摘要
The measurement of the crustal stress at depth is a difficult problem in geological engineering projects. The crustal stress is hard to determine or measuring data are not ideal because of the limitation of unduly simple research means and measuring techniques. On the other hand, satisfying results can be achieved by artificial neural network (ANN) even though the data have deficiencies such as data noise, partial absence, lack of cognition because of the native advantages: self-learning, self-adaptability, robust, error tolerance and generalization. Based on the BP artificial neural network method, this paper provides a prediction model for the crustal stress values using 6 factors: depth, field density, elastic modulus, triaxial compressive strength, acoustic emission?? stress measurements and fissure density. The authors made hydrofracturing stress measurements in the Qinling deep-buried long tunnel by using the BP artificial neural network model, performed a fitting analysis of the measured data and predicted the crustal stress at depth.