A novel locally connected recurrent neural network for identification of nonlinear dynamical system
R. Shobana, Rajesh Kumar, Bhavnesh Jaint · 2023
In this work, a novel Locally Connected Recurrent Neural Network structure (LLCRNN) is designed for the identification of nonlinear systems. The structure is an extension of a local recurrent neural Network with better dynamic extraction capability. The weights are updated using the offline mode of the gradient descent algorithm. The proposed structure is found to give better identification as compared to other considered neural Network structures such as Feed Forward Neural Network (FFNN) and Local Recurrent Neural Network structure (LRNN). The efficiency of the proposed structure is proved against the selected structures.