 17 and an automatic ticket gate 17 are connected in the station premises 10. , and a CCD camera 19 are installed. Depending on the circumstances, the pedestrian 2 gets off the train that has arrived at the platform 11, passes through the automatic ticket gate 17 through the internal passageway 15 inside the ticket gate, passes through the automatic ticket gate 17 through the internal passageway 15 outside the ticket gate, and exits the train. Head towards station entrance/exit 13. Alternatively, the user can enter through any station entrance/exit 13, pass through the automatic ticket gate 17, and board the train from any platform 11. Of course, there are cases where it is not possible. OD traffic volume measurement is to measure the flow of departures and arrivals in the station premises 10 as seen when focusing on individual pedestrians 2, for the entire station premises 10.

Note that the example in FIG. 1 shows a simplified example for the purpose of explanation. In reality, the number and relative connections of the platforms 11, station entrances 13, passageways 15, automatic ticket gates 17, CCD cameras 19, etc. are determined depending on the actual configuration of the station premises 10. do.

FIG. 2 is a diagram of a pedestrian traffic network 20 that models the configuration of FIG.
A user using the layout plan creation support system performs a predetermined input operation to enter nodes NDx (x=1, 2,...; node number) and links Cy (y=1, 2,...; link number). , a pedestrian traffic network 20 is set up.

The node NDx (x=1, 2, . . . ; node number) is set at the intersection of the platform 11, which is the starting point/arrival point of the pedestrian 2, and the internal passageway 15. In the example of FIG. 2, nodes ND1 and ND2 correspond to the station entrances and exits 13 (13a, 13b), and nodes ND3 to ND6 correspond to the platforms 11 (11a, 11b, . . . ). Nodes ND1 to ND6 are OD nodes that are the origin/destination of OD traffic. Nodes ND7 to ND10 correspond to intersections of the internal passageway 15.

Link Cy (y=1, 2, . . . ; link number) represents the flow of pedestrians 2 (flow of people) between nodes NDx. Therefore, the link Cy has a "direction" represented by the node NDx serving as the starting point and the node NDx serving as the ending point. For example, if we focus on the flow between two adjacent nodes ND1 and ND7, there are two combinations in which one is the starting point and the other is the ending point. A link C1 whose end point is the node ND7, and a link C2 whose start point is the node ND7 and whose end point is the node ND1 are set.

Note that the number of nodes NDx and the direction and number of links Cy are appropriately set according to the structure of the actual station premises 10 to be measured. For example, depending on the architectural structure of the station premises 10, there may be one-way passageways. In that case, in the part of the pedestrian traffic network 20 related to the one-way passage, one link Cy corresponding to the one-way street is set between two adjacent nodes NDx.

A plurality of candidate positions Pz (z=1, 2, 3, . . . ) including at least a position where a measurement sensor for measuring the link Cy can be installed are set in the pedestrian traffic network 20. The candidate position Pz is a position in the station premises 10 where a measurement sensor can be installed. During actual operation, settings should be made according to the actual situation at the target station.

The candidate positions Pz include locations where devices that can be used as measurement sensors have already been set. The automatic ticket gate 17 can detect the direction of travel of the passenger who is checking the ticket and whether or not the passenger is passing through the ticket gate, so it can be used as a measurement sensor and is included in the candidate position Pz. Furthermore, if the CCD camera 19 has already been installed, the existing CCD camera 19 can detect the traveling direction and passage of the passenger by image analysis, etc., so it can be used as a measurement sensor, and the existing CCD camera 19 can be used as a measurement sensor. include. In the example of FIG. 2, candidate position P9 corresponding to automatic ticket gate 17 and candidate position P3 corresponding to CCD camera 19 correspond to this.

Note that the equipment that can be used as a measurement sensor, that is, the existing sensor, is not limited to the automatic ticket gate 17 or the CCD camera 19, but can be of any type as long as it can be used/used to detect the passage of a person. For example, existing sensors such as sensors installed at human gates to detect when people pass (such as photoelectric passing sensors) and mat sensors that detect when people pass by sensing their weight can also be used as existing sensors. can.
Further, candidate positions for which it is clear that traffic volume should be measured in the method of estimating OD traffic volume, which will be described later, can be set to be always included in the candidate positions Pz.

The correspondence between which links Cy can be measured at each candidate position Pz is created and stored as candidate position setting data 512 as shown in FIG. 3, for example. The candidate position setting data 512 stores a candidate position ID 513 for each candidate position and a measurable link ID 514 indicating a link Cy that can be measured at the corresponding candidate position in association with each other.

Note that in the example of FIG. 2, in order to clarify the correspondence between the candidate position Pz and the measurable link Cy, a black circle representing the candidate position Pz is drawn on the line segment between the nodes. For example, candidate position P1 indicates a black circle between node ND1 and node ND7 as a position for measuring link C1 and link C2. However, the actual candidate position P1 is not necessarily on the path between the nodes ND1 and ND7.

