EXPERIENCES WITH OPTIMIZERS IN STRUCTURAL DESIGN Adaptive Computing in Engineering Design and Control

The control of structural vibration in cars, aeroplanes, ships, etc., is of great importance in achieving low noise targets. Currently, such control is effected using viscoelastic coating materials although much current research is concerned with active, anti-noise based control measures. This paper is concerned with a third approach: that of using advanced structural analysis methods and modern optimization techniques to produce designs with inherently superior vibration performance. It gives a brief review of a ten year programme of work, illustrating some of the difficulties that have been encountered and the progress that has been made. Central to this process has been the availability of a library of optimization techniques that can be readily applied to the problems of interest. In recent years this suite has been expanded to include a number of evolutionary computing methods such as the genetic algorithm, simulated annealing, evolutionary programming, etc. These have shown significant advantages over more classical methods. By way of example this paper considers three rather different problems: an extremely simple structure consisting of two rods and a spring, a moderately complex satellite boom and a computationally efficient but optimizationally hard, test problem that allows study and comparison of the various optimization methods available.