WO2011138911A1 - 障害分析装置、障害分析方法および記録媒体 - Google Patents
障害分析装置、障害分析方法および記録媒体 Download PDFInfo
- Publication number
- WO2011138911A1 WO2011138911A1 PCT/JP2011/060108 JP2011060108W WO2011138911A1 WO 2011138911 A1 WO2011138911 A1 WO 2011138911A1 JP 2011060108 W JP2011060108 W JP 2011060108W WO 2011138911 A1 WO2011138911 A1 WO 2011138911A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- failure
- analysis target
- value
- target data
- values
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/0703—Error or fault processing not based on redundancy, i.e. by taking additional measures to deal with the error or fault not making use of redundancy in operation, in hardware, or in data representation
- G06F11/079—Root cause analysis, i.e. error or fault diagnosis
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
- G05B23/024—Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
- G05B23/0275—Fault isolation and identification, e.g. classify fault; estimate cause or root of failure
Definitions
- the present invention relates to a failure analysis apparatus, failure analysis method, and recording medium for a mechanical system.
- attention attributes As support technology for investigating the cause of this, various sensor information and log information related to the failure that has occurred, from sensor information and log information with different attributes acquired from the mechanical system (hereinafter referred to as attention attributes) A technology to narrow down and provide to users has been proposed.
- the present invention has been made in view of the above circumstances, and an object thereof is to provide a failure analysis device, a failure analysis method, and a recording medium that can extract a target attribute with high accuracy.
- a failure analysis apparatus includes: The analysis target based on a relative relationship between analysis target data whose elements are values generated based on values of a plurality of indices of the analysis target and representative values of the plurality of indices for each of a plurality of obstacles An obstacle contribution degree calculation means for obtaining an obstacle contribution degree indicating how much an individual obstacle (hereinafter referred to as an obstacle element) contributes to an obstacle occurring in an object; Based on the failure contribution degree obtained by the failure contribution degree calculation means, a failure element specifying means for specifying a failure element that has occurred, The failure element specified by the failure element specifying means based on the representative value taken by the plurality of indices and the value of each element of the analysis target data when a failure combining the failure elements occurs A factor indicator identifying means for identifying an indicator estimated to be a factor of Output means for outputting at least one of the failure element identified by the failure element identification means and the index identified by the factor index identification means; It is
- a failure analysis method includes: The analysis target based on a relative relationship between analysis target data whose elements are values generated based on values of a plurality of indices of the analysis target and representative values of the plurality of indices for each of a plurality of obstacles An obstacle contribution calculating step for obtaining an obstacle contribution indicating how much an individual obstacle (hereinafter referred to as an obstacle element) contributes to an obstacle occurring in the object; Based on the failure contribution determined in the failure contribution calculation step, a failure element identification step that identifies a failure element that has occurred, The failure element identified in the failure element identification step based on the representative value taken by the plurality of indices and the value of each element of the analysis target data when a failure in which the failure elements are combined has occurred A factor indicator identifying step for identifying an indicator estimated to be a factor of An output step of outputting at least one of the failure element identified in the failure element identification step and the index identified in the factor index identification step; It is characterized
- a computer-readable recording medium provides: Computer The analysis target based on a relative relationship between analysis target data whose elements are values generated based on values of a plurality of indices of the analysis target and representative values of the plurality of indices for each of a plurality of obstacles Obstacle contribution degree calculation means for obtaining a degree of obstacle contribution indicating how much an individual obstacle (hereinafter referred to as the obstacle element) contributes to the obstacle occurring in the object, A failure element identification unit that identifies a failure element that has occurred, based on the failure contribution degree obtained by the failure contribution degree calculation unit; The failure element specified by the failure element specifying means based on the representative value taken by the plurality of indices and the value of each element of the analysis target data when a failure combining the failure elements occurs Factor indicator identification means to identify the indicator estimated as the factor of Output means for outputting at least one of the failure element identified by the failure element identification means and the index identified by the factor index identification means; Record
- the present invention it is possible to provide a failure analysis device, a failure analysis method, and a recording medium that can extract a target attribute with high accuracy.
- the failure analysis apparatus 100 identifies a failure that has occurred from sensor information or log information with different attributes acquired from a machine system that is an analysis target, and extracts a target attribute in the failure It is a computer that operates under control.
- the attribute indicates the type of various sensors and the like.
- the failure analysis apparatus 100 includes an input unit 101, a communication unit 102, a control unit 103, a storage unit 104, and an output unit 105.
- the input unit 101 is an input device represented by a keyboard and a mouse.
- the input unit 101 includes a control unit 103 that includes an element for identifying a fault that has occurred, an element for extracting a target attribute, and a start signal for operating the fault analysis apparatus 100 set by the user.
- a control unit 103 that includes an element for identifying a fault that has occurred, an element for extracting a target attribute, and a start signal for operating the fault analysis apparatus 100 set by the user.
- the communication unit 102 includes an interface for connecting to an external database system.
- data regarding each attribute to be analyzed is stored as log information or sensor information.
- Log information and sensor information are supplied from the external database system to the control unit 103 through the communication unit 102.
- the control unit 103 operates according to the program 106 stored in the storage unit 104, receives an input from the input unit 101, and provides a function necessary for analysis.
- the control unit 103 functions as a failure analysis processing unit 107 and an attribute extraction processing unit 108 by operating according to the program 106.
- the failure analysis processing unit 107 includes an internal memory that temporarily stores an element received from the input unit 101 for identifying a failure.
- the internal memory stores in advance a correspondence table for identifying a failure that has occurred by associating a failure contribution type described later with failure identification information.
- the attribute extraction processing unit 108 includes an internal memory that temporarily stores an element for extracting an attribute received from the input unit 101.
- the storage unit 104 includes a storage device such as a hard disk or a semiconductor memory.
- the storage unit 104 stores a program 106 for operating the failure analysis apparatus 100, a failure pattern matrix 110, and an attribute pattern database 111.
- the failure pattern matrix 110 is a matrix of total number of failures ⁇ total number of attributes (M rows and D columns), as shown in FIG.
- the element Q md of the failure pattern matrix 110 indicates a representative value of the attribute d for the failure m.
- the representative value is an average value of attribute values corresponding to failures that have occurred in the past.
- the attribute pattern database 111 includes a plurality of failure pattern tables 112.
- the failure pattern table 112 is a table showing a combination of failures and representative values of each attribute when a failure of the combination occurs.
- the representative value is an average value of attribute values corresponding to combinations of failures that have occurred in the past.
- a number of failure pattern tables 112 obtained by adding the total number of failures and the total number of combinations of failures is stored in the attribute pattern database 111 in advance.
- the failure pattern table 112 corresponding to all failure combinations may be stored, or the failure pattern table 112 corresponding to a specific combination of failures narrowed down in advance may be stored. Good.
- the output unit 105 is an output device represented by a display device or a printer.
- the failure analysis processing according to the present embodiment is started upon reception of a start signal from the input unit 101.
- the communication unit 102 as data input means reads out sensor information and log information (hereinafter referred to as input data) as analysis target data from an external database system (step S100).
- the failure analysis processing unit 107 receives an element for specifying a failure from the input unit 101 and input data from the communication unit 102, and stores an element for specifying the failure in the internal memory.
- the attribute extraction processing unit 108 receives an element for extracting an attribute from the input unit 101 and stores it in the internal memory.
