KR102263608B1 - System for guiding multi road and method thereof - Google Patents

Abstract

및 i 에서 j 로의 서비스 링크의 새도우 프라이스

를 도출하며, k-shortest routing path 알고리즘에서 도출된 k- 최단 경로에 대해, 링크 기반의 하이퍼 패스 알고리즘의 수학적 모델에서의 Pricing 문제의 해로부터 감소된 코스트 Z' 를 도출하고, 도출된 새도우 프라이스

와 감소된 코스트 Z' 의 비교 결과를 토대로 하이퍼 경로를 도출함에 따라 대기 시간, 이동 시간 및 이동 시간의 리스크를 반영된 하이퍼 경로를 도출할 수 있고, 최단 경로 P1 를 포함하는 서브 셋에 포함된 모든 경로에 대해 각 경로 마다 소정 회 반복 수행함에 따라, 도출된 하이퍼 경로에 대한 신뢰성이 향상될 수 있다.The present technology discloses a multi-path recommendation system and method. According to a specific implementation example of the present technology, the shadow price from the solution of the Master problem in the mathematical model of the path-based hyperpath algorithm for the shortest distance P1 derived from the Dijkstra algorithm (Dijkstra's algorithm)

and the shadow price of the service link from i to j

For the k-shortest path derived from the k-shortest routing path algorithm, the reduced cost Z' is derived from the solution of the Pricing problem in the mathematical model of the link-based hyperpath algorithm, and the derived shadow price

By deriving a hyper route based on the comparison result between and reduced cost Z', it is possible to derive a hyper route that reflects the risks of waiting time, travel time, and travel time, and all routes included in the subset including the shortest route P1 By repeatedly performing a predetermined number of times for each path, the reliability of the derived hyper path may be improved.

Description

SYSTEM FOR GUIDING MULTI ROAD AND METHOD THEREOF

The present invention relates to a multi-path recommendation system and method, and more particularly, to the hyper-path searched by traversing the hyper-route of public transportation in consideration of the travel time, waiting time, and risk of travel time of public transportation. It relates to a technology that can further improve reliability.

The existing wayfinding system using public transportation not only predicts the waiting time but does not provide a navigation service for public transportation that can minimize the waiting time, and similar inventions also provide only the average travel time. it was only

In order to solve the above problems, the present invention provides a travel time, waiting time and movement of public transportation from a departure point to a destination based on a user terminal possessed by a user who has requested a multi-path search service and an application that provides a multi-path search service. An object of the present invention is to provide a multi-path recommendation system and method that can provide a user with a multi-path of public transportation reflecting time uncertainty.

Accordingly, it is an object of the present invention to provide a multi-path recommendation system and method that can reduce the travel time of public transportation and the waiting time at the boarding stop to a minimum, and use public transportation easily and conveniently.

According to an aspect of an embodiment, in a multi-path recommendation method for searching a multi-path of a departure point and a destination of a public transportation means on a web-based network and delivering it to a user terminal requesting a service,

The multi-path recommendation method is

With respect to the multi-path searched between the origin and the destination, the hyper-path of the shortest route among a plurality of multi-paths is derived and provided to the user terminal by reflecting the risk of travel time, transfer stop waiting time, and travel time characterized.

Preferably, the method comprises:

For each of the multiple routes of public transportation between the destination and the departure point, the Master step of minimizing the transit time of the public transportation and the waiting time of the transfer stop as a solution to the Master problem of the mathematical model of the established route-based hyperpath algorithm;

Pricing step of minimizing the risk of non-linear travel time as a solution to the pricing problem of the mathematical model of the pre-established link-based hyperpath algorithm; and

It may include a hyper-path deriving step of deriving a hyper-path that minimizes the risk of the travel time, the waiting time of the transfer stop, and the travel time.

Preferably, the Master step is

For each of the multiple routes of public transport between the destination and the origin, the shortest distance is derived through the Dijkstra algorithm,

Derives the solution of the master problem of the mathematical model of the path-based hyperpath algorithm derived by the heuristic method for a subset of multiple paths including the derived shortest distance,

As a solution to the derived Master problem, it may be provided to derive a shadow price for each path of the subset and a service shadow price between links i to j.

Preferably, the pricing step is

Update the arc cost with the sum of the service shadow price and the previous arc cost between links i to j, and then derive k shortest path through k shortest routing path algorithm with arc cost,

For the derived shortest path, it may be provided to output a reduced cost by deriving a solution to the pricing problem of the mathematical model of the link-based hyper-path algorithm derived by the column generation technique.

Preferably, the hyper-path derivation step comprises:

It may be provided to derive the hyper path based on a comparison result of the shadow price and the reduced cost.

