Hybrid Computer Approach to Parametric Optimization of Dynamic Systems Under Random Disturbances

The paper deals with application of hybrid computer to parametric optimization of dynamic systems subjected to random disturbances. A procedure is used which significantly simplifies the design of a statistically optimal control. Parametrization of control actions reduces the search for a minimum in an infinitely-dimensional space to search in a finite-dimensional space. Computation of optimal characteristics for a linear electron accelerator waveguide buncher electromagnetic field is used as an example to illustrate the advantages of hybrid over digital computers or special-purpose analog optimizers. Computer time is choosen as the efficiency criterion.