- the failure analysis processing unit 107 reads the failure pattern matrix 110 from the storage unit 104 (step S101).
- the failure analysis processing unit 107 uses the input data received from the communication unit 102 and the failure pattern matrix 110 read from the storage unit 104, as shown below, for each failure element for the failure that has occurred. Is calculated (hereinafter referred to as a failure contribution degree) (step S102).
- the failure analysis processing unit 107 represents the predicted value of the input data as a function of a row vector in the failure pattern matrix 110 read from the storage unit 104 using the parameter vector A.
- the parameter vector A is a failure contribution degree.
- the value of each element of A is obtained by solving the following equation (2).
- A argmin_A (XX ′) 2 (2)
- Equation (2) represents an operation for obtaining A that minimizes (XX ′) 2 .
- the failure analysis processing unit 107 identifies the failure that has occurred, based on the elements stored in the internal memory, from each element A obtained above (step S103).
- a threshold value is stored in the internal memory as an element for identifying the fault that has occurred, received from the input unit 101.
- the failure analysis processing unit 107 compares the value of the element A with the threshold value stored in the internal memory for each element A obtained above, and the value of the element A is equal to or greater than the threshold value. , The fault corresponding to the element is specified as the fault that has occurred.
- the failure analysis processing unit 107 compares the value of each element of A with a threshold value, and extracts A 1 and A 2 whose A value is 0.4 or more.
- the failure analysis processing unit 107 compares the extracted A 1 and A 2 with the correspondence table stored in advance in the internal memory, and generates the failure 1 and the failure 2 corresponding to A 1 and A 2. Identify as a failure.
- the attribute extraction processing unit 108 receives, from the failure analysis processing unit 107, the input data and the failure and the failure contribution degree specified and calculated by the failure analysis processing unit 107.
- the attribute extraction processing unit 108 extracts notable attributes from each attribute of the input data based on the failure specified by the failure analysis processing unit 107 as follows.
- the attribute extraction processing unit 108 reads the attribute pattern table 112 corresponding to the specified failure from the attribute pattern database 111 of the storage unit 104 using the specified failure as a key (step S104).
- the attribute extraction processing unit 108 reads the attribute pattern table 112 as shown in FIG.
- the attribute extraction processing unit 108 compares the values of the elements of the received input data with the corresponding values of the read attribute pattern table 112, and extracts the attribute of interest based on the elements stored in the internal memory. (Step S105).
- the attribute extraction processing section 108 compares the value of the input data elements, the absolute value of the difference between the values of the elements of the attribute pattern table 112 shown in FIG. 4 (the value of the element Z 1) Ask for. Then, the attribute extraction processing unit 108 compares the calculated absolute value of the difference with the threshold value stored in the internal memory. If the calculated absolute value of the difference is equal to or less than the threshold value, the attribute extraction processing unit 108 corresponds to each element. Attributes to be extracted as attention attributes. The attribute extraction processing unit 108 performs this operation for each element of the input data, thereby extracting a notable attribute from the input data.
- the attribute extraction processing unit 108 passes the fault identification information, the fault contribution degree, and the extracted attention attribute to the output unit 105 in association with the input data.
- the output unit 105 outputs the received information (step S106).
- the user investigates the cause of the failure that occurred in the analysis target based on the information output by the output unit 105.
- the failure analysis apparatus 100 pays attention to high accuracy by simultaneously providing the user with information regarding the failure related to the analysis target data, the magnitude of the contribution, and which attribute the failure affects.
- the attribute can be extracted, which leads to shortening of the time for investigating the cause of the failure.
- the present invention is not limited to the above embodiment, and various modifications and applications are possible.
- the failure analysis processing unit 107 may receive the analysis target data input from the input unit 101 by the user as input data and perform the analysis process, and the method of acquiring the input data is arbitrary.
- the representative value of the failure pattern matrix 110 is an average value of attribute values corresponding to failures that have occurred in the past
- the representative value of the attribute pattern table 112 is an attribute that corresponds to combinations of failures that have occurred in the past.
- an average value is shown, the present invention is not necessarily limited to this.
- any value such as a value defined by an expert in advance or a value estimated by a predetermined method from past failure data can be used.
- the attribute pattern table 112 corresponding to the failure identified by the failure analysis processing unit 107 may be generated from the representative value of each element of the failure pattern matrix 110 based on the failure identified by the failure analysis processing unit 107.
- the representative value for fault 1 in the fault pattern matrix 111 and the representative value for the corresponding fault 2 may be used.
- the attribute pattern table 112 may be generated by an arbitrary method.
- failure analysis processing unit 107 uses a linear function as f (Q, A)
- the present invention is not necessarily limited thereto.
- the failure analysis processing unit 107 can use an arbitrary function such as a polynomial function as f (Q, A).
- the failure analysis processing unit 107 uses a function representing the distance between X and X ′ as the loss function L, but this is an example, and the present invention is not necessarily limited thereto.
- the failure analysis processing unit 107 can use a logarithmic loss function as the loss function L when the predicted value of the input data is regarded as a probability distribution.
- the failure analysis processing unit 107 can use an arbitrary function as the loss function L, such as structured risk minimization, maximum likelihood estimation, Bayesian estimation, or the like.
- the elements set by the input unit 101 for the failure analysis processing unit 107 to identify a failure and the elements for the attribute extraction processing unit 108 to extract the attention attribute are performed using threshold values.
- An example of setting was shown. However, this is an example, and the elements set in the input unit 101 are arbitrary. In addition, an element may not be set. In this case, for each value whose failure contribution is non-negative, a corresponding failure is specified, and an attribute corresponding to this failure is extracted as an attention attribute.
- the failure analysis apparatus 200 extracts a failure that has occurred and an attribute related to the failure as a target attribute without storing the failure pattern matrix 110 in the first embodiment in the storage unit 104 in advance. It is a failure analysis device.
- the failure analysis apparatus 200 further includes a data matrix generation processing unit 205 and a failure occurrence matrix generation processing unit 206 in the control unit 103 as shown in FIG. Yes.
- the storage unit 104 is different from the failure analysis device 100 in that the failure pattern matrix 110 and the attribute pattern database 111 are not provided. Note that the same members and configurations as those in the first embodiment are omitted.
- the communication unit 102 reads a plurality of input data (step S200).
- the input data in the present embodiment simultaneously includes the types of related failures in addition to the sensor information and log information.
- the data matrix generation processing unit 205 and the failure occurrence matrix generation processing unit 206 read a plurality of input data from the communication unit 102 which is a data input unit.
- the data matrix generation processing unit 205 obtains attribute values from a plurality of read input data, as shown in FIG. 6, the number of data times the number of attributes (N rows and D columns) (N is the total number of data; the same applies hereinafter).
- the data matrix 208 is arranged and stored in the storage unit 104 (step S201).
- the data matrix 208 is G.
- Each row vector Gn is a row vector indicating one input data.
- the failure occurrence matrix generation processing unit 206 arranges the occurrence of failure from the read input data as a failure occurrence matrix 209 of the number of data ⁇ total number of failures (N rows and M columns) as shown in FIG.
- the information is stored in the unit 104 (step S202).
- the failure matrix 209 and F each row vector of F and F n.
- the element F ij indicates whether or not a failure j has occurred for the data i, and indicates 0 when the failure j does not occur and 1 when it occurs.
- the failure analysis processing unit 107 reads the data matrix 208 and the failure occurrence matrix 209 from the storage unit 104.