According to another aspect of an embodiment, in the multi-path recommendation system for searching for multiple routes of the source and destination of public transportation on a network based on the web and delivering it to the user terminal that requested the service, for the multi-path searched between the source and the destination, , by reflecting the risk of travel time, transfer stop waiting time, and travel time to derive a hyper-path of the shortest path among a plurality of multi-paths and provide it to the user terminal,

For each of the multiple routes of public transportation between the destination and the departure point, a Master unit that minimizes the transit time of the public transportation and the waiting time of the transfer stop as a solution to the Master problem of the mathematical model of the established route-based hyperpath algorithm;

A pricing unit that minimizes the risk of non-linear travel time as a solution to the pricing problem of the mathematical model of the link-based hyperpath algorithm that has been established; and

It may include a hyper route derivation unit for deriving a hyper route that minimizes the risk of the travel time, the waiting time of the transfer stop, and the travel time.

및 i 에서 j 로의 서비스 링크의 새도우 프라이스

를 도출하며, k-shortest routing path 알고리즘에서 도출된 k- 최단 경로에 대해, 링크 기반의 하이퍼 패스 알고리즘의 수학적 모델에서의 Pricing 문제의 해로부터 감소된 코스트 Z' 를 도출하고, 도출된 새도우 프라이스

와 감소된 코스트 Z' 의 비교 결과를 토대로 하이퍼 경로를 도출함에 따라 대기 시간, 이동 시간 및 이동 시간의 리스크를 반영된 하이퍼 경로를 도출할 수 있다. According to an embodiment, the shadow price from the solution of the Master problem in the mathematical model of the path-based hyperpath algorithm for the shortest distance P1 derived from the Dijkstra algorithm (Dijkstra algorithm).

and the shadow price of the service link from i to j

For the k-shortest path derived from the k-shortest routing path algorithm, the reduced cost Z' is derived from the solution of the Pricing problem in the mathematical model of the link-based hyperpath algorithm, and the derived shadow price

By deriving a hyper route based on the comparison result between and the reduced cost Z', it is possible to derive a hyper route reflecting the risks of waiting time, travel time, and travel time.

In addition, according to an embodiment, since all paths included in the subset including the shortest path P1 are repeatedly performed a predetermined number of times for each path, reliability of the derived hyper path may be improved.

The following drawings attached to this specification illustrate preferred embodiments of the present invention, and serve to further understand the technical spirit of the present invention together with the detailed description of the present invention to be described later, so the present invention is a matter described in such drawings should not be construed as being limited only to
1 is an overall flowchart illustrating a process of a multi-path recommendation method according to an embodiment.
2 is a detailed configuration diagram of a Master step of a method according to an embodiment.
3 is a detailed configuration diagram of a Pricing step of a method according to an embodiment.
4 is a detailed configuration diagram of a hyper-path derivation step of the method according to an embodiment.
5 is a block diagram of a multi-path recommendation system according to an embodiment.
6 is an exemplary view showing a public transportation network in Seoul and Seongnam according to an embodiment.
7 is a comparison diagram of performance ratios of a single shortest path and a hyper path.
8 is a comparison diagram of performance ratios of equal probabilities of a single shortest path and a hyper path.
9 is a graph showing the sensitivity analysis of the total required time to the variation of the path k value.
10 is a graph showing the sensitivity analysis of the calculation speed of the probability change and the total required time.
11 is a graph of the probability of arrival within a total estimated time required according to an embodiment.
12 is a graph showing the probability in which the probability and the covariance are reflected according to an embodiment.
13 is a diagram showing a path using a hyper path and another application according to an embodiment.
14 is a graph showing the travel time required for a hyper path and a path of another application according to an embodiment.

Advantages and features of the present invention and methods of achieving them will become apparent with reference to the embodiments described below in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only these embodiments allow the disclosure of the present invention to be complete, and common knowledge in the art to which the present invention pertains It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims.

Terms used in this specification will be briefly described, and the present invention will be described in detail.

The terms used in the present invention have been selected as currently widely used general terms as possible while considering the functions in the present invention, but these may vary depending on the intention or precedent of a person skilled in the art, the emergence of new technology, and the like. In addition, in a specific case, there is a term arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the description of the corresponding invention. Therefore, the term used in the present invention should be defined based on the meaning of the term and the overall content of the present invention, rather than the name of a simple term.

In the entire specification, when a part "includes" a certain component, it means that other components may be further included, rather than excluding other components, unless otherwise stated. Also, as used herein, the term “unit” refers to a hardware component such as software, FPGA, or ASIC, and “unit” performs certain roles. However, "part" is not meant to be limited to software or hardware. A “unit” may be configured to reside on an addressable storage medium and may be configured to refresh one or more processors.

