python - Park and Ramirez's Bioreactor using TensorFlow - An optimization problem - Stack Overflow

admin管理员组

文章数量:1276853

I have been tasked with optimizing the productivity of Park and Ramirez's Bioreactor using TensorFlow. To achieve this, I generate a dataset by creating random values for the "Feed" variable, and based on these values, the other variables are calculated and stored. I then train an RNN to optimize the value of Pm * V (t = t_f), which I was informed should be 34, along with its corresponding feed value. However, I'm currently only able to optimize for a single feed value at a time, whereas I need to optimize for multiple feed values simultaneously. Despite my efforts, I have not been able to achieve the target value of 34, which I suspect is due to the fixed feed value.

I have been advised to use random feed values, as this is supposedly the only way to reach a productivity of 34. However, I am uncertain whether the feed values should vary in an increasing pattern or truly randomly. I was only told that the feed should not be constant.

Currently, I am using only 1 timestep in the RNN, but the dataset contains 16 time measurements per experiment. I have tried using 16 timesteps and shaping the variables to optimize productivity, but again, my efforts were in vain. Any guidance on addressing these challenges would be greatly appreciated.

Notebook:

本文标签： pythonPark and Ramirez39s Bioreactor using TensorFlowAn optimization problemStack Overflow