Modelling and optimization of the firing process for roller kiln using GAP-RBF neutral network

The firing process of roller kiln consists of several sub-processes and there exists unknown complex nonlinear mapping between the sub-process set points and the final firing quality. To meet this demand, a training algorithm for the radial basis function (RBF) network using GAP method based on the “significance” of a specified neuron is proposed in the paper. The training algorithm which uses GAP method to train the network has a number of advantages such as could be constructed and updated based on the new data sequentially collected from the real process in order to optimize the set point of each sub-process dynamically. Simulation results shows that this training system can work accurately and reliably.