Real-time adaptive control of rotationally-induced vibration

• US 6,145,381 A
• Filed: 11/12/1998
• Issued: 11/14/2000
• Est. Priority Date: 11/12/1998
• Status: Expired due to Term

• Current Assignee: Alliedsignal Inc.
• Original Assignee: Alliedsignal Inc.
• Inventors: Grogan, Robert L., Mathisen, Eric C.
• Primary Examiner(s): Williams, Hezron
• Assistant Examiner(s): Kwok, Helen C.
• Application Number:
  US09/190,677
• Time in Patent Office: 733 Days
• Field of Search: 73/659, 73/660, 73/593, 702/190, 702/191, 702/194, 702/189, 381/71.1, 381/71.2, 381/71.8, 381/71.13, 381/71.14, 381/94.1, 381/94.9
• US Class Current: 73/660
• CPC Class Codes: F16F 15/002 characterised by the control method or circuitry control of mechanical oscillations per se G05D19/00
• CPC Class Codes: G01H 1/003 of rotating machines G01H1/10 takes precedence

Abstract

Disturbances of a periodic nature, such as those occurring in a rotating device can be attenuated by a cancellation function. A cancellation signal is produced by extracting one or more frequency components (harmonics) from a disturbance. A set of weighting coefficients are generated by an artificial neural network based on information derived from the selected frequency components of the periodic disturbance signal. The artificial neural network algorithm adapts in real time to shifts in the magnitude and phase of the disturbance frequencies selected for attenuation at the sensor location. In turn, these coefficients are applied to a function of a similar number of components and applied to the rotating device. Over a period of time, feedback and adaptation will attenuate the disturbance at the selected frequencies.

Claims

1. An apparatus for attenuating periodic vibration in a rotating device, where the periodic vibration comprises a series of time-variable coefficients and functions, said apparatus comprising means for detecting the periodic vibration in the rotating device;

means responsive to said means for detecting for extracting one or more of the time-variable coefficients and functions from the periodic vibration;

an artificial neural network having input neurons, output neurons and a hidden neuron, the connections from the input neurons to the hidden neuron constituting a weight layer having weights, said input neurons being connected to said means for extracting;

means providing a learning algorithm and responsive to said means for extracting for adapting in real time to adjust the weights in said artificial neural network;

means responsive to the weights in said artificial neural network for generating a cancellation signal; and

means responsive to said cancellation signal for imparting a force to the rotating device.

2. An apparatus as set forth in claim 1, where the means for extracting comprises means for extracting one or more harmonics of the periodic vibration.

3. An apparatus as set forth in claim 1, where the means for adjusting the weights in said artificial neural network comprises means, responsive to the means for extracting, for generating the sum of the squares.

4. An apparatus as set forth in claim 3, where the means for adjusting the weights in said artificial neural network further comprises means, responsive to the means for generating the sum of the squares, for filtering and, optionally, integrating.

5. A method for attenuating periodic vibration in a rotating device, where the periodic vibration comprises a series of time-variable coefficients and functions, said method comprising the steps of:

detecting the periodic vibration in the rotating device;

in response to said detecting, extracting one or more of the time-variable coefficients and functions from the periodic vibration;

in response to said extracting one or more of the time-variable coefficients and functions from the periodic vibration, training an artificial neural network comprising input neurons, output neurons, and a hidden neuron, the connections from the input neurons to the hidden neuron constituting a weight layer having weights and said training comprising adjusting the values of the weights in real time;

generating a cancellation signal using the adjusted weights; and

in response to said generating a cancellation signal, imparting a force to the rotating device.

6. A method as set forth in claim 5, where the step of extracting comprises the step of extracting one or more harmonics of the periodic vibration.

7. A method as set forth in claim 6, where the step of adjusting the weights comprises the step of generating the sum of the squares.

8. A method as set forth in claim 7, where the step of adjusting the weights futher comprises the steps of filtering and, optionally, integrating.

9. A method of attenuating a periodic disturbance in a rotating device, comprising the steps of:

sensing the periodic disturbance;

extracting selected components of the periodic disturbance;

applying the selected components to an artificial neural network to generate a set of weighting coefficients;

generating a function corresponding to the selected components and employing the weighting coefficients in the function;

applying the function to the device through a transducer; and

adapting the weighting coefficients in response to changes sensed over time in the periodic disturbance.