System Identification and Control using Diagonal Recurrent Neural Networks

This paper presents an approach for control and system identification using Diagonal Recurrent Neural Networks (DRNN). An unknown plant is identified by a Diagonal Recurrent Neuroidentifier (DRNI), and provides the sensitivity information of the plant to a Diagonal Recurrent Neurocontroller (DRNC). A dynamic backpropagation (DBP), is developed to train both DRNC and DRNI. The DRNN captures the dynamic nature of a system and since it is not fully connected, training is much faster than a fully connected recurrent neural network. The architecture of DRNN is a modified model of the fully connected recurrent neural network with one hidden layer. The hidden layer is comprised of self-recurrent neurons, each feeding its output only into itself. The dynamic backpropagation algorithm (DBP) is developed to train the recurrent neural networks for dynamic mappings, which are found in applications such as control, speech processing and sequential cognition. The proposed DRNN architecture is applied to numerical problems and the simulation results are included.