- the failure analysis processing unit 107 optimizes the loss function as in the first embodiment. In S102 in the first embodiment, the failure analysis processing unit 107 optimizes the loss function for A. However, in the present embodiment, the failure analysis processing unit 107 optimizes A and Q by mathematical programming to obtain the failure contribution level, and at the same time, obtains each element of the corresponding failure pattern matrix 110 (step S203).
- the failure analysis processing unit 107 generates a data matrix G of the following equation (3) and a failure occurrence matrix F of the following equation (4).
- the distance between the data matrix G and the predicted value g is expressed by the following formula (5) and the following formula (6).
- the failure analysis processing unit 107 calculates the sum of Expression (5) and Expression (6).
- the sum of equations (5) and (6) is a function of A and Q.
- the failure analysis processing unit 107 obtains the values of the elements A and Q that minimize the sum of the expressions (5) and (6) using mathematical programming.
- the failure analysis processing unit 107 identifies a failure from each element of the failure contribution degree A obtained above as in the first embodiment (step S204).
- the failure analysis processing unit 107 uses the respective elements of Q obtained above to display the attribute pattern table 112 indicating the failure identified above and the representative value of each attribute corresponding to the number of attributes of the input data. Obtained (step S205).
- Failure analysis unit 107 for example, as shown in FIG. 4, a representative value Z 1 of the attribute 1 when a failure 1 and 2 has occurred, a representative value Q 11 of each attribute when only fault 1 has occurred , determined by the sum of the representative value Q 21 of each attribute when only failure 2 has occurred.
- the failure analysis processing unit 107 obtains the attribute pattern table 112 for the identified failure by repeating this process for each attribute of Q obtained above.
- the attribute extraction processing unit 108 reads out from the failure analysis processing unit 107 the data matrix 208 and the failure, failure contribution level, and attribute pattern table 112 specified by the failure analysis processing unit 107.
- the attribute extraction processing unit 108 compares each element of the data matrix 208 with each element of the attribute pattern table 112, and extracts the attribute of interest as in the first embodiment (step S206).
- the attribute extraction processing unit 108 passes the failure, the degree of failure contribution, and the extracted attention attribute to the output unit 105 in association with the data matrix 208. Then, the output unit 105 outputs the received information (step S207).
- the failure analysis apparatus 200 gives the user the failure related to the analysis target data, the magnitude of the contribution, and which attribute the failure affects. It is possible to provide information on whether or not it is given at the same time. As a result, the attribute of interest can be extracted with high accuracy without requiring prior knowledge of the failure to be analyzed.
- the said embodiment is an example and various deformation
- the attribute pattern table 112 is obtained by using the sum of the representative values of each attribute for each failure. However, this is an example, and an arbitrary method can be used.
- the failure analysis processing unit 107 generates the failure pattern matrix 110, and calculates each element of the attribute pattern table 112 corresponding to the specified failure based on each element of the generated failure pattern matrix 110.
- the attribute pattern table 112 may be stored in advance.
- the value of each element can be an arbitrary value such as a value defined by an expert in advance or estimated from past data.
- the failure analysis apparatus 300 uses time-series data as input data to calculate the degree of change of each attribute value when a failure occurs, thereby causing the failure analysis apparatus 100 according to the first embodiment.
- the failure analysis device outputs change information generated in the attribute of interest.
- the failure analysis apparatus 300 further includes a data change degree calculation processing unit 301 in the control unit 103 as shown in FIG. Further, the storage unit 104 is different from the failure analysis apparatus 100 in that the failure change pattern matrix 302 is provided instead of the failure pattern matrix 110 and the attribute change pattern database 303 is provided instead of the attribute pattern database 111. Note that the same members and configurations as those in the first embodiment are omitted.
- the failure change pattern matrix 302 is a matrix of the total number of failures ⁇ the total number of attributes (M rows and D columns), similarly to the failure pattern matrix according to the first embodiment.
- the element C md of the failure change pattern matrix 302 indicates a representative value of the degree of change of the attribute d with respect to the failure m.
- the representative value is an average value of the degree of change of the attribute corresponding to the failure that has occurred in the past.
- the attribute change pattern database 303 includes an attribute change pattern table 304.
- the attribute change pattern table 304 is a table similar to the attribute pattern table according to the first embodiment as shown in FIG.
- the value of each element of the attribute change pattern table 304 indicates a combination of each failure and a representative value of the degree of change of each attribute corresponding to the combination.
- the representative value is an average value of the degree of change of the attribute corresponding to the combination of failures that occurred in the past.
- the communication unit 102 which is a data input unit, reads input data that is analysis target data from an external database system (step S300).
- the input data in this embodiment is time-series data of sensor information and log information with different attributes acquired from the machine system that is the analysis target.
- the time-series data in the present embodiment is a continuous value such as the rotation speed of an automobile engine or the value of a temperature sensor.
- the data change degree calculation processing unit 301 receives the input data from the communication unit 102, and calculates a change degree corresponding to each attribute of the input data using an arbitrary change degree calculation technique (step S301).
- a technique such as ChangeFinder (Non-patent Document 1) or IBM method (Patent Document 1) is used.
- the failure analysis processing unit 107 receives the change degree calculated by the data change degree calculation processing unit 301 from the data change degree calculation processing unit 301, and reads the failure change pattern matrix 302 from the storage unit 104 (step S302). .
- the failure analysis processing unit 107 uses the received change degree and the failure change pattern matrix 302 to obtain a failure contribution degree by the same processing as in the first embodiment (step S303), and specifies a failure (step S304).
- the degree of change in this embodiment corresponds to the input data in the first embodiment
- the failure change pattern matrix 302 in this embodiment corresponds to the failure pattern matrix 110 in the first embodiment.
- the attribute extraction processing unit 108 receives the degree of change, the failure, and the degree of failure contribution from the failure analysis processing unit 107, and from the attribute change pattern database 303 of the storage unit 104 in the same manner as in the first embodiment, the attribute change pattern table 304. Is read (step S305).
- the attribute extraction processing unit 108 extracts the attention attribute by the same processing as in the first embodiment (step S306), and outputs the failure, the failure contribution degree, and the attention attribute corresponding to the failure in association with the degree of change. It passes to the part 105.
- the attribute change pattern table 304 in the present embodiment corresponds to the attribute pattern table 112 in the first embodiment.
- the output unit 105 outputs the received information (step S307).
- the failure analysis apparatus 300 can provide the user with detailed information such as what change caused which failure caused by which failure by using time-series data as analysis target data.
- the attribute of interest can be extracted with high accuracy, detailed information when the failure occurs can be provided to the user, and the cause of the failure can be identified with higher accuracy.
- time-series data may be discrete values such as ON / OFF of various functions.
- the attribute change pattern table 304 is stored in the attribute change pattern database 303 in the storage unit 104 in advance, but the present invention is not necessarily limited thereto. Similar to the first embodiment, the attribute change pattern table 304 may be generated from a failure change pattern matrix or the like.
- the failure analysis apparatus 400 uses the time series data to calculate the degree of change in each attribute value when a failure occurs, similarly to the analysis apparatus 300 according to the third embodiment. Furthermore, the failure analysis device 400 specifies necessary information without requiring prior knowledge about the failure, as with the analysis device 200 according to the second embodiment.