Furthermore, for example, when assuming a tenth candidate position for node ND7, it can be said that four links Cy, links C1 to C4, can be measured at the candidate position. Depending on how the measurement sensors are installed and their capabilities, a configuration in which six links C1 to C6 can be measured at the candidate position is also possible. In this case, the links Cy associated with the tenth candidate position are links C1 to C4, or links C1 to C6.

Once the pedestrian traffic network 20 and candidate positions Pz are set, the layout plan creation support system selects measurement sensors for each candidate position Pz from among the plurality of candidate positions Pz set in the pedestrian traffic network 20. Recommended combinations of candidate locations for measurement sensors, taking into account the cost required to install the sensor, the upper limit of the total placement cost (the upper limit of the budget for installing measurement sensors), and the measurement accuracy of OD traffic measurement. and present it to the user.

The placement planning support system uses a given combination optimization algorithm to find recommended combinations of candidate positions. In this embodiment, a genetic algorithm is used as the combination optimization algorithm. First, as a prerequisite for finding recommended combinations, the placement plan creation support system requests the user to set the placement cost for each candidate position Pz and the total placement cost.

The user performs a predetermined input operation to set and input the placement cost for each candidate position Pz and the total placement cost. The placement plan creation support system creates and stores (1) placement cost setting data 516 and (2) total placement cost upper limit value 520 as shown in FIG. 4 in response to input.

The placement cost setting data 516 stores a unique candidate position ID 517 for each candidate position Pz and a placement cost 518 required for installing a measurement sensor at the position in association with each other. Note that the placement cost 518 corresponding to the existing measurement sensor and the existing device that can be used as the measurement sensor is set to "0" or a relatively inexpensive value. In the example of FIG. 4, this applies to the placement cost 518="5" of the candidate position P3 and the placement cost 518="0" of the candidate position P9.

Once the preconditions are given, the placement plan creation support system selects multiple combinations of where to place the measurement sensor and where not to place it among the multiple candidate positions Pz set in the pedestrian traffic network 20. create. Eventually, one or some of the created combinations will be presented to the user as a recommended combination, so the created combinations will become "solution candidates." The placement plan creation support system creates and stores solution candidate setting data 530 as shown in FIG. 5 for each combination.

As solution candidates, all combinations (in the example of the pedestrian traffic network 20 in FIG. 2, the candidate position P3 is the CCD camera 19, and the candidate position P9 is the automatic ticket gate, so the seven locations are determined). Calculating and evaluating ²⁷ possible combinations of candidate positions requires an enormous amount of calculation, and there is a possibility that no results will be obtained within a practical amount of time. Therefore, instead of all combinations, a limited number of m (m is an integer) solution candidates are first created.

One solution candidate setting data 530 includes a unique solution candidate ID 531, a placement setting 534 (“1” = placed, “0” = not placed) that is in one-to-one correspondence with a candidate position ID 533, and the corresponding solution. It includes a total placement cost 537 for the candidate and an evaluation index value 539 for the solution candidate.

The total placement cost 537 is the total value of the placement costs 518 (see FIG. 4) of the candidate positions Pz for which the placement setting 534 is set to "1" in the solution candidate.

The evaluation index value 539 indicates the evaluation in terms of the measurement accuracy of the OD traffic volume of the "solution candidate application network model" determined by setting the measurement sensor indicated by the solution candidate to the pedestrian traffic network 20. When the solution candidate setting data 530 is generated, it is set to an initial value indicating undetermined.

After creating m solution candidates, the placement planning support system randomly selects α solution candidates (α is a natural number of 2 or more) from among them, considers them as the current generation, and selects the next generation solution candidates. (for example, solution candidate setting data 530b, 530c). That is, each candidate position ID 533 is regarded as one locus, and the arrangement setting 534 of one or more of the loci is changed from that in the current generation. In other words, we create mutants of the current generation solution candidates. However, existing measurement sensors or existing devices that can be used as measurement sensors are excluded from the changes (candidate positions P3 and P9 in FIG. 2 correspond to these excluded positions).

The “solution candidate application network model” based on the created m current generation solution candidates and α next generation solution candidates will be the subject of measurement accuracy verification this time.

For verification, the layout plan creation support system creates verification traffic volume setting data 540 as shown in FIG. The verification traffic setting data 540 includes a unique traffic pattern ID 541, an OD table 543, and input data 545.

The OD table 543 is a matrix in which nodes NDx are combined as departure nodes/arrival nodes, and the value shown at the intersection position of the departure node and the arrival node is the verification OD traffic volume T _ij between the departure node and the arrival node. (i=node number indicating the departure node, j=node number indicating the arrival node). The layout plan creation support system randomly sets these OD traffic volumes T _ij for verification, adds up the OD traffic volumes T _ij for verification related to each node NDx, and calculates the total departure traffic volume 544 for verification and the OD traffic volume T ij for verification. The total arriving traffic volume 546 is determined. These constitute input data 545 for verification.