Thus, by way of example, “part” includes components such as software components, object-oriented software components, class components and task components, processes, functions, properties, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuitry, data, databases, data structures, tables, arrays and variables. The functionality provided within components and “parts” may be combined into a smaller number of components and “parts” or further divided into additional components and “parts”.

Hereinafter, with reference to the accompanying drawings, embodiments of the present invention will be described in detail so that those of ordinary skill in the art can easily implement them. And in order to clearly explain the present invention in the drawings, parts irrelevant to the description will be omitted.

According to an embodiment, by applying the mathematical model of the path-based hyper-path algorithm to the mathematical model of the link-based hyper-path algorithm, the hyper-path (α-reliable routing) having a probability (α%) that can be derived within the expected arrival time (α-reliable routing) Path) to the user terminal, it is possible to minimize the risk of travel time, waiting time, and travel time of public transportation.

That is, the mathematical model of the path-based hyperpath algorithm satisfies Equation 1 below, and the mathematical model of the link-based hyperpath algorithm to which the path with probability (α%) (α-reliable Path) is applied satisfies the following Equation 2 do.

[Equation 1]

와 환승 정류소의 대기 시간의 합

에 대한 최소화로 도출된다. 이에 경로 기반의 하이퍼 패스 알고리즘의 수학적 모델을 이용하여 대중 교통 수단의 이동 시간 및 대기 시간을 최소화할 수 있다.Here, the sum of the travel times of the links used by the path-based hyperpath algorithm solution (total estimated time required)

and the sum of the waiting time at the transfer stop

It is derived as a minimization of Accordingly, the travel time and waiting time of public transportation can be minimized by using a mathematical model of the route-based hyperpass algorithm.

This path-based hyperpath algorithm satisfies the following condition 1.

[Condition 1]

[Equation 2]

Here, the following condition 2 of the mathematical model for the link-based hyper-path algorithm to which the α-reliable path is applied with a probability (α%) is as follows.

[Condition 2]

The parameters and decision variables of Equations 1 and 2 are shown in Table 1 below.

[Table 1]

Accordingly, one embodiment derives the solution of the Master problem from the mathematical model of the hyperpath algorithm of Equation 1 for the shortest distance derived using the Dijkstra algorithm (Dijkstra algorithm), and k-shortest routing Path (k-shortest path routing). ) for the k-shortest path derived by the algorithm, the hyperpath that can minimize the risk of waiting time, travel time, and travel time of public transportation by deriving the solution of the Pricing problem of the mathematical model of the hyperpath algorithm of Equation 2 Derive a path (Hyper Path).

Accordingly, according to an embodiment, a plurality of hyper-paths for the input source and destination are delivered to the user in consideration of uncertainty about the travel time, waiting time, and travel time of public transportation.

1 is a flowchart of a multi-path recommendation method according to an embodiment, and may include a master step 100 , a pricing step 200 , and a hyper-path derivation step 300 .

Here, the Master step 100 is the solution of the Master problem of the mathematical model of the path-based hyper-path algorithm derived using the heuristic technique for the shortest distance P1 derived from the Dijkstra algorithm (Dijkstra algorithm). can be derived. That is, the amount of change in the total estimated time required according to the change of the path of the subset is derived by the solution of the master problem.

That is, the solution (total estimated time required) of the master problem of the mathematical model of the path-based hyper-path algorithm derived using the heuristic technique is expressed by Equation 3 below.

[Equation 3]

는 Path p 의 볼륨값(Volume of Path P),

는 노드(node) i의 대기 시간(waiting time of node i),

는 Path p 의 이동 시간의 표준편차값(standard deviation of the travel time),

는 링크 a 의 주파수(Frequency of Link a),

는 노드 i 의 나가는 서비스 링크 셋(Set of Outgoing Service Link in Node i) ,

는 대기 노드의 셋(set of Waiting Node),

는 α신뢰 레벨에서 표준 정규 분포의 역 누적 밀도 함수(Inverse accumulative density function of standard normal distribution at A confidence level)이다.Here, P1 is a total Path Set (set of total Path), P ' is the subset (subset of total Path), Cp is a traveling time (travel time Expected of Path P) of the total p Path Path,

is the volume value of Path p (Volume of Path P ),

The node (node) i waiting time (waiting time of node i) of,

is the standard deviation of the travel time of Path p,

The frequency of link a (Frequency of Link a),

It is out of service link set (Set of Outgoing Service Link in Node i) of node i,

is a set of Waiting Nodes,

is the inverse accumulative density function of standard normal distribution at A confidence level at α confidence level.

