Relibility enhance powertrain using ANFIS base prognostics model

In the past decades power electronics have become more interested devices for underpinning research towards the feasibility of new generation of electrical vehicle (EV) which helping to reduce the reliance on fossil fuels. Power electronic semiconductor devices play an important role in power electronic converter and inverter and rectification systems and design enhance the efficiency of EV performance as well as lowering the cost of electric power propulsion systems. The aim of this paper is to develop a prognostics capability for estimating remaining useful life (RUL) of power electronics components. There is a need for an efficient prognostics algorithm that is embeddable and able to improve on the current prognostic models. A positive aspect of this approach is that the IGBT failure model develops using fuzzy logic adapts prognostic model with the fuzzy nature of failure mechanism. Actually, this method is like adaptive neuro-fuzzy inference system (ANFIS). We also compare the results from the proposed prognostic model with stochastic Monte-Carlo approach which can efficiently estimate the remaining useful life of Insulated Gate Bipolar Transistor (IGBT). The RUL (i.e. mean and confident bounds) is then calculated from the simulated of the estimated degradation states to support on-board real-time decision-making. The prognostics results are evaluated using RMSE prognostics evaluation metrics.