The placement plan creation support system sets a "solution candidate application network" for each solution candidate. The solution candidate application network is a partial network in which a part of the pedestrian traffic network 20 (see FIG. 2) is restricted by applying the arrangement of measurement sensors set in the solution candidate in the pedestrian traffic network 20 (see FIG. 2). The layout plan creation support system performs verification by estimating the estimated OD traffic volume T _ij in the solution candidate application network based on the input data 545.

For example, FIG. 7 is a diagram showing an example of the solution candidate application network 22, which is a partial network of the pedestrian traffic network 20 based on the solution candidate setting data 530a of FIG.
According to the placement setting 534 of the solution candidate setting data 530a and the candidate position setting data 512, measurement sensors are not placed at the candidate positions P2, P5, P6, and P8 of the pedestrian traffic network 20 (see FIG. 2). , links C3, C4, and C13 to C18 will not be measured. Therefore, when verifying the solution candidate application network 22 based on the solution candidate setting data 530a, the placement plan creation support system creates links C1, C2, C5 that realize the entry and exit of people at each OD node specified by the input data 545. The estimated OD traffic volume T _ij of ~C12 will be estimated. The details of the estimation method will be explained along with the flowchart described later.

Then, when the solution candidate application network 22 estimates the estimated OD traffic volume T _ij for all solution candidates, the layout plan creation support system calculates the estimated OD traffic volume T _ij and the verification OD traffic volume T _ij for each solution candidate. and calculates the evaluation index value for each solution candidate. Details of the method for calculating the evaluation index value will be explained with reference to a flowchart described later.

The placement planning support system uses a given combination optimization algorithm to search for a solution candidate with a higher evaluation index value, and uses a genetic algorithm in this embodiment. Once the placement plan creation support system has determined the evaluation index value for each solution candidate whose measurement accuracy was verified this time, it performs elite selection based on the evaluation index value and deletes solution candidates that have not been selected. In other words, weed out. Then, a mutant species (next generation) of the selected solution candidate (current generation) with a part of the placement setting of the candidate position Pz (genetic locus) changed is created, and it is compared with the elite selected current generation. , the evaluation index value is calculated for the next generation solution candidate as a target for verification of measurement accuracy again. This series of processes is repeated. Then, when a predetermined verification end condition is met, the layout plan creation support system presents a predetermined number of solution candidates with the highest evaluation index value at that time to the user as a recommended combination.

Verification end conditions can be set as appropriate. For example, if a predetermined number of solution candidates whose evaluation index value satisfies a predetermined high evaluation condition are searched, or if a solution candidate whose evaluation index value satisfies a predetermined high evaluation condition is not searched within a predetermined time, iterative control When the number of times reaches a predetermined value, etc. can be set as one or more AND conditions or OR conditions.

FIG. 8 is a diagram showing an example of the hardware configuration, functional configuration, and programs and data stored by the system 1000.

The layout plan creation support system 1000 is a computer system, and includes, for example, a main unit 1101, a keyboard 1106, a touch panel 1108, and a storage 1140, and a control board 1150 is mounted on the main unit 1101.

The control board 1150 is equipped with various microprocessors such as a CPU (Central Processing Unit) 1151, a GPU (Graphics Processing Unit), and a DSP (Digital Signal Processor), various IC memories 1152 such as VRAM, RAM, and ROM, and a communication device 1153. has been done. Note that part or all of the control board 1150 may be realized using an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or an SoC (System on a Chip).

The layout plan creation support system 1000 realizes various functions for implementing layout plan creation support by the control board 1150 performing arithmetic processing based on predetermined programs and data.

Specifically, the placement plan creation support system 1000 stores the support program 501 and executes it on the CPU 1151, so that the network creation section 202, candidate position setting section 204, placement cost setting section 206, and overall A placement cost upper limit setting unit 208, a solution candidate creation unit 210, a total placement cost evaluation unit 212, a verification traffic volume setting unit 216, an estimation unit 218, a measurement accuracy evaluation unit 220, a repetition control unit 224, It realizes the function of the presentation control section 226.

The network creation unit 202 sets the passage points of the pedestrian 2 in the station premises 10 including the OD node that is the departure point/destination of the OD traffic volume as a node NDx, and creates a pedestrian traffic network 20 connecting each node NDx. (See Figure 2). Specifically, the layout plan creation support system 1000 displays an input screen for network creation on the touch panel 1108, and creates and stores pedestrian traffic network setting data 510 in response to operation input from the keyboard 1106 or the like.

