Prediction/ optimization in Knime ?


I’m trying to study a heat transfer in a wall (with several materials) on Knime. I built a database (xlsx file) with these datas:
-outside temperature
-mean temperature in material 1
-mean temperature in material 2
-mean temperature in material 6
-surface temperature of the wall
-heat power

At each timestep, the outside temperature changes and influences the mean temperatures in the wall and the surface temperature. The heat power depends on the difference between surface temperature (Tsurface) and the target temperature for the room (for exemple Ttarget = 20°C).

The input variables are timestep and temperatures, the target is the heat power. I’m able to create the predictive model with few algorithms, select the best of them, and do the model deployment with a new set of datas.

For instance, for the deployment, I use a set of parameters (timestep1, mean temperatures, and surface temperature), the model makes a prediction of the heat power. I want to use the prediction in my heat transfer simulation to compute the new temperature for the new timestep and use these results to make another heat power prediction and so on.

I’d like to optimize the heat power in order to minimize the difference (Tsurface - Ttarget) at each timestep. I’m kinda stuck on this, is there any way to make this optimization?

Hoping my request is clear. Thank you

Hallo @yrobert,

I’m not sure if I understand the problem that you are trying to solve. It sounds a little bit complicated. Do you maybe have some example data or maybe even a workflow that you can share? That would make it a little bit easier for me to understand the problem and help you.


This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.