Multi-level tool break detection using multi-mode sensing

• US 4,918,427 A
• Filed: 03/27/1989
• Issued: 04/17/1990
• Est. Priority Date: 03/27/1989
• Status: Expired due to Term

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Wildes, Douglas G., Thomas, Charles E., Hayashi, Steven R.
• Primary Examiner(s): Orsino, Joseph A.
• Assistant Examiner(s): Hofsass, Jeffery A.
• Application Number:
  US07/329,044
• Time in Patent Office: 386 Days
• Field of Search: 340/680, 340/683, 73/104, 73/660
• US Class Current: 340/680
• CPC Class Codes: B23Q 17/0957 Detection of tool breakage detecting failure of drills B23B49/001
• CPC Class Codes: B23Q 17/0971 by measuring mechanical vibrations of parts of the machine arrangements for measuring vibrations B23Q17/12
• CPC Class Codes: G08B 21/187 Machine fault alarms

Abstract

A system and method for monitoring machine tool operations provides a multi-level tool break alarm and uses information from multiple sensors of different types. Signal processing and pattern recognition logic techniques are applied to a cutting process high frequency vibration signal to detect major tool breaks requiring prompt stoppage of the cut. False alarm resistant detection of minor tool breaks, for which the response may not be an immediate stop of the cutting process, is obtained with information from a low frequency vibration sensor, an axis drive current sensor, or an axis velocity sensor. A minor tool break alarm is generated when signal transients in both a high frequency and low frequency channel signal are in close time coincidence, and a major alarm when the high frequency channel transient is followed by a persisting mean vibration level shift.

Claims

1. A machine tool monitoring system for detecting cutting tool breakage that does and does not cause an immediate major change in cutting conditions, comprising:

means for sensing vibrations generated by interaction of a cutting tool and workpiece and deriving a high frequency vibration signal having a frequency in a range between 30 KHz and 100 KHz;

means for sensing at least one parameter related to forces acting on the tool during a cutting process and producing a low frequency signal having a frequency less than about 1 KHz;

means for preprocessing, sampling and digitizing each of said signals and respectively computing first and second tracking mean signals from a given number of previous samples;

pattern recognition means for detecting signal transients that may be caused by a tool break; and

alarm means for determining that transients in said first and second mean signals are in close time coincidence and occur in a designated time window, and generating a minor tool break alarm.

2. The system of claim 1 wherein said pattern recognition means further comprises logic for determining that the signal transient in said first mean signal derived from said high frequency vibration signal is followed by a substantial and persisting shift in mean level due to a major change in cutting conditions, and said alarm means comprises means for generating a major tool break alarm requiring an immediate stop of the cutting process.

3. The system of claim 2 wherein said time window to detect close coincidence between transients in said first and second mean signals is a small fraction of one workpiece revolution, and said persisting level shift of said first mean signal to trigger the second type alarm lasts longer than one workpiece revolution.

4. The system of claim 2 wherein said means for sensing vibrations is a broadband vibration sensor and said means for sensing at least one parameter related to forces acting on the tool is a low frequency vibration sensor.

5. The system of claim 2 wherein said means for sensing vibrations is a broadband vibration sensor and said means for sensing at least one parameter related to forces acting on the tool comprises two orthogonally mounted low frequency vibration sensors.

6. The system of claim 2 wherein said means for sensing vibrations is a broadband vibration sensor and said means for sensing at least one parameter related to forces acting on the tool is a machine tool axis drive current sensor.

7. The system of claim 2 wherein said means for sensing vibrations is a broadband vibration sensor and said means for sensing at least one parameter related to forces acting on the tool is a machine tool axis velocity sensor.

8. A machine tool monitoring system for detecting cutting tool breakage that does and does not cause an immediate major change in cutting conditions, comprising:

a broadband vibration sensor for sensing vibrations generated by interaction of a cutting tool and workpiece and producing an electrical signal having a frequency less than about 100 KHz, and means for extracting therefrom high and low frequency vibration signals and preprocessing the same;

means for sampling and digitizing each of said vibration signals and respectively computing first and second tracking mean vibration signals from a given number of previous samples;

pattern recognition means for detecting signal transients that may be caused by a tool break; and

alarm means for determining that the transients in said first and second mean vibration signals are in close time coincidence and occur in a designated time window, and generating a tool break alarm.

9. The system of claim 8 wherein said pattern recognition means further comprises logic for determining that the signal transient in said first mean signal derived from said high frequency vibration signal is followed by a substantial and persisting shift in mean level due to a major change in cutting conditions, and said alarm means comprises means for generating a major tool break alarm requiring an immediate stop of the cutting process.

10. A method of detecting different classes of tool breaks comprising:

sensing vibrations generated at a cutting tool-workpiece interface on a machine tool during a cutting operation and generating an electrical signal, from which is derived a high frequency channel signal indicative of the amplitude of vibrations in a chosen high frequency band;

sensing at least one parameter related to forces acting on the cutting tool and producing a low frequency channel signal;

sampling and digitizing each of said signals and respectively computing high and low frequency channel tracking mean signals from a given number of previous samples;

detecting a transient in said high frequency channel mean signal which is a suspicion of a tool break, and another transient in said low frequency channel mean signal which may be a confirmation of the tool break;

determining that said first-recited and second-recited transients are in close time coincidence, and generating a minor tool break alarm.

11. The method of claim 10 further comprising digitally processing said high frequency channel mean signal following detection of the suspicious transient to check for an abrupt and substantial shift in mean signal level that persists for a preselected period due to a major change in cutting conditions, and generating a major tool break alarm.

12. The method of claim 11 wherein said sensing at least one parameter related to forces acting on the cutting tool comprises sensing low frequency vibrations generated at the tool-workpiece interface and generating a second electrical signal from which said low frequency channel signal is derived.

13. The method of claim 11 wherein said sensing at least one parameter related to forces acting on the cutting tool comprises sensing a machine tool axis drive current from which said low frequency channel signal is derived.

14. The method of claim 11 wherein said sensing at least one parameter related to forces acting on the cutting tool comprises sensing a machine tool axis velocity from which said low frequency channel signal is derived.

15. The method of claim 11 wherein said sensing at least one parameter related to forces acting on the cutting tool comprises using a broadband accelerometer to sense vibrations and filtering the electrical signal to extract a low frequency signal from which said low frequency channel signal is derived.

16. The method of claim 11 wherein said chosen high frequency band is above 30 KHz but below 100 KHz, and said low frequency channel signal is derived from a dc to 1 KHz signal.