JP6183252B2 - Failure information presentation system - Google Patents

Description

  The present invention relates to a technique for presenting information relating to a vehicle failure to a repairer or the like.

  Using a database in which information indicating vehicle status at the time of a failure that occurred in the past and information (repair information) indicating the contents of repairs performed for the failure are stored in association with each other, A technique is known in which repair information is searched based on the degree of similarity between the detected vehicle state and the vehicle state stored in a database, and the search result is presented to a repairer or the like.

  In addition, it has been proposed to use freeze frame data (FFD) as a feature value that represents the vehicle state at the time of failure, and to use a multidimensional vector that represents whether or not each sensor value constituting the FFD is expressed in binary. (For example, refer to Patent Document 1).

Japanese Patent No. 4640475

  The feature amount based on the FFD described above represents a pattern of sensor values measured at temporally local timings before and after the occurrence of an abnormality based on the measured values of a plurality of in-vehicle sensors. For this reason, it affects the abnormality that causes a time-series change in the sensor value, for example, the abnormality in which the fluctuation pattern of the sensor value changes, the abnormality in which the fluctuation frequency changes, or the order in which the sensor values fluctuate between specific sensors. There has been a problem that it is impossible to detect such abnormalities.

  The present invention has been made in view of such problems, and an object of the present invention is to provide a technique that realizes accurate presentation of failure information for failures that cause time-series abnormalities in the measurement values of the in-vehicle sensor. .

  The failure information presentation system of the present invention includes information acquisition means, case storage means, case extraction means, and information presentation means. The information acquisition unit acquires vehicle state information in which time-series data of a plurality of measurement values representing a vehicle state is expressed by a predetermined feature amount. The case storage means includes state identification information in which time series data of measured values measured at the time of a vehicle failure is expressed by a feature amount, repair information indicating a cause of the failure corresponding to the state identification information and a repair method of the failure, A plurality of case information associated with is stored. The case extraction means extracts one or more case information according to the similarity between the vehicle state information and the state identification information obtained for each case information stored in the case storage means. The information presenting means presents repair information related to the case information extracted by the case extracting means. In the present invention, a symbol string obtained by symbolizing a pattern that repeatedly appears in time-series data of measured values is used as the feature amount.

  As described above, in the present invention, since the characteristic value representing the state identification information and the vehicle state information is obtained by converting the time-series data of the measurement values into a symbol string, it is locally localized in terms of individual symbols. It is possible not only to express the events, but also to efficiently represent the occurrence pattern and frequency of occurrence of these events by the symbol string, so it is possible to accurately grasp failures that cause time-series abnormalities in these events. it can.

  In addition, by using a symbol string as a feature amount, robustness against a time lag or a scale change of each event can be obtained, and a time series of feature amounts (multi-dimensional vectors) based on a conventional FFD is used. Compared to the case, it is possible to reduce the memory capacity required for storing information and the processing amount required for searching the stored information.

  In addition, the code | symbol in the parenthesis described in the claim shows the correspondence with the specific means as described in embodiment mentioned later as one aspect, Comprising: The technical scope of this invention is limited is not.

It is a block diagram which shows the whole structure of a failure information presentation system. It is explanatory drawing which shows the content of the failure case information accumulate | stored in a failure case information server. It is explanatory drawing about the production | generation of the symbol row | line | column used as a vehicle state symbol. It is a flowchart which shows the content of the process which an information gathering apparatus performs. It is a flowchart which shows the content of the failure diagnosis process which the information processing part of a failure information presentation apparatus performs. It is explanatory drawing about the Levenshtein distance used for calculation of the similarity between symbol strings. It is explanatory drawing which illustrated the similarity calculation result and the case candidate extraction result. It is a flowchart which shows the content of the information registration process which the information processing part of a failure information presentation apparatus performs. It is explanatory drawing which illustrates the effect by using a symbol row | line. It is explanatory drawing which shows the other structural example of failure example information.

Embodiments to which the present invention is applied will be described below with reference to the drawings.
[overall structure]
As shown in FIG. 1, a failure information presentation system 1 to which the present invention is applied includes a vehicle-mounted device group 10 mounted on a vehicle, and failure information that presents information about a failure based on information acquired from the vehicle-mounted device group 10. A presentation device 20 and a failure case information server 30 that provides information on failure cases are provided.

