Fast adaptive observers for battery management systems
Author(s)Jenkins, Benjamin Michael
Massachusetts Institute of Technology. Department of Mechanical Engineering.
Anuradha M. Annaswamy.
MetadataShow full item record
With the proliferation of batteries in transportation, mobile devices and, more recently, large scale energy storage, the demand for new efficient and safe algorithms for battery management has surged. More complex chemistry cells, such as lithium-ion batteries, with their sensitivity to mishandling, misuse and defects as is evident with recent device recalls due to fires, have historically been treated more conservatively. Maximizing performance of these cells safely requires knowledge of internal variables of interest which are not directly measurable. Therefore, accurate models which estimate these variables are needed. The focus of this thesis will be on a modified Single Particle Model (SPM), specifically its internal state estimates. Unfortunately, while the model structure is known, internal parameters which specify it are not, hence, state estimation alone is not enough. This motivates simultaneous state estimation and parameter identification of the electrochemical model. Existing solutions to this task are minimal in the literature. Hence this thesis. This thesis enumerates multiple developments in electrochemical modeling and adaptive observers in general. The first and fundamental component is a modification of the SPM with attractive features such as the encapsulation of lithium diffusion as a linear dynamical system independent of nonlinearities and decoupling of the nonlinear relationships defining the kinetic properties of lithium ion transfer and open circuit potential respectively. A second development defines a set of guidelines reducing the design parameters for adaptive observers to a single tuning parameter, enabling rapid implementation and prototyping. Third, a new variant of adaptive observer, using multiple simultaneous equivalent system representations, is derived for fast parameter convergence. A novel selection of observer design variables and augmentation of the underlying equivalent system with nonlinear basis functions constitutes a fourth development extensively validated through numerical simulation and theory. This adaptive observer combined with an independent offline algorithm to update effective electrode capacity and available lithium adapts every parameter of the modified SP model to account for aging or manufacturing differences. Validation of this observer in hardware using commercially available Panasonic 18650 cells completes the goals originally set forth for this research. The developments presented pave the way for fast, computationally efficient, advanced battery management systems with the potential to increase the effective capacity of a battery or alternatively reduce the size, and therefore cost, of batteries in various applications.
Thesis: Ph. D., Massachusetts Institute of Technology, Department of Mechanical Engineering, 2017.Cataloged from PDF version of thesis. Page 145 blank.Includes bibliographical references (pages 139-144).
DepartmentMassachusetts Institute of Technology. Department of Mechanical Engineering.
Massachusetts Institute of Technology