The candidate position setting unit 204 sets a candidate position Pz that includes at least a position on the pedestrian traffic network 20 where a measurement sensor can be placed. Specifically, the layout plan creation support system 1000 displays the pedestrian traffic network 20 on the touch panel 1108, displays a setting input screen for the candidate position Pz, and selects the candidate position in response to an operation input using the keyboard 1106 or the like. Setting data 512 is created and stored (see FIG. 3).

The placement cost setting unit 206 sets the placement cost of the measurement sensor for each candidate position Pz. Specifically, the placement plan creation support system 1000 displays an input screen for setting the placement cost for each Pz of candidate positions on the touch panel 1108, and inputs the placement cost setting data 516 in response to an operation input using the keyboard 1106 or the like. Create and store (see Figure 4).

Note that the placement cost setting unit 206 receives a setting input of a candidate position Pz corresponding to the installation position of an existing sensor such as the automatic ticket gate 17 or an existing CCD camera 19 on the input screen for setting the placement cost, and selects the candidate position Pz. As for Pz, the placement cost is set to zero (or a predetermined low cost).

The total placement cost upper limit setting unit 208 sets an upper limit value of the total placement cost of measurement sensors for each candidate position Pz. Specifically, the placement plan creation support system 1000 displays an input screen for setting the total placement cost upper limit value 520 on the touch panel 1108, and creates the total placement cost upper limit value 520 in response to operation input using the keyboard 1106 or the like. - Store (see Figure 4).

The solution candidate creation unit 210 creates solution candidates that define combinations of candidate positions of measurement sensors from among the candidate positions Pz. Then, the solution candidate creation unit 210 creates solution candidate setting data 530 for each created solution candidate (see FIG. 5).

The total placement cost evaluation unit 212 functions as a second evaluation means that evaluates solution candidates based on the total placement cost. Specifically, the total placement cost evaluation unit 212 includes a total placement cost calculation unit 214, and the total placement cost calculation unit 214 determines the placement based on the placement setting 534 (see FIG. 5) of the solution candidate to be evaluated. The total placement cost 537 is calculated by adding up the costs 518 (see FIG. 4). Then, the total placement cost evaluation unit 212 compares the calculated total placement cost 537 with the total placement cost upper limit 520 (see FIG. 3), and if the calculated total placement cost 537 is less than or equal to the total placement cost upper limit 520, Evaluate the evaluation target solution candidate as "eligible" and leave it as a verification target. However, if the calculated total placement cost exceeds the upper limit, the solution candidate to be evaluated is evaluated as "disqualified." Then, delete it and exclude it from verification.

The verification traffic volume setting unit 216 creates data for verification of the measurement accuracy of the solution candidate application network 22 based on the solution candidates. Specifically, the verification traffic volume setting unit 216 sets the verification OD traffic volume T _ij between each OD node, thereby setting the verification total departure traffic volume 544 and verification total arrival traffic volume for each OD node. 546 to create verification traffic volume setting data 540 (see FIG. 6).

The estimation unit 218 applies a solution candidate consisting of a total departure traffic volume 544 for verification and a total arrival traffic volume 546 for verification for each OD node, and a partial network in which traffic volume can be measured based on the solution candidate of the pedestrian traffic network 20. The estimated traffic volume between each OD node (estimated OD traffic volume T _ij of link Cy) is estimated based on the network 22 and . The estimation unit 218 performs estimation for each solution candidate in the solution candidate setting data 530, and stores the result as estimated traffic volume data 550 for each solution candidate.

The measurement accuracy evaluation unit 220 functions as a first evaluation means that evaluates solution candidates by comparing the verification OD traffic volume T _ij and the estimated OD traffic volume T _ij between the OD nodes.
Specifically, the measurement accuracy evaluation unit 220 includes an evaluation index value calculation unit 222, and calculates an evaluation index value for each solution candidate based on the estimation result of OD traffic volume, and calculates the evaluation index value based on the solution candidate setting data 530 to be calculated. is stored as an evaluation index value 539 (see FIG. 5). Then, a predetermined number of elites are selected from the high evaluation side of the evaluation index value 539, and solution candidate IDs 531 that are not selected are registered as a weeding solution candidate ID list 558. The weeded-out solution candidate ID list 558 is used for checking to ensure that solution candidates that have been weeded out due to insufficient accuracy are not included in the verification target again.

The repetition control unit 224 causes the solution candidate generation unit 210 to change a part of the solution candidates to create a new solution candidate, and performs estimation by the estimation unit 218 and measurement accuracy evaluation unit 220 (first evaluation method).

