Fuzzy inference apparatus and methods, systems and apparatuses using such inference apparatus

• US 8,892,496 B2
• Filed: 08/19/2010
• Issued: 11/18/2014
• Est. Priority Date: 08/19/2009
• Status: Expired due to Fees

• Current Assignee: BAE Systems Plc, University of Leicester
• Original Assignee: BAE Systems Plc, University of Leicester
• Inventors: Yao, Liqun, Gu, Da-Wei, Postlethwaite, Ian
• Primary Examiner(s): Starks, Wilbert L
• Application Number:
  US13/391,206
• Publication Number:
  US 20130080376A1
• Time in Patent Office: 1,552 Days
• Field of Search: None
• US Class Current: 706/52
• CPC Class Codes: G06N 3/043 based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
• CPC Class Codes: G06N 7/02 using fuzzy logic computing arrangements based on biological models G06N3/00; computing arrangements using knowledge-based models G06N5/00
• CPC Class Codes: G06N 7/023 Learning or tuning the parameters of a fuzzy system

Abstract

A health monitoring system for complex networked apparatus includes a number of neuro-fuzzy inference apparatuses feeding inference results into a data fusion hierarchy. At each level in the hierarchy, fuzzy inference is applied to generate a desired output signal by processing selected input signals in accordance with a knowledge base defining fuzzy membership functions and fuzzy inference rules defined in advance. The knowledge base includes alternative definitions of membership functions and/or inference rules. The apparatus selects which definition to use according to environmental or other conditions, and predetermined selection criteria.

Claims

1. A fuzzy inference apparatus comprising:

a plurality of inputs for receiving time-varying input signals;

at least one output for outputting a time-varying output signal dependent on the input signals; and

a processing apparatus configured to;

perform, on the time-varying input signals, a fuzzification process, thereby providing fuzzified data, the fuzzification process being performed in accordance with fuzzy membership functions contained within a knowledge base; and

perform, on the fuzzified data, fuzzy inference to generate the output signal, the fuzzy inference being performed in accordance with fuzzy inference rules contained within the knowledge base;

wherein

the knowledge base comprises fuzzy membership functions and fuzzy inference rules, the fuzzy membership functions and fuzzy inference rules having been defined in advance;

the knowledge base comprises at least two alternative definitions of the membership functions and inference rules, and the processing apparatus is operable to apply different ones of the definitions at different times; and

the processing apparatus is arranged to optimize its fuzzy rule set automatically during operation, between fuzzification of the input signals and completion of the fuzzy inference.

2. An apparatus as claimed in claim 1, in combination with a monitoring system for a monitored apparatus, the output signal representing a measure of health of the monitored apparatus.

3. An apparatus as claimed in claim 2, wherein the inference apparatus and the monitored apparatus are housed together as one unit, together with a communications interface for delivering the output signal via the communications interface to a remote apparatus.

4. An apparatus as claimed claim 2, wherein the monitored apparatus comprises:

a plurality of subsystems, wherein the inference apparatus is part of a hierarchy of inference apparatuses monitoring the subsystems, the hierarchy including a higher-level inference apparatus arranged to receive, as inputs, outputs of lower-level inference apparatuses monitoring respective subsystems, and wherein at least one of the lower-level and higher-level inference apparatuses has the knowledge base comprising alternative definitions.

5. An apparatus as claimed in claim 1, comprising:

a selector for selecting which definition to use automatically in response to selection criteria specifying ranges of operating conditions in which each definition is used.

6. An apparatus as claimed in claim 5, wherein the selection criteria define ranges of operating conditions by reference at least partially to externally received signals.

7. An apparatus as claimed in claim 6, wherein the selection criteria define ranges of operating conditions at least partially by references to certain of the input signals.

8. An apparatus as claimed claim 1, wherein the alternative definitions comprise:

a first definition for normal operating conditions and one or more further definitions for anomalous operating conditions.

9. An apparatus as claimed in claim 1, wherein the apparatus is responsive to an externally supplied selection signal, commanding application of a particular definition.

10. An apparatus as claimed in claim 1, wherein the apparatus is responsive to a selection value loaded into memory prior to operation.

11. An apparatus as claimed in claim 1, wherein the processing apparatus is arranged to optimize a fuzzy rule set automatically during operation, between fuzzification of the input signals and completion of a fuzzy inference.

