改进小波包与长短时记忆组合模型的短时交通流预测

为克服非稳定交通流状态下短时交通流预测精度不高、过分依赖大样本历史数据的缺陷，提出一种改进小波包分析和长短时记忆神经网络组合(IWPA-LSTM)的短时交通流预测方法. 利用功率谱细化的思想改进小波包分析算法对小样本交通流时间序列进行多尺度分解和单支重构. 对低频序列和高频序列进行相空间重构，完成长短时记忆模型的逐层构建，实现本地保存并根据预测精度进行自适应更新，将重构的子序列输入模型训练和预测. 将各子序列的预测值叠加输出IWPA-LSTM最终预测值. 实验结果表明，提出的IWPA-LSTM模型在小样本情况下的预测精度优于经典深度学习模型，具有较强的实用性.

Short-term Traffic Flow Forecast Based on Improved Wavelet Packet and Long Short-term Memory Combination Model

This paper proposes a short- term traffic flow prediction method based on improved wavelet packet analysis and long short-term memory neural network combination (IWPA-LSTM) to overcome the shortcomings of short-term traffic flow prediction under the condition of unsteady traffic flow, such as low precision and overreliance on large sample historical data. First, the wavelet packet analysis algorithm was improved by the idea of power spectrum refinement. The wavelet packet algorithm was used to perform multi- scale decomposition and single-branch reconstruction of small sample traffic time series. Next, the phase space of low-frequency sequence and high-frequency sequence is heavy. At the same time, the layer-by-layer construction of the long and short-term memory model was completed, local preservation was performed and adaptively updated according to prediction accuracy. Then the reconstructed subsequence was input into the model for training and prediction. In the last step, the predicted values of each subsequence were superimposed to output the final predicted value of IWPA-LSTM. It was proved that the proposed IWPA-LSTM model produced higher prediction accuracy than the classical deep learning model for small sample size problems, and the practicability of the model was improved significantly.