A New Neural Network Model Based on Attention Mechanism that Embeds LSTM into RNN for Nonlinear Time-Lag System Identification

Concerning the problem of nonlinear system identification under the condition of unknown time-delay, a system identification model for unknown time-delay based on system identification with self attention and temporal attention was proposed, namely Enhance-Attention mechanism Lstm insert Rnn(E-ALIR). On the basis of system identification model for the unknown time-delay system, a combined system identification neural network model was proposed, which can better learn time-delay and capture the dynamic changes pattern of the system. In order to further explore the influence of time lag, the self-attention mechanism and temporal attention mechanism are introduced into the model to capture the dependencies between the data. The experimental results prove that it has a fine effect on the identification of nonlinear systems with unknown time delays.