JP6453784B2 - Reliability evaluation apparatus, reliability evaluation method, and program - Google Patents

Description

  The present invention relates to a method for evaluating in advance the effects of natural disasters such as earth and sand disasters in various networks including information and communication networks.

There are various methods for evaluating network reliability in information communication networks. For example, Non-Patent Document 1 discloses a method for calculating the communication probability between src-dst of interest (between the start node and the end node) using the probability that each link constituting the network will fail. Non-Patent Document 1 is directed to a method of evaluating reliability for dealing with an unpredictable event such as a device failure.
As another related technique, Patent Document 1 discloses a method of defining reliability in consideration of node importance (user scale, traffic, etc.) and the like as reliability evaluation in a disaster. Patent Document 2 discloses a method for determining an upper node in consideration of which lower node is connected to an upper node and reliability is increased when the network has an upper / lower hierarchical structure. ing. Non-Patent Document 4 proposes a robust network design method for earthquakes.

JP 2014-23064 A JP 2014-93743 A

Hayashi, Abe, "Reliability of communication networks," The Institute of Electronics, Information and Communication Engineers, 2010. Hiroshi Saito, Ryoichi Kawahara, and Takeshi Fukumoto, Proposal of Disaster Avoidance Control, Networks 2014, 2014. Kuramoto et al., `` Study on the setting of non-linear landslide generation limit rainfall line using RBF network, '' JSCE Proceedings (672), 117-132, 2001. H. Saito. Spatial design of physical network robust against earthquakes.IEEE Journal of Lightwave Technology, 33 (2): 443-458, Jan. 2015. Sugaya et al., Operational control method against rainfall using effective rainfall index, Geotechnical Society, 60-3. http://www.jma.go.jp/jma/kishou/know/bosai/doshakeikai.html, January 12, 2016 search http://www.sabo.pref.hiroshima.lg.jp/portal/kaisetsu/kikenhelp/about_2.htm, January 12, 2016 search

  On the other hand, Non-Patent Document 2 proposes a method for performing disaster avoidance control by evaluating in advance the impact of a disaster on a network that can be predicted, such as heavy rain. In this case, it is necessary to evaluate the reliability of the network in consideration of the probability that a network facility (link or node) belonging to a specific area (for example, an area where a heavy rain warning is issued) will be damaged. At this time, for example, how to estimate the failure probability of the link in the area where the warning is issued, using weather information, warning information, or the like becomes a problem.

  The present invention has been made in view of the above points, and by identifying an area that may be damaged by using weather information and the like, after estimating the link failure probability in the area appropriately, It is an object to provide a technology that makes it possible to evaluate in advance the impact on the network when it occurs.

According to an embodiment of the present invention, in a geographical network composed of a link set and a node set, a part of the network is damaged within an alarm area specified based on information about an event that can cause a disaster. A reliability evaluation device for evaluating the impact of
Based on the metric r of the alarm area M_i related to the event at a certain time t, an intra-area link failure rate, which is a rate at which a link existing in the alarm area M_i fails, is calculated for the link j constituting the network. By acquiring link section information L_i, j that may be damaged in the alarm area M_i, and using the link section information L_i, j and the link failure rate in the area for each alarm area M_i, the link j has failed. a reliability evaluation device comprising a link failure probability calculation means for calculating a link failure probability is the probability that,
In the reliability evaluation device, a function for calculating an expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and g (r) Using the total area Sa of the hazard area in the area used to determine the area, the area Sb1 damaged per disaster occurrence, and g (r), the damage occurrence rate is g (r) x Sb1 / Sa It further comprises means for calculating the damage occurrence rate in the warning area to be calculated,
The link failure probability calculation means calculates the intra-area link failure rate using the damage occurrence rate and the probability that the damaged link is disconnected.
The reliability evaluation apparatus characterized by this is provided.

In addition, according to the embodiment of the present invention, in a geographical network composed of a link set and a node set, a part of the network within an alarm area specified based on information on an event that may cause a disaster. A reliability evaluation method executed by a reliability evaluation device that evaluates the impact of a disaster
Based on the metric r of the alarm area M_i related to the event at a certain time t, an intra-area link failure rate, which is a rate at which a link existing in the alarm area M_i fails, is calculated for the link j constituting the network. By acquiring link section information L_i, j that may be damaged in the alarm area M_i, and using the link section information L_i, j and the link failure rate in the area for each alarm area M_i, the link j has failed. a reliability evaluation method comprising a link failure probability calculating step of calculating a link failure probability is the probability that,
In the reliability evaluation method, a function for calculating an expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and g (r) Using the total area Sa of the hazard area in the area used to determine the area, the area Sb1 damaged per disaster occurrence, and g (r), the damage occurrence rate is g (r) x Sb1 / Sa It further includes a step of estimating the damage occurrence rate in the alarm area to calculate,
In the link failure probability calculation step, the reliability evaluation apparatus calculates the in-area link failure rate using the damage occurrence rate and the probability that the damaged link is disconnected.
The reliability evaluation method characterized by this is provided.

  According to the embodiment of the present invention, an area that may be damaged by using weather information or the like is specified, and a failure probability of a link in the area is appropriately estimated, and then a network when a disaster occurs It is possible to evaluate the impact on

It is a figure which shows the system configuration example in embodiment of this invention. It is a figure which shows the structural example of the reliability evaluation server 400 in embodiment of this invention. It is a flowchart of the calculation procedure which the reliability evaluation server 400 performs. It is a figure which shows the example of the alarm area displayed on the network. It is a figure which shows the example which analyzed the data regarding the index | exponent regarding rainfall, and disaster occurrence / non-occurrence (cited from nonpatent literature 3). It is a figure for demonstrating the example of failure rate calculation in an area. It is a figure for demonstrating the example of link failure probability calculation. It is a figure which shows the example of a graph-like network. FIG. 10 is a diagram for explaining a calculation example of a src-dst disconnection probability in the third embodiment. FIG. 10 is a diagram for explaining a calculation example of a src-dst disconnection probability in the third embodiment.

