Multipoint temperature model predictive control with improved MOEA/D

Model predictive control (MPC) algorithm which can predict future states and generate accurate control sequences is adopted to control multiple measurement points collaboratively in a temperature system. To carry out the multiobjective optimization problem (MOP) according to MPC, multiobjective evolutionary algorithms based on decomposition (MOEA/D) and its improved version have been employed. The improved MOEA/D algorithm was compared with its original version through simulation, and the practical effectiveness is verified by tests on a self-made dual-point temperature control system. Finally, the improved MOEA/D algorithm was embedded into the control system of the experimental device, and temperature-tracking control experiments were conducted. The results show that the algorithm applied here can simultaneously control the temperature of two points and keep them close to the target temperature curve. It helps optimize the multiobjective control by virtue of MPC, and particularly opens up new possibilities to achieve optimal temperature control in different engineering applications.