论文部分内容阅读
对时间序列随机性与神经网络预测精度的关系进行了研究．并将序列模糊滑动平均法用于神经网络预测中的训练样本预处理．以减小序列随机性成分。实例结果表明时间序列经模糊滑动平均处理后，随机波动对神经网络预测精度的影响可大大减小。
The relationship between the randomness of time series and the prediction accuracy of neural networks is studied. The sequence fuzzy sliding average method is used to pre-process the training samples in the neural network prediction. To reduce the randomness of the sequence components. The case study shows that the influence of stochastic fluctuation on the prediction accuracy of neural network can be greatly reduced after the time series is processed by fuzzy sliding average.