PSI - Issue 38

Florian Grober et al. / Procedia Structural Integrity 38 (2022) 352–361 Grober, Janßen, Küçükay / Structural Integrity Procedia 00 (2021) 000 – 000

360

9

Regarding the longitudinal acceleration spectra in Fig. 7 it can be determined that in particular the branch of the forward accelerations (upper half of the spectrum) is reproduced very well. Overall, the number of load cycles is slightly reduced, but this is due to a lack of small amplitudes. Since they cause typically only few damage, they can be omitted. In addition, the deceleration branch (lower half of the spectrum) overreaches its target, thereby compensating for the loss of damage, so that the deviation is within the usual scatter for driving operation measurements.

Fig. 8. Comparison of lateral acceleration spectra (target versus test drive).

The comparison of the lateral acceleration spectra in Fig. 8 shows a slight deviation in the peak values. The latter are generated in a few very sharp turns. Obviously, even small deviations in velocity can have a large effect here. However, in a real durability road test, which is significantly longer than ten kilometers, it can be assumed that these influences may balance themselves. Furthermore, the driver guidance system induced significantly less level crossings in the area of small lateral accelerations. A closer examination shows that the target spectrum included some drives on uneven ground (e.g. cobblestone). They created a shaking of the vehicle body and thus a noise in the lateral acceleration. In the test drive with the guidance system, such road roughness was not present. Therefore, this effect was not taken into account. In the sense of omission, such small load amplitudes would be negligible anyway. All in all, it can therefore be concluded that the generated lateral acceleration spectrum is of acceptable quality. 7. Conclusion and outlook This article has assessed the questions how general usage field data can be transferred to load data for testing and how an effective load monitoring in the durability road test experiment can be carried out. The presented driver guidance system solves both problems by using the customer-based target spectra as input, recording the induced loads during the testing and using artificial intelligence to select the further route guidance in such a way that the target spectra are reached as well as possible. The experience database with the expected loads per road edge is continuously updated, thereby adapting to driving style and changed road conditions. By means of the designed user interface, a comprehensive and ergonomic communication with the driver is realized. The test with the prototypical implementation of the system shows a principle applicability with good results. It must be taken into account that the validation measurement was only ten kilometers long, whereas a real durability road test takes many thousands of kilometers. Therefore, it can be assumed that the result quality increases with longer test duration, since individual deviations from the driving specification (see example with curve speeds) or single unfavorable decisions of the heuristic algorithm become less important overall. For the resulting enhancement in test quality, a benefit can be expected especially for testing on public roads since the load deviations there are usually greater than on proving grounds due to a varying traffic flow.