محمد بهارشاهي

Determining the state of degradation, diagnosis and estimation of remaining useful life (RUL) are two main activities in prognostics and health management (PHM). Analyzing the results of these two activities can lead to increasing the reliability and safety of the system and reducing unnecessary maintenance costs. In this thesis, new methods based on evidence theory are presented for diagnostics and simultaneously determining the state of degradation and estimation of remaining useful life. In the proposed method for diagnostics, the belief model based on rules and evidence theory was used to detect and isolate the types of turbine faults, as well as estimate their severity for gas turbine. In the second model, to estimate the remaining useful life, the behavior of the turbofan was investigated by using clustering techniques and artificial neural networks, and its similarity with other damaged turbines was calculated. The final RULs of the turbine was obtained from the extension of the Dempster-Shafer integration rules, based on similarity. The presented method is effective for problems in which a huge set of data with large dimensions is available. In order to eva‎luate the proposed methods, the real data set of the TurboTec Co. has been used for the fault diagnosis model and the turbofan engine data set provided by NASA for the RUL estimation model. The second data set is widely used in the literature. The results show that the performance of the proposed methods in diagnostic and predicting the RUl is significantly superior to the previous approaches in the literature.

عمر مفيد باقيمانده , تلفيق اطلاعات , پيش¬بيني , عيب يابي , تئوري شواهد , دمپستر-شفر , توربين گاز

Remaining useful life , integration of information , prognostic , diagnostic , evidence theory , Dempster-Shafer , gas turbine

Mohammad Baharshahi

seyedmohammad seyedhosseini