The presentation control unit 226 selects solution candidates whose evaluation results by the measurement accuracy evaluation unit 220 (first evaluation means) and evaluation results by the total placement cost evaluation unit 212 (second evaluation means) satisfy a predetermined high evaluation condition. , is presented as a recommended combination of candidate positions for measurement sensors. Specifically, the presentation control unit 226 sorts the solution candidates by their respective evaluation index values 539 (see FIG. 5), selects a predetermined number of top solution candidates from the high-performing side as an elite, and assigns their solution candidate IDs 531 is set in the recommended solution candidate ID list 560. Then, the presentation control unit 226 associates the placement settings of the solution candidates registered in the recommended solution candidate ID list 560 with the total placement cost 537 at the time when the verification end condition is satisfied, and displays the information on the touch panel 1108. Control the display.

FIG. 9 is a flowchart for explaining the flow of processing in the layout plan creation support system 1000.
The layout plan creation support system 1000 first performs creation processing of the pedestrian traffic network 20 (step S2). Specifically, an input screen for creating the pedestrian traffic network 20 is displayed on the touch panel 1108, and the pedestrian traffic network setting data 510 is created in accordance with the creation operation input by the user.

Next, the placement plan creation support system 1000 performs a process of setting candidate positions Pz (step S4). Specifically, an input screen for setting the candidate position Pz is displayed on the touch panel 1108, and candidate position setting data 512 is created in response to the setting operation input by the user.

Next, the placement plan creation support system 1000 executes a process of setting the placement cost 518 for each candidate position and the total placement cost upper limit 520 (step S6). Specifically, an input screen for placement cost 518 and total placement cost upper limit 520 for each candidate position Pz is displayed on touch panel 1108, and placement cost setting data 516 and total placement cost upper limit are displayed in accordance with the setting operation input by the user. 520 is created.

Next, the layout plan creation support system 1000 creates verification traffic volume setting data 540 (step S8), and generates m+α solution candidates (step S10; see FIG. 5). That is, the placement plan creation support system 1000 randomly creates m solution candidates and further creates α solution candidates that are mutations of the m solution candidates.

Then, the placement plan creation support system 1000 calculates the total placement cost 537 for all current solution candidates (step S12), and performs selection, weeding, and supplementation for the current solution candidates based on the total placement cost (step S14). ). Specifically, the placement plan creation support system 1000 selects solution candidates whose total placement cost 537 is less than or equal to the total placement cost upper limit 520, and selects solution candidates whose total placement cost 537 exceeds the total placement cost upper limit 520. Perform deletion (culling). That is, a second evaluation process is performed in which each solution candidate is evaluated based on the total placement cost 537.

As a result of selection and culling, if the number of remaining solution candidates is less than m+α, the placement plan creation support system 1000 again creates additional solution candidates in the number short of m+α. Similarly to the m solution candidates, additional candidates may be created at random, or α mutation solution candidates may be additionally created. Alternatively, both may be additionally created at a predetermined ratio. Then, the placement plan creation support system 1000 calculates the total placement cost 537 of each additionally created solution candidate, and deletes the solution candidate whose total placement cost 537 exceeds the total placement cost upper limit 520. By repeating these steps, the placement plan creation support system 1000 prepares m+α solution candidates whose total placement cost 537 is equal to or less than the total placement cost upper limit 520. The prepared solution candidates constitute a verification target group.

Next, the layout plan creation support system 1000 executes loop A for each solution candidate to be verified to estimate the estimated OD traffic volume T _ij (steps S20 to S28).

In loop A, the placement plan creation support system 1000 creates a solution candidate application network 22 for solution candidates to be processed (step S22; see FIG. 7), and determines the location of each OD node in the placement of measurement sensors in the network. The estimated OD traffic volume T _ij that realizes the verification total departure traffic volume 544 and the verification total arrival traffic volume 546 is estimated (step S24).

Specifically, the layout planning support system 1000 estimates the estimated OD traffic volume T _ij using the formulation and solution of an OD traffic volume estimation model based on entropy maximization. The formulation and solution of the OD traffic estimation model used by the layout planning support system are described in, for example, "Henk J. Van Zuylen and Luice G. Willumsen, The Most Likely Trip Matrix Estimated from Traffic Counts, Transportation Research, 14B, No.3, pp281-293, 1980.

ｉ：出発ノード番号
ｊ：到着ノード番号
Ｔ_ｉｊ：ノードｉとノードｊ間のＯＤ交通量 According to the above source, the entropy maximization model is defined as Equation 1.

i: departure node number
j: Arrival node number
T _ij : OD traffic volume between node i and node j

Equation 1 is given by the number of combinations that occur in the OD table 543 (see FIG. 6), and the estimated OD traffic volume T _ij is obtained by maximizing W (T _ij ). This is interpreted as maximizing the likelihood (probability of occurrence) of the pattern (distributed traffic pattern) of the estimated OD traffic volume T _ij .

Ｔ：総交通量 Here, when equation (1) is transformed logarithmically, equation (2) is obtained.

