Time-series interval prediction under uncertainty using modified double multiplicative neuron network

•A hybrid approach for prediction intervals (PIs) estimation of terrain profiles is proposed.•It integrates the double multiplicative neuron (DMN) model and the modified PSO (MPSO).•MPSO adjusts parameters of DMN by minimizing the value of the proposed cost function.•It improves the prediction accuracy for terrain trends by 15.4% in the testing data,•It reduces the computational burden by 8% in the testing data over the LUBE method.