KR100923232B1 - An apparatus and method to analyze causality graphs - Google Patents

Abstract

The present invention relates to an apparatus for analyzing causality graphs and an analysis method. The apparatus for analyzing causality graphs according to the present invention relates to a node constituting a causality graph and the degree of influence on other nodes connected to each node. Input unit for receiving the dependence according to the error tree using the Boolean logic to convert the causal relationship graph using the data input to the input unit, the fault tree conversion unit, the connection relationship between each node included in the fault tree Path generation unit for generating a path by analyzing, path frequency-node resource matrix generator for generating a matrix including the relative frequency and relative number of resources for the path, and node importance calculation for calculating node importance using the matrix It can be configured to include critical parts in a timely manner, so that critical nodes in complex causal systems can be It can be systematically found, thereby improving effective system reliability and improving system reliability by providing effective countermeasures for protecting critical nodes.

Causality, Graphs, Trees, Node Importance

Description

Causality graph analysis device and analysis method {AN APPARATUS AND METHOD TO ANALYZE CAUSALITY GRAPHS}

The present invention relates to an apparatus and an analysis method for causality graph analysis. In particular, the causality graph is analyzed in consideration of the relative incidence of the connection paths, the commonality of the devices, and the required resources in the causality graphs of the devices constituting the system. It relates to an apparatus and an analysis method.

One effective way to ensure the reliability of nuclear power plants, telecommunications networks and other complex systems is to identify critical parts of the system and then provide a response to protect it.

At this time, one of the basic methods for identifying the important part of a complex system is to consider the devices included in the system as a node, and then to construct a causal graph between the devices to distinguish the most important nodes.

This most important node is the node that has the greatest impact in the event of a failure of the node in the system.

Conventional methods used to analyze a causality graph have a disadvantage in that a large amount of time is required to calculate an important node when the graph becomes large.

The present invention has been made to solve the above-described problems of the prior art, and while having a relatively low influence on the size of the causality graph, a causality graph analysis capable of systematic analysis capable of identifying important nodes in a short time is possible. Its purpose is to provide an apparatus and analytical method.

Causality graph analysis apparatus according to the present invention for solving the above problems is an input unit for receiving a dependency on the nodes constituting the causality graph, and the degree of influence on each node connected to each node, the Fault tree converting unit converting the causality graph into fault tree using Boolean logic using data inputted to the input unit, and generating a path by analyzing a connection relationship between each node included in the fault tree The apparatus may include a path frequency-node resource matrix generator for generating a matrix including a relative frequency of occurrence and relative number of resources of the path, and a node importance calculator for calculating node importance using the matrix.

In this case, the dependency is expressed as a number between 0 and 1 depending on the degree to which one of the two nodes connected affects the other node.

Here, the fault tree is rearranged in a tree form by applying a Boolean algebra according to the connection relationship and fault relationship step by step from the top node to the bottom node of the entire node constituting the system.

In addition, the path generation unit finds a path according to a connection relationship from the lowest node to the highest node of the fault tree, and identifies the nodes existing in each path.

The path frequency-node resource matrix generation unit sets a path as a row and a node as a column, and uses the dependence of each node as an element of a matrix, which affects the next node of the path. Relative frequency, the last row creates a matrix with the relative number of resources for each node.

In addition, the node importance calculator calculates the importance of each node by the following equation.

Where P is the sum of the relative incidence of all paths containing the node, P _y is the sum of the product of the relative incidence of each path and the dependence of the node for all the paths containing the node as P Divided, P _n is 1-P _y , C _j is the relative number of resources of the node.

In addition, a causality graph analysis method according to the present invention for solving the above problems is a first step of matching the device to the node in a system consisting of a plurality of devices, according to the degree of influence on the operation of each other between the nodes A second step of generating a causality graph showing dependence, a third step of converting the causality graph into a fault tree using Boolean logic, a fourth step of identifying a path between each node included in the fault tree, And a fifth step of generating a path frequency-node resource matrix including a relative frequency of occurrence and relative number of resources for the path, and a sixth step of calculating node importance using the path frequency-node resource matrix.