- the failure analysis apparatus 400 further includes a data change degree calculation processing unit 301 according to the third embodiment in the control unit 103 as shown in FIG. Yes. Note that the same members and configurations as those of the second embodiment are omitted.
- the communication unit 102 reads input data that is time-series data from the external database system (step S400), and uses the read input data as a data change rate calculation processing unit 301 and a failure occurrence matrix generation.
- the data is passed to the processing unit 206.
- the data change degree calculation processing unit 301 calculates the degree of change corresponding to each attribute of the input data by the same process as in the third embodiment (step S401).
- the data matrix generation processing unit 205 receives the degree of change calculated by the data change degree calculation processing unit 301, arranges the degree of change as a data matrix 208 of the number of data ⁇ the number of attributes (N rows and D columns), and stores it in the storage unit 104. Store (step S402).
- the failure occurrence matrix generation processing unit 206 generates a failure occurrence matrix 209 by the same processing as in the second embodiment, and stores it in the storage unit 104 (step S403).
- the failure analysis processing unit 107 receives the data matrix 208 and the failure occurrence matrix 209 from the storage unit 104, and calculates the failure contribution degree and the failure change pattern by the same processing as in the second embodiment (step S404).
- the fault is specified (step S405).
- the failure analysis processing unit 107 obtains the attribute change pattern table 304 similar to that of the third embodiment by the same processing as that of the second embodiment (Step S406) and extracts the attention attribute (Step S407). .
- the attribute extraction processing unit 108 passes the failure, the degree of failure contribution, and the attribute of interest corresponding to the failure to the output unit 105 in association with the degree of change.
- the output unit 105 outputs the received information (step S408).
- the failure analysis apparatus 400 is more detailed such as what change caused which failure caused which failure even if the failure change pattern indicating the representative value of the change degree of the attribute in each failure is unknown. Information can be provided to the user. As a result, the attribute of interest can be extracted with high accuracy without requiring prior knowledge of the failure to be analyzed, and detailed information at the time of failure can be provided to the user, and the cause of the failure can be identified with higher accuracy. be able to.
- the said embodiment is an example and various deformation
- the attribute change pattern table 304 may be stored in the attribute change pattern database 303 of the storage unit 104 in advance.
- the functions of the failure analysis apparatuses 100 to 400 according to the above embodiment can be realized by dedicated hardware or by a normal computer system.
- the program 106 stored in the storage unit 104 of the failure analysis apparatus 100 to 400 is stored in a flexible disk, a CD-ROM (Compact Disk Read-Only Memory), a DVD (Digital Versatile Disk), MO ( By storing and distributing in a computer-readable recording medium such as Magneto-Optical disk, and installing the program in a computer, an apparatus that executes the above-described processing can be configured.
- a computer-readable recording medium such as Magneto-Optical disk
- the program may be stored in a disk device or the like of a predetermined server device on a communication network such as the Internet, and may be downloaded onto a computer by being superimposed on a carrier wave, for example.
- the above-described processing can also be achieved by starting and executing a program while transferring it via a communication network. Furthermore, the above-described processing can also be achieved by executing all or part of the program on the server device and executing the program while the computer transmits and receives information regarding the processing via the communication network.
- An obstacle contribution degree calculation means for obtaining an obstacle contribution degree indicating how much an individual obstacle (hereinafter referred to as an obstacle element) contributes to an obstacle occurring in an object; Based on the failure contribution degree obtained by the failure contribution degree calculation means, a failure element specifying means for specifying a failure element that has occurred, The failure element specified by the failure element specifying means based on the representative value taken by the plurality of indices and the value of each element of the analysis target data when a failure combining the failure elements occurs
- a factor indicator identifying means for identifying an indicator estimated to be a factor of Output means for outputting at least one of the failure element identified by the failure element identification means and the index identified by the factor index identification means;
- a failure analysis apparatus comprising:
- the obstacle contribution calculation means Analysis target data storage means for storing analysis target data whose elements are values generated based on the values of a plurality of indicators of the analysis target; Correspondence between a representative value of an element of a failure pattern indicating a representative value of the plurality of indices for each of a plurality of failures stored in advance, and a value of an element of analysis target data stored in the analysis target data storage unit Based on the relative relationship between what to do, a means for determining the degree of contribution to the obstacle that indicates how much each obstacle (hereinafter referred to as the failure element) contributes to the failure that has occurred,
- the factor indicator specifying means includes a value of an element of an attribute pattern stored in advance indicating a set of representative values taken by the plurality of indicators when a failure combining the failure elements occurs, and the analysis target Based on the value of each element of the analysis target data stored in the data storage means, the index that is estimated as the factor of the failure element specified by the failure element specification means is specified.
- the obstacle contribution calculation means Analysis target data storage means for storing analysis target data whose elements are values generated based on the values of a plurality of indicators of the analysis target; Correspondence between a representative value of an element of a failure pattern indicating a representative value of the plurality of indices for each of a plurality of failures stored in advance, and a value of an element of analysis target data stored in the analysis target data storage unit Based on the relative relationship between what to do, a means for determining the degree of contribution to the obstacle that indicates how much each obstacle (hereinafter referred to as the failure element) contributes to the failure that has occurred, With
- the factor indicator specifying means indicates a value of an element of an attribute pattern indicating a set of representative values taken by the plurality of indicators when a failure in which the failure elements are combined occurs, and a representative value of the element of the failure pattern
- the failure element identified by the failure element identification unit based on the calculated element value of the attribute pattern and the value of each element of the analysis target data stored in the analysis target data storage
- (Appendix 4) Data matrix generating means for generating a first matrix by arranging the index values of the analysis target data from the analysis target data; Data matrix storage means for storing the value of each element of each row of the first matrix generated by the data matrix generation means as the analysis target data; A failure occurrence matrix generating means for generating, from the analysis target data, a second matrix by arranging whether or not a failure has occurred in the analysis target data; Further comprising The obstacle contribution calculation means A fault that stores a value of each element of the second matrix generated by the fault occurrence matrix generation unit and an unknown fault pattern indicating a representative value of the plurality of indices for each of the plurality of faults Pattern storage means; The failure based on the relative relationship between corresponding values of the values of the elements of the plurality of analysis target data stored in the data matrix storage means and the values of the elements stored in the failure pattern storage means Means for calculating a contribution and a value of the element of the obstacle pattern; With The factor indicator specifying means includes a value of an element of an attribute pattern stored
- (Appendix 5) Data matrix generating means for generating a first matrix by arranging the index values of the analysis target data from the analysis target data; Data matrix storage means for storing the value of each element of each row of the first matrix generated by the data matrix generation means as the analysis target data; A failure occurrence matrix generating means for generating, from the analysis target data, a second matrix by arranging whether or not a failure has occurred in the analysis target data; Further comprising The obstacle contribution calculation means A fault that stores a value of each element of the second matrix generated by the fault occurrence matrix generation unit and an unknown fault pattern indicating a representative value of the plurality of indices for each of the plurality of faults Pattern storage means; The failure based on the relative relationship between corresponding values of the values of the elements of the plurality of analysis target data stored in the data matrix storage means and the values of the elements stored in the failure pattern storage means Means for calculating a contribution and a value of the element of the obstacle pattern; With The factor indicator specifying means converts an element value of the attribute pattern indicating
- the analysis object data storage means stores the values of a plurality of indicators of the analysis object as analysis object data as they are.
