Reliable optimal control of a fed-batch fermentation process using ant colony optimisation and bootstrap aggregated neural network models

This paper presents a reliable optimal control strategy for a fed-batch fermentation process using ant colony optimisation and bootstrap aggregated neural network models. Bootstrap aggregated neural networks are used to enhance model accuracy and reliability. A further advantage of bootstrap aggregated neural network is that model prediction confidence bounds can be calculated from individual network predictions. The objective function of fed-batch fermentation process optimisation based on neural network models typically contains multiple local minima and traditional gradient based optimisation may be trapped in a local minimum. In order to overcome this problem, ant colony optimisation is used. The optimisation objective function is modified to incorporate model prediction confidence in order to enhance the reliability of the calculated “optimal” control policy. Application results on a simulated fed-batch fermentation process demonstrate that the proposed strategy is very effective.