Here, in the second step, the dependency is expressed as a number between 0 and 1 depending on how much one of the two nodes connected affects the other node.

In the third step, all nodes constituting the causality graph are rearranged in tree form using Boolean algebra according to the connection relationship and the failure relationship from the highest node to the lowest node in the stepwise manner to convert the tree into failure trees.

The fourth step is to find a path according to a connection relationship from the lowest node to the highest node of the fault tree and to identify the nodes present in each path.

In the fifth step, a path is set as a row, a node is set as a column, the dependence is set as an element of a matrix, and the last column is a relative occurrence frequency of each path, and the last row is an element of relative resources of each node. Create a matrix.

In addition, the sixth step calculates the importance of each node by the following equation.

Where P is the sum of the relative incidence of all paths containing the node, P _y is the sum of the product of the relative incidence of each path and the dependence of the node for all the paths containing the node as P Divided, P _n is 1-P _y , C _j is the relative number of resources of the node.

The causality graph analysis method may further include a seventh step of outputting the calculation result of the sixth step on a screen.

The causality graph analysis apparatus and analysis method according to the present invention configured as described above can systematically find important nodes in a system having a complex causal relationship in a short time, thereby providing an effective countermeasure for protecting important nodes. It has the effect of improving the reliability of the system and improving the stability of the system.

Hereinafter, a configuration and an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram illustrating an example of a causality graph according to the present invention.

Referring to FIG. 1, a causality graph is a graph in which devices constituting a system are considered as nodes, and a connection relationship is shown according to dependencies and influence relationships between nodes. That is, the causality graph refers to a graph expressed as a number between 0 and 1 on the edges connected between nodes in consideration of the degree (effect) of one node affecting another node.

In the present invention, the degree of influence expressed as a number between 0 and 1 is expressed as a dependency. When the dependency is 0, it means that there is no influence relationship between two connected nodes. This means that any one of the nodes is completely dependent on the other node.

1 shows a causality graph composed of all 13 nodes as an example.

Here, nodes 1 to 3, node 8, and node 9 are nodes that are not affected by other nodes and only affect other nodes. For convenience, these nodes are called start nodes.

The direction in which each node is affected is indicated by an arrow, and the number on each edge indicates the dependency.

Node 13 is a node that is only affected by other nodes and does not affect other nodes, which is referred to as a final node for convenience.

Each node constituting the causality graph may be affected by, or may affect, one or more nodes.

A causality analysis device according to the present invention for analyzing a causality graph created according to causality and dependence of a device configuring a system as described above is as follows.

2 is a block diagram showing the configuration of a causality analysis apparatus according to the present invention.

The causality analysis apparatus according to the present invention first receives a component of a causality graph through the input unit 10. Mainly, dependencies are input according to the influence of the nodes constituting the causality graph and the other nodes connected to each node.

In this case, the dependency is expressed as a number between 0 and 1 according to the extent to which any one of the two nodes connected affects the other node as described above.

The greater the impact, the closer the dependence is to 1; the less the impact, the closer the dependence is to 0. If the dependency is 1, one of the two nodes is completely dependent on the other node. If the dependency is 0, the two nodes are independent of each other.

If the dependency is 0, the edge between two nodes is not displayed in the causal graph.

After inputting data such as the node and dependency on the causality graph from the input unit 10, the fault tree converting unit 20 converts the causality graph into the form of the fault tree.

In the present invention, a fault tree analysis technique is used to solve the problem of increasing computation time according to the graph size.

Fault trees are the most commonly used technique for evaluating the reliability of large and complex systems such as nuclear power plants, aviation, railways, chemical plants and tankers. In the present invention, Boolean logic (AND, OR, NOT, etc.) is used. Model the structure of the system. By applying Boolean Algebra to the fault tree, the path of occurrence of system failure can be found and finally the reliability of the system can be evaluated.

