Contributions to Systems Science and Engineering by Prof. Kazantzis Are Taught at Harvard
Department(s):
Chemical EngineeringThe accurate reconstruction of the state (and the reliable characterization of the dynamic behavior) of a complex process system assumes central importance in the fields of systems science and engineering. In particular, the design of advanced process system performance monitoring and automatic control strategies, as well as chemical risk assessment/management protocols and environmental regulatory policy development depend on this scientific endeavor. For such a task, a powerful tool known as state observer is appropriately designed and digitally implemented with the aid of a computer code, offering an accurate reconstruction of performance-critical variable profiles. A method developed by Prof. Kazantzis and collaborators, led to a rigorous development of a nonlinear analogue of the well-known Luenberger observer in linear systems theory, and in particular, an insightful design template of a nonlinear state observer capable of reliably reconstructing the state of a dynamical system in the presence of irreducible nonlinearities. It should be pointed out that the conventional “linear approximation approach” exhibits limited validity and often leads to unsatisfactory process performance monitoring and/or control strategies. The novel nonlinear state observer design method (currently recognized as the Kazantzis-Kravaris-Luenberger (KKL)-observer in the nonlinear systems literature) has been recently integrated into various machine-learning algorithmic frameworks through the creative efforts of numerous other research groups around the world. It has been also recently incorporated into a textbook and a set of notes developed by Prof. H. Yang for a graduate course entitled: “Optimal Control and Estimation” offered at the J. A. Paulson School of Engineering and Applied Sciences (SEAS), Harvard University: