Spark knock detection system for an internal combustion engine

• US 5,483,936 A
• Filed: 07/05/1994
• Issued: 01/16/1996
• Est. Priority Date: 07/05/1994
• Status: Expired due to Fees

• Current Assignee: Visteon Global Technologies Incorporated (Visteon Corporation)
• Original Assignee: Ford Motor Company
• Inventors: Grider, Duane M., Dunn, Brett A., Simmons, Stephen C., Kerstein, Scott M.
• Primary Examiner(s): Dolinar, Andrew M.
• Application Number:
  US08/270,982
• Time in Patent Office: 560 Days
• Field of Search: 123/425, 123/435, 73/35, 364/431.08
• US Class Current: 123/406.16
• CPC Class Codes: G01L 23/225 circuit arrangements therefor

Abstract

A spark knock detection system is disclosed which has a vibration detector connected in proximity to the engine for converting a vibration into an electrical output signal representative of the vibration. A sampler which is connected to the vibration detector samples the electrical output signal and converts the electrical output signal into a numerical data pattern corresponding to the electrical output signal. A feature extractor which is connected to the sampler converts the numerical data pattern into frequency domain signal portions which have a magnitude. A classifier which is connected to the feature extractor receives the frequency domain signal portions. The classifier has predetermined coefficients for classifying the frequency domain signal portions to indicate the presence of spark knock.

Claims

1. A system for detecting spark knock in an internal combustion engine comprising:

vibration detection means connected in proximity to the engine for converting a vibration into an electrical output signal representative of said vibration;

sampling means connected to said vibration detection means for sampling said electrical output signal and converting said electrical output signal into a digital data pattern corresponding to said electrical output signal;

feature extraction means connected to said sampling means for converting said digital data pattern into frequency domain signal values forming a frequency domain pattern, each value having a magnitude;

trained classification means connected to said feature extraction means for receiving said frequency domain pattern and having predetermined coefficients determined during training for classifying said frequency domain pattern to indicate the presence of spark knock.

2. A system for detecting spark knock as recited in claim 1 further comprising an output means for providing an electrical signal corresponding to said spark knock.

3. A system for detecting spark knock as recited in claim 2 wherein said output means further provides a signal corresponding to sensor failure.

4. A system for detecting spark knock as recited in claim 2 wherein said electrical signal is a digital signal indicating spark knock and no spark knock.

5. A system for detecting spark knock as recited in claim 2 further comprising an output means for providing an electrical signal corresponding to severity of spark knock.

6. A system for detecting spark knock as recited in claim 5 wherein said electrical signal is a serial signal.

7. A system for detecting spark knock as recited in claim 5 wherein said electrical signal is a pulse width modulated signal having a pulse width indicative of the severity of knock.

8. A system for detecting spark knock as recited in claim 5 wherein said electrical signal is an analog signal having an amplitude indicative of the severity of spark knock.

9. A system for detecting spark knock as recited in claim 5 wherein said electrical signal is frequency signal having an frequency indicative of the severity of spark knock.

10. A system for detecting spark knock as recited in claim 1 wherein said predetermined coefficients are multiplied by the magnitude of said frequency domain signals.

11. A system for detecting spark knock as recited in claim 1 wherein said sampling means sampling at a predetermined frequency, said predetermined frequency is continuously variable.

12. A system for detecting spark knock as recited in claim 1 wherein said sampling means sampling at a predetermined frequency, said predetermined frequency is constant.

13. A system for detecting spark knock as recited in claim 1 wherein the internal combustion engine comprises an engine block, said sensing means mounted within said engine block.

14. A system for detecting spark knock as recited in claim 1 wherein said engine comprises a predetermined number of cylinders, a respective sensing means being mounted in proximity to each respective cylinder.

15. A system for detecting spark knock in an internal combustion engine comprising:

engine control means for controlling the operation of the engine based on predetermined inputs;

sensing means connected in proximity of the engine for converting a vibratory waveform into an electrical output signal indicative of said vibratory waveform;

sampling means connected to said sensing means for sampling a portion of said electrical output signal and converting said electrical output signal into a digital data pattern;

trained classification means connected to said sampling means for receiving said digital data pattern from said sampling means and classifying said digital data pattern based on a plurality of trained coefficients; and

output means connected to said trained classification means for providing an electrical signal for indicating the presence of spark knock to said engine control means so that the engine control means can adjust said predetermined outputs to eliminate spark knock.

16. A system for detecting spark knock as recited in claim 15 wherein said electrical signal provided to output means indicates severity of spark knock.

17. A system for detecting spark knock as recited in claim 15 wherein said output means further provides a signal corresponding to sensor failure.

18. Method of determining spark knock in an internal combustion engine comprising the steps of:

converting a vibratory waveform into an electrical output signal indicative of said vibratory waveform;

sampling and converting said electrical output signal into a digital data pattern;

converting said digital data pattern into frequency domain signal pattern;

classifying said frequency domain signal pattern based on a weighted coefficients using said frequency domain signal pattern and coefficients determined by training; and

indicating the presence of spark knock based on said weighted calculation.

19. A method of determining spark knock as recited in claim 18 wherein said steps of sampling and converting are performed at fixed time intervals.

20. A method of determining spark knock as recited on claim 18 wherein said steps of sampling and converting are performed at variable time intervals.