A multi-step predictor for dynamic system property forecasting

A reliable multi-step predictor is very useful to a wide array of industries to forecast the behavior of dynamic systems. In this paper, an adaptive predictor is developed based on a novel weighted recurrent neuro-fuzzy paradigm to forecast properties of dynamic systems. An online training technique is proposed to improve forecasting convergence and accommodate different operating conditions. The viability of the developed predictor is firstly evaluated based on benchmark data sets, and then it is implemented for real-time machinery system monitoring. The monitoring index is derived from measurement based on a beta kurtosis reference function. The investigation results show that the developed adaptive predictor is a reliable forecasting tool and is able to accommodate different system conditions. It can capture the system's dynamic behavior quickly and track the system's characteristics accurately. Its performance is superior to other classical forecasting schemes.