Paper Title
An Adaptive NF Predictor For Gear System State Forecasting

Abstract
A reliable predictor is very useful in industries to forecast the future states of a dynamic system. In this paper, an adaptive predictor is developed based on a weighted neuro-fuzzy (NF) system to forecast properties of dynamic gear systems. An online training technique is adopted to improve forecasting convergence and adaptive capability. The effectiveness of the developed predictor is evaluated based on some benchmark data set; then it is implemented for gear system monitoring. The monitoring index is derived from measurement based on a beta kurtosis reference function. The investigation results show that the developed adaptive NF predictor is a reliable forecasting tool. It can capture the system’s dynamic behavior quickly and track the system’s characteristics accurately. Index Terms - Neuro-fuzzy (NF) system, system state prediction, machine learning, gear system condition monitoring.