Neural, Modular, Adaptive Structures

The success of any strategic defense system depends on the control of space structures. Space structures facilitate precise and rapid retargeting maneuvers without adverse jitter and to provide decision making in the presence of uncertainties and failures. Two enabling technologies; piezoceramic sensor/actuator and neural networks will be blended in a modular, miniaturized, light-weight, adaptive structure in this project to meet the challenge for space applications. The adaptive structurure can either be embedded in the structural member or mounted on the surface of the host surface. Use of miniaturized electronics, distributed sensor/actuator designs, and advanced control concepts, the adaptive structure patch meets the weight, size, power, and reliability requirements for space applications. Another innovation is the use of dual neural networks to perform stabilization and reconfiguration in a real-time environment. A back propagation type neural network is used for stabilization and precision control. A cerebullar model articulation computer is used for health monitoring and reconfiguration control. The mutual-aid concept can be exploited to reinforce learning and enhance decision making capability for the adaptive structure.