Neural Network Macromodeling of Nonlinear Electrical Circuit for Variable Frequency Inputs using Karhunen-Loeve Decomposition

Nagaraj, S.; Seshachalam, D.

doi:10.1109/csitss.2018.8768788

Cited by 2 publications

(1 citation statement)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Considering the calculation multifaceted nature, the NARX neural system with two input layer node, one hidden layer consisting of ten nodes are utilized for the displaying of nonlinear reduced order model behaviour. The model is prepared on Levenberg-Marquardt backpropagation calculation because of its fast convergence capability [28].…”

Section: Machine Learning Through Artificial Neural Network Modelmentioning

confidence: 99%

Variable Frequency Signal Carrying Nonlinear Transmission Line - Modeling using Machine Learning

S.,

2019

IJEAT

View full text Add to dashboard Cite

in modeling of complex systems, manual creation and maintenance of the appropriate behavior is found to be the key problem. Behavior modeling using machine learning has found successful in modeling and simulation. This paper presents artificial neural network (ANN) modeling of transmission line carrying frequency varying signal using machine learning. This work uses proper orthogonal decomposition (POD) based reduced order modeling. In this proposed work, snapshot sets of complex mathematical model of nonlinear transmission line and also linear model are obtained at different time interval. These snapshot sets are arranged in matrix form separately for nonlinear and linear models. POD method is applied on both the matrices separately. This reduces the order of the matrix which is used as input and output data set for neural network training through machine learning technique. Trained neural network model has been verified using different untrained data set. The proposed algorithm determines the dimension of the interpolation space prompting a considerable decrease in the computational expense. The proposed algorithm doesn't force any imperatives on the topology of the appropriate circuit or kind of the nonlinear segments and hence relevant to general nonlinear systems.

show abstract

Section: Machine Learning Through Artificial Neural Network Modelmentioning

confidence: 99%