Dynamic-Order-Extended Time-Delay Dynamic Neural Units

Abstract

The paper introduces a linear dynamic-order-extended time-delay dynamic neural unit, which is one possible modification of novel class of artificial neurons called time-delay dynamic neural units (TmD-DNU). In standalone implementations, these artificial dynamic neural architectures can be understood as an analogy to continuous time-delay differential equations. TmD-DNU is capable of identification of all parameters of continuous time differential equation including unknown time delays both in the unit's inputs as well as in its state variable. A modification of dynamic backpropagation learning algorithm is shown. Results on system identification of an unknown system with dynamics of higher-order including unknown time delays are shown in comparison to achievements by common identification methods applied to the same system. Robust identification capabilities and network implementations of TmD-DNU are briefly discussed