Application of optimal and robust design methods to a MEMS accelerometer

This paper reports on the optimization and robust design of a MEMS accelerometer using the genetic algorithm technique. Performance requirements and MEMS process factors are taken into account. Objective functions are created to optimize each performance aspect and to penalise designs that are not sufficiently robust to manufacturing variations. The variation of performance with robustness is also studied, and increasing the requirement for robustness is shown to have a negative effect on performance.