- the failure analyzer according to any one of appendices 2 to 5, characterized in that:
- a degree-of-change calculating means for obtaining a degree of change in the value of each index of the analysis target data; Means for storing a set of values indicating the degree of change obtained by the degree-of-change calculating means as the analysis target data; Further comprising
- the failure pattern is a value indicating a representative value of the degree of change in the value of each index corresponding to each failure,
- the attribute pattern is a value indicating a representative value of the degree of change in the value of each index corresponding to each combination of obstacles.
- the analysis target based on a relative relationship between analysis target data whose elements are values generated based on values of a plurality of indices of the analysis target and representative values of the plurality of indices for each of a plurality of obstacles
- An obstacle contribution calculating step for obtaining an obstacle contribution indicating how much an individual obstacle (hereinafter referred to as an obstacle element) contributes to an obstacle occurring in the object;
- a failure element identification step Based on the failure contribution determined in the failure contribution calculation step, a failure element identification step that identifies a failure element that has occurred, The failure element identified in the failure element identification step based on the representative value taken by the plurality of indices and the value of each element of the analysis target data when a failure in which the failure elements are combined has occurred
- a factor indicator identifying step for identifying an indicator estimated to be a factor of
- a failure analysis method comprising:
- (Appendix 9) Computer The analysis target based on a relative relationship between analysis target data whose elements are values generated based on values of a plurality of indices of the analysis target and representative values of the plurality of indices for each of a plurality of obstacles Obstacle contribution degree calculation means for obtaining a degree of obstacle contribution indicating how much an individual obstacle (hereinafter referred to as the obstacle element) contributes to the obstacle occurring in the object,
- a failure element identification unit that identifies a failure element that has occurred based on the failure contribution degree obtained by the failure contribution degree calculation unit;
- the failure element specified by the failure element specifying means based on the representative value taken by the plurality of indices and the value of each element of the analysis target data when a failure combining the failure elements occurs
- Factor indicator identification means to identify the indicator estimated as the factor of Output means for outputting at least one of the failure element identified by the failure element identification means and the index identified by the factor index identification means;
- the present invention is preferably applied to an application for analyzing a failure of a mechanical system.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Automation & Control Theory (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Quality & Reliability (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
- Debugging And Monitoring (AREA)
Abstract
Description
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出手段と、
前記障害寄与度算出手段により求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定手段と、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する要因指標特定手段と、
前記障害要素特定手段により特定された障害要素と前記要因指標特定手段により特定された指標との少なくとも一方を出力する出力手段と、
を備えることを特徴とする。
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出ステップと、
前記障害寄与度算出ステップで求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定ステップと、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定ステップで特定された障害要素の要因と推定される指標を特定する要因指標特定ステップと、
前記障害要素特定ステップで特定された障害要素と前記要因指標特定ステップで特定された指標との少なくとも一方を出力する出力ステップと、
を備えることを特徴とする。
コンピュータを、
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出手段、
前記障害寄与度算出手段により求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定手段、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する要因指標特定手段、
前記障害要素特定手段により特定された障害要素と前記要因指標特定手段により特定された指標との少なくとも一方を出力する出力手段、
として機能させるためのプログラムを記録する。
本実施形態に係る障害分析装置100は、分析対象である機械システムから取得される、異なる属性のセンサ情報やログ情報等から、発生した障害を特定し、その障害における注目属性を抽出する、プログラム制御により動作するコンピュータである。ここで、属性とは、各種センサ等の種別を指し、例えば、エアコンにおける室外機では、外気温センサや、室外熱交換センサ等である。障害分析装置100は、図1に示すように、入力部101と、通信部102と、制御部103と、記憶部104と、出力部105と、から構成されている。
f(Q,A)=X’=A1Q1+A2Q2+・・・+AMQM ・・・(1)
A=argmin_A(X-X’)2 ・・・(2)
式(2)は、(X-X’)2を最小にするようなAを求める演算を表す。
この発明は、上記実施形態に限定されず、種々の変形及び応用が可能である。例えば、上記実施形態では、外部データベースシステムから分析対象データである入力データを読み出し、分析処理を行う例を示した。しかし、これは一例であり、ユーザーが入力部101から入力した分析対象データを入力データとして障害分析処理部107が受け取り、分析処理を行ってもよく、入力データを取得する方法は任意である。
本実施形態に係る障害分析装置200は、第1の実施形態における障害パタン行列110を予め記憶部104に格納することなく、発生している障害と、その障害に関連する属性を注目属性として抽出する障害分析装置である。
なお、上記実施形態は一例であり、第1の実施形態と同様に、種々の変形及び応用が可能である。上記実施形態では、属性パタンテーブル112は、各障害に対する各属性の代表値の和を利用して求められたが、これは一例であり、任意の方法を利用することができる。
本実施形態に係る障害分析装置300は、入力データとして時系列データを用いて、障害が発生した際の各属性値の変化度を算出することにより、第1の実施形態に係る障害分析装置100で出力する情報に加え、注目属性に生じた変化情報を出力する障害分析装置である。
なお、上記実施形態は一例であり、種々の変形及び応用が可能である。
