Forecasting World Petroleum Fuel Crisis by Nonlinear Autoregressive Network

Petroleum is an essential commodity in today’s world for the survival of mankind. The petroleum reserve in the world is limited and it is expected to get exhausted in the coming future, and the world is witnessing the crisis of this oil. The prediction of petroleum crisis in the world is a challenging problem to deal with. In this paper, we are using a novel and suitable time series prediction approach based on artificial neural network (ANN) for the forecast of future instances of a time series data. The prediction of petroleum crisis in the near future is obtained by a nonlinear and multistep method known as nonlinear autoregressive network (NARnet). The data set is obtained from different government sources of ten countries (like USA, Middle East countries, China and India, etc.) for over a period of more than 30 years and contains three features, viz. population, petroleum production, and petroleum consumption. The NARnet model requires known instances which are employed for the dynamic multistep prediction for the ahead time where one can predict as many numbers of future data instances as looked-for. The normalized mean square error (NMSE) and T-test of our prediction method, i.e., NARnet, have been verified by different standard predictive methods to handle the data set and found to be better in its performance. It has been forecasted that by 2050, hydrogen fuel (which can reduce pollution and boost the fuel efficiency at lower cost) could be thought of as an appropriate replacement for petroleum.