论文部分内容阅读
介绍了一种基于STFT ,WVD和Wavelet变换的混合神经系统在特征提取和信号分类中的应用 .它运用时频信号处理技术与混合神经网络结构以及高级训练算法相结合的设计思想 ,辅之以主元分析和全局决策融合策略 ,对传统的模式识别技术进行了有效地改进 .本文将所述基于WT ,STFT和WVD的混合神经系统信号分类器(WSWHNS)的算法程序嵌入一汽车实时智能故障诊断软件包中做了现场实验 ,获得了非常满意的诊断效果
A hybrid neural network based on STFT, WVD and Wavelet transform is introduced in feature extraction and signal classification.It uses the combination of time-frequency signal processing technology and hybrid neural network structure and advanced training algorithm, Principal component analysis and global decision fusion strategy, the traditional pattern recognition technology is effectively improved.In this paper, the algorithm program of WT, STFT and WVD based hybrid neural system signal classifier (WSWHNS) is embedded into a real-time intelligent fault Diagnostic package made a field experiment, obtained a very satisfactory diagnostic results