This paper describes the development and testing of a new algorithm to identify faulty sensors, based on a statistical model using quantitative statistical process history. Two different mathematical models were used and the results were analyzed to highlight the impact of model approximation and random error. Furthermore, a case study was developed based on a real micro gas turbine facility, located at the University of Genoa. The diagnostic sensor algorithm aims at early detection of measurement errors such as drift, bias, and accuracy degradation (increase of noise). The process description is assured by a database containing the measurements selected under steady state condition and without faults during the operating life of the plant. Using an invertible statistical model and a combinatorial approach, the algorithm is able to identify sensor fault. This algorithm could be applied to plants in which historical data are available and quasi steady state conditions are common (e.g. Nuclear, Coal Fired, Combined Cycle).
A statistical model to predict the steady-state creep rate
