Problems with look up table

Hi. I'm having some problems with using a look-up table in Matlab using the function fixpt_interp1(xdata,ydata,x,sfix(8),1 (x scale), sfix(16), 1, 'Floor').

xdata = 0:1:40

ydata = [0.0 0.0 55.0 .......... 7580 (at rated) .... 0 (cutoff)]

My xdata and ydata is a power curve for a wind turbine, with xdata being wind speeds and ydata being Power at this particular wind speed. My x-variable is a time series of wind speeds between 0 - 35 m/s. I want to insert my wind speed time series so that the output of the function is a power time series.

My problem is that once a wind speed is between two integers, for example 9.1 m/s, the output gives me the equivalent power of 9 m/s (3750 kW). I know that 'Floor' makes this happen, because 'Ceiling' will make it go to the power of 10 m/s (4850 kW). Still, I thought this function would interpolate between the two points, giving me a power of 3860 kW, but it doesn't seem so.

Therefore I tried to interpolate the curve using:

yi = interp1(xdata,ydata,xi), where

xi = 0:0.1:40

When I then insert yi and xi in fixpt_interp1, I get the right amount of power for 9.1 m/s, but it messes up other things, like putting all wind speeds over 12.7 m/s to 0 kW, when 25 m/s is the cut-off speed.

Does anyone have a solution for this problem?

If you need more info, please let me know.

Jonas from Norway

[Jan]

What is the problem exactly? Why do you use the fixed point type? Why is 0kW replied for speeds over 12.7, when not 12.7 is the cut-off but 25? How are "other things messed up" exactly? I think, that the solution is to avoid that these "other things" are "messed up", therefore I suggest to describe this point with any details.

[Thorsten]

Please provide your full ydata such that we can see how interp1 "messes up other things".

Answer 1

[Surya Vamsi Krishna Chivukula]

I had a similar problem with fixpt_itnerp1. Try using the following:

y = fixpt_interp1(xdata, ydata, x, float('single'), [], float('single'), [], 'Floor');

Setting the 'x' and 'y' data types as float('single') and letting MATLAB choose it's own 'x' and 'y' scaling does the trick.