本文在频域内,通过两正交方向振动信号的合成,抽取反应实际振动情况的特征值,建立标准样本,然后利用模糊识别的方法进行分类。通过近一年生产实际数据的计算,可以看出机组运行状态变化的连续性。掌握和监测这种连续性,可以对许多过去归结为突发性的故障找到来龙去脉,从而为正确诊断和预报创造了条件。 In the frequency domain, through the synthesis of the vibration signals in two orthogonal directions, the eigenvalues ​​of the actual vibration of the reaction are extracted, and the standard samples are established, then the classification is made by the method of fuzzy recognition. Through the calculation of the actual production data in the past year, we can see the continuity of the unit operation status. Mastering and monitoring this continuity can find the context of many past failures that have been classified as sudden, thus creating the conditions for proper diagnosis and forecasting.