JP2009211344A - Method for specifying supposed problem by hierarchical causality matrix - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve a problem by using causality matrices even in a complicated system configuration. <P>SOLUTION: Causality matrices are hierarchically arranged, a supposed problem is specified from a supposed phenomenon input to a lower causality matrix, and the specified supposed problem is allocated and input to a supposed phenomenon of an upper causality matrix. The upper causality matrix sets supposed problem candidate outputs from a plurality of lower causality matrices as supposed phenomena. By repeating the processing, a final problem candidate can be narrowed in the uppermost causality matrix. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an assumed problem identification method using a hierarchical causal relationship matrix that can easily identify problems such as failures in a wide variety of systems such as a ubiquitous network.

  Conventionally, a causal relationship matrix that can identify a single assumed problem based on a plurality of assumed events from a plurality of assumed problems in order to identify the cause of the failure from information on the failure that has occurred in an apparatus or network system. Is used.

  For example, according to Patent Document 1, in a plant abnormality diagnosis system, for process data input from a plant, the failure detection and the degree of the event are estimated by threshold determination of the input data, and the cause and data are weighted. The causal relationship performed is read and the cause is estimated, and the detected event and the estimated cause are displayed on the display device.

Japanese Patent Laid-Open No. 8-21932

  However, in a large-scale system such as the ubiquitous network that is proposed in recent years and is an information communication network that can be used everywhere in life and society, the number of devices at the end becomes enormous, and the installation locations can be widely dispersed. In many systems for detecting and isolating problems used to realize stable movement of such a system, the number of objects to be monitored, the number of events, and judgment items are extremely large. Since system problems are detected using a single causal relationship matrix, the number of events input to the causal relationship matrix is too large, making the scale huge and complex, and maintenance work such as maintenance is very important. It becomes difficult.

  Conventionally, a method for detecting and isolating a problem in a predetermined range as a system and connecting them by human work to detect the entire problem has been performed, but it belongs to a part that relies on human work. Humanity occurred, and there were difficulties in ensuring quality.

  The present invention has been made in view of the above problems, and it is possible to automatically detect problems in a wide variety of complicated events, and to identify an assumed problem using a hierarchical causal relationship matrix that can easily perform maintenance work. It aims to provide a method.

The assumed problem identification method by the hierarchical causal relationship matrix according to claim 1 includes a plurality of event monitoring means for monitoring a change in an event,
A plurality of problem candidates that are assumed in advance (hereinafter referred to as assumed problems) are stored, and a plurality of causal relationship matrices that specify the assumed problems from the input assumed events are included.
The causal relationship matrix is arranged in at least two layers, the event monitoring unit is connected to the lowest causal relationship matrix, and a change in the event detected by the event monitoring unit is input as an assumed event. The assumed problem is identified, and the assumed problem is output as an assumed event of a causal relation matrix that is higher than the lowest causal relation matrix. It is characterized in that it is repeatedly output as an assumed event of the matrix, and the assumed problem is specified by the highest causal relationship matrix.

  According to the method of claim 1, the change in the event detected by the event monitoring means is input as an assumed event to the lowest causal relationship matrix, the assumed problem is identified, and this is further assumed in the upper causal relationship matrix. By repeating the process of inputting as an event and identifying the expected problem, it is easy to filter unnecessary information and to identify the expected problem that matches based on the input event from a wide variety of complicated and large number of assumed problems It becomes.

  The assumed problem specifying method using the hierarchical causal relationship matrix according to claim 2 is the assumed problem specifying method using the hierarchical causal relationship matrix according to claim 1, wherein any of the causal relationship matrices is included in the assumed problem output from the causal relationship matrix. It is characterized in that a code indicating whether it is output from is attached.

  According to the method of claim 2, in each causal relationship matrix, a code for identifying itself is added to the assumed problem that has been output, so that the assumption problem identified as one by the highest causal relationship matrix is added. Based on this, it is possible to determine which causal relationship matrix has passed.