  Embodiments of the present invention will be described below with reference to the drawings. The embodiment described below is only an example, and the embodiment to which the present invention is applied is not limited to the following embodiment.

(System configuration)
FIG. 1A is a configuration diagram showing an example of a system configuration in the embodiment of the present invention (common to Examples 1 to 3). As shown in FIG. 1A, the system according to the embodiment of the present invention includes a network 100 to be managed, a network / geographic information management server 200, a weather data collection server 300, and a reliability evaluation server 400. FIG. 1A shows a configuration in which the network / geographic information management server 200, the weather data collection server 300, and the reliability evaluation server 400 are connected to the network 100 to be managed. Is an example, and the network 100 to be managed may not be connected to any or all of the network / geographic information management server 200, the weather data collection server 300, and the reliability evaluation server 400. In the embodiment of the present invention, it is assumed that the reliability evaluation server 400, the network / geographic information management server 200, and the meteorological data collection server 300 are connected by some kind of network. . The network / geographic information management server 200 and the meteorological data collection server 300 are also connected by a network and can communicate with each other.

  The network / geographic information management server 200 manages (stores) information related to devices (links and nodes) constituting the network 100. Specifically, as shown in FIG. 1B, as node information, the id and geographic position (for example, latitude / longitude) of each node v are managed. As shown in FIG. 1B, the link information includes an end point node 1, an end point node 2, and a geographical position of each link.

In the example of FIG. 1B, the geographical position of the link is expressed by a combination of the latitude and longitude of both end points of the link and the latitude and longitude of each dividing point when the link is divided into a plurality of sections. . For example, the first link in FIG. 1 (b) is divided into three sections,
[(lat1, lon1), (lat11, lon11), (lat12, lon12), (lat2, lon2)]. That is, the positions of both end points of the link are (lat1, lon1) and (lat2, lon2), and the two locations (lat11, lon11), (lat12, lon12) are division positions.

Below, each link section divided | segmented as mentioned above is called a link segment.
The network / geographic information management server 200 holds geographic mesh information. It is assumed that weather information (specifically, metrics (R_x, R_y)) to be described later is given for each geographic mesh represented by the geographic mesh information, and the geographic mesh information exists in the position of each mesh and the mesh. Has hazard area information.

Specifically, as shown in FIG. 1B, the geographic mesh information includes a mesh ID, a mesh position, and hazard area information. The mesh position is represented by a set of coordinates (latitude and longitude) of the end points (four places) of the mesh. Here, specific numerical values are described as an example, for example, (latitude, longitude) = (500.0, 500.0). However, for the purpose of image only, fictitious latitude and longitude are illustrated. The hazard area information indicates the position of one or a plurality of hazard areas corresponding to, for example, a sediment-related disaster hazard location, and holds the position of the hazard area existing in the corresponding mesh expressed by a combination of latitude and longitude. Among the “geographic mesh information” in FIG. 1B, for example, as hazard area information in the mesh 1111,
{(501.1,502.3), (501.0, 503.2), (502.3, 502.8), (502.3, 502.5), (501.1, 502.3)}, but the first hazard area is these This means that the point is expressed by a polygon connecting the points in this order.

  The network / geographic information management server 200 further holds link length data. The link length data is data indicating which geographic mesh and how long each link intersects. For example, in the link length data shown in FIG. 1B, the link [100, 200] indicates that the mesh 1111 and 1.2km overlap with the 2222 and 3.2km. This length (for example, 1.2 km) means that if the mesh 1111 becomes a warning area, a section length of 1.2 km may be damaged. More specifically, for example, the k-th hazard area existing in the target geographic mesh is H_k (k = 1, ..., n_i, where n_i is the number of hazard areas), and the geographic mesh and ∪H_k overlap Use the link length that intersects “∪H_k” is a union of the hazard areas existing in the geographic mesh. Alternatively, the link length intersecting with the overlapping portion of the geographic mesh and H_k is obtained for each k, and the sum of the link lengths for k = 1 to n_i is set as the link length. This link length can be calculated by using the hazard information in the link information and geographic mesh information.

  The set of nodes defined here is denoted as V, the set of links as E, and this geographical network as (V, E).

The meteorological data collection server 300 calculates a metric related to the weather condition of each geographic mesh for each fixed period. For example, the weather data collection server 300 uses a short-term index (hourly rainfall R_y [mm / h] in the last 60 minutes) and a long-term index (effective rainfall R_x [mm] at a half-life of 72 hours) for rainfall data. calculate. R_x is calculated according to the procedure described in Non-Patent Document 5, for example, using the current unit-time rainfall R and the previous period R_x (referred to as R_x_pre) by R_x = R + 0.5 ^{(1 / H)} R_x_pre . H is a parameter called half-life, which is set to 72 hours. In addition, as described in Non-Patent Document 5, the effective rainfall is a mathematical model of a temporal change in which the effect is greater in the previous rain with respect to the current time, and the effect is smaller as the time goes back. It is.

  In the present embodiment, a rain metric is used as a metric (index) related to a disaster. However, this is merely an example, and another index may be used.

  A configuration example of the reliability evaluation server 400 is shown in FIG. As illustrated in FIG. 2, the reliability evaluation server 400 includes an alarm area detection unit 401, an alarm area damage occurrence rate estimation unit 402, a link failure probability calculation unit 403, and a disconnection probability calculation unit 404. The contents of the processing of each unit will be described in detail in each embodiment described later.

  The reliability evaluation server 400 according to the embodiment of the present invention can be realized, for example, by causing a computer to execute a program describing the processing contents described in the embodiment of the present invention. That is, the function of the reliability evaluation server 400 is realized by executing a program corresponding to the process executed by the reliability evaluation server 400 using hardware resources such as a CPU and a memory built in the computer. Is possible.