3 is a diagram illustrating an example of creating a fault tree for an arbitrary system.

Referring to FIG. 3, the system shown in FIG. 3 is mainly composed of two devices, A and B. Of these two devices, assume that the event that device A fails is called a, the event that device B fails is called b, and the event that the entire system fails is T.

In this case, if event a and b occur at the same time and event T occurs, then T is the intersection of a and b, and in the fault tree, the relationship is AND and expressed in the form of T = ab. That is, the path where T occurs is composed of a and b.

These fault trees can be solved mathematically using Boolean algebra, so even in very complex systems, all paths of failure can be calculated very quickly. For example, Fault Tree Reliability Evaluation eXpert (FTREX), a program developed by the Korea Atomic Energy Research Institute, provides the fastest fault tree analysis results in the world.

As described above, in the present invention, the fault tree rearranges all nodes constituting the system into a tree type using Boolean algebra according to the connection relationship and the fault relationship from the highest node to the lowest node step by step.

Here, the highest node refers to the last node described above, and is arranged in stages from the last node to the lower nodes in a tree form according to the connection relationship.

In addition, the failure relationship means an AND relationship or an OR relationship of Boolean numbers. The AND relationship means that the upper node fails only when all of the lower nodes of the fault tree have failed, and the OR relationship means that even if any one node among the lower nodes of the fault tree fails. It is a case of failure.

4 is a graph showing a part of the results of converting the causality graph of FIG. 1 into a fault tree.

Next, the path generation unit 30 generates a path by analyzing a connection relationship between each node included in the fault tree. That is, the path is found from the lowest node to the highest node of the fault tree according to the connection relationship, and the nodes existing in each path are identified.

FIG. 5 is a diagram illustrating all paths existing in the causality graph of FIG. 1.

The paths shown in FIG. 5 are more easily obtained by analyzing the fault tree of FIG. 4 than directly looking at the causality graph of FIG.

In other words, if the connected nodes are connected one by one up from the lowest node to the highest node of the fault tree, it becomes one path.

In Fig. 5, P ₁ to P ₇ are found paths, and all 7 paths were identified.

Next, in order to find the important node, the path frequency-node resource matrix generator 40 generates a matrix including the relative occurrence frequency and the relative number of resources for the path.

That is, the path frequency-node resource matrix generator 40 sets a path as a row and a node as a column, sets the dependency as an element of the matrix, the last column is a relative occurrence frequency of each path, and the last row is a node of each node. Create a matrix with relative numbers of resources.

Here, the relative frequency refers to the relative frequency at which a given path is actually generated in an actual system. In general, all paths have the same frequency, but if there is a special restriction, the user may randomly assign the frequency.

Therefore, if the relative incidence frequency is constant for all the paths, the same probability is set. Therefore, it is not necessary to input them separately, but may be set to the same probability in advance. It should be entered separately by 10).

In addition, the relative resource number (Relative Resource) represents the relative size of resources such as time, manpower, cost, etc. that must be put in order to maintain or manage each node included in a given path.

The relative resource size may be obtained through expert judgment, the sum of actual resources consumed, or the simulation result. Therefore, in order to analyze the system, the number of relative resources required for each node may be set in advance or may be input for each node by the input unit 10.

FIG. 6 is a diagram illustrating a matrix including relative frequency of occurrence and relative number of resources for the path of FIG. 4.

The matrix is also called a path frequency-node resource matrix, where the paths are arranged vertically so that each path takes care of the rows, and the nodes are placed horizontally so that each node takes care of the columns.

In the last column of the horizontal column, the relative frequency of occurrence is not assigned to the last node, but the relative number of resources is allocated to the column corresponding to each node in the last row.

