Hybrid Concurrent Learning for Hybrid Linear Regression

We consider the problem of estimating a vector of unknown constant parameters for a linear regression model whose input and output signals are hybrid - that is, they exhibit both continuous (flow) and discrete (jump) evolution. Using a hybrid systems framework, we propose a hybrid algorithm capable...

Full description

Saved in:
Bibliographic Details
Main Authors: Johnson, Ryan S., Sanfelice, Ricardo G.
Format: Conference Proceeding
Language:English
Subjects:
Adaptation models
Current measurement
Heuristic algorithms
Linear regression
Noise measurement
Robustness
Simulation
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We consider the problem of estimating a vector of unknown constant parameters for a linear regression model whose input and output signals are hybrid - that is, they exhibit both continuous (flow) and discrete (jump) evolution. Using a hybrid systems framework, we propose a hybrid algorithm capable of operating during both flows and jumps, that utilizes current measurements alongside stored data for adaptation. We show that our algorithm guarantees exponential convergence of the parameter estimate to the true value under a new notion of excitation that relaxes both the classical continuous-time and discrete-time persistence of excitation conditions and a recently proposed hybrid persistence of excitation condition. Simulation results show the merits of our proposed approach.
ISSN:2576-2370
DOI:10.1109/CDC51059.2022.9992473