A Hybrid Forecasting Model for Non-Stationary Time Series

Abstract

In time series analysis, an important problem is how to extract the information hidden in the non-stationary and noise data and combine it into a model for forecasting. In this paper, the authors propose a TEI@I based hybrid forecasting model. A novel feed forward neural network is developed based on the improved particle swarm optimization with adaptive genetic operator (IPSO-FNN) for forecasting. In the proposed IPSO, inertia weight is dynamically adjusted according to the feedback from particles’ best memories, and acceleration coefficients are controlled by a declining arccosine and an increasing arccosine function. Subsequently, a crossover rate which only depends on generation and an adaptive mutation rate based on individual fitness are designed. The parameters of FNN are optimized by binary and decimal particle swarm optimization. Further, the forecast results of IPSO-FNN are adjusted with the knowledge from text mining and an expert system. The empirical results on the container throughput forecast of Tianjin Port show that forecasts with the proposed method are much better than some other methods.