大机组起停车状态在线监测方法研究

在大机组起停车过程中，影响转子动力行为的因素众多，工况难以把握。为此，基于信息熵提出了一个机组起停车过程转子运行状态的评估指标，并在此基础上，为实现机组运行故障早期预诊的目的，进而基于Ｊ散度及多参数评估理论，提出了信号频谱结构分布散度的概念，以描述实测信号与机组典型故障信号间的伪空间距离，从而实现了机组运行故障的初步预诊。上述方法计算简捷，在线实现能力强，评估、预诊结果可靠，可适用于机组起停车等任

On-Line Diagnostic Approach to Machine Start and Stop Conditions

The dynamic behavior of large rotors is studied during the stage of start and stop. The concept of entropy is applied to define an index on evaluating the abnormal condition. Prediagnosis of machine fault is made using the spectral structure distribution divergence as a criterion, in which the J divergence concept and multi parameter evalution theory that describe the pseudo distances between the sampling signals and the corresponding standard faulty signals are used. These methods have simply computational algorithms, on line realization capacity and can be applied to different working conditions. Their reliability is verified experimently.