Optimise a loop with interpolation

Hello dear forum:

I am working on an battary optimization problem. As part of the problem, an electrical resistance should be dependent on T, C-Rate and SOC. I have a lot of sample values in between which it should be interpolated. Because the optimization problem is designed as a shortest path model, it iterates around 200 million times. My current approach is shown below and is currently very slow due to the interpolation. I used the griddata-command for interpolation.

Is there a faster way? My ideas:

• Interpolate a lot of values in the first rotation and then only select the closest one etc.
• Turn the sample values into a function. (I have heard about this but didnt find it).

I am glad for any suggestions!

Thank you and merry easter.

Here my current function:

function R = calculateResistance_neu_neu(R0,I,SOC,T,Q)
T = T;
C_Rate = I/Q;
SOC = SOC;
matrix = [5, 0.5, 0.1, 2.25, 135;
            .....
          45, 3, 0.8, 1, 60;
          45, 3, 0.9, 1, 60];

Multiplikator = griddata(matrix(:,1),matrix(:,2),matrix(:,3),matrix(:,4),T,C_Rate,SOC);

R = R0*Multiplikator;

[埃博拉酱]

Try to set the matrix as persistent. You'd better show more code outside this function. Optimization is hardly possible without your outer code.

[Fabian Haslwanter]

@埃博拉酱

Thank you. Tried making it persistent. However most of the time is consumed by the interpolation.

Here is the full code. I tried to put in some comments to explain it but it bases on the shortest path method. I hope it's somewhat understandable.

[Matt J]

Your mfile contains no function called calculateResistance_neu_neu()

[Fabian Haslwanter]

Line 47

I just tried running it and it works for me. Here again with the function. Hope thats what you mean :)

EDIT: Sorry i should have mentioned i have the function in a different file...

Answer 1

[Vanshika Vaishnav]

Setting the matrix as persistent will resolve the issue, however most of the time will be consumed by the interpolation.