本実施形態に係る障害分析装置400は、第3の実施形態に係る分析装置300と同様に、時系列データを用いて、障害が発生した際の各属性値の変化度を算出する。さらに、障害分析装置400は、第2の実施形態に係る分析装置200と同様に、障害に関する予備知識を必要とすることなく、必要な情報を特定する。
更に、プログラムの全部又は一部をサーバ装置上で実行させ、その処理に関する情報をコンピュータが通信ネットワークを介して送受信しながらプログラムを実行することによっても、上述の処理を達成することができる。
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出手段と、
前記障害寄与度算出手段により求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定手段と、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する要因指標特定手段と、
前記障害要素特定手段により特定された障害要素と前記要因指標特定手段により特定された指標との少なくとも一方を出力する出力手段と、
を備えることを特徴とする障害分析装置。
前記障害寄与度算出手段は、
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データを記憶する分析対象データ記憶手段と、
予め記憶されている、複数の障害それぞれに対する前記複数の指標の代表値を示す障害パタンの要素の代表値と、前記分析対象データ記憶手段に記憶された分析対象データの要素の値と、の対応するもの同士の相対的な関係に基づいて、発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める手段と、
を備え、
前記要因指標特定手段は、障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値の組を示す、予め記憶されている属性パタンの要素の値と、前記分析対象データ記憶手段に記憶された分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する、
ことを特徴とする付記1に記載の障害分析装置。
前記障害寄与度算出手段は、
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データを記憶する分析対象データ記憶手段と、
予め記憶されている、複数の障害それぞれに対する前記複数の指標の代表値を示す障害パタンの要素の代表値と、前記分析対象データ記憶手段に記憶された分析対象データの要素の値と、の対応するもの同士の相対的な関係に基づいて、発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める手段と、
を備え、
前記要因指標特定手段は、障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値の組を示す、属性パタンの要素の値を、前記障害パタンの要素の代表値に基づいて算出し、算出した前記属性パタンの要素の値と、前記分析対象データ記憶手段に記憶された分析対象データの各要素の値に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する、
ことを特徴とする付記1に記載の障害分析装置。
前記分析対象データから、前記分析対象データの各指標値を配列して第1行列を生成するデータ行列生成手段と、
前記データ行列生成手段により生成された前記第1行列の各行の各要素の値を前記分析対象データとして記憶するデータ行列記憶手段と、
前記分析対象データから、前記分析対象データに対する障害の発生の有無を配列して第2行列を生成する障害発生行列生成手段と、
をさらに備え、
前記障害寄与度算出手段は、
前記障害発生行列生成手段により生成された前記第2行列と、未知の、複数の障害それぞれに対する前記複数の指標の代表値を示す障害パタンと、を対応付けたものの各要素の値を記憶する障害パタン記憶手段と、
前記データ行列記憶手段に記憶された複数の分析対象データの要素の値と、前記障害パタン記憶手段に記憶された各要素の値との、対応するもの同士の相対的な関係に基づいて前記障害寄与度および前記障害パタンの要素の値を算出する手段と、
を備え、
前記要因指標特定手段は、障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値の組を示す、予め記憶されている属性パタンの要素の値と、前記分析対象データ記憶手段に記憶された分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する、
ことを特徴とする付記1に記載の障害分析装置。
前記分析対象データから、前記分析対象データの各指標値を配列して第1行列を生成するデータ行列生成手段と、
前記データ行列生成手段により生成された前記第1行列の各行の各要素の値を前記分析対象データとして記憶するデータ行列記憶手段と、
前記分析対象データから、前記分析対象データに対する障害の発生の有無を配列して第2行列を生成する障害発生行列生成手段と、
をさらに備え、
前記障害寄与度算出手段は、
前記障害発生行列生成手段により生成された前記第2行列と、未知の、複数の障害それぞれに対する前記複数の指標の代表値を示す障害パタンと、を対応付けたものの各要素の値を記憶する障害パタン記憶手段と、
前記データ行列記憶手段に記憶された複数の分析対象データの要素の値と、前記障害パタン記憶手段に記憶された各要素の値との、対応するもの同士の相対的な関係に基づいて前記障害寄与度および前記障害パタンの要素の値を算出する手段と、
を備え、
前記要因指標特定手段は、障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値の組を示す、属性パタンの要素の値を、前記障害パタンの要素の値に基づいて算出し、算出した前記属性パタンの要素の値と、前記分析対象データ記憶手段に記憶された分析対象データの各要素の値に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する、
ことを特徴とする付記1に記載の障害分析装置。
前記分析対象データ記憶手段が、分析対象物の複数の指標の値をそのまま分析対象データとして記憶する、
ことを特徴とする付記2乃至5のいずれか一個に記載の障害分析装置。
前記分析対象データの各指標の値の変化度を求める変化度算出手段と、
前記変化度算出手段により求められた変化度を示す値の組を前記分析対象データとして記憶する手段と、
をさらに備え、
前記障害パタンが、各障害に対応する各指標の値の変化度の代表値を示す値であり、
前記属性パタンが、各障害の組み合わせに対応する各指標の値の変化度の代表値を示す値である、
ことを特徴とする付記2乃至5のいずれか一個に記載の障害分析装置。
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出ステップと、
前記障害寄与度算出ステップで求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定ステップと、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定ステップで特定された障害要素の要因と推定される指標を特定する要因指標特定ステップと、
前記障害要素特定ステップで特定された障害要素と前記要因指標特定ステップで特定された指標との少なくとも一方を出力する出力ステップと、
を備えることを特徴とする障害分析方法。
コンピュータを、
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出手段、
前記障害寄与度算出手段により求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定手段、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する要因指標特定手段、
前記障害要素特定手段により特定された障害要素と前記要因指標特定手段により特定された指標との少なくとも一方を出力する出力手段、
として機能させるためのプログラムを記録した、コンピュータ読み取り可能な記録媒体。
101 入力部
102 通信部
103 制御部
104 記憶部
105 出力部
106 プログラム
107 障害分析処理部
108 属性抽出処理部
110 障害パタン行列
111 属性パタンデータベース
112 属性パタンテーブル
200 障害分析装置
205 データ行列生成処理部
206 障害発生行列生成処理部
207 障害パタン推定処理部
208 データ行列
209 障害発生行列
300 障害分析装置
301 データ変化度算出処理部
302 障害変化パタン行列
303 属性変化パタンデータベース
304 属性変化パタンテーブル
400 障害分析装置
Claims (9)
- 分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出手段と、
前記障害寄与度算出手段により求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定手段と、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する要因指標特定手段と、
前記障害要素特定手段により特定された障害要素と前記要因指標特定手段により特定された指標との少なくとも一方を出力する出力手段と、
を備えることを特徴とする障害分析装置。 - 前記障害寄与度算出手段は、
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データを記憶する分析対象データ記憶手段と、
予め記憶されている、複数の障害それぞれに対する前記複数の指標の代表値を示す障害パタンの要素の代表値と、前記分析対象データ記憶手段に記憶された分析対象データの要素の値と、の対応するもの同士の相対的な関係に基づいて、発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める手段と、
を備え、
前記要因指標特定手段は、障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値の組を示す、予め記憶されている属性パタンの要素の値と、前記分析対象データ記憶手段に記憶された分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する、
ことを特徴とする請求項1に記載の障害分析装置。 - 前記障害寄与度算出手段は、
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データを記憶する分析対象データ記憶手段と、
予め記憶されている、複数の障害それぞれに対する前記複数の指標の代表値を示す障害パタンの要素の代表値と、前記分析対象データ記憶手段に記憶された分析対象データの要素の値と、の対応するもの同士の相対的な関係に基づいて、発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める手段と、
を備え、
前記要因指標特定手段は、障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値の組を示す、属性パタンの要素の値を、前記障害パタンの要素の代表値に基づいて算出し、算出した前記属性パタンの要素の値と、前記分析対象データ記憶手段に記憶された分析対象データの各要素の値に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する、
ことを特徴とする請求項1に記載の障害分析装置。 - 前記分析対象データから、前記分析対象データの各指標値を配列して第1行列を生成するデータ行列生成手段と、
前記データ行列生成手段により生成された前記第1行列の各行の各要素の値を前記分析対象データとして記憶するデータ行列記憶手段と、
前記分析対象データから、前記分析対象データに対する障害の発生の有無を配列して第2行列を生成する障害発生行列生成手段と、
をさらに備え、
前記障害寄与度算出手段は、
