SPACE STRUCTURES

A MAJOR DIFFICULTY WITH CURRENT DESIGN METHODS FOR LARGE SPACE STRUCTURES IS THEIR INABILITY TO COPE WITH TWO FUNDAMENTAL CHARACTERISTICS OF SUCH SYSTEMS: (A) LARGE MODELINACCURACIES, AND (B) MULTIPLICITY OF CONTROL LOOPS IN A HIERARCHICAL CONTROL ORGANIZATION. WE PROPOSE TO DEVELOP HIGHLY INTERACTIVE CAD METHODOLOGY FOR LARGE SPACE STRUTURESBASED ON FREQUENCY DOMAIN MODELS. THE LATTER OFFER THE FOLLOWING ADVANTAGES: (1) CAN BE VERIFIED BY EXPERIMENTAL DATA, (II) ALLOW DEVELOPMENT OF ROBUST CONTROLLERS IN A NATURAL WAY, (III) CAN TOLERATE "LARGE" MODEL INACCURACIES OR VARIATIONS AND (IV) PERMIT THE ESTABLISHMENT OF A QUANTITATIVE TRADEOFF BETWEEN COMPLEXITY OF THE HIERARCHICALCONTROL MULTILOOP STRUCTURE AND PERFORMANCE OF THE CONTROLLER. THE APPROACH PROPOSED HERE BUILDS ON THREE FUNDAMENTAL TECHNIQUES: (A) TRANSFER FUNCTION MODELS FOR DISTRIBUTED SYSTEMS AND THEIR APPROXIMATIONS; (B) RESULTS ONFACTORIZATIONS AND REPRESENTATIONS FOR IRRATIONAL FUNCTIONS;(C) INTERACTIVE OPTIMIZATION PACKAGES (DELIGHT) IN THE SEARCH FOR FEASIBLE DESIGNS SATISFYING PRACTICAL ENGINEERINGSPECIFICATIONS. OF PARTICULAR INTEREST TO US IS (C) BECAUSEIT PERMITS THE TREATMENT OF SPECIFICATIONS DIRECTLY PLACED ON THE SHAPE OF THE CLOSED LOOP TRANSFER FUNCTION AND/OR UNIT STEP RESPONSE. WE SHALL INVESTIGATE SIMPLE EXAMPLES INCLUDING INTERCONNECTED BEAMS AND/OR PLATES WITH RIGID BODIES.