T: Total traffic volume

The optimal solution T _ij ' of the maximization problem of equation (2) matches the maximization problem of equation (2). Since the first term on the right side of equation (2) is a constant, the maximization problem on the left side of equation (2) is equivalent to the minimization problem on the right side of equation (2). Also, from Stirling's formula, InT _ij ! can be transformed as shown in equation (3).

Ｏ_ｉ：ノードｉの出発交通量
Ｄ_ｊ：ノードｊの到着交通量 On the other hand, the constraint conditions for satisfying the consistency of the OD table 543 are the verification total departure traffic volume 544 and the verification total arrival traffic volume 546 that are input to each node by the input data 545. Therefore, this problem can be defined as an optimization problem of equations (4) to (6).

O _i : Departure traffic volume of node i
D _j : Arrival traffic volume of node j

Ｃ：リンク番号

Here, since it is assumed that one OD has one route in the solution candidate application network 22, if formula (7) is defined, the traffic volume Vc of link C can be written as formula (8).

C: Link number

When a constraint condition based on the observed traffic volume of link C is added, the estimation of the estimated OD traffic volume T _ij in the solution candidate application network 22 can be redefined as an optimization problem of equations (9) to (12).

λ_ｉ，γ_ｊ，β_ｃ：ラグランジュ乗数 Defining the Lagrangian function L from equations (9) to (12) results in equation (13).

λ _i , γ _j , β _c : Lagrangian multiplier

よって、式（１５）が導かれる。

Next, if the partial differentiation of the Lagrangian function L with respect to the decision variable (estimated OD traffic volume T _ij ) is set to "0", equation (14) is obtained.

Therefore, equation (15) is derived.

The Lagrange multiplier can be obtained by substituting equation (15) into equation (13) and searching for a Lagrange multiplier that minimizes the value of L in equation (13). T _ij obtained by substituting the obtained Lagrange multiplier into equation (15) is the optimal solution that minimizes the objective function, and becomes the estimated value of the estimated OD traffic volume T _ij .

Next, the placement plan creation support system 1000 calculates the evaluation index value 539 of the solution candidate to be processed in loop A (step S26).

ｉ：ＯＤノードの要素数（推計ＯＤ交通量Ｔ_ｉｊの数）
ｆ_ｓｉj：解候補ｓの推計ＯＤ交通量Ｔ_ｉｊの値
ｙ_ｉj：検証用ＯＤ交通量Ｔ_ｉｊの値 Specifically, the reciprocal of the RMSE (Root Mean Square Error) of the verification OD traffic volume T _ij set in the OD table 543 of the verification traffic volume setting data 540 and the estimated OD traffic volume T _ij is calculated as follows. It is calculated and set as the evaluation index value EVALs of the solution candidate s. The RMSEs of the solution candidates are defined by Equation (16), and the evaluation index value EVALs is defined by Equation (17). The larger the evaluation index value EVALs, the higher the measurement accuracy of the OD traffic measurement system created by adopting the layout setting of the solution candidate.

i: Number of elements of OD node (number of estimated OD traffic volume _Tij )
f _sij : Value of estimated OD traffic volume T _ij of solution candidate s
y _ij : Value of verification OD traffic volume T _ij

After executing loop A for all solution candidates to be verified, the layout plan creation support system 1000 next determines whether the verification end condition is satisfied (step S40). For example, "If a solution candidate whose evaluation index value 539 reaches a predetermined high evaluation standard value is searched," "If a solution candidate whose evaluation index value 539 reaches a predetermined high evaluation standard value is not searched within a predetermined time from the start." It is determined that the verification end condition is satisfied in either of the following cases.

If it is not satisfied (NO in step S40), as a first evaluation, the placement plan creation support system 1000 selects an elite number of solution candidates with a high evaluation index value 539 from the top and solves the problem. The recommended solution candidate ID list 560 is updated with the candidate ID 531. Regarding solution candidates that are not selected, their solution candidate IDs 531 are set in the weeding solution candidate ID list 558, and the solution candidates are excluded from the verification target from here on, thereby weeding them out (step S42).

Then, the process returns to step S14. At this stage, there remain solution candidates whose total placement cost 537 is less than or equal to the total placement cost upper limit 520 and whose evaluation index value 539 is highly evaluated. The total number is less than m+α. The placement plan creation support system 1000 additionally creates solution candidates that are short of m+α as mutations of the elite selected solution candidates. Of course, the total placement cost 537 of the additionally created solution candidates must be less than or equal to the total placement cost upper limit 520. Note that the solution candidates to be added are checked against the solution candidate setting data 530 of the solution candidates indicated in the weeded-out solution candidate ID list 558 to confirm that they have not been weeded out in the past.

After m+α verification target groups have been prepared again, the layout plan creation support system 1000 executes loop A for all solution candidates to be verified (steps S20 to S28).

