Structured synthesis of MEMS using evolutionary approaches
Introduction
Due to the complexity and mixed-domain intricacy involved in MEMS design, designing MEMS remains an art in most applications, requiring a large amount of investment of human resources, time and money. Much of the investment is consumed in the iterative trial-and-error design process. Automated design synthesis helps engineers to rapidly develop optimal configurations for a given set of performance and constraint guidelines, and thus to shorten typical development cycles for MEMS (with a given fabrication technology) by a large factor and to enable design of far more complex MEMS than can be handled today. Electronic Design Automation (EDA) has achieved great success in both industry and academia. However, analogous research in design automation for MEMS seems to lag far behind, although considering the close affinity of MEMS and VLSI – MEMS actually evolved from microelectronics and inherited the fabrication techniques of VLSI – the potential for successful applications of design automation to MEMS appears to be promising. Translating the key insights of silicon evolution success into MEMS technologies is a much more challenging task than most people have expected. Major research topics to be addressed include:
- 1.
developing a broad base of building blocks in MEMS technologies so that huge networks of micro-devices can be assembled into arbitrary architectures with desirable functionalities;
- 2.
hierarchically decomposing the MEMS design process to reduce design complexity, making it more amenable to automation;
- 3.
improving models of computation and extending current synthesis methodologies to facilitate generation of viable design candidates and smoother transitions from conceptual and embodied designs to process fabrication;
- 4.
combining MEMS component layout extraction and lumped-parameter simulation and design synthesis to provide MEMS designers with a VLSI-like environment, enabling faster design cycles and improved design productivity.
This paper aims to partially address these challenges. The proposed hierarchical and evolutionary design framework for MEMS seeks to eliminate tedious and repetitive design tasks, facilitate hierarchical problem decomposition, and combine the power of multiple evolutionary computation algorithms working sequentially to generate and identify better product designs and process solutions. In particular, we divide design representations of MEMS into two levels, the system-level behavioral macromodel and the detailed-level physical geometric layout model. At the system level, we use a combination of genetic programming and bond graphs to automatically generate and search for viable design candidates represented by behavioral macromodels satisfying high-level design specifications. At the second detailed (layout) level, constrained genetic algorithms are used to optimize the geometric parameters that relate the physical device model to the behavioral macromodel and meet more detailed design objectives.
This paper is organized as follows: Section 2 introduces a structured method of MEMS synthesis presented by Antonsson [1]. Section 3 describes the GPBG approach that combines genetic programming and bond graphs for system-level behavioral synthesis. Section 4 explains how a constrained GA approach can be used to solve the second-level physical layout synthesis problem. Concluding remarks are presented in Section 5.
Section snippets
Structured MEMS design methodology
In MEMS, there are a number of levels of designs that need to be synthesized [10]. Usually the design process starts with basic capture of the schematic of the overall system, and then goes on through layout and construction of a 3-D solid model. So the first design level is the system level, which includes selection and configuration of a repertoire of planar devices or subsystems. The second level is 2-D layout of basic structures like beams to form the elementary planar devices. In some
Genetic programming for system-level sythesis
In this research, genetic programming (GP) is used as a strong search tool to explore the topologically open-ended design space for system-level behavioral models of MEMS. The bond graphs (BG) is also used as a modeling tool to unify representations of mixed energy domains of MEMS. The overall approach is called GPBG approach. The next section gives a brief introduction to bond graph.