The assumed problem identification method using the hierarchical causal relationship matrix according to claim 1 is:
A plurality of event monitoring means for monitoring event changes;
A plurality of problem candidates that are assumed in advance (hereinafter referred to as assumed problems) are stored, and a plurality of causal relationship matrices that specify the assumed problems from the input assumed events are included.
The causal relationship matrix is arranged in at least two layers, the event monitoring unit is connected to the lowest causal relationship matrix, and a change in the event detected by the event monitoring unit is input as an assumed event. The assumed problem is identified, and the assumed problem is output as an assumed event of a causal relation matrix that is higher than the lowest causal relation matrix. Since it repeats outputting as an assumed event of the matrix and identifies the assumed problem by the highest causal relationship matrix,
In systems where there are a large number of assumed events and the assumption problems are complicated and diverse, information that can be noise can be filtered by the causal relationship matrix arranged in a hierarchical manner, and the assumption problems can be easily identified. It can be performed. In addition, by separating complex assumption problems and configuring them with multiple causal relationship matrices, maintainability is maintained higher than when multiple assumed events and assumption problems are configured with a single causal relationship matrix. Can do.

  The assumed problem specifying method using the hierarchical causal relationship matrix according to claim 2 is the assumed problem specifying method using the hierarchical causal relationship matrix according to claim 1, wherein any of the causal relationship matrices is included in the assumed problem output from the causal relationship matrix. Since the code indicating whether or not the data has been output from is assigned, the event that causes the problem can be easily identified by the assumed problem output by the highest causal relationship matrix. That is, the event monitoring means can be identified quickly.

  Hereinafter, an embodiment as the best mode for carrying out the present invention will be described with reference to FIGS. Of course, it goes without saying that the present invention can be easily applied to other than those described in the embodiments without departing from the spirit of the invention.

  The causal relationship matrix used in the assumed problem identification method using the hierarchical causal relationship matrix in this embodiment will be described with reference to FIGS. In the causal relationship matrix, for example, as shown in FIG. 1, an assumed event is arranged on the vertical axis and an assumed problem is arranged on the horizontal axis. An event detected by an after-mentioned sensor or an assumed problem output from a lower-level causal relationship matrix is input as an event to the monitoring device as an event monitoring means. For example, it is assumed that events A1 and A6 are simultaneously input to the monitoring device. P2 is the only assumption problem in which both event A1 and event A6 are established at the same time. This P2 is output as an assumed event of the higher-level causal relationship matrix. In the expected problem to be output, a code for specifying the causal relationship matrix and the input expected event are recorded. As a result, it is possible to specify which causal relationship matrix has passed through the assumption problem uniquely specified in the upper and highest causal relationship matrices.

  Such a causal relationship matrix is continuously constructed as a pyramidal hierarchical structure as shown in FIG. The assumed problem output from the lower causal relationship matrix is handled as an assumed event input to the upper causal relationship matrix. Changes in events detected from sensors (not shown) are input to the lowest causal relationship matrices 21a to 21c as assumed events. The input assumption event is processed as described above, and one assumption problem is selected. The selected assumption problem is input as an assumption event to the lower-level causal relationship matrix 22a-b. Similar processing is also performed in the higher-level causal relationship matrixes 23a-b and the highest-level causal relationship matrix 24a, and the assumed problem finally selected in the highest-level causal relationship matrix 24a is assumed as an assumed problem for the assumed event detected by the sensor. Is output. As a result, a plurality of events detected by a large number of monitoring devices can be output as unique assumption problems by the highest causal relationship matrix 24a.

  A configuration example of a system using the above problem detection method will be described with reference to FIG. 31A to N, 31a to n, 31i to iii are sensors serving as event detecting means for detecting different events. 32a to n are lower edge controllers, and 33 is an upper edge controller. Reference numeral 5 denotes a public network. Reference numerals 34a to 34n denote a district center that receives an input of an assumed event from a higher-level edge controller and identifies an assumed problem.

  Each of the edge controllers 42a-n, 43, the district centers 44a-n, and the supercenter 45 has a causal relationship matrix, and the assumed problem output in the causal relationship matrix of the lower-level device is an assumed event. Are input to the causal relationship matrix of the higher-level device.

  An application example of the above configuration as an IT system monitoring system will be described. The lower edge controller 42a is used for monitoring the IT system. The sensor 41A for determining whether the server and the storage device are alive, the sensor 41B for monitoring the utilization rate of the server and the storage device, and the room temperature of the server room in which the server and the storage device are installed. A sensor 41N to be monitored is connected. Similarly, the lower edge controller 42b is used for monitoring the local environment system, the sensor 41a monitors the temperature, the sensor 41b detects the acceleration, and the sensor 41n monitors the humidity. The lower edge controller 42c monitors an IT system different from that of the lower controller 42a, and detects a sensor 41i that determines whether the server and the storage device are alive, a sensor 41ii that monitors the usage rate of the server and the storage device, the server and the storage device A sensor 41iii for monitoring the room temperature of the server room in which is installed.

