Applications of polynomial neural networks to FDIE and reconfigurable flight control

Fault detection, isolation, and estimation (FDIE) functions and reconfiguration strategies for flight control systems present major technical challenges, primarily because of uncertainties resulting from limited observability and an almost unlimited variety of malfunction and damage scenarios. Attention is focused on a portion of the problem, i.e. global FDIE for single impairments of control effectors. Polynomial neural networks are synthesized using a constrained error criterion to obtain pairwise discrimination between impaired and no-fail conditions and isolation between impairment classes. The pairwise discriminators are then combined in a form of voting logic. Polynomial networks are also synthesized to obtain estimates of the amount of effector impairment. The algorithm for synthesis of polynomial networks (ASPN) and related methods are used to create the networks, which are high-order, linear or nonlinear, analytic, multivariate functions of the in-flight observables. The authors outline the design procedure, including database preparation, extraction of waveform features, network synthesis techniques, and the architecture of the FDIE system that has been studied for control reconfigurable combat aircraft (CRCA). Single-look (25-ms response time) simulation results are presented.<<ETX>>

[1]  H. Akaike,et al.  Information Theory and an Extension of the Maximum Likelihood Principle , 1973 .

[2]  Colin Giles,et al.  Learning, invariance, and generalization in high-order neural networks. , 1987, Applied optics.

[3]  R. J. Lee,et al.  Generalization of learning in a machine , 1959, ACM '59.

[4]  J J Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities. , 1982, Proceedings of the National Academy of Sciences of the United States of America.

[5]  W. Pitts,et al.  A Logical Calculus of the Ideas Immanent in Nervous Activity (1943) , 2021, Ideas That Created the Future.

[6]  J. Rissanen.  A UNIVERSAL PRIOR FOR INTEGERS AND ESTIMATION BY MINIMUM DESCRIPTION LENGTH , 1983 .

[7]  Dennis Gabor,et al.  A universal nonlinear filter, predictor and simulator which optimizes itself by a learning process , 1961 .

[8]  Roger L. Barron,et al.  Automated Design of Continuously-Adaptive Control: The "Super-Controller" Strategy for Reconfigurable Systems , 1988, 1988 American Control Conference.

[9]  A. Barron,et al.  Statistical properties of artificial neural networks , 1989, Proceedings of the 28th IEEE Conference on Decision and Control,.

[10]  Bernard Widrow,et al.  Adaptive switching circuits , 1988 .

[11]  A. G. Ivakhnenko,et al.  Polynomial Theory of Complex Systems , 1971, IEEE Trans. Syst. Man Cybern..

[12]  Donald F. Specht,et al.  Generation of Polynomial Discriminant Functions for Pattern Recognition , 1967, IEEE Trans. Electron. Comput..

[13]  Dennis Gabor,et al.  Communication Theory and Cybernetics , 1954 .

[14]  John W. Tukey,et al.  A Projection Pursuit Algorithm for Exploratory Data Analysis , 1974, IEEE Transactions on Computers.

[15]  Harry N. Gross,et al.  Application of Supercontroller to Fighter Aircraft Reconfiguration , 1988, 1988 American Control Conference.

[16]  J. Friedman,et al.  Projection Pursuit Regression , 1981 .

[17]  Geoffrey E. Hinton,et al.  Learning internal representations by error propagation , 1986 .

[18]  W. A. Clark,et al.  Simulation of self-organizing systems by digital computer , 1954, Trans. IRE Prof. Group Inf. Theory.