[Failure case information server]
The failure case information server 30 includes a storage unit that accumulates a lot of failure case information that is information related to vehicle failure cases that have occurred in the past, and a communication unit that is used for communication with the failure information presentation device 20.

  As shown in FIG. 2, the failure case information includes a case number for identifying a failure case, a failure case classification that classifies the failure case based on its cause, etc., and various information and failure cases necessary for failure cause analysis. It has repair information consisting of details of repairs performed, etc., and vehicle state symbols (corresponding to state device-specific information in the present invention) consisting of features representing the vehicle state measured when the failure case occurred.

  Specifically, the repair information includes the model and year of the vehicle in which the failure occurred, the region and time at which the abnormality was detected, the failure location and inspection items performed, the parts replaced by the repair, the location and date of repair, etc. Is included.

  Here, as the feature amount representing the vehicle state, as shown in FIG. 3, it is generated by sequentially assigning symbols (C1, C2,...) To a multi-dimensional measurement value pattern that repeatedly appears in the vehicle state information. Use symbol sequences. For generation of the symbol string, an sHDP-HMM (sticky Hierarchical Dirichlet Process-Hidden Markov Model) obtained by extending a hidden Markov model (HMM: Hidden Markov Model) often used for analyzing time series data is used. The HMM converts time-series data into a hidden state transition sequence, and the sHDP-HMM can automatically determine the number of hidden states that had to be set in advance in the HMM from the data. It is what has.

[In-vehicle device group]
Returning to FIG. 1, the in-vehicle device group 10 includes an in-vehicle sensor group 11, an input device 12, an information collection device 13, a state symbol storage device 14, and an interface (IF) unit 15. For example, they are connected to each other via CAN).

  The in-vehicle sensor group 11 is mounted on the vehicle and measures the state of each part of the vehicle, and detects, for example, a throttle opening, an engine speed, an intake air amount, and the like. However, the present invention is not limited to these, and any signal may be used as long as the influence appears in some form when any trouble occurs.

  The input device 12 is used to input various commands (at least a measurement start command and a measurement end command) to the information collecting device 13, and includes a switch, a touch panel, and the like disposed at a position where the driver can operate.

The state symbol storage device 14 stores the symbol sequence (state symbol sequence) calculated by the information collection device 13 together with the measurement time and the like.
The IF unit 15 is for connecting an external device such as the failure information presentation device 20 to the in-vehicle network, and the external device connected to the IF unit 15 can communicate with the in-vehicle device connected to the in-vehicle network. It becomes.

The information collection device 13 is a known microcomputer having a CPU, a ROM, and a RAM, and repeatedly executes information collection processing while the device is powered on.
<Information collection processing>
The contents of the information collection process will be described with reference to the flowchart shown in FIG.

  When this process is started, the CPU of the information collection device 13 first determines whether or not a collection condition for starting the collection of information is satisfied in S110. Specifically, from when the measurement start command is input via the input device 12 until the measurement end command is input, or when the vehicle is in a predetermined state (for example, at idling or at a predetermined speed or higher). At the constant speed), it is determined that the collection condition is satisfied.

If it is determined that the collection condition is not satisfied (S110: NO), this process is temporarily terminated. When it is determined that the collection condition is satisfied (S110: YES), the process proceeds to S120.
In S120, the measurement value by the vehicle-mounted sensor group 11 is repeatedly acquired at a predetermined period during a predetermined measurement period. Multidimensional time-series data of measurement values acquired from a plurality of in-vehicle sensors in this way is referred to as vehicle state information.

  In subsequent S130, the vehicle state information acquired in S120 is converted into a symbol string using sHDP-HMM (see FIG. 3). In subsequent S140, the symbol string generated in S130 is stored in the state symbol storage device 14 in association with additional information such as the acquisition time and acquisition position of the vehicle state information that is the generation source, and this processing is performed. Exit once.

