Abstract
The problem of in-flight failure-origin diagnosis is addressed by combining aspects of analytical redundancy and artificial intelligence theory. The objective is to use the mathematical model designed to simulate aircraft behavior as a supplement to the knowledge used for diagnosis. A method is developed whereby qualitative causal information about a dynamic system is drawn from its model. Based on sensitivities of the equations of motion to worst-case failure modes, a measure of the relative capacity of system elements to affect one another is derived. A diagnosis procedure combining problem reduction and backward-chaining ordered search uses this knowledge to reduce a list of elements capable of failure to a relatively small list of elements suspected of failure.
Original language | English (US) |
---|---|
Pages (from-to) | 366-375 |
Number of pages | 10 |
Journal | AIAA Paper |
State | Published - 1985 |
All Science Journal Classification (ASJC) codes
- General Engineering