12. A distributed data processing system for monitoring a condition of a system of discrete apparatuses over time, the monitoring system comprising:

a hierarchy in which a higher level monitoring apparatus is supplied with input signals from a plurality of lower level monitoring apparatus, each lower level monitoring apparatus comprising one or more fuzzy inference apparatuses for processing a plurality of input signals relating to its respective monitored apparatus and for generating an output signal representing the condition of its monitored apparatus; and

the higher level monitoring apparatus comprising at least one fuzzy inference apparatus for receiving as inputs the output signals of plural lower level monitoring apparatuses, and for generating a higher level output signal representing the condition of the system of discrete apparatuses as a whole, wherein the at least one fuzzy inference apparatus is an apparatus as claimed in claim 1.

13. A distributed processing system as claimed in claim 12, wherein at least one of the monitored apparatuses is an unmanned autonomous vehicle.

14. A computer program product containing instructions for causing a programmable processing apparatus to implement functions of the processing apparatus of a fuzzy inference apparatus as claimed in claim 1.

15. The computer program product of claim 14, comprising:

the knowledge base including the alternative definitions.

16. The computer program product of claim 14, comprising:

a definition of selection criteria to be applied automatically for selection between the alternative definitions.

17. A method of generating a time-varying output signal in response to a plurality of time-varying input signals, the method comprising:

(a) storing in a processing apparatus a knowledge base comprising definitions of membership functions for the input and output signals and fuzzy inference rules for use in fuzzy inference processing;

(b) receiving the input signals; and

(c) operating the processing apparatus to;

perform, on the time-varying input signals, a fuzzification process, thereby providing fuzzified data, the fuzzification process being performed in accordance with fuzzy membership functions contained within the knowledge base; and

generate the time-varying output signal by performing fuzzy inference processing on the fuzzified data, the fuzzy inference being performed in accordance with the fuzzy inference rules stored in the knowledge base;

wherein step (a) includes storing at least two alternative definitions of the membership functions and inference rules, and step (c) includes making a selection between the alternative definitions so as to perform the inference processing using different definitions at different times; and

the processing apparatus is arranged to optimize its fuzzy rule set automatically during operation, between fuzzification of the input signals and completion of the fuzzy inference.

18. A method as claimed in claim 17, wherein the step (a) comprises:

generating the knowledge base by a training process using a training data, representing potential input signals and desired output signals, the different definitions being a result of training using different subsets of the training data.

19. A method as claimed in claim 18, wherein the different subsets of training data include some data common to both subsets.

20. A method as claimed in claim 18, wherein a first subset of training data is selected to represent normal operating conditions while one or more further subsets representing anomalous conditions expected to arise relatively infrequently during operation of the method.

21. A method as claimed in claim 17, comprising:

producing the output signal based on input signals used to monitor the health of another apparatus with which the processing apparatus is associated.

22. A method as claimed in claim 17, wherein the step (a) comprises:

storing selection criteria specifying ranges of operating conditions in which each definition is used, the selection in step (c) then being made automatically in response to changes detected in operating conditions.

23. A method as claimed in claim 22, wherein the selection criteria define ranges of operating conditions by reference to externally received signals, to certain of the input signals already received as inputs or to a combination of these.

24. A method as claimed in claim 17, wherein the selection is made in response to an externally supplied selection signal, commanding application of a particular definition.

25. A method as claimed in claim 17, wherein the selection is made by a selection value loaded into memory prior to operation.

26. An automated method of distributed data processing for monitoring a condition of a system of discrete apparatuses over time, the method comprising:

establishing and operating a hierarchy of monitoring apparatuses in which a higher level monitoring apparatus is supplied with input signals from a plurality of lower level monitoring apparatus;

in each lower level monitoring apparatus operating one or more fuzzy inference apparatuses to process a plurality of input signals relating to its respective monitored apparatus and generate an output signal representing the condition of its monitored apparatus; and

in the higher level monitoring apparatus, operating at least one fuzzy inference processor to receive as inputs the output signals of plural lower level monitoring apparatuses, and to generate a higher level output signal representing the condition of the system of discrete apparatuses as a whole,

wherein at least one of the fuzzy inference apparatuses is an apparatus performing a method as claimed in claim 17.

27. A method as claimed in claim 26, wherein at least one of the monitored apparatuses is an unmanned autonomous vehicle.

28. A computer program product containing instructions for causing a programmable processing apparatus to implement a method as claimed in claim 17.