Elamn Network using Simulated Annealing Algorithm and its Application in Thermal Processes Modeling

This paper obtained the connection values of both feedforward and feedback connections of improved Elman networks through simulated annealing algorithm, which can avoid the network trapped into local minimum. The connection values of the feedback don´t need modified manually. The paper established the nonlinear model of thermal processes with the Elman networks. And then the simulation study proved that the Elman network model using simulated annealing algorithm does not need complicated architecture, which make a good match for the thermal processes plant and has good generalization ability.