  In the reliability evaluation server 400, data such as nodes, links, geographic meshes, link lengths, and the like acquired from the network / geographic information management server 200 and data such as metrics acquired from the weather data collection server 300 are stored in memory (stored). In accordance with the program, the CPU reads the data from the memory and executes the process, thereby calculating the disconnection probability between the start point node src and the end point node dst. As will be described later, the reliability evaluation server 400 calculates various functions, but the procedure of the function may be incorporated in the above program, or separately from the entire (main) program, A function processing module may be stored in the memory, and the function processing module may be used when function processing becomes necessary.

  The above-mentioned program can be recorded on a computer-readable recording medium (portable memory or the like), stored, or distributed. It is also possible to provide the program through a network such as the Internet or electronic mail.

(Processing performed by the reliability evaluation server 400)
Below, the content of the process which the reliability evaluation server 400 performs is demonstrated in detail as Example 1, Example 2, and Example 3. FIG.

  In the following embodiments, a landslide disaster caused by rain, which is a weather condition, is described as an example, but this is only an example.

  FIG. 3 is a flowchart of the disaster impact assessment procedure executed by the reliability assessment server 400 in the first embodiment. The process executed by the reliability evaluation server 400 will be described along the procedure shown in FIG.

<Step S1: Identification of alarm area M_i>
In step S1, the alarm area detection unit 401 of the reliability evaluation server 400 detects a geographic mesh where an alarm is issued from weather information or the like as an alarm area M_i. The weather information is an example of information related to an event that can cause a disaster. FIG. 4 shows an example of an alarm area in a geographical area including the network 100 to be managed. The shaded area in FIG. 4 indicates the alarm area. In this example, alarms are generated at five locations. For example, in the case of landslide disasters, landslide disaster alert information has been issued every 5 km mesh based on rainfall data (see Non-Patent Document 6), and in the case of landslide disasters, it has reached a predetermined standard for landslide disaster alert information. The area can be M_i.

<Step S2: Estimating the rate of damage in the area>
If the alarm area M_i is detected in step S1, the alarm occurrence rate estimation unit 402 in the alarm area reads the metric R_t, i related to the weather condition at the current time t in the alarm area M_i from the weather data collection server 300 in step S2. Calculate the damage occurrence rate in the alarm area M_i. As an example, the metric (R_x, R_y) at the current time t is used as R_t, i. That is, the metric (R_x, R_y) calculated for the geographic mesh of the alarm area M_i is used as R_t, i. Below, the calculation procedure of a disaster occurrence rate is demonstrated in detail.

  First, the damage occurrence rate estimation unit 402 in the alarm area collects a set of past disaster situations (occurrence / non-occurrence) and metric (R_x, R_y) values in a certain reference area. Here, the reference area is an area for analyzing the relationship between the metric and the disaster occurrence rate. For example, by specifying a certain geographic mesh (or a certain geographic area having the same size) from each geographic mesh as a reference area and accumulating metric values in that area, the metric and the disaster rate To analyze the relationship. Then, the analysis result in the reference area is used to estimate the disaster occurrence rate in the warning area where the warning is currently issued. Note that the analysis in the reference area may be executed by the reliability evaluation server 400, or an apparatus other than the reliability evaluation server 400 performs in advance, and the reliability evaluation server 400 performs the analysis result (that is, described later). Data for realizing the function G to be performed) may be stored in the storage means, and the analysis result may be used. The analysis method is not limited to a specific method, but the method disclosed in Non-Patent Document 3 is used in the present embodiment.

  Non-Patent Document 3 discloses collecting and analyzing metrics in a certain area. An example of collecting and analyzing such data is shown in FIG. 7B in Non-Patent Document 3, and FIG. 5 of the present application shows the figure.

  In FIG. 5 (FIG. 7B in Non-Patent Document 3), a curve having a value between 1 and −1 is plotted. This was learned from the collected data using RBFN (Radial Basis Function Network), where (R_x, R_y) is the input value, non-disaster occurrence is 1, and the disaster output is -1. It is a result. The output value in this learning is called RBFN value. A function that returns an RBFN value when (R_x, R_y) is input is described as G (R_x, R_y). Note that G (R_x, R_y) is considered to correspond to the expected value of a random variable that takes a value of -1 (disaster occurrence) or 1 (disaster occurrence) for the probability p of occurrence of a disaster. That is, it is considered that there is a relationship of “1 × (1-p) + (− 1) × p = G (R_x, R_y)”.

  The damage occurrence rate estimation unit 402 in the alarm area uses G (R_x, R_y) to calculate the probability p that a disaster will occur when given (R_x, R_y), p = (1-G ( R_x, R_y)) / 2. This equation is obtained by solving “1 × (1−p) + (− 1) × p = G (R_x, R_y)” with respect to p.

  Also, let Z1 be the number of disasters that occurred when a disaster occurred in the reference area. If the number of occurrences of rainfall defined in Non-Patent Document 3 (the number of rainfall data plotted as “occurrence” in FIG. 5) is Za and the number of landslides is Zb, the number of disaster occurrences Z1 = Estimated as Zb / Za. The method for estimating the number of disaster occurrences is not limited to this, and other methods may be used. The damage occurrence rate estimation unit 402 in the alarm area calculates the expected value Es of the number of disaster occurrences when (R_x, R_y) is given as Es = (1-G (R_x, R_y)) / 2 × Z1. .

  Next, the disaster area damage occurrence rate estimating unit 402 sets the area of the disaster occurrence area as Sb1 per disaster (eg, earth and sand disaster), and Sa as the total area of the hazard area in the alarm area , Es × Sb1 / Sa is defined as a ratio (referred to as “disaster occurrence rate”) that the point is included in the disaster occurrence area under the condition that the certain point is included in the hazard area.

  A calculation example of the damage occurrence rate will be specifically described with reference to FIG. As shown in FIG. 6A, it is assumed that there are four landslide disaster risk locations in the reference area. In addition, the specific description of the landslide disaster danger part is described in the nonpatent literature 7, for example.

