Due to the quickly developing economy and improving living standards, it has become a problem that urban traffic roads are not able to meet the needs of such a great amount of motor vehicles. The condition of a traffic system is sensitive to the distribution of traffic flow, which can be directly led by the signal lamps. In this study, we propose a novel architecture of neuron network, CNN-LSTM (convolution neuron network-long short-term neuron network), which puts both spatial and temporal corresponding into consideration. A deep convolutional neuron network is utilized to capture the features among data in different lanes and a long short-term memory neuron network is used to capture the temporal features in time sequences. A classifier is applied to determine which assignment strategy to choose. A comparison with other models suggests that our deep learning method is superior to other methods with high accuracy.
展开▼
