Gaussian Processes are a powerful tool that is ﬁnding successful applications in the Control Systems ﬁeld due to their high ﬂexibility and ability to identify dynamical systems. In a general context, Gaussian Process Regression is a Bayesian nonparametric method which makes predictions based on the posterior probability density given training data. The behavior of the predictions (regressed function) is completely characterized by the covariance function, whose properties are directly inherited by the regressed function. In many problems, such function is chosen ad hoc and depends on prior knowledge of the underlying phenomena, i.e. depends on human intervention. LearningBased NonLinear Model Predictive Control (LBNMPC) scheme is a datadriven implementation of the MPC used to control nonlinear dynamical systems while satisfying plant constraints, it is aimed at improving the nominal model available to the controller by approximating underlying missing dynamics with a Gaussian process using past plant observations as training data. We implement the socalled Generalized Kernels to the Gaussian process within the LBNMPC, exploiting the spectral representation of a positivedeﬁnite function that is modelled as mixture of a basis function. Such basis function can have desired properties such as smoothness or degree of diﬀerentiability, generating richer model classes that can be used to identify more complex dynamical systems. This procedure automatizes the selection of the kernel function, while still maintaining an analytic expression of the regressed function, allowing its implementation on the MPC scheme and, in particular, it has a strong link with RKHS framework that is useful to analyze the model class rigorously.
Gaussian Processes are a powerful tool that is ﬁnding successful applications in the Control Systems ﬁeld due to their high ﬂexibility and ability to identify dynamical systems. In a general context, Gaussian Process Regression is a Bayesian nonparametric method which makes predictions based on the posterior probability density given training data. The behavior of the predictions (regressed function) is completely characterized by the covariance function, whose properties are directly inherited by the regressed function. In many problems, such function is chosen ad hoc and depends on prior knowledge of the underlying phenomena, i.e. depends on human intervention. LearningBased NonLinear Model Predictive Control (LBNMPC) scheme is a datadriven implementation of the MPC used to control nonlinear dynamical systems while satisfying plant constraints, it is aimed at improving the nominal model available to the controller by approximating underlying missing dynamics with a Gaussian process using past plant observations as training data. We implement the socalled Generalized Kernels to the Gaussian process within the LBNMPC, exploiting the spectral representation of a positivedeﬁnite function that is modelled as mixture of a basis function. Such basis function can have desired properties such as smoothness or degree of diﬀerentiability, generating richer model classes that can be used to identify more complex dynamical systems. This procedure automatizes the selection of the kernel function, while still maintaining an analytic expression of the regressed function, allowing its implementation on the MPC scheme and, in particular, it has a strong link with RKHS framework that is useful to analyze the model class rigorously.
Applicazione di Kernel Generalizzati al LearningBased NonLinear Model Predictive Control
FERNANDEZ GALINDO, DIEGO
2021/2022
Abstract
Gaussian Processes are a powerful tool that is ﬁnding successful applications in the Control Systems ﬁeld due to their high ﬂexibility and ability to identify dynamical systems. In a general context, Gaussian Process Regression is a Bayesian nonparametric method which makes predictions based on the posterior probability density given training data. The behavior of the predictions (regressed function) is completely characterized by the covariance function, whose properties are directly inherited by the regressed function. In many problems, such function is chosen ad hoc and depends on prior knowledge of the underlying phenomena, i.e. depends on human intervention. LearningBased NonLinear Model Predictive Control (LBNMPC) scheme is a datadriven implementation of the MPC used to control nonlinear dynamical systems while satisfying plant constraints, it is aimed at improving the nominal model available to the controller by approximating underlying missing dynamics with a Gaussian process using past plant observations as training data. We implement the socalled Generalized Kernels to the Gaussian process within the LBNMPC, exploiting the spectral representation of a positivedeﬁnite function that is modelled as mixture of a basis function. Such basis function can have desired properties such as smoothness or degree of diﬀerentiability, generating richer model classes that can be used to identify more complex dynamical systems. This procedure automatizes the selection of the kernel function, while still maintaining an analytic expression of the regressed function, allowing its implementation on the MPC scheme and, in particular, it has a strong link with RKHS framework that is useful to analyze the model class rigorously.File  Dimensione  Formato  

Fernandez_Diego.pdf
accesso aperto
Dimensione
12.61 MB
Formato
Adobe PDF

12.61 MB  Adobe PDF  Visualizza/Apri 
The text of this website © Università degli studi di Padova. Full Text are published under a nonexclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/40466