Sensitivity Analysis in Large Scale Multi-Objective Systems

No matter what method is employed in the programming of a transducer array, in practice the performance predicted by the model will not be achieved due to error in construction and operation of the array. Furthermore, the degree to which a given amount of error degrades an array design (i.e., design sensitivity) will differ greatly from one design to the next. A measure of sensitivity is proposed, based on the maximum impact which a given amount of error can produce on a particular design. The adoption of this measure allows the sensitivity problem to be solved through the use of mathematical programming techniques. The best of these techniques for the general array design case is shown to be Nonlinear Goal Programming. The sensitivity values for five planar symmetric array designs are determined and compared.