Prediction for Rock Conditions in a Tunnel Area Using Advanced Geological Drilling Predictions Based on Multiwavelet Analysis and Modified Evidence Reasoning

Abstract

Advanced geological prediction plays an essential role in disaster prevention and safety guarantee. In this paper, by combining multiwavelet analysis with modified evidence reasoning, a new geological prediction method for surrounding rock in a tunnel considering the parameters’ coupling effect was proposed. Based on the key drilling parameters’ curves that were denoised in the No. 1 hole of the Xiaochuan tunnel in Gansu Province, China, the coupling effect of these parameters were analyzed and the power–speed ratio (PSR) concept was thereby proposed to deal with the denoised curves and obtain a PSR curve. Through continuous wavelet transform (CWT) of the PSR curve, the lithology and structure were preliminarily determined by sectional matching and optimization of multiple wavelet coefficient curves under the optimal scales and PSR curve. Further, the evidence reasoning was introduced to solve the uncertainty parts in the preceding result. To avoid misjudgment caused by the limitation of Dempster Shafer (D-S) theory in the case of complete conflict, it was modified, in which the basic probability assignment (BPA) was determined by the interval number, and a binary group was then defined by combining the traditional conflict coefficient with pignistic probability distance to measure the degree of conflict. The division result that settled by this method showed greater accuracy than tunnel seismic prediction (TSP) and the drilling core, also could give a clear division for both lithology and structure. Importantly, by applying the established method to Chengzhou tunnel, the result fitted the real condition well. This approach provides references for rock prediction in tunnels.