FUNCTIONAL CAPABILITY OF NEUROMINE NETWORKS FOR USE IN ATTITUDE STABILIZATION SYSTEMS.
Final rept. 15 Mar 62-15 Mar 63,
ADAPTRONICS INC MCLEAN VA
Pagination or Media Count:
An adaptive network illustrating the use of Neurotrons neuromimes with the ability to learn both analog and logical functions and Neurotron goals in an adaptive control system is presented, performande data being provided by computer simulation. The use of this system for satellite attitude stabilization is described. Computer flow diagrams and programming data for simulating a Neurotron, Neurotron networks, and goal systems on a digital computer are described. This report also includes mathematical development of a suitable way to digitize the analog Neurotron functions on the computer, a mathematical demonstration that standard electrical equations for RC circuits can be applied to given Neurotron analog functions, and a hardware-oriented discussion of Neurotrons in terms of detailed block diagrams. Based on the computer simulation data and mathematical analyses of Neurotron function, it is concluded that Neurotrons are feasible for providing not only adaptive control functions and stabilization, but that they provide a new type of reliability advantage which stems from their flexibility in terms of learning. This flexibility will not only take care of gradual or catastrophic failures of Neurotrons, but will also compensate for performance degradation in the feedback loop. Author