  In addition, when acquiring vehicle state information by inputting a measurement start command and a measurement end command, the driver performs a driving operation according to a predetermined test pattern for failure diagnosis while the collection condition is satisfied. It is desirable to make it. Since the vehicle state information obtained in this way is acquired in a running state suitable for diagnosis, it is possible to suppress variations in measurement values due to reasons other than failure. The execution of the test driving pattern is usually an operation performed at a repair shop or the like. For example, the test driving pattern is presented to the driver via an in-vehicle display device or the like to collect vehicle state information to the user. May be performed. When the driving operation is performed according to the test travel pattern, it is desirable to store the identification information of the test travel pattern in association with the additional information stored in the state symbol storage device 14. Further, when the vehicle condition information is acquired because the collection condition is satisfied because the vehicle is in a predetermined state, the predetermined state (for example, idling or constant speed traveling at a predetermined speed or higher) is identified. The identification information is also preferably stored as additional information in the state symbol storage device 14.

[Failure information presentation device]
Returning to FIG. 1, the failure information presentation device 20 includes an interface (IF) unit 21, an input unit 22, a display unit 23, an information processing unit 24, a failure case storage unit 25, and a communication unit 26.

The IF unit 21 is used to connect the IF unit 15 of the in-vehicle device group 10 and connects the failure information presentation device 20 to the in-vehicle network.
The input unit 22 is for an operator of the failure information presentation device 20 to input various commands, and includes a keyboard, a touch panel, and the like.

The display unit 23 is for visually presenting guidance regarding an input operation via the input unit 22 and a processing result in the information processing unit 24.
The communication unit 26 is for communicating with the failure case information server 30 using a public wireless communication line or the like.

The failure case storage unit 25 stores failure case information acquired from the failure case information server 30 via the communication unit 26 and information generated by the processing of the information processing unit 24.
The information processing unit 24 includes a well-known microcomputer including a CPU, a ROM, and a RAM, and executes at least a failure diagnosis process and an information registration process. Among these, the failure diagnosis process is activated when a diagnosis start command is input via the input unit 22 in a state where the IF unit 21 is connected to the vehicle-side IF unit 15. On the other hand, the information registration process starts when a registration command is input via the input unit 22 regardless of the connection state of the IF unit 21.

<Fault diagnosis processing>
The contents of the failure diagnosis process will be described with reference to the flowchart shown in FIG.
When this process is activated, the CPU of the information processing unit 24 first updates the failure case information stored in the failure case storage unit 25 as necessary in S210. Specifically, the failure case information stored in the failure case storage unit 25 matches the information stored in the failure case information server 30 through communication with the failure case information server 30 via the communication unit 26. If they do not match, the information is acquired from the failure case information server 30 and the stored contents of the failure case storage unit 25 are updated. In addition to this, the update may be performed when a predetermined time or more has elapsed since the last update, or when an update command is input via the input unit 22.

  In subsequent S220, the symbol string stored in the state symbol storage device 14 is acquired via the IF unit 21. At this time, information related to the above-described repair information such as the vehicle type and year of the acquisition destination vehicle is also acquired.

  In subsequent S230, one of the failure case information stored in the failure case storage unit 25 is selected, and in S240, a symbol string (as a vehicle state symbol of the failure case information selected in S230) ( The similarity between the example symbol string) and the symbol string acquired in S220 (acquired symbol string) is calculated.

  Note that the normalized Levenstein distance is used as the similarity. As shown in FIG. 6, the Levenshtein distance is calculated based on the number of editing operations (“one character replacement”, “one character deletion”, “one character addition”) from one symbol row to the other symbol row. Distance. Here, the similarity is obtained by normalizing the number of edits (Levenshtein distance) by the longer number of symbols in both symbol sequences.

  In subsequent S250, it is determined whether or not the processing of S240 has been performed for all the failure cases stored in the failure case storage unit 25. If there is a failure case in which the processing is not performed, the process returns to S230. .

  On the other hand, if processing has been performed for all the failure cases, failure case candidates are extracted based on the obtained similarity in S260. Specifically, as shown in FIG. 7, those having a degree of similarity of a predetermined value (for example, 0.8) or more are extracted as failure case candidates. However, when there are a plurality of failure case candidates having the same failure case classification, only those having the highest similarity are extracted.

