Robust methodology for steady state measurements estimation based framework for a reliable long term thermal power plant operation performance monitoring

• Noise variance estimator based in discrete wavelets transform is provided.
• Method for steady state identification calibration is provided.
• A reference model able to detect slowly developing disturbances was built.
• It was applied it to a particular case study, tracking effects of fouling.

Steady state identification is a process control research approximating the successive values of samples in steady state into its average values. According to the plant-wide control hierarchical model, these results implement monitoring and optimizing functions. Thermal power plant operates into a wide range of mean value active power. Systematic plant-wide slow developing disturbances affect the power plant operation performance through deviations of each process variable between its current true process value and the expected good performance relative value. Supervised records are realizations contaminated with stationary correlated noise carrying successive steady state deviations. Long term thermal power plant operation performance monitoring depending on (i) accuracy and precision of steady state identification method and (ii) fitness approximation per process variable versus mean value active power. This paper bases: (i) a computational experiment design to calibrate a steady state identification before to embed into a real system, and (ii) a solution for curve structure to capture good performance relative value per process variable with few knots availability right after the start-up of the plant at base load regime. A case study tracking the cumulative effects of degradation due to fouling on a heat exchanger was performed.