Current Research
GOALI: Model-Based System Fault Diagnosis and Prognosis - Passive Robustness and Aging Prediction.
Principal investigators:
Pierluigi Pisu, Clemson University, Mechanical Engineering
Giorgio Rizzoni, Ohio State University, Director of the Center for Automotive Research and
Professor of Mechanical and Electrical Engineering
Graduate Students:
Sponsors:
NSF
Brief abstract:
This project is part of a collaborative effort with Ohio State University and GM. Clemson University will provide the following major contributions to the research: a) the formulation of a method to reduce the computational complexity and efficiently implement adaptive thresholds; b) the development of an approach for the calibration of adaptive thresholds based on the type of uncertainties; c) the identification of a minimal test procedure for an automotive electrical system or minimal series of virtual and physical tests, based on the uncertainty, that starting from a complex simulator or the physical system allows to automatically determine suitable upper-bounds for un-modeled dynamics.
This collaboration will result in a three-fold contribution to the state of the art in the field: 1) we will develop realistic and viable model-based, modular residual evaluation algorithms based on adaptive thresholds to account for unavoidable model uncertainty; 2) we will develop a systematic methodology for the calibration of these algorithms in a large number of systems; and 3) we will develop a methodology for a quantitative evaluation of system aging by way of model-based prognostic algorithms.
Impact:
The outcome of this work will have a broad impact on the theory, methodology and application of fault diagnosis and prognosis to engineering systems, beyond the application considered more specifically in this proposal. The participation of GM is critical in that it provides an opportunity to validate and implement the proposed methodology in a real-world setting that presents many important challenges such as cost constraints, the application to a large number of systems (potentially millions of vehicles), and the need to achieve a degree of robustness that is consistent with today’s consumer expectations of warranty.
Project schedule:
Year 1
1. Theoretical study of order reduction in adaptive threshold and sensitivity trade-off.
2. Design of residual generation units for single fault isolation for integrated diagnosis of the electrical power generation storage system, including alternator and battery.
3. Adaptive thresholds generation for the electrical system.
4. Discrete-event realization of adaptive thresholds.
5. Simulations.
Year 2
1. Implementation of discrete-event realization for adaptive thresholds.
2. Theoretical study of threshold calibration/synchronization
3. Extension of adaptive thresholds generation for different uncertainty types.
4. Simulations.
Year 3
1. Design of tests to characterize parameter bounds.
2. Integration of diagnosis and prognosis.
3. Development of a joint graduate course between OSU and CU.
4. Publications of the research results in journals and conferences.
