Showing papers by "Raghunathan Rengaswamy published in 2013"

Receding-Horizon Nonlinear Kalman (RNK) Filter for State Estimation

Abstract: This technical note presents a new Receding-horizon Nonlinear Kalman (RNK) filter for state estimation in nonlinear systems with state constraints. Such problems appear in almost all engineering disciplines. Unlike the Moving Horizon Estimation (MHE) approach, the RNK Filter formulation follows the Kalman Filter (KF) predictor-corrector framework. The corrector step is solved as an optimization problem that handles constraints effectively. The performance improvement and robustness of the proposed estimator vis-a-vis the extended Kalman filter (EKF) are demonstrated through nonlinear examples. These examples also demonstrate the computational advantages of the proposed approach over the MHE formulation. The computational gain is due to the fact that the proposed RNK formulation avoids the repeated integration within an optimization loop that is required in an MHE formulation. Further, the proposed formulation results in a quadratic program (QP) problem for the corrector step when the measurement model is linear, irrespective of the state propagation model. In contrast, a nonlinear programming problem (NLP) needs to be solved when an MHE formulation is used for such problems. Also, the proposed filter for unconstrained linear systems results in a KF estimate for the current instant and smoothed estimates for the other instants of the receding horizon.

Traffic of pairs of drops in microfluidic ladder networks with fore-aft structural asymmetry

Abstract: We investigate the dynamics of pairs of drops in microfluidic ladder networks with slanted bypasses, which break the fore-aft structural symmetry of the network. Our analytical results indicate that unlike symmetric ladder networks, structural asymmetry introduced by a single slanted bypass can be used to modulate the relative drop spacing, enabling them to contract, synchronize, expand, or even flip at the ladder exit. Our experiments confirm all these behaviors predicted by theory. Numerical analysis further shows that while ladder networks containing several identical bypasses are limited to nearly linear transformation of input delay between drops, combination of forward and backward slanted bypasses can cause significant nonlinear transformation enabling coding and decoding of input delays.

Online fault diagnostics and impedance signature mapping of High Temperature PEM fuel cells using rapid small signal injection

Abstract: The recent development of electric vehicles and electronic equipment has resulted in a demand for portable power supply systems with high power density and life expectancy. High Temperature (HT) Proton Exchange Membrane (PEM) fuel cells have emerged as a possible solution to this requirement. Recent advances in the materials allow the cells to operate at higher temperatures resulting in a number of advantages. These systems still suffer from degradation mechanisms that can degrade performance and decrease the life expectancy. In order to diagnose the fuel cell performance and degradation mechanisms, impedance analysis can be performed. This paper presents a design for the power electronic hardware that can perform these analyses online for single cell applications, where the output voltage is very low. Special consideration must be given to the hardware selection and small signal injection capability. The Nyquist plots and small amplitude current step voltage response of the HTPEM fuel cell is presented along with those captured after catalyst degradation has occurred. It is shown how the online system can capture these changes by using equivalent circuits.

Electrical circuit analysis of CO poisoning in high temperature PEM fuel cells for rapid fault diagnostics

Abstract: High Temperature PEM fuel cells are able to withstand a substantial amount of CO poisoning while still maintaining stable output. High concentrations of CO will however degrade performance and can lead to instability. This paper presents a detailed study on the impact of CO poisoning on the performance of High Temperature PEM fuel cells. A dedicated test setup was constructed to perform both steady state and dynamic analysis on the fuel cell under a wide range of operating conditions and variations in CO content. The drop in performance captured in the polarization curves is modeled using a simple circuit model with a dedicated fault element. The captured impedance spectra from the electrochemical impedance spectroscopy (EIS) tests provide insight to the changes in the electrochemical circuit parameters during CO poisoning that can be used to track the fuel cell state of health. The results from this study will allow for the development of a rapid diagnostic control system that can be implemented in the power conditioning unit of the fuel cell.

Degradation of high temperature PEM fuel cells and the impact on electrical performance

Abstract: This paper presents the effect different degradation mechanisms have on the electrical performance of a high temperature PEM fuel cell. The changes to the fuel cell composition and internal chemistry is related the drop in electrical performance. In particular, three main effects will be discussed. This includes the impact of carbon support and catalyst degradation, loss of phosphoric acid in the membrane and CO poisoning of the catalyst layer.

A generative approach to qualitative trend analysis for batch process fault diagnosis

Abstract: Most of the existing methods for qualitative trend analysis are based on discriminative models. A disadvantage of such models is that many heuristic rules or local search methods are needed. Recently, an effort has been made to develop a globally optimal method for qualitative trend analysis. This method is based on a generative (rather than discriminative) model and has shown to lead to excellent performance. However, this method comes at an extreme computational demand which renders the methods unlikely for on-line application. In this work, an alternative method, while still generative in nature, is proposed which is shown to deliver the same performance while reducing the computational demand considerably.