REG NASA-LLIS-1525--2005 Lessons Learned - Problems Associated With the Use of a Dynamic Inversion Controller in Conjunction With an Adaptive Neural Network in the Presence of Surf.pdf

《REG NASA-LLIS-1525--2005 Lessons Learned - Problems Associated With the Use of a Dynamic Inversion Controller in Conjunction With an Adaptive Neural Network in the Presence of Surf.pdf》由会员分享，可在线阅读，更多相关《REG NASA-LLIS-1525--2005 Lessons Learned - Problems Associated With the Use of a Dynamic Inversion Controller in Conjunction With an Adaptive Neural Network in the Presence of Surf.pdf（3页珍藏版）》请在麦多课文档分享上搜索。

1、Lessons Learned Entry: 1525Lesson Info:a71 Lesson Number: 1525a71 Lesson Date: 2005-02-02a71 Submitting Organization: DFRCa71 Submitted by: Peggy S. Williams-HayesSubject: Problems Associated With the Use of a Dynamic Inversion Controller in Conjunction With an Adaptive Neural Network in the Presenc

2、e of Surface Failures Abstract: The NASA F-15 Intelligent Flight Control System (IFCS) project team has developed a series of flight control concepts designed to demonstrate the benefits of a neural network-based adaptive controller. The objective of the team is to develop and flight-test control sy

3、stems that use neural network technology to optimize the performance of the aircraft under nominal conditions as well as stabilize the aircraft under failure conditions. Failure conditions include locked or failed control surfaces as well as unforeseen damage that might occur to the aircraft in flig

4、ht. The IFCS team is currently in the process of implementing a second generation control scheme, collectively known as Gen 2, for flight testing on the NASA F-15 aircraft. The research controller is a dynamic inverse controller based on the Versatile Control Augmentation System (VCAS) control laws

5、developed by James Buckley at McDonnell Douglas Aircraft. The baseline control laws were taken from the F-15 ACTIVE ILTV program, but the IFCS team has removed the thrust vectoring controls for the Gen 2 implementation. For the original dynamic inverse controller, the proportional, integral, & deriv

6、ative (PID) gains were tuned to achieve linear stability robustness, ASE mode attenuation, and non-linear system command following for the nominal (no failure) case. Description of Driving Event: While testing the original dynamic inverse controller with locked stabilator failure simulations, signif

7、icant lateral acceleration (ny) and angle of sideslip (Beta) excursions as a result of lateral stick inputs were noted. One inch of roll stick input would cause lateral accelerations of -0.5g and Beta values as high as 5 degrees. Because subsequent re-designs of the neural network were not able to m

8、odify this behavior and pilot comments from simulation sessions continued to be negative, it was decided that the research controller be modified so that the lateral and directional axis would be de-Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-cou

9、pled in the research controller. This was accomplished by using a Beta-dot classical controller for the yaw axis, while continuing to use the original dynamic inverse control for the pitch and roll axes.Lesson(s) Learned: Dynamic inversion controller contributes significantly to cross-coupled respon

10、se in presence of surface failures (locked surface).Recommendation(s): The Beta-dot controller was able to eliminate the lateral-directional coupling issues noted previously. The Beta-dot controller was chosen because it was straightforward to implement and had been used successfully in previous pro

11、grams. The Beta-dot controller uses the rudder pedal input, as well as information from the aircraft sensors for ny, velocity, stability axis yaw rate, pitch angle, and bank angle as inputs. The Beta-dot controller does not use Beta as an input due to the fact that the only source available is from

12、the nose boom measurement of Beta, which is a single-string source. The research controller necessary for flight test is a hybrid controller using dynamic inversion in the longitudinal and lateral axes, with a classical control used in the directional axis. This modification was necessary in order t

13、o obtain reasonable flying qualities for the nominal research controller with no neural network adaptation present. Evidence of Recurrence Control Effectiveness: N/ADocuments Related to Lesson: N/AMission Directorate(s): a71 Exploration Systemsa71 Aeronautics ResearchAdditional Key Phrase(s): a71 Ai

14、rcraftAdditional Info: Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-Approval Info: a71 Approval Date: 2005-07-12a71 Approval Name: Mike Reljaa71 Approval Organization: DFRCa71 Approval Phone Number: 661-276-3621Provided by IHSNot for ResaleNo reproduction or networking permitted without license from IHS-,-,-