基于灰箱模型的中速磨煤机故障诊断方法

Fault Diagnosis of a Medium-speed Coal Mill Based on Grey Box Model

采用数据与机理分析相结合的方法建立了中速磨煤机系统的灰箱模型,该建模方法既克服了纯机理建模过于复杂、耗时较长的问题,同时比纯数据建模具有更好的精确性和鲁棒性。然后利用该灰箱模型得到磨煤机输出量的残差数据,并通过小波变换提取残差的变化趋势,提出了一种基于斜率阈值的故障检测方法,根据随机森林算法的原理对故障数据进行训练,建立了一个用于故障类型识别的故障分类器。结果表明:所提故障诊断方法能够实现对磨煤机故障的早期诊断,并具有较高的故障识别率和识别精度。

A grey box model was established for medium-speed coal mill by combining the mechanism anal- ysis method with data identification algorithm, which has higher accuracy and stronger robustness and overcomes the deficiencies of pure mechanism modeling, such as complicated modeling process and long modeling time, etc. The grey box model is then used to calculate the residual of mill output, and by the way of wavelet transform, the residual trend is extracted. In addition, a fault detection method is proposed based on the slope threshold of residual trend signals, while the random forests algorithm is employed to create a fault classifier to identify the type of mill faults. Results show that the fault diagnosis method proposed can realize early fault detection, which has high recognition rates.