4. Factors Affecting the Identified System Parameters Before presenting the results of the measurements and data analyses, the factors which affect the identified system parameters are briefly discussed below. 4.1. Measurement Accuracy The first step in extracting the dynamic properties of a system is ensuring a reliable and accurate measurement of its response. High noise to signal ratios or other physical factors limiting the accuracy of the measurements may cause an inherent problem for the rest of the analyses and have a negative effect on the results which are generally very difficult to detect and eliminate. The strain gauges installed in the turbine are used as a part of a long term health monitoring program and checked regularly. Therefore, the data acquired from these sensors were considered to be accurate enough and mostly used as a reference measurement in this work. Provided that the signal quality is high, laser Doppler vibrometers have very high accuracies (in micron scale). The laser measurements mentioned in this paper were taken on the special locations where retro-reflective markers were placed. As mentioned before, the advantage of this material is that it is 1000 times more reflective than the background blade material and it reflects the light directly back to its source. During the tests the signal level was continuously checked and verified to be within acceptable range. However, photogrammetric measurements were performed as a feasibility study aiming at testing whether they can be used for monitoring large turbines or not. Photogrammetry was applied for the very first time on a MW scale turbine within the scope of this work. Therefore, the measurement accuracy has not been verified for such a large measurement volume before (please see Figure 4). Since the conventional methods used to estimate the accuracy of the small scale photogrammetric measurements were not appropriate to be applied for the field tests, all the critical pre or post-processing operations and the corresponding error estimation analyses have been done by using the measurements themselves i.e. by verifying the consistency of the data recorded. Figure 5 shows a typical flapwise direction tip marker displacement measured by photogrammetry. An important point that should be paid attention to was the synchronization of the different measurement systems having different sampling frequencies and data acquisition systems. The data recorded by the LDV (sampling frequency 1280 Hz) photogrammetric measurement system (sampling frequency 28 Hz) and the strain gauges were later synchronized by using a GPS clock. The estimated accuracy of the absolute time recorded by these GPS systems was in the range of 10 milliseconds. Considering the fact that the sampling frequency of the strain gauges were 32 Hz and that the dominant frequencies that were expected to be identified were mainly lower than 5 Hz, this accuracy can be said to be sufficient. 261
RkJQdWJsaXNoZXIy MTMzNzEzMQ==