论文部分内容阅读
采用基于神经网络的重要性权值调整的粒子滤波算法对预处理后的振动信号进行降噪、小波包能量谱提取,将提取到的能量谱作为特征向量用BP神经网络对其进行故障模式分类识别。对经过粒子滤波降噪的数据和没有经过处理的数据分别用BP神经网络进行诊断,之后进行训练、测试和诊断。结果表明经过粒子滤波降噪后的数据诊断效果比较好,也证明了基于神经网络粒子滤波降噪的效果较好。
The preprocessed vibration signal is denoised and the wavelet packet energy spectrum is extracted by using the particle filter algorithm based on importance weight of neural network. The extracted energy spectrum is used as eigenvector to classify the fault pattern by BP neural network Recognize. After the particle filter noise reduction data and untreated data were used to diagnose BP neural network, followed by training, testing and diagnosis. The results show that the data after particle filter noise reduction is better, and also shows that the particle filter based on neural network noise reduction is better.