- About CTTC
- Invention Disclosure
- News & Events
Transcend: Qualitative Diagnosis System & Method
The Modeling and Analysis of Complex Systems group at Vanderbilt University has developed a system and method for qualitative diagnosis for complex dynamic systems. This systematic approach combines qualitative and quantitative information to improve diagnostic accuracy and generate more informed predictions.
• There is a strong demand for real-time and online monitoring, fault isolation and control of complex physical systems for safety reasons and performance issues
• Real-time online requirements create computational burdens and require simplified models
• Parameter and time scale abstraction are used to simplify models; however, they make it difficult to address individual component parameter changes
• In some cases, physical laws governing discrete behavior may be violated by the models when discontinuities occur
This model-based diagnosis of continuous systems combines dynamic system modeling with three integrated modules: monitoring, generation and refinement of fault hypotheses, and prediction. The initial model starts from bond graphs and derives a temporal causal graph of dynamic system behavior. This is used to identify system faults from deviating measurements and predict future behavior. The system compares predictions to observations, monitoring the engineering system until the true fault is isolated. The present system correctly handles algebraic loops with negative feedback effects in order to analyze and predict the behavior of complex physical systems. In the future, the system will include multiple fault and structural fault diagnoses.
• Diagnosis of fault detection and isolation (FDI) for analyzing mixed continuous and discrete behavior of engineering systems (eg: fast breeder reactors, automobile engines, etc.)
Unique Properties and Competitive Advantages
• More informative analysis than traditional methods
• Improved resolution of the prediction
• Combines qualitative and quantitative information to improve diagnostic accuracy
• Real-time online monitoring without the simplification limitations
Intellectual Property Status
US Patent 7,181,374 was issued February 20th, 2007.
For more information go to: http://macs.isis.vanderbilt.edu/
Inventors:Pieter MostermanEric-Jan MandersGautam Biswas