A Multi-step Prediction Model Based on Interpolation and Adaptive Time Delay Neural Network for Time Series

The drawback of indirect multi-step-ahead prediction is error accumulation. In order to tackle this problem and improve the capacity of adaptive time delay neural network (ATNN) for prediction, a three-stage prediction model SATNN based on spline interpolation and ATNN is presented. With spline interpolation and ATNN, the impact of last prediction errors that would be iterated into the model for the next step prediction is decreased, and then the better prediction can be obtained. The annual sunspot, considered as the benchmark chaotic nonlinear systems, is selected to test the multi-step prediction model. Validation studies indicate that the proposed model is quite effective in multi-step prediction.