Neuro-Dynamic Programming
Integrating Guidance into Relational Reinforcement Learning
Machine Learning
Hi-index | 0.00 |
A fed-batch fermentation process is examined in this paper for experimental and further dynamic optimization. The optimization of the initial process conditions is developed for to be found out the optimal initial concentrations of the basic biochemical variables --- biomass, substrate and feed substrate concentration. For this aim, the method of dynamic programming is used. After that, these initial values are used for the dynamic optimization carried out by neuro-dynamic programming. The general advantage of this method is that the number of the iterations in the cost approximation part is decreased.