Short⁃term traffic flow state prediction plays an important role in realizing urban intelligent transportation system. Many neural network models have been proposed to predict traffic flow in the past,and the effects are unsatisfied. The reason for this is that most models learning uses shallow model. Since shallow model is liable to sink into local extremum and unable to simulate more complicated arithmetical operation,it is not suitable for simulating actual traffic condition. As a new branch of ma⁃chine learning,deep learning has made great success in the field of voice and image processing. It can learn valid features for prediction from data sets in an unsupervised way. Deep learning is applied to prediction urban main road traffic flow by modeling. The experimental results show that this method has achieved better traffic flow prediction effect.%短时交通流状态预测对于实现城市智能交通系统至关重要。在过去，很多神经网络模型被提出来用以预测交通流，但是效果并不是很显著。究其原因，是因为大多数都是利用浅层模型在学习，浅层模型由于容易陷入局部极值而且不能模拟更复杂的数学运算，所以并不适合于模拟现实的交通状况。深度学习作为机器学习的新兴学科，在语音与图像处理方面取得了显著的成效，它能够非监督地从数据中学习出有效的特征用以预测，故在此利用深度学习进行建模用以城市主干道交通流预测。实验表明，模型取得了不错的交通流预测效果。
展开▼