  These four places are designated as hazard areas (that is, areas where sediment disasters may occur), and the total area (that is, the sum of the areas of the four places) is designated as Sa. Also, it is assumed that a landslide disaster occurs in any one of these four locations (for example, a hatched portion in the landslide disaster risk location in the drawing in FIG. 6A). This location is called a disaster occurrence area. Since it is not known in advance where the landslide disaster will occur, it is assumed that the area affected by the landslide disaster is one of these four locations, and Sb1 is the four locations. It is set as the representative value (one of average, median, etc.) of the area. Alternatively, a representative value of the area of one earth and sand disaster may be used as Sb1 from past data or the like without being limited to the reference area.

  In the example of FIG. 6A, any landslide disaster risk location is included in the area, but if there is a landslide disaster risk location where only a part of the area overlaps the reference area, the landslide disaster risk location Only the overlapping part of the dangerous part and the area is used for the calculation of Sa and Sb1, or the entire area of the sediment disaster dangerous part is used for the calculation of Sa and Sb1.

  The alarm occurrence rate estimation unit 402 in the alarm area calculates Es × Sb1 / Sa (Es = (1−G (R_x, R_y)) / 2 × Z1) for each alarm area. Here, the functions G, Z1, Sb1 (representative values of the area of each hazard area, etc.) and Sa (total area of the hazard area) are data obtained from the analysis of the reference area, and for each alarm area It is common. Metrics (R_x, R_y) are data for each alarm area.

<Step S3: Link failure probability calculation>
In step S3, the link failure probability calculation unit 403 uses the in-area damage occurrence rate Es × Sb1 / Sa to calculate the in-area link failure rate beta_t, i of the alarm area M_i at time t, as follows: beta_t, i = (Es * Sb1 / Sa) * pL / delta is calculated. The intra-area link failure rate is a rate at which links existing in the alarm area M_i fail. Here, pL represents the probability that the link will be disconnected when the point u on the link is included in the disaster occurrence area, and delta represents a predetermined section length. delta is set to a value corresponding to the link section length that is simultaneously affected by one earth and sand disaster (for example, set to the average length of the link length that intersects with one hazard area). Note that pL is set based on, for example, data at the time of a past disaster, or pL = 1 if there is no such data, and is considered to be disconnected at the time of a disaster.

  Specifically, the link failure probability calculation unit 403 reads link length data information from the network / geographic information management server 200 and identifies a link that overlaps the alarm area. For example, it is assumed that alarms are currently issued at five locations of mesh IDs 1111, 2222, 4444, 5555, and 6666. In this case, among the links belonging to the link length data shown in FIG. 1B, the link [100, 200] intersects with two locations 1111, 2222.

  When beta_t, i of mesh IDs 1111, 2222 are written as beta1111 and beta2222, respectively, the link failure probability calculation unit 403 calculates the link failure probability q of the link [100,200] as follows.

q = 1-(1-beta1111 * delta) ^ (1.2 / delta) x (1-beta2222 * delta) ^ (3.2 / delta)
More generally, the link failure probability calculation unit 403 calculates the failure probability q_j of the link j as follows.

q_j = 1-(1-beta_t, 1 * delta) ^ (L_1, j / delta) × (1-beta_t, 2 * delta) ^ (L_2, j / delta) ×… × (1-beta_t, NL * delta ) ^ (L_NL, j / delta)
Here, NL is the number of warning areas (geographic mesh number), and L_i, j is a link that may be damaged in M_i (M_i is the i-th area (i = 1 to NL) where the warning is issued). The section length. For example, the k-th hazard area existing in M_i is H_k (k = 1, ..., n_i, n_i is the number of hazard areas that overlap M_i), and the link length that intersects the overlapping part of M_i and ∪H_k is L_i, j And Alternatively, the link length that intersects the overlapping part of M_i and H_k is obtained for each k, and the sum of the link lengths for k = 1 to n_i is L_i, j. If the link j does not intersect any hazard area in M_i, L_i, j = 0.

A specific example will be described with reference to FIG. Of the five alarm areas shown in FIG. 7, in the example of the alarm area M_5, the link y intersects with two landslide hazard areas (hazard areas). Here, assuming that the first dangerous location intersects with 0.2 km and the next dangerous location intersects with 0.5 km, L_i, j of link y is L_5, y = 0.2 + 0.5 = 0.7 km.
The link failure probability calculation unit 403 performs the above calculation for all links. If there is a link that does not intersect any alarm area, the failure probability q of that link is set to zero.

<Step S4: src-dst cut probability calculation>
In step S4, the disconnection probability calculation unit 404 calculates the disconnection probability between a given node src-dst (start node-end node) using the failure probability q_j of each link j calculated in step S3.

  Specifically, first, the disconnection probability calculation unit 404, based on the link information of the network / geographic information management server 200, changes from a geographical network as shown in FIG. 4 to a graphical network (geographical) as shown in FIG. Excluding the information, only the connection relationship between nodes is narrowed down). The link failure probability obtained in step S3 is given to each link in the graph network.

  The disconnection probability calculation unit 404 calculates the disconnection probability between src and dst in the graph network. Existing techniques can be used for this calculation. As an example, the cutting probability calculation unit 404 calculates the probability that communication between src and dst cannot be performed using any of the methods described in Chapter 3 of Non-Patent Document 1, and outputs the probability as a cutting probability between src and dst. To do.

  The functional classification and processing procedure of the reliability evaluation apparatus 400 described above are examples. For example, after calculating Es and the area damage occurrence rate for each alarm area, the link failure probability for each link may be calculated using the calculation result, or the link failure probability for each link may be calculated. Among them, it is also possible to calculate Es and the damage occurrence rate in the area. Moreover, the reliability evaluation apparatus 400 may calculate a cutting probability as, for example, the following damage influence evaluation method. In addition, the method demonstrated below summarizes the description so far.

  The disaster impact assessment method assumes that there is a geographical network (V, E) composed of a link set E and a node set V, and designates two nodes on the network as a start node src and an end node dst. This is a disaster impact evaluation method for calculating a probability (disconnection probability) that communication between src-dst cannot be performed when a part of a network is damaged in an alarm area specified based on information or the like.

