摘要
对旋转机械设备进行状态评估具有重要的科学理论意义和工程应用价值;然而,强背景噪声干扰和特征指标适应范围的局限性大大增加了评估的难度。首先提出了一种基于自适应提升多小波的改进相邻系数降噪方法,利用该方法增强振动信号的微弱特征;然后提取振动信号及其包络解调信号的几种典型特征指标;最后利用自组织神经网络和小波包实现特征映射和状态趋势信息提取,构建一种能够真实反映设备不同运行阶段内在性能变化的状态评估指标。设计了船载天线传动机构加速疲劳实验系统,验证了上述方法的有效性。
State evaluation for rolling machinery is a meaningful task. But huge background noise and localiza- tion of features make the task difficult. An improved denoising method based on adaptive multiwavelets via lifting scheme is presented, to enhance feeble character of vibration signal. After that, representative features are extrac- ted from vibration signal and its envelope demodulation signal. SOM and wavelet packets are used to map features and extract state trend. State evaluation criterion which reflects machinery state in different stages is constructed. Run-to-failure test system for ship-based antennas is designed to validate effectiveness of methods above.
出处
《科学技术与工程》
北大核心
2014年第21期280-284,共5页
Science Technology and Engineering