前記障害発生行列生成手段により生成された前記第2行列と、未知の、複数の障害それぞれに対する前記複数の指標の代表値を示す障害パタンと、を対応付けたものの各要素の値を記憶する障害パタン記憶手段と、
前記データ行列記憶手段に記憶された複数の分析対象データの要素の値と、前記障害パタン記憶手段に記憶された各要素の値との、対応するもの同士の相対的な関係に基づいて前記障害寄与度および前記障害パタンの要素の値を算出する手段と、
を備え、
前記要因指標特定手段は、障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値の組を示す、予め記憶されている属性パタンの要素の値と、前記分析対象データ記憶手段に記憶された分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する、
ことを特徴とする請求項1に記載の障害分析装置。 - 前記分析対象データから、前記分析対象データの各指標値を配列して第1行列を生成するデータ行列生成手段と、
前記データ行列生成手段により生成された前記第1行列の各行の各要素の値を前記分析対象データとして記憶するデータ行列記憶手段と、
前記分析対象データから、前記分析対象データに対する障害の発生の有無を配列して第2行列を生成する障害発生行列生成手段と、
をさらに備え、
前記障害寄与度算出手段は、
前記障害発生行列生成手段により生成された前記第2行列と、未知の、複数の障害それぞれに対する前記複数の指標の代表値を示す障害パタンと、を対応付けたものの各要素の値を記憶する障害パタン記憶手段と、
前記データ行列記憶手段に記憶された複数の分析対象データの要素の値と、前記障害パタン記憶手段に記憶された各要素の値との、対応するもの同士の相対的な関係に基づいて前記障害寄与度および前記障害パタンの要素の値を算出する手段と、
を備え、
前記要因指標特定手段は、障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値の組を示す、属性パタンの要素の値を、前記障害パタンの要素の値に基づいて算出し、算出した前記属性パタンの要素の値と、前記分析対象データ記憶手段に記憶された分析対象データの各要素の値に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する、
ことを特徴とする請求項1に記載の障害分析装置。 - 前記分析対象データ記憶手段が、分析対象物の複数の指標の値をそのまま分析対象データとして記憶する、
ことを特徴とする請求項2乃至5のいずれか一項に記載の障害分析装置。 - 前記分析対象データの各指標の値の変化度を求める変化度算出手段と、
前記変化度算出手段により求められた変化度を示す値の組を前記分析対象データとして記憶する手段と、
をさらに備え、
前記障害パタンが、各障害に対応する各指標の値の変化度の代表値を示す値であり、
前記属性パタンが、各障害の組み合わせに対応する各指標の値の変化度の代表値を示す値である、
ことを特徴とする請求項2乃至5のいずれか一項に記載の障害分析装置。 - 分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出ステップと、
前記障害寄与度算出ステップで求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定ステップと、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定ステップで特定された障害要素の要因と推定される指標を特定する要因指標特定ステップと、
前記障害要素特定ステップで特定された障害要素と前記要因指標特定ステップで特定された指標との少なくとも一方を出力する出力ステップと、
を備えることを特徴とする障害分析方法。 - コンピュータを、
分析対象物の複数の指標の値に基づいて生成された値を要素とする分析対象データと、複数の障害それぞれに対する前記複数の指標の代表値との相対的な関係に基づいて、前記分析対象物に発生している障害に個々の障害(以下、障害要素という)がどの程度寄与しているかを示す障害寄与度を求める障害寄与度算出手段、
前記障害寄与度算出手段により求められた障害寄与度に基づいて、発生している障害要素を特定する障害要素特定手段、
障害要素を組み合わせた障害が発生しているときに、前記複数の指標がとる代表値と、前記分析対象データの各要素の値と、に基づいて、前記障害要素特定手段により特定された障害要素の要因と推定される指標を特定する要因指標特定手段、
前記障害要素特定手段により特定された障害要素と前記要因指標特定手段により特定された指標との少なくとも一方を出力する出力手段、
として機能させるためのプログラムを記録した、コンピュータ読み取り可能な記録媒体。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP11777430.7A EP2568433A4 (en) | 2010-05-06 | 2011-04-26 | Malfunction analysis apparatus, malfunction analysis method, and recording medium |
CN201180022767.9A CN102884486B (zh) | 2010-05-06 | 2011-04-26 | 故障分析装置、故障分析方法 |
JP2012513806A JP5765336B2 (ja) | 2010-05-06 | 2011-04-26 | 障害分析装置、障害分析方法およびプログラム |
US13/643,391 US9043645B2 (en) | 2010-05-06 | 2011-04-26 | Malfunction analysis apparatus, malfunction analysis method, and recording medium |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2010106810 | 2010-05-06 | ||
JP2010-106810 | 2010-05-06 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2011138911A1 true WO2011138911A1 (ja) | 2011-11-10 |
Family
ID=44903763
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2011/060108 WO2011138911A1 (ja) | 2010-05-06 | 2011-04-26 | 障害分析装置、障害分析方法および記録媒体 |
Country Status (5)
Country | Link |
---|---|
US (1) | US9043645B2 (ja) |
EP (1) | EP2568433A4 (ja) |
JP (1) | JP5765336B2 (ja) |
CN (1) | CN102884486B (ja) |
WO (1) | WO2011138911A1 (ja) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2013120475A (ja) * | 2011-12-07 | 2013-06-17 | Shunji Sugaya | 診断対処サーバ、診断対処方法、及び、診断対処サーバ用プログラム |
JP2020135242A (ja) * | 2019-02-15 | 2020-08-31 | オムロン株式会社 | 異常検知システム、異常検知方法およびプログラム |
JP2020173491A (ja) * | 2019-04-08 | 2020-10-22 | 株式会社日立製作所 | 判断支援装置および判断支援方法 |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP5765336B2 (ja) * | 2010-05-06 | 2015-08-19 | 日本電気株式会社 | 障害分析装置、障害分析方法およびプログラム |
CN107203450B (zh) * | 2016-03-16 | 2020-06-02 | 伊姆西Ip控股有限责任公司 | 故障的分类方法和设备 |
JP6579995B2 (ja) * | 2016-04-26 | 2019-09-25 | 三菱電機株式会社 | 静観候補特定装置、静観候補特定方法及び静観候補特定プログラム |
DE102016225081A1 (de) * | 2016-12-15 | 2018-06-21 | Robert Bosch Gmbh | Vorrichtung und Verfahren zum Bestimmen der Pinpoint-Fähigkeit möglicher Fehler einer oder mehrerer Komponenten |
WO2020039610A1 (ja) * | 2018-08-20 | 2020-02-27 | 日本電信電話株式会社 | 異常要因推定装置、異常要因推定方法、及びプログラム |
CN111609526B (zh) | 2019-02-25 | 2023-11-14 | 开利公司 | Hvac系统不舒适指数和显示 |
CN111026570B (zh) * | 2019-11-01 | 2022-05-31 | 支付宝(杭州)信息技术有限公司 | 用于确定业务系统异常原因的方法和装置 |
US11676071B2 (en) * | 2020-06-30 | 2023-06-13 | Oracle International Corporation | Identifying and ranking anomalous measurements to identify faulty data sources in a multi-source environment |
CN112199226B (zh) * | 2020-10-14 | 2022-12-27 | 北京紫光展锐通信技术有限公司 | 问题确定方法及相关产品 |
CN112650608B (zh) * | 2020-12-28 | 2024-03-01 | 广州虎牙科技有限公司 | 异常根因定位方法以及相关装置、设备 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0546428A (ja) * | 1991-08-08 | 1993-02-26 | Nec Software Ltd | フアジー診断プロセツサ |
JPH05151373A (ja) * | 1991-11-26 | 1993-06-18 | Komatsu Ltd | ニユーラルネツトワークを用いたフアジイ逆診断方法 |
JP2005257416A (ja) | 2004-03-10 | 2005-09-22 | Internatl Business Mach Corp <Ibm> | 診断装置、検出装置、制御方法、検出方法、プログラム、及び記録媒体 |
JP2007328645A (ja) * | 2006-06-09 | 2007-12-20 | Fuji Xerox Co Ltd | 故障診断システム、画像形成装置および故障診断プログラム |
JP2010106810A (ja) | 2008-10-31 | 2010-05-13 | Honda Motor Co Ltd | V型内燃機関 |
Family Cites Families (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3224226B2 (ja) * | 1989-09-22 | 2001-10-29 | 株式会社リコー | 故障診断エキスパートシステム |
JPH0816956B2 (ja) * | 1990-01-18 | 1996-02-21 | 株式会社日立製作所 | フォールト・トリー表示方法、フォールト・トリー表示装置およびプロセス診断支援システム |
