Stabilized adaptive neural network based control system

• US 5,493,631 A
• Filed: 11/17/1993
• Issued: 02/20/1996
• Est. Priority Date: 11/17/1993
• Status: Expired due to Fees

• Current Assignee: Grumman Aerospace Corporation (Northrop Grumman Corporation)
• Original Assignee: Northrop Grumman Corporation
• Inventors: Eilbert, James L., Engel, Stephen J., Huang, Chien Y.
• Primary Examiner(s): Moore, David K.
• Assistant Examiner(s): HAFIZ, TARIQ R
• Application Number:
  US08/153,096
• Time in Patent Office: 825 Days
• Field of Search: 395/20-25, 395/27, 395/906, 204/435, 204/433, 364/157, 382/153-159
• US Class Current: 706/23
• CPC Class Codes: G05B 13/027 using neural networks only

Abstract

A neural network based control system includes a nominal control system augmented by adaptive control such as a neuro-controller which generates additional compensating control signals based on differences between a model and actual system output. The nominal control system provides basic stability and performance, while the adaptive controller provides performance enhancement. The adaptive control can rely on any neural network that can encode a-priori knowledge and employs a high resolution pattern discrimination capability suitable for real-time changes. To prevent unbounded adaptation, the output of the adaptive controller is constrained by a limiter, thus ensuring safety of the overall system.

Claims

1. A control system, comprising:

means for generating a command control signal in response to a command and supplying the command control signal to an actuator to cause the actuator to initiate an action in a plant in response to the command control signal;

means for sensing results of the action and outputting a first sensor signal and a second sensor signal based thereon;

a nominal control system, including;

means responsive to the first sensor signal for generating a first control signal in response to sensing of the results of the action; and

an auxiliary adaptive control system, including;

means for comparing the action as indicated by the second sensor signal with a model of the action based on said command and generating a second control signal in response to the comparison;

means for combining said first and second control signals with said command control signal to modify said command control signal supplied to the actuator,

wherein said auxiliary adaptive control system includes a neuro-controller which generates a corrective control signal responsive to said second sensor signal and to a learned condition, and

further comprising a limiter which limits the second control signal within a gain margin and a phase margin of the nominal control system in order to generate said second control signal, and means for changing an amount by which the limiter limits the second control signal in accordance with a confidence level generated by the neuro-controller.

2. A system as claimed in claim 1, wherein said plant is an aircraft.

3. A control system, comprising:

means for generating a command control signal in response to a command and supplying the command control signal to an aerator to cause the actuator to initiate an action in a plant in response to the command control signal;

means for sensing results of the action and outputting a first sensor signal and a second sensor signal based thereon;

a nominal control system including;

means responsive to the first sensor signal for generating a first control signal in response to sensing of the results of the action; and

an auxiliary adaptive control system, including;

means for comparing the action as indicated by the second sensor signal with a model of the action based on said command and generating a second control signal in response to the comparison;

means for combining said first and second control signals with said command control signal to modify said command control signal supplied to the actuator.

wherein said auxiliary adaptive control system includes a neuro-controller, and

wherein said neuro-controller is arranged to compare the action as indicated by the second sensor signal, generate a second control signal in response to the comparison while on-line in said plant, and be trained off-line.