Journal of the Korean GEO-environmental Society (한국지반환경공학회 논문집)

Korean Geo-Environmental Society (한국지반환경공학회)
권호 표지
DOI QR코드

DOI QR Code

Comparison of Ordinary Kriging and Artificial Neural Network for Estimation of Ground Profile Information in Unboring Region

미시추 구간의 지반 층상정보 예측을 위한 정규 크리깅 및 인공신경망 기법의 비교

  • Chun, Chanjun (Construction Automation Research Center, Korea Institute of Civil Engineering and Building Technology) ;
  • Choi, Changho (Construction Automation Research Center, Korea Institute of Civil Engineering and Building Technology) ;
  • Cho, Jinwoo (Construction Automation Research Center, Korea Institute of Civil Engineering and Building Technology)
  • Received : 2018.11.13
  • Accepted : 2019.02.18
  • Published : 2019.03.01
Download PDF
⟨ Previous Next ⟩

Abstract

A large amount of site investigation data is essential to obtain reliable design value. However, site investigations are generally insufficient due to economic problems. It is important to estimate the ground profile information in unboring region for accurate earthwork-volume prediction, and such ground profile information can be estimated by using the geo-statistical approach. Furthermore, the ground profile information in unboring region can be estimated by training a model via machine learning technique such as artificial neural network. In this paper, artificial neural network-based model estimated the ground profile information in unboring region, and this results were compared with that of ordinary kriging technique, which is referred to the geo-statistical approach. Accordingly, a total of 84 ground profile information in an actual bridge environment was split into 75 training and 9 test databases. The observed ground profile information of the test database was compared with those of the ordinary kriging technique and artificial neural network.

확한 토공량 설계를 위해서는 충분한 량의 지반조사 자료가 필요하나 비용적인 문제로 인하여 제한적인 지반조사가 수행되고 있다. 정확한 토공량 예측을 위해서 지반의 층상정보를 추정하는 것은 중요한 사항이며, 이러한 제한적인 지반조사 데이터로부터 정확한 토공량 예측을 위해서는 지구통계학적(geo-statistical) 분석방법으로 지반 층상정보를 예측할 수 있다. 또한, 기시추된 지반 층상정보를 활용하여 기계학습을 통하여 모델을 학습하여 미시추된 지반 층상정보를 예측할 수도 있는데, 본 논문에서는 인공신경망을 통하여 미시추된 지반 층상정보를 예측하고 기존의 정규 크리깅 기법과 성능을 비교한다. 이를 위하여, 84공의 지반 층상정보를 활용한다. 84공의 지반 층상정보의 데이터셋 중에서 75공을 학습 데이터셋으로 활용하였고, 나머지 9공을 검증 데이터셋으로 활용하였다. 검증 데이터셋의 실측된 지반 층상정보와 정규 크리깅 기법과 인공신경망으로 예측된 지반 층상정보를 비교 분석한다.

Keywords

HJHGC7_2019_v20n3_15_f0001.png 이미지

Fig. 1. Overview of variogram modeling

HJHGC7_2019_v20n3_15_f0002.png 이미지

Fig. 2. Examples of variogram modeling depending on distribution types

HJHGC7_2019_v20n3_15_f0003.png 이미지

Fig. 3. Overview of artificial neural network

HJHGC7_2019_v20n3_15_f0004.png 이미지

Fig. 4. Ground profile information in boring region

HJHGC7_2019_v20n3_15_f0005.png 이미지

Fig. 5. Result of variogram modeling

HJHGC7_2019_v20n3_15_f0006.png 이미지

Fig. 6. Architecture of artificial neural network

HJHGC7_2019_v20n3_15_f0007.png 이미지

Fig. 7. Result of variogram modeling

Table 1. RMSE results of estimated data

HJHGC7_2019_v20n3_15_t0001.png 이미지

References

  1. Dai, F., Zhou, Q., Lv, Z., Wang, X. and Liu, G. (2014), Spatial prediction of soil organic matter content integrating artificial neural network and ordinary kriging in Tibetan Plateau, Ecological Indicators 45, pp. 184-194. https://doi.org/10.1016/j.ecolind.2014.04.003
  2. Deutsch, C. V. (1996), Correcting for negative weights in ordinary kriging, Computers & Geosciences, 22(7), pp. 765-773. https://doi.org/10.1016/0098-3004(96)00005-2
  3. Goodfellow, I., Bengio, Y. and Courville, A. (2016), Deep Learning, MIT Press, Cambridge, MA. pp. 161-263.
  4. Han, W., Wu, C., Zhang, X., Sun, M. and Min, G. (2016), Speech enhancement based on improved deep neural networks with MMSE pretreatment features, in Proc. IEEE 13th International Conference on Signal Processing (ICSP), Chengdu, China, DOI: 10.1109/ICSP.2016.7878006.
  5. Ioffe, S. and Szegedy, C. (2015), Batch normalization: accelerating deep network training by reducing internal covariate shift, in Proc. the 32nd International Conference on Machine Learning (ICML), Lille, France, pp. 448-456.
  6. Isaaks, E. H. and Srivastava, R. M. (1989), An Introduction to Applied Geostatistics, Oxford University Press, New York, NY. pp. 278-322.
  7. Kingma, D. P. and Ba, J. L. (2015), ADAM: a method for stochastic optimization, in Proc. 3rd International Conference on Learning Representations (ICLR), San Diego, CA, pp. 1-15.
  8. Min, D. H. and Yoon, H. K. (2018), Method for Establishment of High Reliable Soil Depth Modeling, J. Korean Soc. Hazard Mitig. 18(2), pp. 307-315. https://doi.org/10.9798/kosham.2018.18.2.307
  9. Nair, V. and Hinton, G. E. (2010), Rectified linear units improve restricted boltzmann machines, in Proc. the 27th International Conference on Machine Learning (ICML), Haifa, Israel, pp. 807-814.
  10. Rahman, H. S., Alireza, K. and Reza, G. (2010), Application of Artificial Neural Network, Kriging, and Inverse Distance Weighting Models for Estimation of Scour Depth around Bridge Pier with Bed Sill, J. Software Engineering & Application, 3, pp. 944-964. https://doi.org/10.4236/jsea.2010.310112
  11. Samui, P. and Sitharam, T. G. (2010), Site Characterization Model Using Artificial Neural Network and Kriging, Int. J. Geomech. 10(5), pp. 171-180. https://doi.org/10.1061/(ASCE)1532-3641(2010)10:5(171)
  12. Seo, Y., Kim, S. and Singh, V. P. (2015), Estimating Spatial Precipitation Using Regression Kriging and Artificial Neural Network Residual Kriging (RKNNRK) Hybrid Approach, Water Resour Manage, 29, pp. 2189-2204. https://doi.org/10.1007/s11269-015-0935-9
  13. Tapoglou, E., Karatzas, G. P., Trichakis, I. C. and Varouchakis, E. A. (2014), A spatio-temporal hybrid neural network-kriging model for groundwater level simulation, Journal of Hydrology, 519, pp. 3193-3203. https://doi.org/10.1016/j.jhydrol.2014.10.040