09.07.2015 Views

Robust Optimization: Design in MEMS - University of California ...

Robust Optimization: Design in MEMS - University of California ...

Robust Optimization: Design in MEMS - University of California ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

39We were able to f<strong>in</strong>d Φ 1000 by m<strong>in</strong>imiz<strong>in</strong>g the squared error between w 2 n and w 2 targetfrom different <strong>in</strong>itial conditions. This can be written as the follow<strong>in</strong>g optimizationm<strong>in</strong>x( )w2n − wtarget2 2s.t. gi (x) ≤ 0 (5.10)This problem is equivalent to m<strong>in</strong>imiz<strong>in</strong>g the c 2 (x) term <strong>in</strong> the robust design problem<strong>of</strong> equation (2.9). The optimization problem posed <strong>in</strong> (5.10) is rational polynomialwith polynomial constra<strong>in</strong>ts, and thus we were able to use our optimization algorithmto f<strong>in</strong>d unique designs. Figure (5.7) illustrates four <strong>of</strong> the more extreme designs fromΦ 1000 . Each <strong>of</strong> these structures has a resonant frequency <strong>of</strong> 200 kHz and satisfiesthe constra<strong>in</strong>ts. They demonstrate that there is significant freedom <strong>in</strong> how the designvariables are chosen. The question we will now ask is “which design is the most robustto geometric uncerta<strong>in</strong>ties, and how can we f<strong>in</strong>d that design?”600600500500400400y (µm)300y (µm)30020020010010000 200 400 600x (µm)00 200 400 600x (µm)600600500500400400y (µm)300y (µm)30020020010010000 200 400 600x (µm)00 200 400 600x (µm)Figure 5.7: Four unique crab-leg resonator designs from Φ 1000 .

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!