直升机信息_直升机_直升飞机_旋翼机_Helicopter

The .rst damage case tested was a sun gear with a sin-gle spall on Sun Tooth ID 9 shown in Figure 26. This spall covers around 75% of the toothface. No other apprecia-ble damage was noticed via visual inspection on any of the other teeth.

The second component tested was the sun gear shown in Figure 27. This gear had four teeth with severe damage. On tooth ID 10 a spall about one-third the facewidth exists. On tooth ID 12, there a chip at the tip which extends about one-sixth the facewidth. Tooth ID 14 has spall covering about 80% of the facewidth and, on tooth ID 15, almost the full facewidth is spalled.

Results

During testing, the transmission ran continuously. The DAQ system waited for a trigger indicating that the plane-tary orientation was in the pre-determined position and pro-ceeded to take data for either 20 or 40 seconds. The DAQ would then wait again for the planetary reset trigger and ac-quire another data set. This continued until multiple runs were stored in individual .les. On average, each test case: baseline, single tooth spall, and multiple tooth fault, took about 50 minutes to complete.
For each test case, each run was loaded, partitioned into carrier cycles, interpolated, notch-.ltered to remove the pinion dynamics and low-pass .ltered to remove dy-namics above the sixth planet mesh frequency. The results were temporarily stored until all of the runs were processed in the same manner. The Assembly Family was produced representing the average of all the runs. Averaging across the runs was performed after interpolation to allow for like points to be averaged. Information for all .ve accelerome-ters were stored in a single structured variable, as described earlier.
Results -SGVS -SASP Method

This section shows SGVS results using the SASP method for both damage cases. Recall that this method takes the measured signal, arranges it into three pass groups, averages within the pass groups, and then assembles the Vi-bration Separation Vector by connecting the three wave-forms. The FM4 metric is then computed on the Vibration Separation Vector signal. The signal from accelerometer A3 as ”seen” through planet P3 is used for the plots. Fig-ure 28 shows the baseline results. Figure 28a shows the individual pass groups. Each pass group represents the collection of 34 waveforms illustrating the high corre-lation between waveforms in the same pass groups but not necessarily between different pass groups. One of the key bene.ts of this technique is that averaging in pass groups preserves the .ne individual tooth mesh waveform details. This would not be the case if all the waveforms were simply averaged together. Figure 28b shows the Vibration Sep-aration Vector as ”seen’ through each planet. The pass groups from Figure 28a were used to create the assembly of the second row. In addition to the FM4 value, 3 other CIs are computed for comparison: NA4, M8A, and RMS. These values are displayed atop each Vibration Separation Vector .
　
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(26)