Robust Design of Control Systems with Physical System Variances
Today’s automotive control system engineering requires precision and accuracy. The cost of a controller designed with conservative margins may increase significantly, causing the design, when produced and marketed, to be less competitive. On the other hand, a design with too little margin may lead to system malfunction under marginal environment conditions or due to component aging. A robust design is one that is immune to the effects of component variance due to tolerance, temperature, and aging, among other factors. Achieving a robust design involves careful analysis of the controller and plant operating together. This paper discusses how MATLAB and Simulink can be leveraged to ensure the robustness of a mechatronic system design. The merits of the network approach as a technique for modeling physical systems as an alternative to the signal flow (block diagram) approach are also discussed.