Multi-Network-Feedback-Error-Learning with Automatic Insertion

This work is devoted to present a control application in an industrial process of iron pellet cooking in an important mining company in Brazil. This work employs an adaptive control in order to improve the performance of the conventional controller already installed in the plant. The main strategy approached here is known as Multi-Network-Feedback-Error-Learning (MNFEL). The basic idea in MNFEL is the progressive addition of neural networks in the Feedback-Error-Learning (FEL) scheme. However, this work brings innovation by proposing a mechanism of automatic insertion of new neural networks in MNFEL. In this work, due to the unknown mathematic model of the iron pellet cooking, the plant is simulated by a previously learned neural model. In such simulation environment, the proposed method is compared against conventional PID, FEL and MNFEL.