A Novel Set-valued Observer Based State Estimation Algorithm for Nonlinear Systems

This study considers the state estimation problem for nonlinear models with unknown but bounded noises. A zonotopic set-valued observer based state estimation algorithm is proposed, and the unknown noise term is wrapped in a zonotope during each recursive step. The second-order polynomial Stirling interpolation improves the linearization accuracy and reduces the calculation amount. The method that combines sequence updating and tightening strips reduces the accumulation of errors and improves the estimation accuracy. Finally, the simulations on the Van der Pol nonlinear model and spring-mass-damper nonlinear model can visually illustrate the feasible parameter set variation process and motion trail of the zonotope, which demonstrates the effectiveness and accuracy of the proposed algorithm.

Automated summary

This study proposes a zonotopic set-valued observer based state estimation algorithm for nonlinear models with unknown but bounded noises. The algorithm wraps the unknown noise term in a zonotope during each recursive step, and uses second-order polynomial Stirling interpolation and a method that combines sequence updating and tightening strips to improve the estimation accuracy and reduce error accumulation.