基于邻接传感器及神经网络的车辆分类算法

Vehicle classification algorithm based on binary proximity sensors and neural networks

为了提高车辆分类的性能,基于邻接传感器网络和BP神经网络提出一个有效的车辆分类算法MSVCA。在本算法中,使用成本相对低廉、灵敏度高的地磁传感器,采集车辆对地磁场的磁扰动特征信号,并根据邻接传感器网络本身的几何特性估计车辆长度,最后采用BP神经网络对车辆进行分类。神经网络的输入包括车辆长度、速度以及特征向量序列,输出为预定义的车辆类型。仿真及路面实验获得了93.61%的准确率。结果表明该算法提高了车辆分类的准确性,且具有较高的精度和顽健性。

To improve the classification accuracy, a new algorithm was developed with binary proximity magnetic sensors and back propagation neural networks. In this algorithm, use the low cost and high sensitive magnetic sensors to detect the magnetic field distortion when vehicle pass by them and estimate vehicle length with the geometrical characteristics of binary proximity networks, and finally classify vehicles via neural networks. The inputs to the neural networks include the vehicle length, velocity and the sequence of features vector set, and the output is predefined vehicle types. Simulation and on-road experiment obtains high recognition rate of 93.61%. It verified that this algorithm enhances the vehicle classification with high accuracy and solid robustness.