Traffic flow prediction based on BP neural network

With the rise of artificial intelligence, artificial neural networks have also begun to develop rapidly. An artificial neural network consists of many artificial neurons, which is a structure that imitates the neurons created by the human brain. BPneural network is a classic algorithm in artificial neural network. Because the BP neural network has a simple structure, is easy to implement and can deal with nonlinear problems, it is widely used in pattern recognition, signal processing, and transportation systems. In this paper, the BP neural network, one of the classic artificial neural networks, is applied to the traffic flow prediction. The data used in this article is the traffic situation on the M4 highway from 2020-1-1 0:14:00 to 2020-1-2 23:59:00. Record the traffic data once every 15 minutes. There are a total of 192 groups of traffic data. From these 192 sets of data, 19 sets of random data are selected as test data, and the remaining 173 sets of data are used as training data. The BP neural network input layer is set to four variables: date, time, the number of cars in this period, the number of cars in this period, and the average speed value, and the output layer is set to the total traffic volume in this period. Use matlab2016a for simulation. The simulation results show that the BP neural network can accurately complete the traffic flow prediction problem. Because the initial weights and thresholds of the BP neural network have a great influence on the prediction accuracy, this paper uses particle swarm optimization to optimize the initial weights and thresholds of the BP neural network. Simulation results show that the optimized BP neural network is more stable than the mean square error before optimization, and the average relative error is smaller.