基于灰色关联及量子门节点神经网络的时间序列预测

Time Series Prediction Based on Grey Relation and Quantum Gate Node Neural Network

为了有效地解决时间序列的波动性、随机性,以及处理难度大等造成的预测不稳定、预测误差大等问题,提出了结合灰色关联分析与量子门节点神经网络的时间序列预测模型。首先,通过灰色关联分析计算时间序列的主特征值与各影响因子的关联度并根据关联度值进行排序,删除低于关联度阈值的影响因子所对应的原始序列并更新时间序列,以降低待处理的数据量。然后,将更新后的时间序列作为输入,采用基于梯度下降算法的量子门节点神经网络,合理分配训练与测试数据,深入学习时间序列的变化规律,得到预测结果和预测误差。实验结果表明,该组合模型所得到的预测误差的稳定性和精度均优于传统的单一时间序列预测模型,为时间序列预测以及其他预测提供了一种新的思路和方法。

In order to effectively solve the problems of time series,such as fluctuation,randomness,difficulty in processing,which causes predictive instability and great prediction error,a time series prediction model based on grey correlation analysis and quantum gated neural network was proposed.Firstly,the correlation degrees between the main eigenvalue of time series and the influencing factors were calculated by grey correlation analysis and sorted.The original sequence of the influencing factors whose correlation degree was lower than the threshold was deleted and the time series was updated to reduce the amount of data.Then,the updated time series was used as input and quantum gated neural network based on descent algorithm was adopted to distribute training and testing data reasonably,and learn the changing rule of time series deeply.At last,prediction results and prediction errors were obtained.The experimental results showed that the combined forecasting model proposed in this paper is superior to the traditional single time series forecasting model in the stability and accuracy of forecasting errors.It also provides a new idea and method for time series forecasting and other forecasting.