  Here, the i-th alarm area is M_i, and the metric regarding the weather condition at the current time t in M_i is R_t, i. On the other hand, when the value of the metric is r, the rate beta [1 / km] at which the link existing in the alarm area fails is defined as a function beta = f (r) of r. Using this function f, the failure rate beta_t, i at time t in M_i is calculated by beta_t, i = f (R_t, i).

  Then, of the links j constituting the network, the link section length that may be damaged in M_i is L_i, j. Using L_i, j and beta_t, i, the probability q_j that the link j will fail is calculated. This is performed for all links, and the probability (disconnection probability) that communication cannot be performed between given nodes src and dst is calculated using the probability q_j of failure of link j.

  When calculating the above L_i, j, for example, the k-th hazard area existing in M_i is H_k (k = 1,..., N_i, n_i is the number of hazard areas overlapping with M_i). Let L_i, j be the link length that intersects the overlapping part of H_k. Alternatively, the link length that intersects the overlapping part of M_i and H_k is obtained for each k, and the sum of the link lengths for k = 1 to n_i is L_i, j.

  As a method for calculating the beta, for example, first, a function for calculating an expected value Es of the number of disaster occurrences in an area is given as Es = g (r) based on data analysis of a certain reference area. r is the metric r of the weather condition. Sa is the total area of the hazard area in the area, Sb1 is the area affected by one disaster, and g (r) x Sb1 / Sa is a condition that a certain point is included in any hazard area Below, it is assumed that the point is included in the disaster occurrence area. Then, beta_t, i at time t is calculated by beta_t, i = (g (r) × Sb1 / Sa) × pL / delta. Here, pL is a probability that the link is disconnected when a point u on the link is included in the disaster occurrence area, and delta is a predetermined section length.

  As a method for obtaining the above g (r), for example, past disaster information and weather conditions at that time are used. For example, in the case of landslide disasters, when using two variables, a short-term index R_y (rainfall for the last 60 minutes, etc.) and a long-term index R_x (an index that takes into account the cumulative value of rainfall over the past several days) as metrics related to weather conditions, Collect past (R_x, R_y) data and the disaster occurrence status (occurrence, non-occurrence) at that time, and from the collected data, calculate the expected value Es of the number of disaster occurrences when a certain (R_x, R_y) is given Configure g (r), which is the function to output.

  Next, Example 2 will be described. Hereinafter, differences from the first embodiment will be described. In the second embodiment, another procedure for calculating the failure probability q_j of the link j using Es = (1-G (R_x, R_y)) / 2 × Z1 calculated in step S2 of the first embodiment will be described.

  In step S2 of the second embodiment, Es / SN = (1-G (R_x, R_y)) / 2 × Z1 / SN is output as a disaster occurrence rate per landslide hazard area (hazard area). Here, SN is the number of landslide hazard points (hazard areas) in the reference area. For example, in the example of FIG. 6A, SN = 4. When only a part of the hazard area overlaps the reference area, SN is counted including that area. Alternatively, counting is performed according to the ratio of the area where the hazard area overlaps the reference area (for example, when the area overlaps only half, counting is performed as 0.5).

  In the second embodiment, the link failure probability calculation unit 403 uses the damage occurrence rate “(1-G (R_x, R_y)) / 2 × Z1 / SN” per the above-mentioned landslide disaster risk location. Instead of step S3, the following step S3 ′ is performed.

  Here, in the second embodiment, in the link length data in the network / geographic information management server 200 shown in FIG. 1B, the link length L_i, j is defined as the number of landslide hazard points that intersect with the link j in M_i. . For example, in the example of the alarm area M_5 in FIG. 7, the link y intersects with two landslide hazard points, so L_i, j of the link y is L_5, y = 2. In the second embodiment, such data on the number of dangerous places for sediment disasters is stored in the network / geographic information management server 200 as “link length data”.

  In the example of the warning area M_5 in FIG. 7, any landslide disaster risk location is included in the warning area. However, as shown in the geographic meshes M_3 and M_4, the landslide disaster risk location spans multiple geographic meshes. Is present, for example, L_i, j is counted by assuming that the landslide disaster risk location belongs to the largest overlapping geographical mesh (M_4 in this example). For example, L_i, j of the link z that intersects with this sediment-related disaster risk location in M_3 is L_4, z = 1. Alternatively, all geographic meshes that overlap with the landslide disaster hazard part are all counted. When there are links z ′ (not shown in FIG. 7) intersecting with the sediment-related disaster risk locations at two locations M_3 and M_4, L_3, z ′ = 1 and L_4, z ′ = 1. Alternatively, when the alarm is issued in both geographic meshes M_3 and M_4, one of them (for example, L_3, z ′) is set to 1 and the other (L_4, z ′) is set to 0, and the subsequent step S3 ′ is performed. Alternatively, it may be counted according to the ratio of the area where the sediment disaster hazard and the geographic mesh overlap (or the ratio of the link length intersecting the geographic mesh to the link length intersecting the sediment disaster risk). . For example, if these sediment-related disaster hazard locations overlap at a rate of M_3 and 0.3, M_4 and 0.7, L_3, z ′ = 0.3 and L_4, z ′ = 0.7.

<Step S3 ′: Link Failure Probability Calculation>
In the second embodiment, the link failure probability calculation unit 403 calculates the failure probability q_j of the link j as follows.

q_j = 1-(1-gamma_t, 1) ^ (L_1, j) × (1-gamma_t, 2) ^ (L_2, j) ×… × (1-gamma_t, NL) ^ (L_NL, j)
Where gamma_t, i = (1-G (R_x, R_y)) / 2 × Z1 / SN × pL '
And (R_x, R_y) uses the value in the alarm area M_i. gamma_t, i indicates the probability that a certain landslide disaster risk location is actually damaged and the link is disconnected in M_i. That is, gamma_t, i in the second embodiment is an example of an intra-area link failure rate that is a rate at which a link existing in the alarm area M_i fails per hazard area.