Referring to FIG. 6, in the first row, the nodes included in the path 1 are assigned the dependencies of the nodes 1, 4, and 11 except for the last node 13 in the column corresponding to each node. Therefore, in FIG. 1, the dependency of Node 1 on Node 4 is 0.2, Node 4 of Node 4 is 0.4, and Node 11 of Node 13 is 0.2. Let it be an element of 1 row 4 columns and 1 row 11 columns.

However, since the last column assigns a relative frequency of occurrence, the elements in row 1 and column 13 are 1/7.

In FIG. 6, only the elements of each row and column are shown in Table 1 below.

0.2 0.2 1/7 0.3 0.2 1/7 0.1 0.2 1/7 0.7 0.1 0.2 1/7 0.2 0.7 0.5 0.3 1/7 0.4 0.3 1/7 0.4 0.3 1/7 One 2 3 3 4 One 6 2 4 3 One 2

The relative number of resources for each node in the last row is an arbitrary value.

Using the path frequency-node resource matrix generated as described above, the node importance calculator 50 calculates the importance of each node.

Node importance calculation section 50 of the present invention is the importance of the node according to the relative frequency of the path, the number of times each node is included in the path (shared) and the amount of resources (required resources) required for the maintenance and repair of the node Determine.

Specifically, this may be expressed by an equation. In other words, the node importance calculator 50 calculates the importance of each node using Equation 1 below.

Here, P is the sum of the relative frequency of occurrence of all paths including the node.

In addition, P _y is the sum of the product of the relative frequency of each path and the dependence of the node for all paths including the node divided by P.

And, P _n means 1-P _y , C _j is the relative number of resources of the node.

Where j refers to each node.

Referring to FIG. 6, when P value is calculated at node 1, node ₁ is included in P ₁ and P ₅ paths, so P at this time is

이 된다.

Becomes

P _y is at node 1 of FIG.

가 된다.

Becomes

Next, since P _n is 1-P _y , 1-0.2 = 0.8.

And C ₁ is 1 since C _j is a relative number of resources.

In this way, the node importance of node 1 becomes 0.143 by Equation 1.

Table 2 below is a table calculating P values for the matrix of Table 1.

The values in the block next to the P value at each node are the relative frequencies of the path each node belongs to.

Table 3 below is a table calculating P _y values for the matrix of Table 1.

The values in the block next to P _y at each node are the product of the relative frequency and dependence of the path to which each node belongs.

Table 4 below is a table that finally calculates the node importance of each node with respect to the matrix of Table 1.

Referring to Table 4, node 11 has the highest importance, and the result is intuitively causal graph because four of the paths to node 13, the final node, must pass through node 11 as shown in FIG. It can be seen that the results are consistent with the results judged by the report.

Accordingly, the node importance calculator 50 may calculate not only the importance of each node but also determine which node has the highest importance.

The causality graph analysis apparatus according to the present invention may further include a display unit 60.

The display unit 60 finally outputs the result value of the node importance calculated by the node importance calculator 50 in the form of a document or an image.

Hereinafter, a causality graph analysis method according to the present invention will be described with reference to FIG. 7.

7 is a flowchart illustrating a causality graph analysis method according to the present invention.

Referring to FIG. 7, first, a device is matched to a node in a system composed of a plurality of devices (S110).

Next, a causality graph in which dependencies are displayed according to the degree of influencing the operation between the nodes is generated (S120).

As shown in FIG. 1, it may be indicated by an arrow according to the direction of influence, and becoming a final node may refer to a device of a system that finally performs an operation, that is, a terminal component of a system or a final performance target. It may be configured.

The dependencies that each node affects other nodes are displayed on the edges connecting each node.

Considering the degree (effect) of one node to another node, it is expressed as a number between 0 and 1 on the edge connected between nodes.

The greater the impact, the closer the dependence is to 1; the less the impact, the closer the dependence is to 0. If the dependency is 1, one of the two nodes is completely dependent on the other node. If the dependency is 0, the two nodes are independent of each other.

