Signal processing method for an engine knock signal

• US 7,676,323 B2
• Filed: 05/08/2007
• Issued: 03/09/2010
• Est. Priority Date: 05/08/2007
• Status: Expired due to Fees

• Current Assignee: PHINIA Holdings Jersey Ltd. (BorgWarner Incorporated)
• Original Assignee: Delphi Technologies, Inc. (BorgWarner Incorporated)
• Inventors: Kearney, Matthew J.
• Primary Examiner(s): WOLFE JR, WILLIS RAY
• Application Number:
  US11/801,026
• Publication Number:
  US 20080276688A1
• Time in Patent Office: 1,036 Days
• Field of Search: 123/406.29, 12340634-40639, 123/435, 123/436, 701/101, 701/102, 701/103, 701/111, 701/114, 701/115, 73 3501- 3504, 731/140.7, 731/141.6
• US Class Current: 701/111
• CPC Class Codes: G01L 23/225 circuit arrangements therefor

Abstract

The duration of a knock detection window over which a filtered knock signal is single-point DFT-processed for engine knock detection is user adjustable. The knock detection window is defined by at least two overlapping identical duration sub-windows, and the extent of the sub-window overlap is changed to adjust the duration of the knock detection window. The DFT processing is performed over each sub-window and the processing results are combined for knock detection.

Claims

1. A method of processing a knock signal produced by an engine knock sensor to determine a magnitude of the knock signal at a specified knock frequency for knock detection, comprising the steps of:

anti-alias filtering the knock signal to pass a range of knock signal frequencies including the specified knock frequency;

sampling and digitizing the filtered knock signal to form a series digital samples for DFT-processing;

characterizing a specified knock detection interval as a combination of at least first and second processing windows of fixed and identical duration with a variable amount of overlap;

DFT-processing said digital samples over said first processing window to form a first knock signal magnitude, and DFT-processing said digital samples over said second processing window to form a second knock signal magnitude; and

combining said first and second knock signal magnitudes to form the knock signal magnitude for knock detection.

2. The method of claim 1, including the step of:

determining an overlap of said first and second processing windows so that their combined duration corresponds to the specified knock detection interval.

3. The method of claim 1, including the steps of:

storing an array of pre-calculated terms for use in DFT-processing said digital samples; and

retrieving the stored pre-calculated terms for DFT-processing the digital samples over both first and second processing windows.

4. The method of claim 1, including the step of:

averaging said first and second knock signal magnitudes to form the knock signal magnitude for knock detection.