Skip to Main Content
This paper presents an integrated method of intelligent decoupling control as a solution to the problem of adjusting the zone temperatures in a regenerative pusher-type reheating furnace. First, a recurrent neural network (NN) for estimating the zone temperatures and a heat transfer model for predicting billet temperatures are built based on data from actual furnace operations. Next, a decoupling strategy in combination with a fuzzy NN is used to control the zone temperatures. The architecture of the controller is based on a fuzzy c-means clustering approach; and the weights are optimized by a hybrid particle swarm optimization (HPSO) algorithm, which integrates the global optimization of density-based selection and the precise search of clonal expansion in an immune system with the fast local search of particle swarm optimization. HPSO is also used to optimize the zone temperature settings to minimize three items: fuel consumption, the temperature gradient within a billet, and the error between the mean and target temperatures of a billet at the furnace exit. The results of actual runs demonstrate the validity of this method.