Acoustic tool break detection system and method

• US 4,642,617 A
• Filed: 12/21/1984
• Issued: 02/10/1987
• Est. Priority Date: 12/21/1984
• Status: Expired due to Term

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Thomas, Charles E., Lee, Minyoung, Hayashi, Steven R., Bedard, James F.
• Primary Examiner(s): Rowland, James L.
• Assistant Examiner(s): Hofsass, Jeffery A.
• Application Number:
  US06/685,005
• Time in Patent Office: 781 Days
• Field of Search: 340/679, 340/680, 340/683, 73/104, 73/660, 73/593, 73/658, 364/475, 364/507, 364/508, 364/550, 364/551
• US Class Current: 340/680
• CPC Class Codes: B23Q 17/0971 by measuring mechanical vibrations of parts of the machine arrangements for measuring vibrations B23Q17/12
• CPC Class Codes: B23Q 17/098 by measuring noise
• CPC Class Codes: G01H 1/00 Measuring characteristics of vibrations in solids by using direct conduction to the detector G01H9/00, G01H11/00 take precedence
• CPC Class Codes: G05B 19/4065 Monitoring tool breakage, life or condition

Abstract

Substantial cutting condition changes that occur gradually, as opposed to the more usual sudden large change, are detected by setting upper and lower cutting noise mean level thresholds. When the mean cutting noise exceeds the upper threshold or stays below the lower threshold for a preset number of signal samples, a tool break alarm is generated. Techniques are given to reduce false alarms at the start and end of the cut and on runout on initial rough surface cuts. The system comprises an accelerometer or other sensor whose signal is preprocessed to attenuate lower frequency machinery noise and detect the signal energy in a band below 100 KHz, then sampled, and the digitized signal samples analyzed by pattern recognition logic.

Claims

1. An acoustic monitoring system to detect cutting tool break events while machining a workpiece on a machine tool comprising:

a broadband vibration sensor which generates an electrical signal representing vibrations at the tool-workpiece interface;

an analog preprocessor including means to discriminate against lower frequency machinery noise and to detect the signal energy in an acoustic frequency band below 100 KHz;

means for sampling the output signal of said analog preprocessor and converting each sample to a digital value;

digital pattern recognition circuitry to detect tool break events that produce gradually occurring, substantial changes in cutting conditions, comprising means for calculating the running mean signal level of N samples, means for determining when the cutting noise and mean signal level drops below a lower break detection threshold which is just above the continuous traverse noise level of said machine tool, and means for generating a tool break alarm after counting a preset number of low level cutting noise samples below the lower threshold.

2. The system of claim 1 and means for setting a flag and activating the lower break detection threshold only after said mean signal level rises above a much higher flat-set threshold and remains there for a preset number of consecutive signal samples.

3. The system of claim 2 and means for determining when the mean signal level exceeds an upper break detection threshold just below the system saturation level and generating a tool break alarm after counting a preset number of very high cutting noise samples above the upper threshold.

4. The system of claim 2 wherein said vibration sensor is an accelerometer which has a resonant frequency above 30 KHz.

5. The system of claim 2 wherein, to prevent false alarms on rough surface cutting, the number of low level cutting noise samples to alarm is set at least equal to the number of samples in a machine tool spindle revolution.

6. An acoustic monitoring system to detect cutting tool break events while machining a workpiece comprising:

a broadband accelerometer which is most sensitive to frequencies around a resonant frequency and is positioned on a machine tool to sense vibrations at the tool-workpiece interface and convert these and other vibrations to an electrical signal;

analog signal processing means including means for high-pass filtering said vibration signal to discriminate against lower frequency machinery noise, and means to rectify and low-pass filter said signal to detect the signal energy in an acoustic frequency band between 30 KHz and 100 KHz;

means for sampling the unipolar output signal of said analog processing means and converting each sample to digital form;

a digital processor to detect major cutting condition changes occurring gradually over a period of time that can damage the workpiece or machine tool;

said digital processor comprising means for calculating the running mean signal level of a selected number of samples, means for activating the lower threshold test of a threshold crossing detector after the mean signal level rises above a much higher flag-set threshold and remains there for a preset number of consecutive signal samples, said detector determining when the mean signal level crosses lower or upper break detection thresholds where the former is just above the continuous traverse noise level of said machine tool and the latter is just below the system saturation level, and means for testing that the mean signal level persistently remains outside of said thresholds, and generating a tool break alarm.

7. The system of claim 6 wherein said last-mentioned means returns control to said threshold crossing detector after failing the persistence test.

8. The method of acoustically detecting tool breakage capable of damaging the workpiece or machine tool comprising:

sensing vibrations at the tool-workpiece interface and converting these and other machine tool vibrations to an electrical signal;

preprocessing said vibration signal to discriminate against lower frequency machinery noise and rectify and low-pass filter said signal to detect the signal energy in an acoustic frequency band below 100 KHz;

sampling the processed vibration signal and converting each sample to digital form; and

analyzing said signal samples to detect gradually occurring, but substantial, cutting noise changes due to tool break events capable of damage, comprising setting upper and lower cutting noise thresholds, determining when the running mean signal level of N samples drops below said lower threshold, which is just above the continuous traverse noise level of said machine tool, and generating a tool break alarm after a preset number of consecutive signal samples and the cutting noise are outside of said thresholds.

9. The method of claim 8 and activating said lower threshold after a much higher flag-set threshold is exceeded by a preset number of consecutive signal samples, to prevent false alarms on low level signals such as those near the start of the cut.

10. The method of claim 9 and preventing false alarms during rough surface cutting by setting the number of consecutive signal samples below the lower threshold, needed to generate an alarm, larger than the number of samples in a machine tool spindle revolution.