In continuing S270, the repair information regarding the failure example extracted by S260 is shown via the display part 23, and this process is complete | finished.
An operator (such as an operator of a repair shop) of the failure information presentation device 20 performs the repair work for the specified failure by specifying the failure location with reference to the presented information.

  In S250, it is determined whether or not the process of S240 has been performed for all the failure cases stored in the failure case storage unit 25, but the additional information of the vehicle state information includes the identification information of the running pattern. In this case, it is desirable to determine whether or not the processing has been performed for all the failure cases having the same identification information among the failure cases stored in the failure case storage unit 25. As a result, failure case information having a low similarity to the vehicle state information can be excluded from the processing in advance, and it is possible to extract an appropriate failure case candidate and shorten the processing time.

<Information registration process>
The contents of the information registration process will be described with reference to the flowchart shown in FIG.
When this process is started, the CPU of the information processing unit 24 first specifies an acquisition symbol string to be registered in S310. In subsequent S320, for the failure case specified from the acquired symbol string specified in S310, the failure case classification indicating the specified failure cause, and the content of the repair performed on the specified failure (inspection item or replacement) Implementation information representing components) is acquired via the input unit 22. These failure case classifications and execution information are input via the input unit 22 by an operator who has performed repairs.

  In subsequent S330, the acquisition symbol string designated in S310 is used as a vehicle state symbol, the implementation information acquired in S320 is used as repair information, and update case information is generated by associating the update case information with the communication unit. 26 is registered in the failure case information server 30 through 26, and this process is terminated.

  The failure case information server 30 is registered when failure case information that matches both the failure case classification and the vehicle state information is already registered based on the update case information acquired from the failure information presentation device 20. New repair information (difference from registered information) is added to the existing failure case information, and if either of the failure case classification and the vehicle state information is different, it is registered as new failure case information. .

[effect]
As described above, in the failure information presentation system 1, a symbol string is generated from the vehicle state information measured when the vehicle is abnormal, and past failure case information accumulated in the failure case information server 30 is generated by this symbol string. By searching, similar failure cases are extracted, and repair information for the failure cases is presented. Thereby, useful information can be provided to a repair person.

  In the failure information presentation system 1, as a feature quantity representing vehicle state symbols (state identification information) and vehicle state information, data obtained by converting time-series data of measurement values obtained by the in-vehicle sensor group 11 into symbol strings is used. For this reason, not only can local events be represented in time by individual symbols, but the occurrence pattern and frequency of such events can be efficiently represented by symbol sequences. It is possible to accurately grasp a failure that causes

  Further, by using a symbol string as a feature amount, robustness against a time lag or a scale change of each event can be obtained. That is, as shown in FIG. 9, even if the failure is caused by the same cause, the duration of each event and the time scale when the event transitions vary depending on the vehicle type, engine size, individual differences in driving operations, and the like. However, it can have universality against such changes.

  Further, by using a symbol string as a feature amount, a memory capacity required for storing information and storage compared to the case of using a time series of feature amounts (multidimensional vectors) based on conventional FFD, It is possible to reduce the amount of processing required to search for the information. In addition, as a method of directly comparing time series in consideration of time lag and scale change, for example, DTW (Dynamic Time Warping) is known. Since the amount of calculation for comparison is small, it is very effective when there are many search targets.

[Other Embodiments]
As mentioned above, although embodiment of this invention was described, it cannot be overemphasized that this invention can take a various form, without being limited to the said embodiment.

  (1) In the above embodiment, sHDP-HMM is used when generating a symbol string from vehicle state information. However, the present invention is not limited to this. For example, sHDM-HMM and NPYLM (Nested Pitman-Yor Language Model) A double segment analyzer using the above may be used (see the parentheses in FIG. 3), or a symbol string may be generated according to a preset quantization table.

  (2) In the above embodiment, the Levenshtein distance is used to calculate the similarity between the case symbol string and the acquired symbol string. However, the present invention is not limited to this. Editing distance may be used.

