Method of fault diagnosis in an apparatus having sensors

• US 5,414,645 A
• Filed: 10/23/1992
• Issued: 05/09/1995
• Est. Priority Date: 10/25/1991
• Status: Expired due to Term

• Current Assignee: Mazda Motor Corporation
• Original Assignee: Mazda Motor Corporation
• Inventors: Hirano, Seiyo
• Primary Examiner(s): ZANELLI, MICHAEL J
• Application Number:
  US07/965,772
• Time in Patent Office: 928 Days
• Field of Search: 364/424.03, 364/424.04, 364/551.01, 395/900, 395/905, 395/913, 73/116, 73/117.2, 73/117.3, 73/118.1, 73/119 R, 371/15.1
• US Class Current: 702/185
• CPC Class Codes: F02D 41/1404 Fuzzy logic control
• CPC Class Codes: F02D 41/22 Safety or indicating devices for abnormal conditions in air/fuel ratio feedback systems F02D41/1495, in electric control linkage F02D11/107, in purge control systems F02M25/0809
• CPC Class Codes: G06F 11/2257 using expert systems
• CPC Class Codes: Y02T 10/40 Engine management systems
• CPC Class Codes: Y10S 706/905 Vehicle or aerospace

Abstract

The relationships between part nodes, including all sensor nodes in an apparatus system are expressed by connecting relations and weighing factors indicating the correlation. Output signals of all the sensor nodes in the system are converted to a state quantity by a membership function using fuzzy reasoning. With a sensor as a starting point, the nodes which are sequentially connected are traced and state quantities of each node are determined. The state quantities of each node are unified in a subsystem comprising this apparatus system and the subsystem in which the value of the state quantity is large is inferred to be faulty.

Claims

1. A fault diagnosis method for diagnosing an apparatus comprising a plurality of sensor nodes and a plurality of part nodes which are directly or indirectly associated with said plurality of sensor nodes, comprising the steps of:

a) before fault diagnosis,

a1) setting linkages based on deliveries of signals among said plurality of sensor nodes and said plurality of part nodes, and establishing weighing factors indicating correlations between said plurality of sensor nodes and said plurality of part nodes forming the linkages;

a2) classifying said plurality of sensor nodes and said plurality of part nodes into a plurality of subsystems;

b) during fault diagnosis;

b1) collecting output signals of said plurality of sensor nodes;

b2) converting the collected output signals of the plurality of sensor nodes to state quantities indicating degree of abnormal operation of the plurality of sensor nodes;

b3) calculating state quantities indicating an abnormal operation of respective part nodes having a linkage with respective sensor nodes;

b4) consolidating the state quantities for the plurality of sensor nodes and the plurality of part nodes obtained at steps b2 and b3 for each of said plurality of subsystems; and

b5) performing fault diagnosis on said plurality of sensor nodes and the plurality of part nodes on the basis of the consolidated results of step b4, wherein a state quantity value is used as a basis for fault diagnosis.

2. The method according to claim 1, wherein said state quantities are represented in a form of a membership function of fuzzy reasoning.

3. The method according to claim 1, wherein said plurality of sensor nodes and said plurality of part nodes have connecting data comprising:

identifying data indicating sensor nodes and part nodes which are connected to each sensor node and part node; and

a weighing factor indicating a degree of correlation between the connected nodes.

4. The method according to claim 1, wherein the calculation of the state quantities in step b3 comprises the steps of:

b3-1;

starting from a sensor node which is included in a subsystem and determining the state quantity of said sensor node;

b3-2;

multiplying the state quantity of said sensor node by a weighing factor which is set for the linkage between said sensor node and the part node connected thereto, and storing the result as the state quantity of said connected part node; and

b3-3;

repeating b3-2 for each part node which is connected to said sensor node.

5. The method according to claim 4, wherein step b3 further comprises the steps of:

b3-4;

checking whether or not the calculation of the state quantity on the linkage to which a sensor node is connected is going to be performed a second time; and

b3-5;

ending the calculation operation for the subsystem, if it is discriminated that the calculation operation is going to be performed the second time.

6. The method according to claim 4, wherein, when the state quantity for a sensor node, determined in step b3-1, is detected as being smaller than a predetermined value, step b3-2 is not performed.

7. The method according to claim 4, wherein, said step b3-1 comprises the steps of:

performing said calculation operation starting from a sensor node indicating a larger value of state quantity, and

ending step b3 when the sensor node is found indicating a state quantity which is less than a predetermined value.

8. The method according to claim 1, wherein the calculation result for each part per every sensor is summed up at step b4.

9. The method according to claim 1, wherein a sensor node or part node having a state quantity over a predetermined threshold is inferred as being a faulty sensor node or part node at step b5.

10. The method according to claim 9, wherein, the faulty sensor node or part node which is inferred at step b5 is displayed as faulty.

11. The method according to claim 10, wherein sensor nodes and part nodes are displayed by predetermined symbols, linkages between the nodes are displayed by lines, and the faulty part inferred at step b5 and its associated linkages to sensor nodes with high state quantities are displayed in forms which are different from displays for normally functioning sensor nodes and part nodes.

12. The method according to claim 1, wherein said sensor nodes are connected to a communication path and the outputs of the sensor nodes are collected via the communication path.

13. A fault diagnosis method for diagnosing an apparatus system which comprises a plurality of sensor nodes and a plurality of part nodes which are directly or indirectly associated with the plurality of sensor nodes, comprising the steps of:

a) before fault diagnosis,

a1) setting linkages based on deliveries of signals among said plurality of sensor nodes and said plurality of part nodes, establishing weighing factors indicating correlations between any of the plurality of sensor nodes and plurality of part nodes forming the linkages, and assigning fault information on the basis of experiences to each sensor and part node;

a2) classifying said plurality of sensor nodes and said plurality of part nodes into a plurality of subsystems;

b) at fault diagnosis time;

b1) collecting output signals of said plurality of sensor nodes;

b2) converting the collected output signals of the sensor nodes to state quantities indicating a degree of abnormality of the sensor nodes;

b3) inferring a faulty state of a sensor node by calculating a state quantity indicating a state of abnormality for every sensor node along with the linkages, while calculating a state quantity for part nodes on the basis of said sensor state quantity indicating the state of abnormality;

b4) processing the two results, obtained in step b3 for their respective subsystems; and

b5) performing fault diagnosis on said plurality of sensor nodes and part nodes on the basis of the processed results of step b4.

14. The method according to claim 13, wherein step b4 further comprises determining an intersection between the two results to identify the faulty node.

15. The method according to claim 13, wherein, said step b4 further comprises the steps of:

narrowing down probable faulty nodes on the basis of one inference result of the two inference results of step b3, and

applying the narrowed down result to the other inference result of b3.

16. The method according to claim 13, wherein the processing of the two results in step b4 comprises the step of summing the state quantity of a sensor node and the state quantity of an associated part node.