Data Mining for Monitoring Plant Devices Using GMDH and Pattern Classification B.R. Upadhyaya and B. Lu University of Tennessee, Knoxville

CONTENTS 15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 15.2 Description of the Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 15.3 Analysis and Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277 15.4 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278

The application of the group method of data handling technique for modeling dynamic measurements to monitor power plant sensors and field devices is presented. The details of the algorithm, with generalization using rational function approximation, are discussed. The data-driven models are used to generate residuals between the measurements and the predictions of state variables and control functions. This information is then processed for fault detection and isolation using both rule-base analysis and pattern classification techniques. The method is demonstrated by applications to the monitoring of selected devices in a nuclear power plant steam generator system.