Design optimization of a microelectromechanical electric field sensor using genetic algorithms
This thesis studies the application of a multi-objective niched Pareto genetic algorithm on the design optimization of an electric field mill sensor. The original sensor requires resonant operation. The objective of the algorithm presented is to optimize the geometry eliminating the need for resonant operation which can be difficult to maintain in the presence of an unpredictable changing environment. The algorithm evaluates each design using finite element simulations. A population of sensor designs is evolved towards an optimal Pareto frontier of solutions. Several candidate solutions are selected that offer superior displacement, frequency, and stress concentrations. These designs were modified for fabrication using the PolyMUMPs abrication process but failed to operate due to the process. In order to fabricate the sensors in-house with a silicon-on-glass process, an anodic bonding apparatus has been designed, built, and tested.
MEMS, Genetic Algorithm, Optimization, Sensors, Evolutionary Computing, Finite Element Analysis, Electric Field Measurement