A New Learning Rate Tuning For Nonlinear System Emulation

An arbitrary choice of neural emulator's learning rate adversely affects the system performances. This paper introduces a new approach to update the learning rate for a Neural Emulator (NE) of the complex processes with unknown behavior. The originality of the proposed method consists in the adaptation of the learning rate by the minimization of a cost function generated by the emulation errors. For carrying out this study, two simulation examples are considered. It is found that performances of the proposed approaches are better than those obtained in previous works.