Accordingly, when the master step 100 sets the destination node I and the original node, the shortest path P1 is derived based on the Dijkstra algorithm (steps 110 and 120), and for the derived shortest path P1 , the shortest path P1 grouped by an arbitrarily determined number of paths is The included subset P' is set (step 130). The subset P' in the initial stage is equal to the shortest distance P1.

, i에서 j로의 서비스 링크의 새도우 프라이스

를 연산한다. 여기서, 새도우 프라이스(Shadow Price)

는 식 3에서 각 경로의 볼륨

의 총합이 1인 조건에서 1의 변동량에 따른 총 예상 소요 시간의 변화량이고, i에서 j로의 서비스 링크의 새도우 프라이스

는 식 3에서 각 링크 별로 형성되는 환승 정류소에서의 대기 시간

을 결정하는 조건에서 0의 변동량에 따른 총 예상 소요 시간의 변화량을 의미한다.Then, the master step 100 derives the solution of the master problem of the mathematical model of the hyper-pass algorithm of Equation 1 for the subset P' (step 140), and the shadow price as a solution of the derived master problem (Shadow Price)

, shadow price of service link from i to j

calculate Here, Shadow Price

is the volume of each path in Equation 3

It is the amount of change in the total estimated time required according to the amount of change of 1 under the condition that the sum of is 1, and the shadow price of the service link from i to j

is the waiting time at the transfer stop formed for each link in Equation 3

It means the amount of change in the total estimated required time according to the amount of change of 0 in the condition for determining .

, i에서 j로의 서비스 링크의 새도우 프라이스

는 Pricing 문제 단계(200)로 전달된다.These shadow prices

, shadow price of service link from i to j

is passed to the Pricing problem step 200 .

, i에서 j로의 서비스 링크의 새도우 프라이스

에 대해, 식 2의 column generation 기법을 이용하여 도출된 링크 기반의 하이퍼 패스 알고리즘의 수학적 모델의 Pricing 문제의 해로 감소된 코스트 Z' 를 연산한다. Here, the Pricing step 200 is a shadow price calculated from the solution of the Master problem in Equation 3

, shadow price of service link from i to j

For , calculate the reduced cost Z' by the solution of the pricing problem of the mathematical model of the link-based hyperpath algorithm derived using the column generation technique of Equation 2.

Here, the pricing problem in the mathematical model of the link-based hyperpath algorithm is expressed by Equation 4 below.

[Equation 4]

Each parameter and decision variable of Equation 3 and Equation 4 is shown in Table 2 below.

[Table 2]

, i에서 j로의 서비스 링크의 새도우 프라이스

에 대해, 네트 워크 내 각 링크 별 평균 이동 시간인 아크 코스트(arc cost)를 이전 아크 코스트 C _ij 와 i에서 j로의 서비스 링크의 새도우 프라이스

의 합으로 업데이트하고(단계 210), 업데이트한 아크 코스트

와 기 정해진 k-shortest routing Path 알고리즘으로 k 최단 거리 경로를 도출한다(단계 220).That is, the Pricing step 200 is a shadow price (Shadow Price)

, shadow price of service link from i to j

For , the arc cost, which is the average travel time for each link in the network, is the previous arc cost C _ij and the shadow price of the service link from i to j

update to the sum of (step 210), and the updated arc cost

and a predetermined k-shortest routing path algorithm to derive k shortest-distance paths (step 220).

And, the pricing step 200 for k shortest path path, Calculate the reduced cost Z' as a solution to the Pricing problem of Equation 4 (step 230), where the reduced cost Z' is transferred to the hyperpath derivation step 300. Here, the reduced cost Z' is It represents the amount of change in the total estimated time required between the origin and destination depending on whether each route or link is used.

에 미만인 경로를 하이퍼 경로로 설정하고, 해당 서브 셋의 모든 경로에 대해 반복 수행된다. In the hyper-path derivation step 300, the reduced cost Z ' is Shadow Price

A path less than . is set as a hyper path, and it is repeated for all paths in the subset.

에 미만인 지를 판단하고(단계 310), 판단 결과 감소된 코스트 Z' 가 새도우 프라이스(Shadow Price)

에 미만인 경우 기 정해진 소정 수를 카운팅하는 카운팅값이 소정 수에 도달하였는 지를 판단하며(단계 320), 판단 결과 카운팅값이 소정 수에 도달하지 아니하면 카운팅값을 증가한 다음 상기 단계(230)으로 진행한다(단계 330). 한편, 판단 결과 카운팅값이 소정 수에 도달하면 해당 경로를 하이퍼 경로로 도출한 다음 본 프로그램은 종료된다(단계 340).That is, the hyper-path derivation step 300 has a reduced cost Z' Shadow Price

It is determined whether it is less than (step 310), and as a result of the determination, the reduced cost Z' is Shadow Price

If it is less than , it is determined whether the counting value for counting a predetermined number reaches a predetermined number (step 320), and if the counting value does not reach the predetermined number as a result of the determination, the counting value is increased and then proceed to step 230 (step 330). On the other hand, if the counting value reaches a predetermined number as a result of the determination, the corresponding path is derived as a hyper path, and then the present program is terminated (step 340).

