CPT-SPT Correlation Analysis Based on BP Artificial Neural Network Associated with Partial Least Square Regression
Most of the correlations of CPT-SPT have been widely investigated based on a statistical method without considering the multicollinearity among the influenced factors. Therefore, one model combining back propagation neural network (BP ANN) with partial le
- PDF / 714,903 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 79 Downloads / 252 Views
2
CCCC Fourth Harbor Engineering Institute Co., Ltd., Guangzhou 51000, China [email protected] Key Laboratory of Environment Protection and Safety of Transportation Foundation Engineer of CCCC, Guangzhou 51000, China
Abstract. Most of the correlations of CPT-SPT have been widely investigated based on a statistical method without considering the multicollinearity among the influenced factors. Therefore, one model combining back propagation neural network (BP ANN) with partial least square regression (PLS), which could consider the multicollinearity influence, is proposed for building CPT-SPT correlation. The sensitivity analysis based on the ANN model and the PLS model is conducted, and the four most sensitive factors are obtained, i.e., cone resistance (qc), soil behavior type (SBT), friction resistance (fs) and soil behavior index (Ic). Further, these four most sensitive factors plus fine content (Fc%) and effective stress ðr0o Þ are adopted as the input factors of the combined model (BP ANN associated PLS). And 362 group data are collected from New Doha Port for building the combined model. The result shows the combined model has a correlation index R2 of 0.83311 and further demonstrates the combined model is effective. Finally, additional 50 group data are applied to verify the combined model. The result indicates that it can improve the correlation coefficient between the predicted (qc/pa)/N value and the measured one from 0.6429 to 0.7523 and reduce the relative error from 28% down to 21% compared with the solely PLS model. Given the advantage of the combined model, it can provide a more effective and reliable method for CPT-SPT correlation analysis in practice. Keywords: Multicollinearity Partial least square regression BP artificial neural network CPT-SPT correlation
1 Introduction The in situ standard penetration test (SPT) has been widely used because of its simplicity and operability; however, it has disadvantages of poor repeatability and low reliability because the N value can be influenced by many factors. Alternatively, the cone penetration test (CPT), as one of the continuous in situ tests with higher repeatability and reliability, is increasingly applied in design aspect [1–4]. Therefore, to © Springer Nature Singapore Pte Ltd. 2018 L. Hu et al. (Eds.): GSIC 2018, Proceedings of GeoShanghai 2018 International Conference: Multi-physics Processes in Soil Mechanics and Advances in Geotechnical Testing, pp. 381–390, 2018. https://doi.org/10.1007/978-981-13-0095-0_43
382
X. Liang et al.
fully use the existing SPT empirical experience, an in-depth study of the correlation between SPT and CPT is required. Based on a detailed referenced review of the literature [5–13], most of the correlation was mainly based on the mathematical statistics correlated with single influence parameters, such as mean grain size (D50), fine content (FC (%)), soil behavior type (SBT) and soil behavior index (Ic). However, these correlations are normally built on multiple factors without considering the multicollinearity
Data Loading...
