Monitoring strategies for a combined cycle electric power generator

Electric power generation systems require continuous monitoring to ensure safe and reliable operation. The data available from plant sensors supplied to the control systems may also be analyzed to verify proper operation and predict future behavior. In this paper, a combined cycle electric power plant has been monitored using limit and trend checking, reconstructed phase planes, and regression curves for transient and steady-state power generation. Representative experimental results are presented and discussed to illustrate the strengths of the proposed analysis strategies on a 510 MW combined-cycle system and a 180 MW steam turbine. The phase space analysis provides a means of visual inspection of operational anomalies and also offers a context for numerical analysis of the anomalous behavior. The statistical prognostic method provided regression errors below 2.0% for two of the three proposed plant signal combinations. However, all signal combinations offered the opportunity for system monitoring and diagnosis in terms of threshold violations which varied from 2.7% to 5.4% for these two signal sets. Overall, the monitoring strategies exhibited great promise for power generation system applications and merit further study.