에 미만이 아닌 경우 서브 셋의 다음 경로 P'+ P _k 에 대해 Master 문제의 해를 도출하는 단계(140)로 진행한다(단계 350).On the other hand, the reduced cost Z' in step 320 is Shadow Price

If it is not less than , the process proceeds to step 140 of deriving the solution of the Master problem for the next path P' + P _{k of the subset (step 350).}

및 i 에서 j 로의 서비스 링크의 새도우 프라이스

를 도출하며, k-shortest routing path 알고리즘에서 도출된 k- 최단 경로에 대해, 링크 기반의 하이퍼 패스 알고리즘의 수학적 모델에서의 Pricing 문제의 해로부터 감소된 코스트 Z' 를 도출하고, 도출된 새도우 프라이스

와 감소된 코스트 Z' 의 비교 결과를 토대로 하이퍼 경로를 도출함에 따라 대기 시간, 이동 시간 및 이동 시간의 리스크를 반영된 하이퍼 경로를 도출할 수 있다. 이러한 일련의 과정은 최단 경로 P1 를 포함하는 서브 셋에 포함된 모든 경로에 대해 각 경로 마다 소정 회 반복 수행함에 따라, 도출된 하이퍼 경로에 대한 신뢰성이 향상될 수 있다.Accordingly, one embodiment is the shadow price from the solution of the Master problem in the mathematical model of the path-based hyper-path algorithm for the shortest distance P1 derived from the Dijkstra algorithm (Dijkstra algorithm).

and the shadow price of the service link from i to j

For the k-shortest path derived from the k-shortest routing path algorithm, the reduced cost Z' is derived from the solution of the Pricing problem in the mathematical model of the link-based hyperpath algorithm, and the derived shadow price

By deriving a hyper route based on the comparison result between and the reduced cost Z', it is possible to derive a hyper route reflecting the risks of waiting time, travel time, and travel time. As this series of processes is repeated a predetermined number of times for all paths included in the subset including the shortest path P1, the reliability of the derived hyper path can be improved.

5 is a view showing the configuration of a multi-path recommendation system according to an embodiment. As shown in FIG. 5 , the multi-path recommendation system S according to an embodiment includes a master unit 510 and a pricing unit 520. ), and a hyper-path derivation unit 530 .

, i에서 j로의 서비스 링크의 새도우 프라이스

를 연산한다. 여기서, 경로 기반의 하이퍼 패스 알고리즘의 수학적 모델의 Master 문제는 휴리스텍(Heuristic) 기법에 의해 도출되므로 연산 복잡도가 감소된다.For a destination node, transmitted from a user terminal (not shown) carried in the person requesting the navigation service, Master unit 510 by performing Dijkstra's algorithm search the shortest distance P1 and containing the found shortest distance P1 Set a subset of multiple paths, and for each path in the subset, the shadow price as a solution to the master problem P' of the mathematical model of the path-based hyperpath algorithm.

, shadow price of service link from i to j

calculate Here, since the master problem of the mathematical model of the path-based hyper-path algorithm is derived by the heuristic technique, the computational complexity is reduced.

, i에서 j로의 서비스 링크의 새도우 프라이스

는 Pricing 부(520)로 전달된다. These shadow prices

, shadow price of service link from i to j

is transmitted to the pricing unit 520 .

와 이전 아크 코스트 C _ij 를 기반으로 아크 코스트

를 설정하고, 설정된 아크 코스트

를 가지는 k shortest routing path 알고리즘을 수행하여 k 최단 경로를 탐색한다. 그리고 Pricing 부(520)는 탐색된 k- 최단 경로에 대해, Column Generation 기법에 의거 도출된 링크 기반의 하이퍼 패스 알고리즘의 수학적 모델에서의 Pricing 문제의 해로부터 감소된 코스트 Z' 를 도출한다. Pricing unit 520 sets the shadow price of the service link from i to j.

and the arc cost based on the previous arc cost C _ij

and set arc cost

The k shortest path is searched by performing the k shortest routing path algorithm with . And the pricing unit 520 derives a reduced cost Z' from the solution of the pricing problem in the mathematical model of the link-based hyper-path algorithm derived based on the column generation technique for the found k-shortest path.

