Publicação

Modeling of complex nonlinear dynamic systems using temporal convolution neural networks

Ver documento

Detalhes bibliográficos
Resumo:An increasingly important class of nonlinear systems includes the nonaffine hybrid systems, in particular those in which the underlying dynamics explicitly depends on a switching signal. When the inherent complexity is treatable and the phenomena governing the system dynamics are known an implicit model can be derived to describe its behaviour over time. Conversely, when these assumptions are not met the system dynamics can still be approximated by regression-based techniques, provided a dataset comprising inputs and outputs collected from the system is available. One approach to deal with data driven modelling relies on computational intelligent frameworks, in which artificial neural networks stand out as a prominent class of universal approximation black box models. This work aims to explore 1D Convolutional Neural Networks capabilities, in which the inputs are represented by regressors and structural configuration parameters, to modelling nonlinear hybrid dynamic systems. Moreover, in order evaluate the intrinsic ability to transparently approximate hybrid dynamics, this deep neural network architecture is compared to a shallow multilayer layer perceptron framework, in which each structural configuration is independently approximated.
Autores principais:Bastos, Vasco Miguel Pessoa
Assunto:Nonlinear hybrid systems switching systems data driven modelling convolutional neural network multilayer perceptron
Ano:2022
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso aberto
Instituição associada:Universidade Nova de Lisboa
Idioma:inglês
Origem:Repositório Institucional da UNL
Descrição
Resumo:An increasingly important class of nonlinear systems includes the nonaffine hybrid systems, in particular those in which the underlying dynamics explicitly depends on a switching signal. When the inherent complexity is treatable and the phenomena governing the system dynamics are known an implicit model can be derived to describe its behaviour over time. Conversely, when these assumptions are not met the system dynamics can still be approximated by regression-based techniques, provided a dataset comprising inputs and outputs collected from the system is available. One approach to deal with data driven modelling relies on computational intelligent frameworks, in which artificial neural networks stand out as a prominent class of universal approximation black box models. This work aims to explore 1D Convolutional Neural Networks capabilities, in which the inputs are represented by regressors and structural configuration parameters, to modelling nonlinear hybrid dynamic systems. Moreover, in order evaluate the intrinsic ability to transparently approximate hybrid dynamics, this deep neural network architecture is compared to a shallow multilayer layer perceptron framework, in which each structural configuration is independently approximated.