机组运行趋势的小波包掩模去噪研究

RESEARCH OF DENOISING RUNNING TRENDS OF ROTATING MACHINERY WITH WAVELET PACKET THRESHOLDING

对机组中长期运行趋势的提取方法进行了研究，提出了基于小波包时频掩模技术去除机组运行趋势噪声的方法，用该方法去除了短期的随机波动，得到了均值意义上的趋势。对该趋势提前进行时间序列预报，可以得到机组未来均值意义上的趋势，显著提高机组运行状态的预报效果，且该方法具有对数据的广泛自适应性。

The extracting method of the long term trend of rotating machinery is investigated. A signal trend extracting method, which is based on wavelet packet thresholding technique, is proposed. The short-term random fluctuate can be removed with this method, and the mean bend can be obtained. The running trend forecasting results are excellent when being used to predict long-term operating condition of large rotating machinery.