와의 비교 결과를 토대로 모든 서브셋의 각 경로에 대해 일련의 과정을 반복 수행하여 하이퍼 경로를 도출하고, 각 경로 별 하이퍼 경로의 도출을 소정 회 반복 수행된다.And the hyper path derivation unit 530 is the reduced cost Z 'of the pricing problem unit 520 and the shadow price of the master problem unit 510 .

Based on the comparison result with , a series of processes are repeatedly performed for each path in all subsets to derive a hyper path, and the derivation of the hyper path for each path is repeated a predetermined number of times.

Accordingly, one embodiment derives the solution of the master problem of the mathematical model of the path-based hyper-path algorithm derived using the heuristic technique, and the mathematical model of the link-based hyper-path algorithm derived using the column generation technique. By deriving a risk of a non-linear form as a solution to the pricing problem, it is possible to derive multiple hyper-paths reflecting the risks of travel time, waiting time, and travel time and provide it to the user terminal.

<Example>

6 is a map of public transportation applied to an embodiment. Referring to FIG. 6, the public transportation networks of Gangnam, Seocho, Songpa, and Seongnam are 108 bus routes in Seongnam and 132 bus routes in Seoul. , 101 stations of the subway, and on foot.

Therefore, as shown in Table 1 below, the data frames collected using the API services of the public data portal and ODAY LAB are 240 bus routes and 101 subway stations, with a total of 5144, and the total number of nodes is 17645 and the number of links. A network network including 44927 is formed.

[Table 1]

Here, busNo: bus number, bus_id: bus ID, stationSeq: stop order, stationName: stop name, regionName: region name, stationId: stop ID, x: x-coordinate of stop, y: y-coordinate of stop, cluster: corresponding cluster , mean: average travel time between stops, std: standard deviation of travel time between stops, bus_Interval_Sat: Saturday dispatch interval, bus_Interval_Sun: Sunday dispatch interval, bus_Interval_Week: weekday dispatch interval.

For each bus route and/or subway station of each of these public transportation means, since the standard deviation of each link varies depending on the surrounding environment, it is necessary to separate the data, and the KL Divergence algorithm indicating the similarity between distributions for each surrounding environment use the

For example, we created 140 clusters according to 20 time zones (5-24:00) and 7 days of the week (Monday-Sunday) for all buses, and computed KL-Divergence values for every pair in each cluster and based on the calculation results. As a result, spectral clustering, a graph-based clustering technique, is performed using the KL-Divergence value for the adjacency matrix weights.

According to these clusters, the average and standard deviation of the travel times for each bus stop, and the covariance between each link are calculated, and the result value of the covariance between these links is used as data used to calculate the required time according to the search time period.

Meanwhile, in one embodiment, after identifying other stops within a range of 250m based on each stop node, calculating the distance between the reference stop and another stop using latitude and longitude, and then setting the average walking speed as 1.2m/s as the standard, the distance between the stops Converts to time to generate a walking link. This walking link enables movement of the shortest distance between stops using walking.

Accordingly, in one embodiment, a stop and a walking link within a radius of 500 m are generated as soon as coordinates are received as input of a departure point and a destination.

<Simulation>

The public transportation network of Seongnam City was targeted, and the total number of nodes in the network was 5554 and the number of links was 13656. After setting α=0.9 and randomly setting the origin and destination, the raw data derived by repeating the experiment 8300 times are shown in Table 2 below.

[Pot 2]

Here, #nodes: number of nodes, #paths: number of discovered paths, Dijk_risk: Dijkstra's algorithm optimal value including risk, Dijk_sol: Dijkstra's algorithm optimal value, Dijk_time: Dijkstra's algorithm solution time, NumsolP: number of derived optimal paths, PF_sol : Optimal value of PathFinder algorithm, PF_time : Solving time of PathFinder algorithm, r : Departure stop ID, s : Arrival stop ID, gap : Difference between Dijkstra's and PathFinder algorithm optimum values, ratio : PathFinder optimum ratio based on Dijkstra's optimum value, ratio_risk : Risk The ratio included, risk: is the risk of Dijkstra's algorithm single path.

Therefore, the performance ratio ratio of the hyper warning derived according to an embodiment and the total estimated time required for the shortest path in which the risk of travel time is not considered can be expressed by the following Equation 5.

[Equation 5]

7 is a graph of the performance ratio ration to the total estimated time derived based on Equation 5. Referring to FIG. 7, the left side of the reference line is shorter than that of a single shortest path, and the right side is the required time of the hyper path. It can be seen that this is longer. Accordingly, it can be confirmed that the total estimated time required for the hyper alert according to an embodiment is shorter than that of a single shortest path.

