Neuro-fuzzy system identification for remaining useful life of electrolytic capacitors
Mohammad Behdad Jamshidi, Neda Alibeigi · 2017
The remaining useful life of electrolytic capacitors is most important to guarantee the safety and reliability of the electric systems. The remaining useful life is a nonlinear function, which dramatically changes by different internal and external effects in the capacitors. Soft computing approaches can be used as a powerful tool to analyze data and identify complex systems. Neuro-fuzzy approach is one of the important and widely popular topics in soft computing that are used for nonlinear system identification of dynamic functions. In this paper, an adaptive neuro fuzzy inference system based on subtractive clustering algorithm with 12 inputs is presented for system identification of remaining useful life in the electrolytic capacitors. An experimental dataset is considered to model the remaining useful life, which was provided by the Prognostics Data Repository of NASA. Comparing between simulations and experimental results has illustrated the accuracy of the purposed method to identify nonlinear systems. According to the results, mean squared error of the training and test data are 3.489×10-4and 1.476×10-2respectively.