If the verification end condition is satisfied (YES in step S40), the layout plan creation support system 1000 displays information on recommended solution candidates on the touch panel 1108 to present it to the user (step S44), and performs a series of processes. end.
Note that if the verification end condition "A solution candidate whose evaluation index value 539 reaches a predetermined high evaluation standard value is not searched within a predetermined time from the start" is met, no recommended solution candidate was found. A notification is presented to the user including that the setting of candidate positions Pz is recommended to be increased.

As described above, according to the present embodiment, the arrangement plan creation support system 1000 can support the creation of a measurement sensor arrangement plan for calculating the pedestrian OD traffic volume. That is, by using the placement plan creation support system 1000, the user can know where to place measurement sensors to enable accurate OD traffic volume estimation.

In addition, when one solution candidate is created, the placement plan creation support system 1000 makes the solution candidate the original model, and changes a part of the combination of where the measurement sensor is installed to make the solution candidate the original model. It is possible to create a variant of , and perform an evaluation (first evaluation) of expected measurement accuracy for the original and variant. In other words, suitable solution candidates can be found with a smaller amount of calculations than when estimation and evaluation are performed in a brute-force manner for all solution candidates.

Furthermore, through repeated control, a new solution candidate is created for each elite-selected solution candidate based on the results of measurement accuracy evaluation (first evaluation), and the elite-selected solution candidate and new solution candidate are By repeating the estimation and measurement accuracy evaluation (first evaluation) again, it is possible to realize an efficient search for an optimal solution or an approximate solution thereof based on a genetic algorithm.

In addition, the placement plan creation support system 1000 performs an evaluation (second evaluation) regarding the total placement cost, and determines the placement cost according to the circumstances of each candidate position and the appropriate placement cost that takes into account the total placement cost (so to speak, the upper limit of the budget). Candidate solutions can be selected.

In addition, the layout plan creation support system 1000 can use measurement sensors including automatic ticket gates and existing CCD cameras, since automatic ticket gates and existing CCD cameras can be used as measurement sensors at zero or low cost. Appropriate solution candidates for placement can be selected.

[Modified example]
Note that the embodiments to which the present invention is applicable are not limited to the above-described embodiments, and components thereof may be changed, added, or omitted as long as the essence of the present invention is not changed.

For example, although the support for creating a layout plan for measuring OD traffic in the station premises 10 has been exemplified, the present invention is also applicable to buildings and facilities other than stations.

Further, for example, although an example is shown in which an entropy maximization model is used as a method for estimating OD traffic volume, a configuration in which other methods are applied is also possible. For example, absorbing Markov chain model (Junichi Takayama, Tomomi Sugiyama, Research on OD estimation method from observed traffic volume using absorbing Markov chain, Proceedings of Japan Society of Civil Engineers, No. 569/IV-36, pp. 75-84, 1997 ), Frater method (Shogo Kawakami, On the goodness of fit of OD traffic prediction models, Proceedings of the Japan Society of Civil Engineers, No. 165, p. 32, 1969 / Toshiyoshi Morisugi, Toshihiko Miyagi: Evaluation of urban transportation projects, pp. 51-53, Coronasha, 1996), etc. can also be applied.

Furthermore, the evaluation method of solution candidates is not limited to the above embodiment, and can be set as appropriate. For example, by using the correlation coefficient R or determination coefficient ^R2 between the verification OD traffic volume and the estimated OD traffic volume as evaluation index values, solution candidates with large correlation coefficients R or determination coefficients ^R2 are treated as highly evaluated. You can also use it as Furthermore, if balance with the budget is important, the reciprocal of the total placement cost 537 may be used as the evaluation index value, and the larger the reciprocal (=the smaller the total placement cost), the higher the evaluation.

Further, in the above embodiment, the layout plan creation support system 1000 is illustrated as a stand-alone computer system (see FIG. 8), but the present invention is not limited to this. For example, it may be realized as a client-server system composed of a user terminal and a server system.

2... Pedestrian 10... Station premises 11... Platform 13... Station entrance/exit 15... Station passage 17... Automatic ticket gate 19... CCD camera 20... Pedestrian traffic network 22... Solution candidate application network 202... Network creation section 204... Candidate position setting Sections 206... Placement cost setting section 208... Total placement cost upper limit setting section 210... Solution candidate creation section 212... Total placement cost evaluation section 214... Total placement cost calculation section 216... Traffic volume setting section for verification 218... Estimation section 220... Measurement Accuracy evaluation section 222... Evaluation index value calculation section 224... Repetition control section 226... Presentation control section 501... Support program 510... Pedestrian traffic network setting data 512... Candidate position setting data 516... Placement cost setting data 518... Placement cost 520... Total placement cost upper limit value 530... Solution candidate setting data 534... Placement setting 537... Total placement cost 539... Evaluation index value 540... Traffic volume setting data for verification 543... OD table 544... Total departure traffic volume for verification 545... Data for input 546...Total arrival traffic volume for verification 550...Estimated traffic volume data 1000...Layout plan creation support system 1150...Control board Cy (y=1, 2,...)...Link NDx (x=1, 2,...)...Node Pz (z=1,2,...)...Candidate position

