Abstract This paper extracted the futures of collected radiated noise of running vehicles by AR model firstly, then opted features by hypothesis test, and carried out classification by BP-NN.

• 摘要 本文首先利用AR参数模型提取采集到的车辆行驶时产生的车外噪声信号的特征，然后利用假设检验进行特征选择，并设计了BP神经网络进行分类识别。