36 Assessment of Large Error Time-Differences for Localization in a Plate Simulation 375 Velocity (inches/sec) 0 200 400 600 800 1000 1200 1400 1600 RMSE (inches) 0 20 40 60 80 100 120 Location (9.26, 5.85) COV = 0.44 Location (12.2, 8.78) COV = 0.12 Location (15.1, 8.78) COV = 0.39 Location (5.85, 5.85) COV = 0.08 Fig. 36.8 RMSE over a portion of the velocity range for several select impacts. The most accurate velocity changes based on impact location: 807 in./s and 681 in./s for impacts at (12.2, 8.78) and (15.1, 8.78) respectively convolution of the reflections could yield an approximation of the time-difference. The SOAs are relatively accurate in all cases, with only a small percentage of signs being interpreted incorrectly, suggesting that when heavy reflections are present in the signal, the SO-TDOA algorithm would still perform well, and could be used to locate the source to a specific area. 36.4 Conclusion This research presents results for time-difference of arrival (TDOA) and cross-correlation methods for a sparsely instrumented plate in a simulation environment where acceleration data is used and reflections are significant. Wave reflections play an important role in the accuracy of localization algorithms in dispersive media. When reflections are significantly large, neither cross-correlation nor peak-difference methods produce accurate localization results. The peak-difference is found to perform better when the sensors are further from the boundaries, while cross-correlation results appear slightly better when the sensors are near the edge. This suggests that cross-correlation may be better tool if reflections are sufficiently large. The SOAs have less than 10 % error for all cases, suggesting good accuracy when despite large TDOA errors. Acknowledgements The authors are thankful for the support and collaborative efforts provided by our sponsors VTI Instruments, PCB Piezotronics, Inc.; Dytran Instruments, Inc.; and Oregano Systems. The authors are particularly thankful for the support provided by the College of Engineering at Virginia Tech through Ed Nelson and Dean Richard Benson as well as Capital Project Manager Todd Shelton. The authors would also like to acknowledge the collaboration with Gilbane, Inc. in particular members David Childress and Eric Hotek. We are especially thankful to the Student Engineering Council at Virginia Tech and their financial support for this project. The work was conducted under the patronage of the Virginia Tech Smart Infrastructure Laboratory and its members. References 1. Schloemann, J., et al.: Vibration event localization in an instrumented building. In: Experimental Techniques, Rotating Machinery, and Acoustics, Volume 8, pp. 265–271. Springer, Cham (2015) 2. Poston, J.D., et al.: Towards indoor localization of pedestrians via smart building vibration sensing. In: Proceedings of the 2015 International Conference on Localization and GNSS (ICL-GNSS). IEEE, Gothenburg, 22–24 June 2015, (2015) 3. Bahroun, R., et al.: New algorithm for footstep localization using seismic sensors in an indoor environment. J. Sound Vib. 333(3), 1046–1066 (2014)
RkJQdWJsaXNoZXIy MTMzNzEzMQ==