Extractions of bifurcation in fractional-order recurrent neural networks under neurons arbitrariness

Abstract

To tackle the subject of a fractional-order neural networks(FONNs) with higher dimensionality and numerous delays, the bifurcations of fractional-order recurrent neural networks(FORNNs) consisting of nonidentical leakage delays and communication delays are developed in this article. The accomplishments with regard to leakage delays and communication delays are extracted, respectively. It discovers that the extraordinary network performance of the devised FORNNs can be acquired when selecting the utilization of a relatively small control delay, while the bifurcations take place and trigger performance degradation once the control delay outweigh its critical value. The proffered measure in this article that multiple delays do not require to be transformed compared with the existing strategies. The correctness of the proposed theoretical analysis is corroborated via picturesque examples.