8 is a graph showing a result of comparing the performance between a single shortest path and a hyper path derived with a probability of 0.9. Referring to FIG. 8 , it can be seen that the performance of the hyper path is superior to that of the single shortest path.

9 is a graph showing the sensitivity analysis result of the total required time to the change in the value of path k. Referring to FIG. 9, as shown in (a), the computational complexity sharply increases as the number of paths k increases. It can be seen that, as shown in (b), it can be seen that the total estimated time required is reduced.

10 is a graph showing the results of analysis of the sensitivity of the calculation speed and the total required time to the variation of the probability α. Referring to FIG. 9, as shown in (a), as the probability α increases, there is little difference in the calculation speed. However, as shown in (b), it can be seen that the total estimated time required is increased.

In one embodiment, for the hyper route, it is assumed that the user takes the bus that comes first among the candidate buses that can be boarded at the stop, the departure point and the destination are a random point in Seongnam-si, and the date is April 16, 2018 - 4, 2018 On the 19th of the month, the time is 12 to 16:00, the network used is the network corresponding to cluster 1, the number of trials is 45 path choices * 150 simulations * 2 times = 13500 times, and the probability value is 0.9. Same as 3

[Table 3]

Here, num_Solpaths: the number of final derived paths, num_paths: the number of searched paths, percent: the probability of arriving within the derived time, solution: the estimated time required through the algorithm, r / s: the origin/destination, solve_time: the problem Solving speed, ant_value : Ideal time required to satisfy 90% probability, gap : (Estimated time taken through algorithm) - (Ideal required time).

11 is a graph showing the probability of arrival within the total estimated time required of the hyper route for each origin and destination. Referring to FIG. 11 , it can be seen that the probability of arrival within the expected time of the hyper route approaches from 84% to 90%. .

12 is a graph showing the probability of arrival within the total expected time of the hyper-path in consideration of the probability and covariance. Referring to FIG. 12, in an embodiment, the variance is considered when the calculation speed of the probability of arriving within the expected arrival time is averaged. Then, the probability of arrival is about 66%, and when both variance and covariance are considered, the probability of arrival is about 84%, confirming that the accuracy is significantly lowered.

In addition, when variance and covariance are considered, the operation speed is 5.32 seconds, and when only variance is taken into account, it is 4.81 seconds, indicating that there is little difference. Accordingly, the computation time of the covariance between links takes longer than the computation time of the variance, and if the covariance between the links is calculated and stored in advance, the computation time of the covariance can be shortened.

For example, when the starting point is the Taepyeong 5-geori stop and the destination is the Hope Grand Park stop, if only the variance is taken into consideration, the estimated time taken is about 35 minutes and the probability of arrival is 60%, but considering the variance and covariance together, the estimated time required is reduced. The probability of arriving in about 40 minutes is 89.2%.

13 is a diagram showing routes of departures and destinations using a plurality of algorithms. Referring to FIG. 13, the departures are searched for latitude-37.388506, longitude-127.093582, and destinations with latitude-37.414286, longitude-127.130832 at 2 pm on weekdays. In the case of map A, walk from the departure point to the Hallim apartment stop, take bus 340, get off at Seongnam City Hall stop, and walk to the destination on foot. In the case of map B, you can see that you are instructed to walk from the departure point to the Pangyo library stop, take bus No. 103, get off at the Rose Village stop, and then walk to the destination. In one embodiment, the hyper route is from the departure point to Pangyo. Walk to the library stop, take bus 350, get off at Yatap Station, and walk to the destination.

Accordingly, it can be seen that in the case of a hyper route in which the risk of travel time is reflected in an existing route, the estimated required time does not increase, but another route is searched and derived.

14 is a graph showing the travel time required for each route shown in FIG. 13. Referring to FIG. 14, it can be seen that for the three routes shown in FIG. 13, the route of the B map has a large deviation in travel time distribution. It can be seen that although the standard deviation of the A path is smaller than that of the B map, the average travel time is longer than that of the hyper path according to an embodiment. Accordingly, it can be confirmed that the hyper path according to an embodiment is a stable path that arrives first with a path having a smaller standard deviation than a path using other applications.

