|
曝光台 注意防骗 网曝天猫店富美金盛家居专营店坑蒙拐骗欺诈消费者
Condition Indicators on TSA data Stewart was the .rst to apply condition indicators to transmissions. (Ref. 22) He observed that in a healthy trans-mission, the amplitude of the sidebands were small and were signi.cantly increased in the presence of a fault. This was especially true for local defects, such as single tooth failure. He introduced a set of algorithms to compute con-dition indicators, termed Figure of Merit. These had names like FM0, FM2, and FM4, and increased in complexity as the order increased. The most promising metric, FM4, works on the premise that the vibration signal for an un-damaged system contains distinctive energy at known mesh and bearing frequencies. If a new signal is synthesized that removes these non-random dynamics, then an assessment can be made on that signal, termed the difference signal. Under undamaged cases, all that should remain is Gaus-sian white noise characterized by a normal distribution. If the difference signal contains distinctive components in the spectrum, then it is assumed that damage is the culprit. n where xi is each data point and ˉx is the mean. The kurtosis value, relative to 3, gives an estimation of the amplitude dis-tribution of the resultant signal, in particular its peakiness. Figure 2 shows three types of distributions. The .rst one Fig. 2. Possible Distribution Functions is termed mesokurtic and has the characteristic bell shape of a normal distribution. A normalized kurtosis value of 3 predicts Gaussian white noise. The other two distributions are discussed relative to the mesokurtic. The leptokurtic is characterized as having a standard deviation value less that that for the normal distribution and the opposite for the platykurtic. Figure 3 gives examples of distributions and their corresponding kurtosis values. It shows the time plot, data distribution relative to a normal distribution, and com-puted kurtosis values for .ve different distributions. The normal distribution curve shares the same mean and stan-dard deviation as the signal, but its amplitude is normalized to enable overlapping. The .rst plot is a normal distribu-tion and is used to provide a baseline for the others. The second distribution is uniform and produces a platykurtic pro.le because it is wider than the normal. The third and fourth signals were generated using only cosine functions. The third had two components and the fourth had a sin-gle component. The distributions becomes more platykurtic 直升机网 www.helicopter.cn 直升机翻译 www.aviation.cn 本文链接地址：Sun Gear Fault Detection on an OH-58C Helicopter Transmission(6) |
