Nonlinear System Identification: Learning while respecting physical models using a sequential Monte Carlo method
arXiv:2210.14684 · doi:10.1109/MCS.2021.3122269
Abstract
Identification of nonlinear systems is a challenging problem. Physical knowledge of the system can be used in the identification process to significantly improve the predictive performance by restricting the space of possible mappings from the input to the output. Typically, the physical models contain unknown parameters that must be learned from data. Classical methods often restrict the possible models or have to resort to approximations of the model that introduce biases. Sequential Monte Carlo methods enable learning without introducing any bias for a more general class of models. In addition, they can also be used to approximate a posterior distribution of the model parameters in a Bayesian setting. This article provides a general introduction to sequential Monte Carlo and shows how it naturally fits in system identification by giving examples of specific algorithms. The methods are illustrated on two systems: a system with two cascaded water tanks with possible overflow in both tanks and a compartmental model for the spreading of a disease.
arXiv comment52 pages, 13 figures
References in corpus (19)
- Discovering governing equations from data: Sparse identification of nonlinear dynamical systems
- The pseudo-marginal approach for efficient Monte Carlo computations
- Pyro: Deep Universal Probabilistic Programming
- Church: a language for generative models
- On Particle Methods for Parameter Estimation in State-Space Models
- Edward: A library for probabilistic modeling, inference, and criticism
- Can local particle filters beat the curse of dimensionality?
- Sequential Monte Carlo smoothing with application to parameter estimation in non-linear state space models
- Particle Gibbs with Ancestor Sampling
- Maximum a Posteriori Estimators as a Limit of Bayes Estimators
- Stochastic (Approximate) Proximal Point Methods: Convergence, Optimality, and Adaptivity
- Uniform ergodicity of the Particle Gibbs sampler
- Variational Sequential Monte Carlo
- Probabilistic learning of nonlinear dynamical systems using sequential Monte Carlo
- Controlled Sequential Monte Carlo
- Twisted particle filters
- Linearly Constrained Neural Networks
- Functional Tensors for Probabilistic Programming
- Learning dynamical systems with particle stochastic approximation EM