및 i 에서 j 로의 서비스 링크의 새도우 프라이스

를 도출하며, k-shortest routing path 알고리즘에서 도출된 k- 최단 경로에 대해, 링크 기반의 하이퍼 패스 알고리즘의 수학적 모델에서의 Pricing 문제의 해로부터 감소된 코스트 Z' 를 도출하고, 도출된 새도우 프라이스

와 감소된 코스트 Z' 의 비교 결과를 토대로 하이퍼 경로를 도출함에 따라 대기 시간, 이동 시간 및 이동 시간의 리스크를 반영된 하이퍼 경로를 도출할 수 있고, 최단 경로 P1 를 포함하는 서브 셋에 포함된 모든 경로에 대해 각 경로 마다 소정 회 반복 수행함에 따라, 도출된 하이퍼 경로에 대한 신뢰성이 향상될 수 있는 다중 경로 추천 시스템 및 방법에 대한 운용의 정확성 및 신뢰도 측면, 더 나아가 성능 효율 면에 매우 큰 진보를 가져올 수 있으며, 길안내 서비스 관련 컨텐츠 또는 오픈 플랫폼의 시판 또는 영업의 가능성이 충분할 뿐만 아니라 현실적으로 명백하게 실시할 수 있는 정도이므로 산업상 이용가능성이 있는 발명이다.Shadow price from the solution of the Master problem in the mathematical model of the path-based hyperpath algorithm for the shortest distance P1 derived from the Dijkstra algorithm.

and the shadow price of the service link from i to j

For the k-shortest path derived from the k-shortest routing path algorithm, the reduced cost Z' is derived from the solution of the Pricing problem in the mathematical model of the link-based hyperpath algorithm, and the derived shadow price

By deriving a hyper route based on the comparison result between and reduced cost Z', it is possible to derive a hyper route that reflects the risks of waiting time, travel time, and travel time, and all routes included in the subset including the shortest route P1 As it iteratively performs a predetermined number of times for each path, the reliability and reliability of the operation of the multi-path recommendation system and method that can improve the reliability of the derived hyper-path will be greatly improved in terms of performance and efficiency. It is an invention that has industrial applicability because the possibility of marketing or sales of content related to a navigation service or an open platform is sufficient, and it can be clearly implemented in reality.

Claims (6)

In the multi-path recommendation method of searching for multiple routes of a departure point and a destination of public transportation on a network based on a web and delivering it to a user terminal that has requested a service,
The multi-path recommendation method is
For the multi-path searched between the origin and the destination, the hyper-path of the shortest path among multiple multi-paths is derived by reflecting the risk of travel time, which is uncertainty about travel time, transfer stop waiting time, and travel time, and the user terminal Multi-path recommendation method, characterized in that provided to provide. The method of claim 1, wherein the method comprises:
For each of the multiple routes of public transportation between the destination and the departure point, the Master step of minimizing the transit time of the public transportation and the waiting time of the transfer stop as a solution to the Master problem of the mathematical model of the established route-based hyperpath algorithm;
Pricing step of minimizing the risk of non-linear travel time as a solution to the pricing problem of the mathematical model of the pre-established link-based hyperpath algorithm; and
and a hyper-path derivation step of deriving a hyper-path that minimizes the risk of travel time, transfer stop waiting time, and travel time. According to claim 2, The Master step,
For each of the multiple routes of public transport between the destination and the origin, the shortest distance is derived through the Dijkstra algorithm,
Derives the solution of the master problem of the mathematical model of the path-based hyper-path algorithm derived by the heuristic method for a subset of multiple paths including the derived shortest distance,
A multi-path recommendation method, characterized in that the shadow price for each path in the subset and the service shadow price between links i to j are derived as a solution to the derived master problem. According to claim 3, wherein the pricing step,
Update the arc cost with the sum of the service shadow price and the previous arc cost between links i to j, and then derive k shortest path through k shortest routing path algorithm with arc cost,
A multi-path recommendation method provided to output a reduced cost by deriving a solution to a pricing problem of a mathematical model of a link-based hyper-path algorithm derived by a column generation technique with respect to the derived shortest path. The method of claim 4, wherein the hyper-path deriving step comprises:
and to derive the hyper-path based on a comparison result of the shadow price and the reduced cost. A multi-path recommendation system that searches multiple routes of a departure point and a destination of public transportation on a web-based network and delivers them to a user terminal that has requested a service, the system comprising:
With respect to the multi-path searched between the origin and the destination, by reflecting the risk of travel time, transfer stop waiting time, and travel time, a hyper-path of the shortest path among a plurality of multi-paths is derived and provided to the user terminal,
For each of the multiple routes of public transportation between the destination and the departure point, a Master unit that minimizes the transit time of the public transportation and the waiting time of the transfer stop as a solution to the Master problem of the mathematical model of the established route-based hyperpath algorithm;
A pricing unit that minimizes the risk of non-linear travel time as a solution to the pricing problem of the mathematical model of the link-based hyperpath algorithm that has been established; and
and a hyper-path derivation unit for deriving a hyper-path that minimizes the risk of travel time, which is uncertainty about the travel time, the waiting time of the transfer stop, and the travel time.

Images