  Here, pL ′ represents the probability that the link will be disconnected under the condition that the disaster has occurred. pL ′ is set based on past disaster data or the like, or pL ′ = 1 if there is no such data.

  Next, Example 3 will be described. Hereinafter, differences from the first and second embodiments will be described. In the third embodiment, the following step S3 ″ is performed instead of step S3 ′ in the second embodiment, and the following step S4 ″ is performed instead of step S4 in the first embodiment.

<Step S3 '': Link failure probability calculation>
In the third embodiment, the link failure probability calculation unit 403 uses a formula obtained by removing pL ′ from the gamma_t, i formula shown in the second embodiment.

  That is, gamma_t, i = (1-G (R_x, R_y)) / 2 × Z1 / SN. This gamma_t, i means the probability that each earth and sand disaster danger point (hazard area) belonging to the warning area M_i will be damaged. That is, it indicates the damage occurrence rate per landslide hazard area.

  This gamma_t, i and pL ′ (probability of link disconnection under the condition of being damaged) are notified to the disconnection probability calculation unit 404 as parameters for determining the link failure probability.

<Step S4 '': src-dst cut probability calculation>
In the third embodiment, the cutting probability calculation unit 404 uses gamma_t, i and pL ′ to simulate whether or not each landslide disaster risk location belonging to the alarm area M_i is damaged. For example, if gamma_t, i = 2/3, a random number is generated and it is determined that the landslide disaster risk area with probability 2/3 is damaged. This is carried out for each sediment-related disaster risk point in each warning area, the link that intersects with the sediment-related disaster risk point determined to be damaged is identified, a random number is generated for each identified link, and the link is set with probability pL ' Is determined to be disconnected. As a result, all links determined to be broken links are removed from the graphical network of FIG. After removing the link, check if there is a path between src-dst.

  In addition, as shown in M_3 and M_4 in FIG. 7, when there are landslide disaster risk locations across multiple warning areas, for example, the landslide disaster risk location in the largest overlapping alarm area (M_4 in this example) You may consider it to belong. That is, using the value of gamma_t, 4 in M_4, it is determined whether or not this landslide disaster risk location is damaged. Alternatively, it may be determined that the device belongs to both of M_3 and M_4, and it is determined whether the disaster is caused by using a probability value of either or both of gamma_t, 3 and gamma_t, 4. For example, it may be determined whether or not the disaster will occur with a probability of max {gamma_t, 3, gamma_t, 4}. Or you may determine whether it suffers from a disaster according to the ratio of the area where this earth and sand disaster danger part and an alarm area overlap. For example, if this sediment-related disaster risk location overlaps at a rate of M_3 and 0.3, M_4 and 0.7, it may be determined that the disaster will occur with a probability of gamma_t, 3 × 0.3 + gamma_t, 4 × 0.7.

  The above procedure is repeated T times. For example, if the number of times that there is no path is T1, the src-dst disconnection probability is output as T1 / T.

  The above procedure will be specifically described using the examples shown in FIGS. As shown in FIG. 9, here, there are five alarm areas, and it is assumed that there are earth and sand disaster risk points (hazard areas) from A to G.

  The cutting probability calculation unit 404 determines whether or not each landslide disaster risk location is damaged by the probability of gamma_t, i in the first trial. In other words, if gamma_t, i = 2/3 for a certain earth and sand disaster risk area, throw a coin that appears in the table with a probability of 2/3, and if the table appears, it will be damaged. Such a decision is made for each landslide hazard point.

  As a result, as shown in FIG. 10, it is assumed that A, D, E, and G are determined to be damaged. Here, as an example, it is assumed that pL ′ = 1 has been set, and links overlapping these disaster areas are disconnected. When a graph-like network excluding these links is constructed and it is checked whether there is a path between src and dst, it can be seen that there is no path as shown in FIG. 10 (first trial). The same procedure is performed for the second trial. As a result, as shown in FIG. 10 (second trial), in this case, there is a path between src and dst. Such a process is repeated T times, the number of times T1 when there is no pass is counted, and the disconnection probability between src and dst is output as T1 / T.

(Summary of embodiment)
As described above, according to the present embodiment, in the geographical network composed of the link set and the node set, one of the networks is included in the alarm area specified based on the information on the event that may cause the disaster. A reliability evaluation device that calculates a disconnection probability that is a probability that communication between a specified start node and an end node cannot be performed when a part is damaged, and an alarm area M_i related to the event at a certain time t Based on the metric r, the link failure rate in the area, which is the rate at which links existing in the alarm area M_i fail, is calculated, and the link j that constitutes the network may be damaged in the alarm area M_i Section information L_i, j is acquired, and the link section information L_i, j and the in-area link failure rate for each alarm area M_i are used. The link failure probability calculation means for calculating the link failure probability that is the probability that the link j will fail, and the link failure probability of each link constituting the network calculated by the link failure probability calculation means, There is provided a reliability evaluation apparatus comprising: a disconnection probability calculating means for calculating a disconnection probability that is a probability that communication between the start point node and the end point node cannot be performed.

  The link failure probability calculation unit 403 is an example of a link failure probability calculation unit, and the disconnection probability calculation unit 404 is an example of a disconnection probability calculation unit.

  The link section information L_i, j is, for example, the total of the link length of the part where the hazard area predetermined in the alarm area M_i and the link j intersect, or the hazard area predetermined in the alarm area M_i The number of hazard areas that intersect with the link j.

  In the reliability evaluation device, a function for calculating an expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and g (r) is Using the total area Sa of the hazard area in the reference area used for setting, the area Sb1 damaged per disaster occurrence, and g (r), the damage occurrence rate is g (r) x Sb1 / Sa The damage occurrence rate in the alarm area to be calculated, or the hazard occurrence rate as g (r) / SN using the number SN of hazard areas in the area used to determine g (r) and g (r) The link failure probability calculating means may further comprise a link failure probability calculating means for calculating the link failure probability in the area using the damage occurrence rate and the probability that the damaged link will be disconnected. The rate may be calculated. The alarm area damage occurrence rate estimation unit 402 is an example of an alarm area damage occurrence rate estimation unit.

