PSI - Issue 5

Alejandro Carvajal-Castrillón et al. / Procedia Structural Integrity 5 (2017) 729–736 Alejandro Carvajal-Castrillón/ Structural Integrity Procedia 00 (2017) 000 – 000

733

5

by the software’s communication module to a ground station computer using the ZEROMQ® request/reply procedure over a 2.4-Ghz WLAN. On the UAV, the communications system is integrated by a 300-mW wireless USB adapter with a 3000-mW signal booster attached to an omnidirectional N-type antenna. The WLAN’s access point is positioned on ground and provided by a 28 dBm outdoor wireless radio with a 15-dBi, long-range omnidirectional antenna. The strain information is received as a string on a ground computer running a MATLAB® software, that decodifies the string into a numeric matrix, in which the arriving strings are added as rows. The data transmission system was tested to determine its influence on the final acquisition frequency at the ground station. The test consisted in sending data packages from the UAV to the ground station during one second, using different amounts of collected data per package sent from the onboard computer and measuring the received data per second on the ground computer. Results are presented on the next section. A general scheme of the strain measurements acquisition and transmission system is presented on Figure 1. One line of sensors, located at one side of the beam, was damaged during the manufacturing and assembly process of the instrumented beam on the UAV’ s wing. As there is another line of sensors on the opposite side intended to measure torsion strains, this damage is not critical and further testing could be performed. Operational ground tests of the acquisition and processing system consisted on placing the complete aircraft’s wing over two expanded polystyrene blocks located at 1.2 m from the center on each side of the wing, accomplishing to have a double supported beam. Flexion and torsion loads were applied on the wing ’s center with different non measured magnitudes, simulating the loads acting on the beam during normal flight operation. Sensor data were collected at 100 Hz due to the very low associated frequency of the applied loads. 12 different wing positions were tested by varying its angle of attack. Once the testing data were collected, several preprocessing techniques were applied to reduce noise and filter outliers. Then, data standardization was performed by using several techniques (see Sierra-Pérez et al. (2005)). These standardization techniques are intended to uncouple load magnitudes and operational conditions, assuming all strains were on the elastic region of the material and no buckling was present with the applied loads. The strain fields of the beam from the 12 experiments, defined as the strain distribution over its length, were processed by an OBS composed by a D2SL-SOM to create multiple baselines and clusters classifying the different operational conditions using MATLAB®. Testing results are presented on the following section.

Fig. 1. System Architecture.

4. Results and Discussion

3.1. Wireless transmission test

The ZEROMQ® requester/receiver procedure, and the wireless data transmission system influence on final data acquisition rates was assessed by performing five tests, sending a different amount of data strings per package sent. As shown in Table 1, final acquisition frequency is directly related to package size, and the process involving string assembly into packages is more demanding than the sending procedure itself, when package size is more than 40