Clustering Algorithm-based Short-term Traffic flow Prediction with CNN and LSTM
To boost traffic volume prediction accuracy and mitigate congestion,a traffic flow prediction model was designed by integrating clustering algorithms with convolutional neural networks(CNN)and long short-term memory networks(LSTM).This model accounts for both the temporal aspects and the connections between traffic patterns and spatial features.Initially,a data tensor X is constructed,and CNN is used to analyze the data X from adjacent intersections,incorporating spatial correlation information.The CNN-processed data is then fed into the LSTM model to predict the next time series values.Next,the LSTM output is categorized into traffic patterns using the K-means clustering algorithm,and clustering centers and weights are calculated.Finally,the LSTM output is weighted to obtain the prediction results.A series of experiments have revealed that the proposed model outperforms other models in terms of prediction accuracy,showcasing the benefits of incorporating spatiotemporal features in short-term traffic flow prediction across different flow patterns.
NETL
NSTL
万方数据
