A Novel Recurrent Neural Network for Dynamic Process Modeling with Inertia and Delay

Abstract

A novel continuous-time multi-layer Recurrent Neural Network (RNN) is presented in this paper. The proposed RNN struture has superiorities of simple structure and parameters with certain physical meanings. A four-neuron dynamic neural network was used to model the water spray desuperheating control system. The test results show that the proposed RNN sturture has good adaptability to the physical process with large inertia and large delay effects.