Claims (6)

A placement plan creation support system that supports the creation of a placement plan for measurement sensors for calculating pedestrian OD (Origin-Destination) traffic volume, the system comprising:
Network creation means for setting pedestrian passage points in the station premises as nodes, including OD nodes that are departure points/destinations of the OD traffic volume, and creating a pedestrian traffic network in which adjacent nodes are connected by links. and,
A candidate position where the measurement sensor can be placed in the pedestrian traffic network , and corresponds to the link where the pedestrian traffic volume can be measured when the measurement sensor is placed at the candidate position. candidate position setting means for setting a plurality of candidate positions;
Solution candidate creation means for creating solution candidates by combining some of the candidate positions from among the candidate positions;
By setting the verification traffic volume between the OD nodes, the total verification departure traffic volume , which is the sum of the traffic volume starting from the OD node, and the OD node as the destination can be set for each OD node. a verification traffic volume setting means that calculates and sets a verification total arrival traffic volume, which is the sum of the traffic volumes, from the verification traffic volumes between the OD nodes ;
The total departure traffic volume for verification and the total arrival traffic volume for verification for each of the OD nodes, and the traffic volume of the link whose traffic volume can be measured at a candidate position related to the solution candidate in the pedestrian traffic network. and a solution candidate application network consisting of a partial network in which measurable links are defined . an estimation means for estimating the estimated traffic volume between each OD node corresponding to the traffic volume measurable link ;
a first evaluation means for evaluating the solution candidate by comparing the verification traffic volume and the estimated traffic volume for each of the OD nodes;
A layout plan creation support system equipped with. Repetitive control means for causing the solution candidate creation means to create a new solution candidate by changing a part of the solution candidate, and for causing the estimation means to estimate the new solution candidate and the first evaluation means to evaluate the new solution candidate. ,
The layout plan creation support system according to claim 1, further comprising: placement cost setting means for setting placement cost of the measurement sensor for each candidate position;
Total placement cost calculation means for calculating a total placement cost when the measurement sensor is placed at a candidate position related to the solution candidate based on the placement cost;
a second evaluation means for evaluating the solution candidates based on the total placement cost;
The layout plan creation support system according to claim 1 or 2, further comprising: The placement plan is a placement plan of measurement sensors for calculating the OD traffic volume of pedestrians in the station premises,
The candidate position setting means regards an automatic ticket gate and/or an existing camera (hereinafter collectively referred to as "existing sensor") in the station premises as the measurement sensor, and sets the installation position of the existing sensor to the candidate position. Set including
The placement cost setting means sets the placement cost to zero or a predetermined low cost for the candidate position corresponding to the installation position of the existing sensor.
The layout plan creation support system according to claim 3. Presentation means for presenting the solution candidates whose evaluation results by the first evaluation means and the evaluation results by the second evaluation means satisfy a predetermined high evaluation condition as a recommended combination of candidate positions of the measurement sensor;
The arrangement plan creation support system according to claim 3 or 4, further comprising: A layout plan creation support method for a computer system to support the creation of a layout plan for measurement sensors for calculating pedestrian OD (Origin-Destination) traffic volume, the method comprising:
Setting the pedestrian passage points including the OD nodes that are the departure points/destinations of the OD traffic volume as nodes, and creating a pedestrian traffic network in which adjacent nodes are connected with links ;
A candidate position where the measurement sensor can be placed in the pedestrian traffic network , and corresponds to the link where the pedestrian traffic volume can be measured when the measurement sensor is placed at the candidate position. By setting multiple candidate positions,
Creating a solution candidate by combining some of the candidate positions from among the candidate positions;
By setting the verification traffic volume between each of the OD nodes, for each OD node, the total verification departure traffic volume, which is the sum of the traffic volume starting from the OD node , and the OD node as the destination Calculating and setting a total arrival traffic volume for verification , which is the sum of the traffic volume , from the traffic volume for verification between each of the OD nodes ;
The total departure traffic volume for verification and the total arrival traffic volume for verification for each of the OD nodes, and the traffic volume that is the link whose traffic volume can be measured at a candidate position related to the solution candidate in the pedestrian traffic network. and a solution candidate application network consisting of a partial network in which measurable links are defined . Estimating the estimated traffic volume between each OD node corresponding to the traffic volume measurable link ;
Evaluating the solution candidate by comparing the verification traffic volume and the estimated traffic volume for each of the OD nodes;
A layout plan creation support method including

Images