Second level physical layout synthesis
Layout synthesis automatically generates valid or optimized geometric sizing parameters for cell components, which in most cases are chosen from micromechanical devices with fixed topologies, according to engineering design objectives. In this research, the cell component is a resonator device in the MEMS domain. The design objectives come from either high-level specifications, such as behavioral model parameters that must be satisfied, or from layout-level objectives such as minimizing the
Conclusion
This paper has suggested a design methodology for automatically synthesizing hierarchical designs for MEMS. While there has been much research using evolutionary computation techniques to synthesize MEMS [23], [40], this is the first work reported that attempts to automates the hierarchical MEMS synthesis process in an integrated framework. Our first step is to synthesize system-level behavioral models using a combination of genetic programming and bond graphs. Then in the second step, we use a
Acknowledgements
The authors gratefully acknowledge the support of the National Science Foundation (NSF) through grant DMI 0084934. The authors are grateful to Ranny Sidhu and Johanna Burgueno at Red Cedar Technology, Inc., for their help in setting up and using the HEEDS software. Financial support from the Natural Sciences and Engineering Research Council of Canada (NSERC) to enable Sofiane Achiche to work at Technical University of Denmark under Post-Doctoral Grant BP-328508-2006 is also gratefully
References (42)
- et al.
Development of a soft computing-based framework for engineering design optimisation with quantitative and qualitative search spaces
Applied Soft Computing
(2007) - et al.
Evolutionary design in a multi-agent design environment
Applied Soft Computing
(2006) Microsystem design synthesis
- F. Broenink, Introduction to Physical Systems Modeling with Bond Graphs, SiE Whitebook on Simulation Methodologies...
- et al.
Agent-based synthesis of electro-mechanical design configurations
J. Mech. Des.
(2000) - et al.
Development of a geometry-based process planning for surface micromachining
International Journal of Production Research
(2002) - et al.
An injection island GA for flywheel design optimization
- Z. Fan, J. Hu, K. Seo, E. Goodman, R. Rosenberg, B. Zhang, Bond Graph Representation and GP for automated analog filter...
- et al.
A novel evolutionary engineering design approach for mixed-domain systems
Engineering Optimization
(2004) - et al.
System-level synthesis of mems via genetic programming and bond graphs
Physical design for surface-micromachined MEMS
A hierarchical circuit-level design methodology for microelectromechanical systems
IEEE Transactions on Circuits and Systems II (TCAS)
Computers and creative design
Genetic Algorithms in Search, Optimization and Machine Learning
Adaptation in Natural and Artificial Systems
System Dynamics A Unified Approach
Target cascading in optimal system design
Genetic Programming III Darwinian Invention and Problem Solving
Genetic Programming: On the Programming of Computers by Means of Natural Selection
Cited by (30)
Multi-level and multi-objective design optimisation of a MEMS bandpass filter
2017, Applied Soft Computing JournalCitation Excerpt :The GAECM approach outlined in this paper was able to provide comparable bandpass filter transmissions to those within the literature for the current state of the art in automated design [43] as well as expand upon it through additional case study examples. In terms of overall performance the approach shows a significant reduction in functional evaluations, 10,000 compared with 2.6 million required to solve this design problem when compared with the state of the art [43]. However it is hard to make a direct comparison as both approaches have their benefits and drawbacks, in particular the GPBG method outlined in [43] utilizes bond graphs which are open to more creative design then a rigid LCR circuit model.
Solving system-level synthesis problem by a multi-objective estimation of distribution algorithm
2014, Expert Systems with ApplicationsCitation Excerpt :Blickle (1996) first developed a single-objective EA, and later Blickle, Teich, and Thiele (1998) introduced a Pareto-ranking technique into the single-objective EA. Fan, Wang, Achiche, Goodman, and Rosenberg (2008) introduced the flow of a structured Micro-Electro-Mechanical Systems (MEMS) design process to emphasize the system-level lumped-parameter model synthesis. To trade off the predefined behavioral specifications for designers, at the system level an approach combining bond graphs and genetic programming can yield satisfactory design candidates.
Hierarchical component-based representations for evolving microelectromechanical systems designs
2011, Artificial Intelligence for Engineering Design, Analysis and Manufacturing: AIEDAMApplications of Evolutionary Computation in the Design Automation of Complex Mechatronic System: A Survey
2021, Zidonghua Xuebao/Acta Automatica SinicaEvolutionary design optimization of MEMS: a review of its history and state-of-the-art
2019, Cluster Computing