Fuzzy microcontroller for complex nonlinear signal recognition

• US 5,673,365 A
• Filed: 03/23/1995
• Issued: 09/30/1997
• Est. Priority Date: 06/12/1991
• Status: Expired due to Fees

• Current Assignee: Microchip Technology Incorporated
• Original Assignee: Microchip Technology Incorporated
• Inventors: Basehore, Paul, Ziemacki, Mike
• Primary Examiner(s): Hafiz, Tariq R.
• Assistant Examiner(s): SMITH, JEFFREY S
• Application Number:
  US08/408,807
• Time in Patent Office: 922 Days
• Field of Search: 364/424.01, 364/424.05, 364/424.1, 364/426.01, 364/426.04, 180/422, 180/282, 180/268, 180/274, 280/728.2, 280/735, 280/737, 395/900, 395/905, 395/3, 395/51, 395/61
• US Class Current: 706/46
• CPC Class Codes: A47L 9/2805 Parameters or conditions being sensed
• CPC Class Codes: A47L 9/2894 Details related to signal transmission in suction cleaners
• CPC Class Codes: B60R 21/0132 responsive to vehicle motion parameters , e.g. to vehicle longitudinal or transversal deceleration or speed value
• CPC Class Codes: B60R 21/01332 by frequency or waveform analysis
• CPC Class Codes: G01P 15/0891 with indication of predetermined acceleration values G01P15/135 takes precedence
• CPC Class Codes: G06N 7/04 Physical realisation
• CPC Class Codes: Y10S 706/90 Fuzzy logic
• CPC Class Codes: Y10S 706/905 Vehicle or aerospace

Abstract

An arrangement (apparatus and method) using fuzzy logic for performing signal processing functions to determine the occurrence of an event, such as a vehicle crash. An acceleration sensor signal is fuzzified by comparison with dynamic membership functions having adjustable centers from feedback inputs. Each crisp input is fuzzified by determining the distance of the crisp input from the center of the membership function and linearly complementing the result with respect to the width of the membership function, thereby eliminating the necessity for determining the shape of the membership function. The dynamic membership functions are used to generate an average level signal, and an accummulated energy signal. A crash detection signal is generated after a comparison between the dynamically-adjusted average level and accummulated energy signals. The uses of the dynamic membership functions provide reliable detection and a low-cost implementation.

Claims

1. An apparatus for detecting a vehicle crash in response to an acceleration sensor signal, comprising:

first means for outputting an average value signal corresponding to an average value of said acceleration sensor signal, said first means adjusting said average value signal on the basis of a first relationship between said acceleration sensor signal and a first group of membership functions, said first group of membership functions each having a center corresponding to said average value signal;

second means for determining a change in said acceleration sensor signal on the basis of a second relationship between said acceleration sensor signal and a second group of membership functions;

third means for determining a deviation in said acceleration sensor signal from said average value signal on the basis of a third relationship between said acceleration sensor signal and a third group of membership functions each having a center corresponding to said average value signal;

fourth means for outputting an accumulated energy signal generated on the basis of a winning rule from a set of rules, each of said rules having rule terms corresponding to at least one of said membership functions of said first, second and third membership functions, said fourth means comprising means for determining said winning rule on the basis of said second and third relationships; and

fifth means for outputting a crash detection signal on the basis of a comparison between said accumulated energy signal and said average value signal.

2. An apparatus as recited in claim 1, wherein said second group of membership functions each have a center corresponding to a time-delayed value of said acceleration sensor signal.

3. An apparatus as recited in claim 2, wherein the membership functions of said first, second, and third membership functions are constant-slope membership functions.

4. An apparatus as recited in claim 3, further comprising means for calculating said corresponding relationship between said acceleration sensor input and said respective centers of said constant-slope membership functions, said calculating means comprising:

a subtractor for outputting a distance signal corresponding to a distance between said acceleration sensor input and said corresponding center of said respective membership function;

means for linearly complementing said distance signal; and

means for selectively truncating said distance signal on the basis of said corresponding width of said respective membership function.

5. An apparatus as recited in claim 2, wherein said second means comprises means for calculating a derivative of said acceleration sensor signal on the basis of said second relationship to said second group of membership functions.

6. An apparatus as recited in claim 1, further comprising:

an input processor for comparing said acceleration sensor signal with said membership functions in accordance with said set of rules to obtain a fuzzified acceleration sensor signal identifying said first, second and third relationships, said winning rule being determined by said fourth means on the basis of said fuzzified acceleration sensor signal; and

an output processor for adjusting said accumulated energy signal in accordance with said winning rule.

