Vehicle failure diagnostic device

• US 8,554,403 B2
• Filed: 12/02/2009
• Issued: 10/08/2013
• Est. Priority Date: 12/10/2008
• Status: Active Grant

• Current Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Original Assignee: Honda Motor Co., Ltd. (Honda Motor Company)
• Inventors: Tsuchikiri, Hideyuki, Sato, Yasushi, Aihara, Jiro
• Primary Examiner(s): Beaulieu, Yonel
• Application Number:
  US13/133,484
• Publication Number:
  US 20110313616A1
• Time in Patent Office: 1,406 Days
• Field of Search: 701 297- 299, 701 336- 368, 701/33.2, 701/33.4, 701/31.4, 701/29.1, 701/29.2, 340/438, 340/500, 702182-185
• US Class Current: 701/29.1
• CPC Class Codes: B60W 2050/005 Sampling
• CPC Class Codes: B60W 2050/0052 Filtering, filters
• CPC Class Codes: B60W 2556/10 Historical data
• CPC Class Codes: B60W 50/0205 Diagnosing or detecting failures; Failure detection models

Abstract

A vehicle failure is diagnosed based on data recorded in an electronic control unit (ECU) of a vehicle. Driving data pertaining to multiple parameters which are stored in the electronic control unit of the vehicle for when a vehicle failure occurs are compared with reference values representing data during normal driving in order to diagnose a failure. The device utilizes an approximate search between inspection data, that is, driving data which are stored in the electronic control unit of the vehicle to be diagnosed for failure, and reference value groups categorized by driving environment, wherein a normal value range is set for each of the aforementioned parameters categorized by driving environment, and selects the reference value group categorized by driving environment which approximates the inspection data. Furthermore, the device compares the inspection data with the normal value ranges of each parameter of the selected reference value group categorized by driving environment so as to identify parameters which deviate significantly from the normal value ranges and find the cause of the failure based on the parameters identified.

Claims

1. A method for making a failure diagnosis by comparing driving data relating to a plurality of parameters stored in an electronic control unit of a vehicle when a failure occurs in the vehicle with reference values as reference driving data during normal time, the method comprising:

selecting, via a failure diagnostic unit, reference value groups categorized by driving environment, which are entirely approximated to inspection data, by approximate search between the inspection data which is the driving data stored in the electronic control unit of the vehicle to be subjected to failure diagnosis and the reference value groups categorized by the driving environment whose normal value ranges are set for the plurality of parameters categorized by the driving environment; and

identifying one of the plurality of parameters having a substantially large deviation from the normal value range by comparing the inspection data with the normal value ranges for the plurality of parameters in the selected reference value groups categorized by the driving environment,

the failure diagnostic unit searching a failure cause based on the identified one of the plurality of parameters.

2. The method according to claim 1, wherein the driving data for the plurality of parameters obtained from many vehicles are clustered by a clustering method to set the normal value ranges of the driving data for the plurality of parameters into respective clusters so that the reference value groups categorized by the driving environment are formed.

3. The method according to claim 2 wherein the inspection data is numerical value vectors having predetermined time intervals of the plurality of parameters and selecting approximate reference value groups categorized by the driving environment calculates a distance between the numerical value vector and a core vector as a center of gravity of a cluster of the reference value group to select the reference value group having a shortest distance per the numerical value vector.

4. The method according to claim 3, wherein the clusters include primary clusters for which partial parameters among the plurality of parameters are clustered and secondary clusters for which the numerical value vectors belonging to the primary clusters are clustered for more parameters than the partial parameters, and the selecting the approximate reference value groups categorized by the driving environment selects the primary clusters for the numerical value vectors of the inspection data and then selects a cluster having the shortest distance from the secondary clusters belonging to the primary clusters.

5. The method according to claim 2, wherein a deviation from the normal value range is judged by comparing the normal value ranges of the clusters selected corresponding to the numerical value vectors of the inspection data with the numerical value vectors and using at least a ratio of the number of numerical value vectors out of the normal value ranges relative to the number of numerical value vectors of the inspection data.

6. The method according to claim 2, wherein a deviation from the normal value range is judged by comparing the normal value ranges of the clusters selected corresponding to the numerical value vectors of the inspection data with the numerical value vectors and using, for a parameter, at least a difference between the numerical values of the numerical value vectors of the inspection data out of the normal value ranges and the normal value ranges.

7. The method according to claim 2, wherein a deviation from the normal value range is judged by comparing the normal value ranges of the clusters selected corresponding to the numerical value vectors of the inspection data with the numerical value vectors and using at least a ratio of the number of numerical value vectors continuously out of the normal value ranges relative to the total number of numerical value vectors.