Linearized Versus Nonlinear Observability Analysis for Lithium-Ion Battery Dynamics: Why Respecting the Nonlinearities Is Key for Proper Observer Design

Abstract

Electrochemical model-based observers for battery state estimation play an important role in advanced battery management system design to provide accurate real-time estimates of unmueasurable critical internal variables. In this work, observability of an electrochemical battery model is studied through a thorough nonlinear observability analysis to quantify the quality of estimated variables by accounting for the rank and the condition number of the observability matrix. The paper highlights the need to carry out observability analysis on the nonlinear dynamics rather than the linearized ones. Moreover, the practical aspects of the electrochemical model's observability are investigated and quantified for different input conditions, electrode's chemistry, and discretization grid points of the partial differential equations describing the governing equations of the electrochemical dynamics.