Hybrid evolutionary optimal MEMS design



A hybrid evolutionary design synthesis and optimization process for microelectromechanical systems (MEMS) devices has been developed. The process integrates a MEMS design component library with multiple simulation modules and two levels of design optimization: global multi-objective genetic algorithms (MOGA) and local gradient-based refinement. During the hybrid evolutionary design process, MOGA randomly searches the design space and approaches the desirable design solutions using probabilistic transition rules, and gradient-based local optimization refines promising design candidates with computational efficiency. To efficiently apply hybrid evolutionary optimization techniques on MEMS designs, a hierarchical tree-structured component-based genotype representation has been developed, which incorporates specific engineering knowledge into the design synthesis and optimization process. The MEMS design component library serves as a source of practical and efficient genotypes for the evolutionary process, with each component associated with its instructions and restrictions on genetic operations. The component-based genotype incorporated with engineering knowledge constrains evolutionary searching in appropriate and promising regions of the search space, allowing a deeper search in a given amount of time. Hybrid evolutionary MEMS design synthesis and optimization are demonstrated with surface-micromachined resonator and accelerometer designs.


Hybrid evolutionary process Multi-objective genetic algorithm Microelectromechanical systems Design synthesis and optimization 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Supplementary material

170_2012_3908_MOESM1_ESM.docx (12 kb)
ESM 1(DOCX 11 kb)

Copyright information

© Springer-Verlag London Limited 2012

Authors and Affiliations

  1. 1.School of Electrical and Computer EngineeringGeorgia Institute of TechnologyAtlantaUSA
  2. 2.Department of Mechanical EngineeringUniversity of CaliforniaBerkeleyUSA

Personalised recommendations