  In this way, a wide variety of systems can be monitored with a single monitoring system, and the scale of the causal relationship matrix of each edge controller, district center, and supercenter can be reduced. Hardware costs required for construction can be reduced, and maintainability can be kept very high as compared with a method of managing and monitoring the entire system using a single causal relationship matrix.

  For example, in the system configuration shown in FIG. 4, a case where the above configuration is applied to a convenience store will be described as a specific application example. The sensor 31 is connected to a lower edge controller 32 in which a refrigerator temperature condition and water leakage detection sensor 31p, a display product out-of-stock detection sensor 31q, and a security device operation detection sensor 31r are installed for each store. The lower edge controller 32 installed in each store is connected to the upper edge controller 33 installed in each city, and is connected to the district center 34 having jurisdiction over the prefecture through the network 5. Each district center is connected to a super center 35 as a headquarters via a network 5.

  According to the above configuration, it is possible to perform monitoring judgment for stable operation of convenience store units in various places throughout the country, and management can be performed for each type of information in more stores.

  Moreover, the case where the said system structure is applied to a landslide detection system is demonstrated based on FIG. When all the sensors 31 are the same type of acceleration sensor 31x and a change in the ground surface such as a landslide occurs, the change is detected in each sensor 31x, and the landslide area or its It becomes possible to specify the scale early.

  As described above, according to the assumed problem identification method based on the hierarchical causal relationship matrix, a certain number of monitoring targets can be monitored even for large-scale systems and ubiquitous networks that have many events to be monitored and need to process many assumed problem events. It is possible to easily identify an assumed problem by processing with the causal relationship matrix of, and hierarchizing it. In addition, since the causal relationship matrix can be distributed as a hierarchy, it is possible to improve the maintainability when implemented as a system.

  As mentioned above, although the Example of this invention was explained in full detail, this invention is not limited to the said Example, A various deformation | transformation implementation is possible within the range of the summary of this invention. For example, in the above embodiment, each device has one causal relationship matrix. However, the present invention is not limited to this. If the device has a plurality of causal relationship matrices and a hierarchy is configured as a system, There is no problem even if it does not have a hierarchical structure. Further, although the case-effect relationship matrix of each case has been described as having four layers, the present invention is not limited to this, and there is no problem even if there are two layers, three layers, or five layers or more. The configuration of the sensor may be appropriately selected and used in addition to the above three types of examples.

It is explanatory drawing which shows the processing concept of the causal relationship matrix in an Example. It is explanatory drawing which shows the concept of the hierarchical arrangement | positioning of a causal relationship matrix same as the above. It is explanatory drawing which shows the example which applied the hierarchy causal relationship matrix to the monitoring system of IT system same as the above. It is explanatory drawing which shows the example which applied the hierarchy causal relationship matrix to the apparatus stable movable system of a convenience store same as the above. It is explanatory drawing which shows the example which applied the hierarchy causal relationship matrix to the landslide detection system same as the above.

Explanation of symbols

1 hierarchy causal relationship matrix 21a-b, 22a, 22b, 23a, 23b, 24a causal relationship matrix 31A-31N, 31a-31n, 31i-31iii, 31p, 31q, 31r, 31x sensor 32a-32n lower edge controller 33 upper edge Controllers 34a to 34n District center 35 Super center 5 Network

Claims (2)

A plurality of event monitoring means for monitoring event changes;
A plurality of problem candidates that are assumed in advance (hereinafter referred to as assumed problems) are stored, and a plurality of causal relationship matrices that specify the assumed problems from the input assumed events are included.
The causal relationship matrix is arranged in at least two layers, the event monitoring unit is connected to the lowest causal relationship matrix, and a change in the event detected by the event monitoring unit is input as an assumed event. The assumed problem is identified, and the assumed problem is output as an assumed event of a causal relation matrix that is higher than the lowest causal relation matrix. A hypothetical problem identification method based on a hierarchical causal relationship matrix, characterized in that it is repeatedly output as an assumed event of a matrix, and an assumption problem is identified by the highest causal relationship matrix.   The method according to claim 1, wherein a code indicating which causal relation matrix is output is assigned to the assumed problem output from the causal relation matrix.

Images