7. An apparatus as recited in claim 6, wherein said third group of membership functions comprise membership functions measuring small, medium, and large deviations of said acceleration sensor signal from said average level signal.

8. An apparatus as recited in claim 6, wherein said input processor comprises means for assigning an input signal as a center of a selected membership function, said assigning means assigning said average value signal as a center for said first and third group of membership functions and a time-delayed value of said acceleration sensor signal as a center for said second group of membership functions.

9. An apparatus as recited in claim 8, wherein said fifth means performs said comparison between said accumulated energy signal and said average value in accordance with a trigger membership function having assigned by said assigning means said average level signal as a center.

10. An apparatus as recited in claim 9, wherein said fourth means comprises a timer for decrementing said accumulated energy signal on the basis of no change between said average value signal and said acceleration sensor signal within a predetermined interval.

11. An apparatus as recited in claim 1, wherein said crash detection signal is output as a trigger for an airbag deployment device.

12. A fuzzy signal processor for measuring accumulated energy from an acceleration sensor signal, comprising:

a fuzzy low-pass filter for outputting a filtered acceleration sensor signal, said fuzzy low-pass filter comprising a first group of membership functions having said filtered acceleration sensor signal as an assigned center;

a fuzzy calculator for determining a change in said acceleration sensor signal with respect to time, said change being calculated on the basis of a second group of membership functions having a time-delayed value of said acceleration sensor signal as an assigned center;

a fuzzy comparator for calculating a deviation in said acceleration sensor signal from said filtered acceleration sensor signal on the basis of a third group of membership functions having said filtered acceleration sensor signal as a center;

means for calculating the accumulated energy in said acceleration sensor signal on the basis of said filtered acceleration sensor signal, said change in said acceleration sensor signal, and said deviation in said acceleration sensor signal over time.

13. A processor as recited in claim 12, wherein said calculating means comprises:

a rule memory for storing a plurality of fuzzy rules associated with said second and third group of membership functions;

an input processor for determining a winning rule from said plurality of fuzzy rules on the basis of a fuzzified representation of said acceleration sensor signal; and

an output processor for adjusting a value of said accumulated energy in accordance with said winning rule.

14. A processor as recited in claim 13, wherein each of said membership functions are constant-slope membership functions.

15. A processor as recited in claim 14, wherein said input processor calculates said fuzzified representation of said acceleration sensor signal on the basis of complemented distance values corresponding to each of said membership functions.

16. A processor as recited in claim 12, wherein each of said membership functions are constant-slope membership functions.

17. A processor as recited in claim 16, wherein said input processor calculates said fuzzified representation of said acceleration sensor signal on the basis of complemented distance values corresponding to each of said membership functions.

18. A processor as recited in claim 12, further comprising means for outputting an airbag trigger signal in response to a comparison between said accumulated energy and a left-exclusive membership function having said filtered acceleration sensor signal as an assigned center.

19. A method for detecting accumulated energy in response to an acceleration sensor signal, comprising the steps of:

adjusting an average value signal corresponding to an average value of said acceleration sensor signal on the basis of a comparison between said acceleration sensor signal and a first group of membership functions each having a center corresponding to said average value signal;

determining changes in said acceleration sensor signal based on a second group of membership functions having a time-delayed value of said acceleration sensor signal as an assigned center;

determining deviations in said acceleration sensor signal from said average value signal on the basis of a third group of membership functions each having a center corresponding to said average value signal;

determining a winning rule from a set of rules stored in a rule memory, each of said rules having rule terms corresponding to said second and third groups of membership functions, said winning rule being determined on the basis of a relational comparison between said set of rules and said changes and said deviations in said acceleration sensor signal;

dynamically adjusting an accumulated energy signal on the basis of said winning rule within a predetermined interval.

20. A method as recited in claim 19, further comprising the steps of:

decrementing said accumulated energy signal if said deviations and said time-delayed value of said acceleration sensor correspond to membership functions identifying zero changes; and

outputting an impact detection signal if said accumulated energy signal reaches a level corresponding to said average value signal.

21. A method as recited in claim 20, wherein said membership functions are constant-slope membership functions.

22. A method as recited in claim 20, wherein said acceleration sensor signal represents acceleration of a vehicle, said impact detect signal corresponding to a crash detection, the method further comprising the step of supplying said impact detection signal to a crash-defensive system.