  (3) In the above embodiment, a single symbol string is generated from the vehicle state information. However, a plurality of symbol strings (acquired symbol strings) are generated stochastically, and these acquired symbol strings, case symbol strings, and all It is also possible to calculate individual similarities for these combinations and use the maximum value or average value among them as the similarity used when extracting failure case candidates. Furthermore, the similarity with the case symbol string may be calculated using a lattice expressing the occurrence probability and transition probability of each symbol instead of a plurality of symbol strings.

  (4) In the above embodiment, when extracting failure case candidates, all failure cases having a similarity equal to or greater than a predetermined value are extracted so that the failure case classifications do not overlap. Alternatively, a predetermined number of failure cases may be extracted.

  (5) In the above embodiment, failure case information is generated for each case symbol string (vehicle state symbol) (see FIG. 2). However, as shown in FIG. 10, failure case information is generated for each failure case classification. However, a plurality of case symbol strings may be associated with one piece of failure case information. In this case, the failure case information server 30 that has received a registration request for new failure case information from the failure information presentation device 20 adds new information to the registered failure case information having the same failure case classification. do it. In this case, when calculating the similarity between the acquired symbol string and the case symbol string, the individual similarity is calculated for each combination of the acquired symbol string and the case symbol string, and the maximum value among them is calculated. Alternatively, an average value may be adopted.

  (6) In the above embodiment, failure cases are handled regardless of the vehicle type. However, only failure cases relating to the same vehicle type may be extracted in the diagnosis process, or failure cases of the same vehicle type may be preferentially extracted. Good. Further, the failure case information itself may be prepared for each vehicle type, or the failure case classification may be set in consideration of the difference in the vehicle type.

(7) In the above embodiment, the conversion from the vehicle state information to the symbol string is performed on the vehicle device group 10 side, but may be configured to be performed on the failure information presentation device 20 side.
(8) In the above embodiment, the failure information presentation device 20 and the failure case information server 30 communicate through the communication unit 26 for each failure diagnosis and information registration, but for updating and updating information in the failure case storage unit 25 Registration of the case information to the failure case information server 30 may be performed using another means. For example, the information in the failure case storage unit 25 may be updated using the update case information instead of updating the information in S210 and registering the update case information in the failure case information server 30 in S330. Then, the information in the failure case storage unit 25 and the information in the failure case information server 30 may be synchronized at a certain timing. In this case, the case information newly registered in the failure case storage unit 25 is transmitted from the failure information presentation device 20 to the failure case information server 30, and the case information not registered in the failure case storage unit 25 is the failure case information server 30. To the failure information presentation device 20. Further, this information synchronization may be performed directly by connecting the failure information presentation device 20 and the failure case information server 30 without using communication, or using an external storage medium (such as a magnetic disk, an optical disk, or a flash memory). .

  (9) Each component of the present invention is conceptual and is not limited to the above embodiment. For example, the functions of one component may be distributed to a plurality of components, or the functions of a plurality of components may be integrated into one component. Further, at least a part of the configuration of the above embodiment may be replaced with a known configuration having the same function. In addition, at least a part of the configuration of the above embodiment may be added to or replaced with the configuration of the other embodiment.

  DESCRIPTION OF SYMBOLS 1 ... Failure information presentation system 10 ... In-vehicle device group 11 ... In-vehicle sensor group 12 ... Input device 13 ... Information collection device 14 ... Status symbol storage device 15, 21 ... Interface (IF) part 20 ... Failure information presentation device 22 ... Input part DESCRIPTION OF SYMBOLS 23 ... Display part 24 ... Information processing part 25 ... Failure case memory | storage part 26 ... Communication part 30 ... Failure case information server

Claims (8)