  In the present embodiment, when a part of the network is damaged in an alarm area specified based on information on an event that may cause a disaster in a geographical network composed of a link set and a node set. And a reliability evaluation device that calculates a disconnection probability, which is a probability that communication between the specified start node and end node cannot be performed, based on a metric r of the alarm area M_i related to the event at a certain time t. Each hazard area is damaged in each warning area M_i using probability calculation means for calculating the probability of damage that is the probability that each hazard area in the warning area M_i will be damaged, and the damage probability calculated by the probability calculation means The link determined to be damaged is identified, and the damaged link is disconnected. Whether or not there is a path between the start point node and the end point node in the network excluding the link determined to be a broken link. There is provided a reliability evaluation apparatus comprising: a determination unit that determines the number of repetitions, and a process for calculating the disconnection probability based on the number of repetitions and the number of times that the path has not been performed. The

  The link failure probability calculation unit 403 according to the third embodiment is an example of probability calculation means. Further, the cutting probability calculation unit 404 of the third embodiment is an example of cutting probability calculation means.

(Section 1)
In a geographical network composed of a link set and a node set, when a part of the network is damaged within an alarm area specified based on information about an event that may cause a disaster, a designated start node A reliability evaluation apparatus that calculates a disconnection probability, which is a probability that communication with an end node cannot be performed,
Based on the metric r of the alarm area M_i related to the event at a certain time t, an intra-area link failure rate, which is a rate at which a link existing in the alarm area M_i fails, is calculated for the link j constituting the network. By acquiring link section information L_i, j that may be damaged in the alarm area M_i, and using the link section information L_i, j and the link failure rate in the area for each alarm area M_i, the link j has failed. A link failure probability calculating means for calculating a link failure probability which is a probability of
A disconnection that calculates a disconnection probability, which is a probability that communication between the start node and the end node cannot be performed, using the link failure probability of each link constituting the network calculated by the link failure probability calculation means. Probability calculation means
A reliability evaluation apparatus comprising:
(Section 2)
The link section information L_i, j is
The total of the link length of the portion where the hazard area predetermined in the alarm area M_i and the link j intersect, or
This is the number of hazard areas that intersect with the link j among the hazard areas that are predetermined in the warning area M_i.
The reliability evaluation apparatus according to item 1, characterized in that:
(Section 3)
A function for calculating the expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and the area used to determine g (r) Using the total area Sa of the hazard area in Sri, the area Sb1 damaged per disaster occurrence, and g (r), the damage occurrence rate in the warning area is calculated as g (r) x Sb1 / Sa Further comprising rate estimating means,
The link failure probability calculation means calculates the intra-area link failure rate using the damage occurrence rate and the probability that the damaged link is disconnected.
The reliability evaluation apparatus according to claim 1 or 2, characterized in that:
(Section 4)
A function for calculating the expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and the area used to determine g (r) Using the number SN of hazard areas and g (r), and further comprising a warning area damage occurrence rate estimating means for calculating the damage occurrence rate as g (r) / SN,
The link failure probability calculation means calculates the intra-area link failure rate using the damage occurrence rate and the probability that the damaged link is disconnected.
The reliability evaluation apparatus according to claim 1 or 2, characterized in that:
(Section 5)
In a geographical network composed of a link set and a node set, when a part of the network is damaged within an alarm area specified based on information about an event that may cause a disaster, a designated start node A reliability evaluation apparatus that calculates a disconnection probability, which is a probability that communication with an end node cannot be performed,
Based on the metric r of the alarm area M_i related to the event at a certain time t, probability calculation means for calculating a disaster probability that is a probability that each hazard area in the alarm area M_i is damaged;
Using the damage probability calculated by the probability calculation means, it is determined by simulation whether each hazard area is damaged in each alarm area M_i, the link determined to be damaged is identified, and the damaged link is disconnected. Whether or not there is a path between the start point node and the end point node in the network excluding the link determined to be a broken link. Determining means for determining
Cutting probability calculating means for repeatedly performing the processing by the determining means, and calculating the cutting probability based on the number of repetitions and the number of times the path has not been;
A reliability evaluation apparatus comprising:
(Section 6)
In a geographical network composed of a link set and a node set, when a part of the network is damaged within an alarm area specified based on information about an event that may cause a disaster, a designated start node A reliability evaluation method executed by a reliability evaluation apparatus that calculates a probability of disconnection, which is a probability that communication with an end node cannot be performed,
Based on the metric r of the alarm area M_i related to the event at a certain time t, an intra-area link failure rate, which is a rate at which a link existing in the alarm area M_i fails, is calculated for the link j constituting the network. By acquiring link section information L_i, j that may be damaged in the alarm area M_i, and using the link section information L_i, j and the link failure rate in the area for each alarm area M_i, the link j has failed. A link failure probability calculation step for calculating a link failure probability that is a probability of
Disconnection that calculates a disconnection probability that is a probability that communication cannot be performed between the start node and the end node using the link failure probability of each link that constitutes the network calculated by the link failure probability calculation step. Probability calculation step and
A reliability evaluation method comprising:
(Section 7)
In a geographical network composed of a link set and a node set, when a part of the network is damaged within an alarm area specified based on information about an event that may cause a disaster, a designated start node A reliability evaluation method executed by a reliability evaluation apparatus that calculates a probability of disconnection, which is a probability that communication with an end node cannot be performed,
Based on the metric r of the alarm area M_i related to the event at a certain time t, a probability calculation step for calculating a disaster probability that is a probability that each hazard area in the alarm area M_i is damaged;
Using the damage probability calculated in the probability calculation step, it is determined by simulation whether each hazard area is damaged in each warning area M_i, the link determined to be damaged is identified, and the damaged link is disconnected. Whether or not there is a path between the start point node and the end point node in the network excluding the link determined to be a broken link. A determination step for determining
A cutting probability calculating step of repeatedly performing the processing by the determination step, and calculating the cutting probability based on the number of repetitions and the number of times the path does not exist;
A reliability evaluation method comprising:
(Section 8)
The program for functioning a computer as each means in the reliability evaluation apparatus of any one of Claim 1 thru | or 5.
The present invention is not limited to the above-described embodiments, and various modifications and applications are possible within the scope of the claims.

