DEVELOPMENT OF COMPUTER-AIDED DESIGN METHODOLOGIES SUITED TO A CLASS OF DC TO DC POWER PROCESSOR SYSTEMS

Current practices for the design of power processor systems separate the design for a set of static performance criteria from the design of the attendant control loop, and do not allow for variations in the system parameters, such as input and load conditions. The computer-aided design methodologies contained in this thesis, which have as their foundation a hierarchical approach to optimization, are formulated and developed in a generalized manner, and have application to a wide range of circuits and systems problems in addition to overcoming the specific power processor design deficiencies above. These methodologies include: a decomposed approach to parameter optimization, based on an interconnected system model; a generalized technique for the coordination of static and dynamic optimization problems to an overall optimum solution; and the formulation and solution of a dynamic fixed tolerance problem, in which the system parameters are allowed to vary within a specified tolerance about their nominal values. Design examples are presented for the coordination of a power processor control loop design with a minimum converter weight objective, and for a converter system subject to load and source variations and component aging, in which certain design constraints may otherwise be violated if the design is effected by conventional means