Model predictive temperature control in long ducts by means of a neural network approximation tool

In this paper, a nonlinear model predictive control (MPC) configuration for hyperbolic distributed thermal systems is presented and applied in the flow-based temperature control in a long duct. At first, a radial basis function neural network is developed to estimate the temperature distribution along the duct with respect to flow velocity, assuming constant ambient temperature. The nonlinear model is then incorporated in the context of an MPC procedure. The use of the neural network model avoids the spatial discretization and decreases significantly the computational effort required to solve the optimization problem that is formulated in real time, compared to conventional modeling approaches. The proposed MPC scheme is able to overcome delay effects and accelerates the outlet temperature response. Reduced tuning effort is another advantage of the proposed control scheme.