If the dependency is 0, the edge between two nodes is not displayed in the causal graph.

Next, the causality graph is converted into a fault tree using Boolean logic (S130).

As explained earlier, these fault trees can be solved mathematically using Boolean algebra, which allows us to quickly calculate all possible paths of failure for very complex systems.

In the present invention, the fault tree rearranges all nodes constituting the system into a tree type using Boolean algebra according to the connection relationship and the fault relationship from the highest node to the lowest node step by step.

In the present invention, using the AND logic or OR logic to convert to the fault tree.

Next, determine the path between each node included in the fault tree (S140).

That is, the path is found from the lowest node to the highest node of the fault tree according to the connection relationship, and the nodes existing in each path are identified.

Stepping up from the lowest node of the fault tree to the highest node one by one leads to one path.

Next, a path frequency-node resource matrix including a relative frequency of occurrence and the number of resources for the path is generated (S150).

As described above, the paths are arranged vertically so that each path takes care of the rows, and the nodes are placed horizontally so that each node takes care of the columns, and the last column of the horizontal columns allocates the final node. Relative frequency is allocated, and the last row is assigned the relative number of resources to a column corresponding to each node, and then the members of each matrix are identified.

Since the final node is the target or end device of the causality graph, it is the main purpose of the present invention to determine the importance at other nodes except this, and thus the final node does not play a special role in this matrix.

After constructing the path-node resource matrix as described above, node importance is calculated using the path-node resource matrix (S160).

The importance of nodes is determined by the relative frequency of occurrence, the number of times each node is included in the path (shared), and the amount of resources (required resources) required for the maintenance and repair of the node.

That is, the node importance is calculated by Equation 1, and the description thereof is omitted since it is substantially the same as the causality graph analysis apparatus described above.

Finally, the calculation result is output to the screen or other output device (S170).

Although the causality graph analysis apparatus and the analysis method according to the present invention have been described with reference to the illustrated drawings, the present invention is not limited by the embodiments and drawings disclosed herein, but within the scope of the technical thoughts protected. Can be applied.

1 is a diagram showing an example of a causality graph according to the present invention;

2 is a block diagram showing the configuration of a causality analysis apparatus according to the present invention;

3 is a diagram showing an example of creating a fault tree for any system;

4 is a graph showing a part of the results of converting the causality graph of FIG.

5 is a diagram illustrating all paths present in the causality graph of FIG.

FIG. 6 illustrates a matrix including relative incidence and relative number of resources for the path of FIG.

7 is a flowchart illustrating a causality graph analysis method according to the present invention.

Claims (24)

An input unit configured to receive a node constituting a causality graph and dependencies according to the degree of influence on each node connected to each node; A fault tree converting unit converting the causality graph into a fault tree using Boolean logic using data inputted to the input unit; A path generation unit for generating a path by analyzing a connection relationship between each node included in the fault tree; A path frequency-node resource matrix generator for generating a matrix including a relative frequency of occurrence of the path and a relative number of resources; And And a node importance calculator for calculating node importance using the matrix. The method according to claim 1, The dependency is a causality graph analysis apparatus, characterized in that expressed as a number between 0 and 1 depending on the degree of influence of any one of the two nodes connected to the other node. The method according to claim 2, If the dependency is 0, it means that there is no influence relationship between the two connected nodes, and if the dependency is 1, the causal relationship means that one node of the two connected nodes is completely dependent on the other node. Graph analysis device. The method according to claim 1, The fault tree is a causality graph analysis apparatus, characterized in that the entire node constituting the system rearranged in a tree form using a Boolean algebra according to the connection relationship and failure relationship from the top node to the bottom node step by step. The method according to claim 4, The fault relationship is an OR relationship in which an upper node fails even when a failure occurs only in any one of the lower nodes of the fault tree, and in a case where the upper node fails only when a fault occurs in all the lower nodes of the fault tree. Causality graph analysis apparatus, characterized in that one of the AND relationship. The method according to claim 1, The path generation unit finds a path according to a connection relationship from the lowest node to the highest node of the fault tree, and identifies the nodes existing in each path. The method according to claim 1, The path frequency-node resource matrix generator sets a path as a row and a node as a column, the dependency as an element of a matrix, the last column as a relative frequency of each path, and the last row as a relative number of resources of each node. A causality graph analysis device, characterized in that to generate a matrix. The method according to claim 1, The relative occurrence frequency is a causality graph analysis apparatus, characterized in that the relative probability that each of the paths generated by the path generation unit is generated. The method according to claim 8, And the input unit further receives the relative occurrence frequency for each path. The method according to claim 1, The relative number of resources is a causality graph analysis apparatus, characterized in that the relative size of resources to be input for the maintenance / management of the nodes included in each path of the causality graph. The method according to claim 10, And the input unit further receives the relative number of resources for each path. The method according to claim 1, The node importance calculator calculates the importance of each node according to the following formula.