Information acquisition means (24: S220) for acquiring vehicle state information in which time-series data of a plurality of measurement values representing a vehicle state is expressed by a predetermined feature amount;
Associating state identification information representing the time-series data of the measured values measured at the time of a vehicle failure with the feature amount and repair information indicating a cause of the failure corresponding to the state identification information and a repair method of the failure Case storage means (25, 30) for storing a plurality of case information;
Case extraction means for extracting one or more case information according to the similarity between the vehicle state information and the state identification information obtained for each case information stored in the case storage means (24: S230 to S260) When,
Information presenting means (24: S270) for presenting the repair information related to the case information extracted by the case extracting means;
With
As the feature amount, a symbol string obtained by symbolizing a pattern that repeatedly appears in the time series data of the measurement value is used ,
In the case information, one or more state identification information is associated with each of the repair information,
The case extraction means uses the maximum value or average value of the similarities obtained for each state identification information as the similarity of the case information . The failure information presentation system according to claim 1 , wherein the case extraction unit obtains an edit distance between symbol strings as the feature amount as the similarity.   Information acquisition means (24: S220) for acquiring vehicle state information in which time-series data of a plurality of measurement values representing a vehicle state is expressed by a predetermined feature amount;
  Associating state identification information representing the time-series data of the measured values measured at the time of a vehicle failure with the feature amount and repair information indicating a cause of the failure corresponding to the state identification information and a repair method of the failure Case storage means (25, 30) for storing a plurality of case information;
  Case extraction means for extracting one or more case information according to the similarity between the vehicle state information and the state identification information obtained for each case information stored in the case storage means (24: S230 to S260) When,
  Information presenting means (24: S270) for presenting the repair information related to the case information extracted by the case extracting means;
  With
  As the feature amount, a symbol string obtained by symbolizing a pattern that repeatedly appears in the time series data of the measurement value is used,
  The case extraction unit obtains an edit distance between symbol strings as the feature amount as the similarity.   The failure according to any one of claims 1 to 3, wherein a plurality of symbol sequences probabilistically generated from time-series data of measurement values of the in-vehicle sensor are used as the vehicle state information. Information presentation system.   The failure information presentation system according to any one of claims 1 to 3, wherein a lattice expressing the occurrence probability of the symbol is used as the vehicle state information.   Information acquisition means (24: S220) for acquiring vehicle state information in which time-series data of a plurality of measurement values representing a vehicle state is expressed by a predetermined feature amount;
  Associating state identification information representing the time-series data of the measured values measured at the time of a vehicle failure with the feature amount and repair information indicating a cause of the failure corresponding to the state identification information and a repair method of the failure Case storage means (25, 30) for storing a plurality of case information;
  Case extraction means for extracting one or more case information according to the similarity between the vehicle state information and the state identification information obtained for each case information stored in the case storage means (24: S230 to S260) When,
  Information presenting means (24: S270) for presenting the repair information related to the case information extracted by the case extracting means;
  With
  As the feature amount, a symbol string obtained by symbolizing a pattern that repeatedly appears in the time series data of the measurement value is used,
  A failure information presentation system using a plurality of symbol sequences stochastically generated from time-series data of measurement values of the in-vehicle sensor as the vehicle state information.   Information acquisition means (24: S220) for acquiring vehicle state information in which time-series data of a plurality of measurement values representing a vehicle state is expressed by a predetermined feature amount;
  Associating state identification information representing the time-series data of the measured values measured at the time of a vehicle failure with the feature amount and repair information indicating a cause of the failure corresponding to the state identification information and a repair method of the failure Case storage means (25, 30) for storing a plurality of case information;
  Case extraction means for extracting one or more case information according to the similarity between the vehicle state information and the state identification information obtained for each case information stored in the case storage means (24: S230 to S260) When,
  Information presenting means (24: S270) for presenting the repair information related to the case information extracted by the case extracting means;
  With
  As the feature amount, a symbol string obtained by symbolizing a pattern that repeatedly appears in the time series data of the measurement value is used,
  A failure information presentation system using a lattice expressing the occurrence probability of the symbol as the vehicle state information. Wherein for said vehicle state information acquired by the information acquisition means, implementation information input means for inputting performance information that indicates the content of repair was performed on the cause and the fault of the failure corresponding to said vehicle state information (24: S320 )When,
The storage content of the case storage unit is updated with the case information for update using the execution information input by the execution information input unit as the repair information and the vehicle state information corresponding to the execution information as the state identification information. Information updating means (24: S330, 30),
The failure information presentation system according to any one of claims 1 to 7 , further comprising: