摘要
详细介绍了一种基于小波包和神经网络的新算法,用于对直升机声音信号检测并且识别机型。具体方法是对采集到的声音样本利用小波包分析提取特征向量,把这些特征向量输入反向误差传播(BackPropagation,BP)神经网络训练,用训练好的检测神经网络进行直升机的检测。检测完毕,证实是直升飞机声信号后,再通过识别神经网络区分不同型号直升机。实验表明,此方法能利用小波包时频局部聚焦分析能力和BP神经网络的自适应能力,较好地对不同型号的直升机声信号进行有效地检测和识别。因此,基于小波包和神经网络的直升机检测和识别算法不仅可靠而且是可行的。
A new algorithm based on wavelet packet and neural network is described in this paper to achieve acoustic detection and recognition of helicopter. Characteristic vectors of helicopters acoustic signal extracted by the wavelet packet analysis are inputted in a BP neural network in order to train the network. And then use the trained BP network to detect helicopters. After the detection procedure of the helicopter acoustic signals, another BP neural network is adopted to recognize different models of helicopters. Because wavelet packet can focus and analyze local time-frequency and neural network has the ability of self-adaptation, the experiment results indicate that the way that makes the best of the frequency characteristic of helicopter acoustic signal is effective to recognize different models of helicopters in low SNR.
出处
《信息与电子工程》
2006年第3期165-169,共5页
information and electronic engineering
关键词
信号检测
直升机声信号
识别
小波包
BP神经网络
signal detection
helicopter acoustic signal
recognition
wavelet packet
BP neural network