关键词: 灰色理论, 反向传播（BP）神经网络, 自回归滑动平均（ARIMA）, 二重时间序列, 预测

Abstract: Neural networks，Autoregressive Integrated Moving Average Model（ARIMA） and other methods have been used comprehensively to predict double trend time series，which have trend change nature and periodic fluctuation characteristic.The prediction performances by a sole model are as still far from satisfactory.Taking into consideration this characteristic，a grey combined model is proposed.After the trend and period are separated from double trend time series，the trend is forecasted by grey G（1，1） model，the period is predicted by the combined model which is comprised by BP neural network model and ARIMA model，the two parts predictive value are combined to the final predictive value by the multiplicative model.Experiments show that the grey combined model is adapted to the characteristics of double trend time series，and it has got the best prediction performance.

Key words: grey theory, Back Propagation（BP） neural network, Autoregressive Integrated Moving Average Model（ARIMA）, double trend time series, forecast