An analysis of the chaotic transition of model muscle tremor mechanism obtained by artificial neural network

This paper describes some experimental results on modeling the chaotic activation underlying certain types of muscular tremor by using artificial neural networks. The existence of chaotic mechanism is confirmed with the appearance of continuous-spectrum via increasing of discrete frequencies by using a method of spectral analysis. The network is a kind of simple recurrent neural network, which is an expansion of an Elman network. The network is simple and useful because it does not require any specific modification of the learning procedure. We examine the behavior of the output signal from the network, and the existence of chaotic activation can be observed. Especially, we show same chaotic phenomena of the network that exhibits chaotic features such as pseudo random vibration, the period doubling bifurcation and Lyapunov exponent. As a result, chaotic behavior in model muscle fiber is obtained by using artificial neural networks.