Title :
Automatic rebuilding of qualitative models for diagnosis
Author :
Vinson, J.M. ; Grantham, Stephen D. ; Ungar, Lyle H.
Author_Institution :
Dept. of Chem. Eng., Pennsylvania Univ., Philadelphia, PA, USA
Abstract :
The diagnosis system with automated rebuilding of models, or Disarm, which uses a hypothesis-and-test strategy to identify faults in chemical plants, is described. The qualitative models in Disarm describe chemical plant behavior at an abstract level, allowing diagnosis of a wide range of faults with less explicit modeling of possible faults. The model-building approach is based on the qualitative process theory in which process units are described by the processes or phenomena that occur within them. Disarm uses these process models to discover what changes have occurred in the plant to cause the observed discrepancies. The experience-based portion of Disarm is a series of general rules that order the hypothesized faults, speeding fault detection.<>
Keywords :
chemical engineering computing; chemical industry; fault location; knowledge based systems; Disarm; automated rebuilding; chemical plant behavior; diagnosis system; fault detection; fault identification; hypothesis-and-test strategy; qualitative models; qualitative process theory; rules; Chemicals; Circuit faults; Coolants; Fault detection; Fault diagnosis; Heat transfer; Inductors; Integrated circuit modeling; Parameter estimation; Temperature;
Journal_Title :
IEEE Expert