Reliable Optimal Control of a Fed-Batch Bio-Reactor Using Ant Colony Optimization and Bootstrap Aggregated Neural Networks

Abstract Optimal control of a fed-batch bio-reactor using ant colony optimisation and bootstrap aggregated neural network models is presented in this paper. In order to overcome the difficulties in developing detailed mechanistic models and to improve the reliability of data based empirical models, bootstrap aggregated neural networks were used to model a fed-batch bio-reactor using process operational data. Bootstrap aggregated neural networks can not only improve model prediction accuracy but also provide prediction confidence bounds. In order to overcome the problem of local minima in the optimisation, ant colony optimisation (ACO) is used. A modified ACO algorithm is proposed for continuous variable optimisation. In the proposed technique, model prediction confidence bounds are incorporated in the optimisation objective function so as to enhance the reliability of the calculated “optimal” control actions.