Model free predictive control for a nonlinear biological system with intelligent optimization approach

Today the importance of life science and its related processes are undeniable. Modeling and control of these kind of processes are too complicate because of existence of delay in growth and also nonlinear behavior of micro-organisms. Model predictive control is one of the most popular advanced controlling strategies in this industry, however its dependence on accurate model for predicting future input and output values is limitating. If there is a way that could predict the future values of the process properly, it is possible to overcome to the existing challenges. In this paper we design a model free predictive controller by using a trained recurrent neural network as a predictor for prediction stage at MPC and using GA for solving the associated optimization problem that result the optimal control signal sequence.