Forward-backward kernel-based state and parameter estimation for linear systems of arbitrary order
Previous results pertaining to algebraic state and parameter estimation of linear systems based on a special construction of a forward-backward kernel representation of linear differential invariants are extended to handle large noise in output measurement. Explicit expressions for the kernel functions for single-input, single-output (SISO) linear time-invariant (LTI) systems of arbitrary order are first stated and proved. A two-stage solution of the estimation problem is proposed next. The parameter estimation sub-problem, which is the task of the first stage, is solved by way of stochastic regression. The parameter estimation method presented in earlier publications is enhanced using multiple regression. The regression model does not satisfy the assumptions of the Gauss-Markov theorem in that the random regressor is heteroskedastic. This does not impede achieving high accuracy of estimation. A recursive version of a feasible generalized least squares with covariance weighting is employed to attenuate adverse effects due to heteroskedasticity. The output of the system and its time derivatives are reconstructed smoothly in the second stage of the approach by way of projection, while minimizing a mean square criterion.
Code (0)
등록된 구현이 없습니다.
Tasks
parameter estimationregressionSimilar Papers 제목 키워드 기반
Dual NUP Representations and Min-Maximization in Factor Graphs
Normals with unknown parameters (NUP) can be used to convert nontrivial model-based estimation problems into iterations of linear least-squares or Gaussian estimation problems. In this paper, we extend this approach by a…
A Kernel Learning Method for Backward SDE Filter
In this paper, we develop a kernel learning backward SDE filter method to estimate the state of a stochastic dynamical system based on its partial noisy observations. A system of forward backward stochastic differential …
Bayesian InferenceHighly Efficient Forward and Backward Propagation of Convolutional Neural Networks for Pixelwise Classification
We present highly efficient algorithms for performing forward and backward propagation of Convolutional Neural Network (CNN) for pixelwise classification on images. For pixelwise classification tasks, such as image segme…
ClassificationGeneral Classificationimage-classificationImage Classification+4Learning Bloch Simulations for MR Fingerprinting by Invertible Neural Networks
Magnetic resonance fingerprinting (MRF) enables fast and multiparametric MR imaging. Despite fast acquisition, the state-of-the-art reconstruction of MRF based on dictionary matching is slow and lacks scalability. To ove…
Magnetic Resonance Fingerprintingparameter estimationConvergence analysis of kernel learning FBSDE filter
Kernel learning forward backward SDE filter is an iterative and adaptive meshfree approach to solve the nonlinear filtering problem. It builds from forward backward SDE for Fokker-Planker equation, which defines evolving…