Where P is the sum of the relative incidences of all paths containing that node, and P _y is the sum of the relative incidence of each path multiplied by the dependence of that node for all paths containing that node. Divided by, P _n is 1-P _y , C _j is the relative number of resources of the node) The method according to claim 1, The causality graph analysis apparatus further includes a display unit configured to output a calculation result of the node importance calculator on a screen. Matching the devices to nodes in a system consisting of a plurality of devices; A second step of generating a causality graph in which the dependencies are displayed according to the degree of influence of the operation between the nodes; Converting the causality graph into a fault tree using Boolean logic; A fourth step of identifying a path between each node included in the fault tree; Generating a path frequency-node resource matrix including a relative frequency of occurrence and relative number of resources for the path; And And a sixth step of calculating node importance using the path-node resource matrix. The method according to claim 14, In the second step, the dependency is a causality graph analysis method, characterized in that expressed as a number between 0 and 1 depending on the degree of influence of any one of the two nodes connected to the other node. The method according to claim 15, If the dependency is 0, it means that there is no influence relationship between the two connected nodes, and if the dependency is 1, the causal relationship means that one node of the two connected nodes is completely dependent on the other node. Graph analysis method. The method according to claim 14, The third step is to rearrange the entire node constituting the causality graph from tree to tree by using Boolean algebra according to the connection relationship and failure relationship from the highest node to the lowest node. Causality graph analysis method. The method according to claim 17, The fault relationship is a case where an upper node fails only when an OR relationship and a lower node of the fault tree fail, even if any one node among the lower nodes of the fault tree fails. Causality graph analysis method, characterized in that one of the AND relationship. The method according to claim 14, The fourth step is to find the path according to the connection relationship from the lowest node to the highest node of the fault tree, and to identify the nodes present in each path causal relationship graph analysis method. The method according to claim 14, The fifth step sets a path as a row and a node as a column, the dependency as an element of a matrix, the last column as a relative frequency of each path, and the last row as a matrix as a relative resource number of each node. Causality graph analysis method, characterized in that generating. The method according to claim 14, The relative occurrence frequency is a causality graph analysis method, characterized in that the relative probability that each of the paths generated by the path generation unit is generated. The method according to claim 14, The relative number of resources is a causality graph analysis method, characterized in that the relative size of resources to be input for the maintenance / management of the nodes included in each path of the causality graph. The method according to claim 14, The sixth step is causality graph analysis method, characterized in that for calculating the importance of each node by the following formula.

Where P is the sum of the relative incidences of all paths containing that node, and P _y is the sum of the relative incidence of each path multiplied by the dependence of that node for all paths containing that node. Divided by, P _n is 1-P _y , C _j is the relative number of resources of the node) The method according to claim 14, The causality graph analysis method further comprises a seventh step of outputting the calculation result of the sixth step on the screen.

Images