Model-Free State Estimation Using Subspace Identification and Kalman Filter

Abstract

<div class="section abstract"><div class="htmlview paragraph">The model-based design is very much prominent in the vehicle level control system design and state estimation algorithms. It gives the edge to understand and interpret the dynamic systems. Three-way catalytic converter is a thermo-chemical device to convert the toxic oxides into carbon dioxide and water vapor, during this conversion reactions it generates the heat over the catalyst surface. Detailed chemical and thermal model of the catalyst will be able to predict the conversion efficiency, state of stored oxygen <i>(SoX)</i> and oxygen storage capacity <i>(OSC)</i>. As the catalyst get aged, the reaction rates of conversion reactions deteriorate, in results the temperature dynamics also varies which wanes the exothermic heat. In this work, a novel perspective is presented to capture the behavior of <i>SoX</i> and health of the catalytic converter using thermal model analysis of TWC. An equivalent second order multi input single output (MISO) linear sub-space model is identified for the complex detailed thermal model. A second order MISO system is obtained using measured temperature sensor signals across the device. Recursive least square method will be updating the system parameters online then Kalman filter is employed for state estimation. Joint estimation of the hidden state is tested and validated on urban drive cycle with differently aged catalytic converters.</div></div>