Artificial Intelligence / Expert Systems Expertise

Development Areas

• AI/Expert systems for the diagnosis and management of thermal-hydraulic process systems
  • Qualitative analysis for IF-THEN rule development
  • Component-characteristic-based neural network representations
  • Two-level hierarchical hybrid expert system/neural network for process fault diagnosis
• Rule generation from process physics modeling
• Development of process models that run in real time
• Neural net development for process simulation

Equipment/Software Tools

• QUINTUS Prolog development system on Sun Microsystems, Inc. workstations

Accomplishments

• Development of the MSET technology for on-line validation of process parameters, plant state verification and fault identification
• Development of process models for real time analysis
• PRODIAG: plant independent AI development system for process fault diagnosis
  The only required plant specific input is the piping and instrumentation diagram information. This benefit is derived from a patented Argonne technique for thermal-hydraulic (T-H) function classification and component characteristics representation, which allows PRODIAG to be applied to T-H systems in power plants, both nuclear and fossil, the chemical processing industry and petroleum refineries with no need for additional verification and validation.
• "System Diagnostics Using Qualitative Analysis and Component Functional Classification", U.S. patent issued
• "Combined Expert System/Neural Networks Method for Process Fault Diagnosis", U.S. patent issued
• "Process Management Using Component Thermal-Hydraulic Function Classes," U.S. patent issued