基于小波分解和ARIMA-GRU混合模型的外贸风险预测预警研究

Early Prediction and Warning of International Trade Risks Based on Wavelet Decomposition and ARIMA-GRU Hybrid Model

全球经济的不确定性加剧了国际贸易风险,高精度贸易风险预测是保障产业安全、促进经济健康发展的有效途径。本文以贸易竞争力指数刻画对外贸易风险,针对贸易数据非线性、非平稳、强波动且样本量相对较少等数据特点,构建了基于小波分解的差分整合移动平均自回归(ARIMA)和门控循环单元(GRU)的混合预测模型。该模型首先通过小波变换将原始贸易时序数据分解为低频数据和高频数据,继而使用ARIMA和GRU对两频段数据分别建模预测,并对预测值融合,得到贸易竞争力指数的最终预测结果。此外,本文构建了便于贸易风险实践管理的预警机制。为了检验该预警系统的有效性、适用性和实用性,本文以机电产品为例进行实证验证。研究结果表明,与LSTM、GRU、RNN等单模型及混合模型相比,本文提出的预测方法具有更高的预测精度,在短期与长期外贸风险预测中具有良好性能,可为我国的外贸风险管理提供更为科学的决策依据。

The uncertainty of the global economy has largely intensified the international trade risk.An early prediction and warning system for international trade risk is designed based on the trade competitiveness index.Considering the non-linearity,non-stationarity,strong volatility and relatively small sample size of international trade data,a hybrid prediction model with autoregressive integrated moving average model(ARIMA)and gated recurrent unit(GRU)after wavelet decomposition is proposed.Specifically,the trade time series data is decomposed into high-frequency sequence data and low-frequency sequence data through wavelet transform.According to the characteristics of data,the ARIMA-GRU hybrid model is constructed,and the prediction results of each frequency data are ensembled together to get the final prediction result of the trade competitiveness index.In addition,an early warning mechanism is proposed to facilitate the practical management of trade risks.To verify the effectiveness,applicability and practicability of the early prediction and warning system,electromechanical products are taken as an example to conduct the empirical analysis.Comparing with other commonly used single models and hybrid models such as LSTM,GRU and RNN,the results indicate that the proposed method has higher prediction accuracy,demonstrating satisfactory performance in long-term and short-term trade risk prediction.