Intelligent control of high-performance microsystems: evolutionary learning and adaptation paradigms for opto-electro-mechanical MEMS in integrated wireless communication networks

摘要

Microelectromechanical systems (MEMS) are composed from microtransducers (actuators and sensors), radiating energy microdevices, microstructures (mirrors, beams, torsions, et cetera) and processing-controlling integrated circuits (ICs). The major challenges in MEMS are to attain the desired performance and functionality while minimizing size, reducing complexity, maximizing efficiency, integrity and robustness. Control of MEMS is a very difficult problem due to model nonlinearities and uncertainties. Microsystems are controlled by ICs which integrate millions of transistors on a single chip. High-fidelity mathematical models are very complex, and the control theory is not yet matured enough to solve a number of challenging problems. Recent developments in modern control theory have demonstrated it capabilities for advanced electro -mechanical systems. Complexity and distinct features of MEMS do not allow the designer to apply the developed methods. We report an intelligent control paradigm, and premium control laws are designed utilizing evolutionary learning and adaptation mechanisms (through reconfiguration, tuning, and optimization) with the ultimate objective being to attain the optimal overall performance. The objective functional is evaluated using measured controls, references and outputs which are the performance variables. It is demonstrated that using evolutionary learning and adaptation, the intelligent control problem can be solved without linguistic or mathematical models. Making use of the performance, requirements, objectives, functionals, and measured information sets, one designs premium controllers. The premium control is derived for an opto-electro-mechanical MEMS to illustrate the results in wireless communication.