A Prognosis Methodology Based on Enhanced Lolimot Algorithm Using Historical Data

Failure and breakdown in advanced machinery can be costly. One of the methods to avoid the failures is to use historical data. Using historical data enables us to predict faults and take necessary precautions to minimize the down-fall times of systems and therefore, optimize the overall performance. Predicting the remaining useful life (RUL) is one of the most convenient ways to assess the risk of system performance. LOLIMOT algorithm benefits from a neuro-fuzzy structure that enables it to provide more accurate RUL estimation results than traditional neural networks or SVMs. To further enhance the LOLIMOT algorithm performance, multiple degradation processes are considered in training the prediction module. In this paper, the enhanced LOLIMOT algorithm is used to perform the prognosis of gas turbine engines using only historical data. In the case study, the effectiveness of our proposed method is demonstrated. The results show that more accurate results can be achieved by using neuro-fuzzy structures in prognosis process.