How to increase weight of a point

Mirhan Ozdemir_21843

Hi everyone,

I am working on an optimization process. I measure the FRF of 9 points from the structure and trying to minimize summation of FRFs as objective of the optimization. however, I realized that a point is more important than others, so I tried to increase its weight in the summation by dequation card. I defined a equation  y(x1,x2,x3,x4,x5,x6,x7,x8,x9) = x1+x2+x3+x4+x5+x5+x5+x5+x6+x7+x8+x9 (to increase weight of the interested point 4 times) and used it in the response. However, I am not sure wheter my approach is correct or not? Or is there another way to do so. 

Thank you for your time. 

Adriano Koga_20259

this is ok.

You could just use 4*x5 in your equation, but it works as well.

You could also try using a minmax formulation, by minimizing the maximum of all your frf resposnes.

you would create a objective reference, selecting your FRF values and giving them some reference (normalization) value. Then assin a objective as minmax (objective reference). But in this case, no weighting is applied.

For weighting you could maybe create a secondary response, multiplying only your x5 by a weight, and use this new one (x5*), instead of the original x5 in the minmax, or even in the original equation.

 

Another option is to use a DSYSID for System Identification. In this case you select all your frf points and give them some target values (known values) and also weighting factors. OS will perform a Min Mean Squares fitting, trying to fit as good as possible your targets.

Mirhan Ozdemir_21843

Adriano Koga_20259 said:

this is ok.

You could just use 4*x5 in your equation, but it works as well.

You could also try using a minmax formulation, by minimizing the maximum of all your frf resposnes.

you would create a objective reference, selecting your FRF values and giving them some reference (normalization) value. Then assin a objective as minmax (objective reference). But in this case, no weighting is applied.

For weighting you could maybe create a secondary response, multiplying only your x5 by a weight, and use this new one (x5*), instead of the original x5 in the minmax, or even in the original equation.

 

Another option is to use a DSYSID for System Identification. In this case you select all your frf points and give them some target values (known values) and also weighting factors. OS will perform a Min Mean Squares fitting, trying to fit as good as possible your targets.

Thank you so much Adriano, I will go through each of your suggestions.