GB0216858D0 (en) * | 2002-07-19 | 2002-08-28 | Bae Systems Plc | Fault diagnosis system |
CN1548286A (zh) | 2003-05-19 | 2004-11-24 | 天津市天锻压力机有限公司 | 液压机异地故障分析诊断系统 |
JP2005222108A (ja) | 2004-02-03 | 2005-08-18 | Matsushita Electric Ind Co Ltd | バグ分析方法および装置 |
JP4347150B2 (ja) * | 2004-07-09 | 2009-10-21 | 三菱重工業株式会社 | 空調サービス支援装置 |
JP4527642B2 (ja) * | 2005-09-29 | 2010-08-18 | 富士通株式会社 | ネットワーク障害診断装置、ネットワーク障害診断方法およびネットワーク障害診断プログラム |
EP2813913A1 (en) * | 2007-03-22 | 2014-12-17 | Nec Corporation | Diagnostic device |
JP5200678B2 (ja) | 2008-06-13 | 2013-06-05 | 日本電気株式会社 | サービシステム、サービスシステム管理方法、及びプログラム |
JP5048625B2 (ja) | 2008-10-09 | 2012-10-17 | 株式会社日立製作所 | 異常検知方法及びシステム |
JP5765336B2 (ja) * | 2010-05-06 | 2015-08-19 | 日本電気株式会社 | 障害分析装置、障害分析方法およびプログラム |
-
2011
- 2011-04-26 JP JP2012513806A patent/JP5765336B2/ja active Active
- 2011-04-26 US US13/643,391 patent/US9043645B2/en active Active
- 2011-04-26 WO PCT/JP2011/060108 patent/WO2011138911A1/ja active Application Filing
- 2011-04-26 EP EP11777430.7A patent/EP2568433A4/en not_active Withdrawn
- 2011-04-26 CN CN201180022767.9A patent/CN102884486B/zh active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0546428A (ja) * | 1991-08-08 | 1993-02-26 | Nec Software Ltd | フアジー診断プロセツサ |
JPH05151373A (ja) * | 1991-11-26 | 1993-06-18 | Komatsu Ltd | ニユーラルネツトワークを用いたフアジイ逆診断方法 |
JP2005257416A (ja) | 2004-03-10 | 2005-09-22 | Internatl Business Mach Corp <Ibm> | 診断装置、検出装置、制御方法、検出方法、プログラム、及び記録媒体 |
JP2007328645A (ja) * | 2006-06-09 | 2007-12-20 | Fuji Xerox Co Ltd | 故障診断システム、画像形成装置および故障診断プログラム |
JP2010106810A (ja) | 2008-10-31 | 2010-05-13 | Honda Motor Co Ltd | V型内燃機関 |
Non-Patent Citations (4)
Title |
---|
A. KYRIAZIS ET AL.: "Enhanced Fault Localization Using Probabilistic Fusion With Gas Path Analysis Algorithms, Transactions of the ASME.", JOURNAL OF ENGINEERING FOR GAS TURBINES AND POWER, vol. 131, 2009, pages 051601.1 - 051601.9, XP008164108 * |
JUN-ICHI TAKEUCHI; KENJI YAMANISHI: "A Unifying Framework for Detecting Outliers and Change Points from time Series", IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING, vol. 18, no. 4, April 2006 (2006-04-01) |
See also references of EP2568433A4 * |
YOICHI FUKUCHI; NORIO HIRAI; KEN SUZAKU; ATSUSHI KONO; TADAHISA HARADA: "VRM solution directed toward the safe and comfortable automotive society", VRM (VEHICLE RELATIONSHIP MANAGEMENT |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2013120475A (ja) * | 2011-12-07 | 2013-06-17 | Shunji Sugaya | 診断対処サーバ、診断対処方法、及び、診断対処サーバ用プログラム |
JP2020135242A (ja) * | 2019-02-15 | 2020-08-31 | オムロン株式会社 | 異常検知システム、異常検知方法およびプログラム |
JP2020173491A (ja) * | 2019-04-08 | 2020-10-22 | 株式会社日立製作所 | 判断支援装置および判断支援方法 |
JP7280094B2 (ja) | 2019-04-08 | 2023-05-23 | 株式会社日立製作所 | 判断支援装置および判断支援方法 |
Also Published As
Publication number | Publication date |
---|---|
US9043645B2 (en) | 2015-05-26 |
JPWO2011138911A1 (ja) | 2013-07-22 |
US20130042148A1 (en) | 2013-02-14 |
JP5765336B2 (ja) | 2015-08-19 |
CN102884486B (zh) | 2014-10-29 |
CN102884486A (zh) | 2013-01-16 |
EP2568433A1 (en) | 2013-03-13 |
EP2568433A4 (en) | 2017-10-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5765336B2 (ja) | 障害分析装置、障害分析方法およびプログラム | |
EP3385889A1 (en) | Abnormality detection system, abnormality detection method, abnormality detection program, and method for generating learned model | |
JP6658250B2 (ja) | 異常診断方法、異常診断装置、及び異常診断プログラム | |
JP5510760B2 (ja) | 計算機システム、及び、規則生成方法 | |
JP4282717B2 (ja) | 定期点検データ分析装置およびその方法 | |
EP3482261A1 (en) | Computer system and method for the dynamic construction and online deployment of an operation-centric first-principles process model for predictive analytics | |
JP2008003920A (ja) | 時系列データの予測・診断装置およびそのプログラム | |
US20080086437A1 (en) | Incremental Learning of Nonlinear Regression Networks For Machine Condition Monitoring | |
JP7292076B2 (ja) | 劣化推定装置、学習装置、劣化推定方法、学習方法、劣化推定プログラム、および、学習プログラム | |
US11748321B2 (en) | Time-series data condensation and graphical signature analysis | |
JP6777142B2 (ja) | システム分析装置、システム分析方法、及び、プログラム | |
US20140188777A1 (en) | Methods and systems for identifying a precursor to a failure of a component in a physical system | |
JP2020035407A (ja) | 異常予兆診断装置及び異常予兆診断方法 | |
JP5695763B2 (ja) | イベント系列のリスク評価値を可視化する方法、装置及びコンピュータプログラム | |
Gitzel | Data Quality in Time Series Data: An Experience Report. | |
JPWO2020148838A1 (ja) | 推定装置、推定方法、及びプログラム | |
JP2016091417A (ja) | 異常診断方法、異常診断装置及び異常診断プログラム | |
JP5771317B1 (ja) | 異常診断装置及び異常診断方法 | |
JP2007323315A (ja) | 協調フィルタリング方法、協調フィルタリング装置、および協調フィルタリングプログラムならびにそのプログラムを記録した記録媒体 | |
SA519410228B1 (ar) | أداة لتقدير المخاطر ونظام لتقدير المخاطر وطريقة لتقدير المخاطر وبرنامج لتقدير المخاطر وبنية بيانات | |
EP4064141A1 (en) | Failure probability evaluation system | |
US8438129B1 (en) | Probabilistic implementation of system health prognosis | |
JP5771318B1 (ja) | 異常診断装置及び異常診断方法 | |
WO2016163008A1 (ja) | 異常診断装置および異常診断方法 | |
JP2020008918A (ja) | 分析フロー作成システムおよび分析フロー作成方法 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 201180022767.9 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 11777430 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13643391 Country of ref document: US |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2012513806 Country of ref document: JP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2011777430 Country of ref document: EP |