Particle filter based prognosis study for predicting remaining useful life of steam generator tubing

Steam Generator (SG) tubes are used to transfer heat from the primary side to pressurized steam on the secondary side in a nuclear power plant. Eddy current inspection of SG tubes is performed periodically and the data is analyzed to detect and characterize a variety of flaw categories in different locations of the SG tube geometry. This paper uses the data analysis results to estimate remaining useful life (RUL) of SG tubes in the nuclear reactor. Knowledge of RUL will ensure corrective action before the occurrence of tube leakage which may lead to an accident. A Monte Carlo based recursive algorithm known as particle filter is used to predict the RUL of the steam generator tubes. The uncertainty involved in prediction of remaining useful life of tubes using particle filter algorithm is controlled by an RUL correction loop based on an auto-regressive algorithm. The technique is applied to actual eddy current inspection data of steam generator tubing acquired at successive intervals to calculate RUL. The predicted RUL is compared to the actual failure time of the tube to show the efficacy of the technique.