100 Network 200 Network / Geographic Information Management Server 300 Weather Data Collection Server 400 Reliability Evaluation Server 401 Alarm Area Detection Unit 402 Alarm Area Damage Rate Estimation Unit 403 Link Failure Probability Calculation Unit 404 Disconnection Probability Calculation Unit

Claims (8)

In a geographical network composed of link sets and node sets, a reliability evaluation that evaluates the impact when a part of the network is damaged within an alarm area identified based on information about events that can cause disasters A device,
Based on the metric r of the alarm area M_i related to the event at a certain time t, an intra-area link failure rate, which is a rate at which a link existing in the alarm area M_i fails, is calculated for the link j constituting the network. By acquiring link section information L_i, j that may be damaged in the alarm area M_i, and using the link section information L_i, j and the link failure rate in the area for each alarm area M_i, the link j has failed. a reliability evaluation device comprising a link failure probability calculation means for calculating a link failure probability is the probability that,
In the reliability evaluation device, a function for calculating an expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and g (r) Using the total area Sa of the hazard area in the area used to determine the area, the area Sb1 damaged per disaster occurrence, and g (r), the damage occurrence rate is g (r) x Sb1 / Sa It further comprises means for calculating the damage occurrence rate in the warning area to be calculated,
The link failure probability calculation means calculates the intra-area link failure rate using the damage occurrence rate and the probability that the damaged link is disconnected.
A reliability evaluation apparatus characterized by that . In a geographical network composed of link sets and node sets, a reliability evaluation that evaluates the impact when a part of the network is damaged within an alarm area identified based on information about events that can cause disasters A device,
Based on the metric r of the alarm area M_i related to the event at a certain time t, an intra-area link failure rate, which is a rate at which a link existing in the alarm area M_i fails, is calculated for the link j constituting the network. By acquiring link section information L_i, j that may be damaged in the alarm area M_i, and using the link section information L_i, j and the link failure rate in the area for each alarm area M_i, the link j has failed. a reliability evaluation device comprising a link failure probability calculation means for calculating a link failure probability is the probability that,
In the reliability evaluation device, a function for calculating an expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and g (r) Using the number of hazard areas SN in the area used to determine the disaster area, and using g (r), further comprising an alarm area damage occurrence rate estimating means for calculating the damage occurrence rate as g (r) / SN,
The reliability evaluation apparatus according to claim 1, wherein the link failure probability calculation means calculates the in-area link failure rate using the damage occurrence rate and the probability that the damaged link is disconnected . A disconnection probability calculation that calculates a disconnection probability that is a probability that communication between the start node and the end node cannot be performed using the link failure probability of each link constituting the network calculated by the link failure probability calculation means. reliability evaluation device according to claim 1 or 2, further comprising a means. The link section information L_i, j is
The total of the link length of the portion where the hazard area predetermined in the alarm area M_i and the link j intersect, or
The reliability evaluation apparatus according to any one of claims 1 to 3 , wherein the number of hazard areas that intersect with the link j in a predetermined hazard area in the alarm area M_i. In a geographical network composed of link sets and node sets, a reliability evaluation that evaluates the impact when a part of the network is damaged within an alarm area identified based on information about events that can cause disasters A reliability evaluation method executed by an apparatus,
Based on the metric r of the alarm area M_i related to the event at a certain time t, an intra-area link failure rate, which is a rate at which a link existing in the alarm area M_i fails, is calculated for the link j constituting the network. By acquiring link section information L_i, j that may be damaged in the alarm area M_i, and using the link section information L_i, j and the link failure rate in the area for each alarm area M_i, the link j has failed. a reliability evaluation method comprising a link failure probability calculating step of calculating a link failure probability is the probability that,
In the reliability evaluation method, a function for calculating an expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and g (r) Using the total area Sa of the hazard area in the area used to determine the area, the area Sb1 damaged per disaster occurrence, and g (r), the damage occurrence rate is g (r) x Sb1 / Sa It further includes a step of estimating the damage occurrence rate in the alarm area to calculate,
In the link failure probability calculation step, the reliability evaluation apparatus calculates the in-area link failure rate using the damage occurrence rate and the probability that the damaged link is disconnected.
A reliability evaluation method characterized by that . In a geographical network composed of link sets and node sets, a reliability evaluation that evaluates the impact when a part of the network is damaged within an alarm area identified based on information about events that can cause disasters A reliability evaluation method executed by an apparatus,
Based on the metric r of the alarm area M_i related to the event at a certain time t, an intra-area link failure rate, which is a rate at which a link existing in the alarm area M_i fails, is calculated for the link j constituting the network. By acquiring link section information L_i, j that may be damaged in the alarm area M_i, and using the link section information L_i, j and the link failure rate in the area for each alarm area M_i, the link j has failed. a reliability evaluation method comprising a link failure probability calculating step of calculating a link failure probability is the probability that,
In the reliability evaluation method, a function for calculating an expected value Es of the number of disaster occurrences in the alarm area M_i when the metric r is given is predetermined as g (r), and g (r) Using the number of hazard areas SN in the area used to determine the disaster area, and using the g (r), further comprising a warning area damage occurrence rate estimation step for calculating the damage occurrence rate as g (r) / SN,
In the link failure probability calculation step, the reliability evaluation apparatus calculates the in-area link failure rate using the damage occurrence rate and the probability that the damaged link is disconnected.
A reliability evaluation method characterized by that . Using the link failure probability of each link constituting the network calculated by the link failure probability calculation step, a disconnection probability calculation for calculating a disconnection probability that is a probability that communication between the start node and the end node cannot be performed. The reliability evaluation method according to claim 6 , further comprising: a step. The program for functioning a computer as each means in the reliability evaluation apparatus of any one of Claims 1 thru | or 4 .