The mechanical parameters of surrounding rock are an essential basis for roadway excavation and support design. Aiming at the difficulty in obtaining the mechanical parameters of surrounding rock and large experimental errors, the optimized BP neural network model is proposed in this paper. The mind evolutionary algorithm can adequately search the optimal initial weights and thresholds, while the neural network has the advantage of strong nonlinear prediction ability. So, the optimized BP neural network model (MEA-BP model) takes advantage of the two models. It can not only avoid the local extreme value problem but also improve the accuracy and reliability of the prediction results. Based on the orthogonal test method and finite element analysis method, training samples and test samples are established. The nonlinear relationship between rock mechanical parameters and roadway deformation is established by the BP model and MEA-BP model, respectively. The importance analysis of the three input variables shows that the ∆D is the most important input variable, while ∆BC has the smallest impact. The comparison of prediction performance between the MEA-BP model and BP model demonstrates that the optimized initial weights and thresholds can improve the accuracy of prediction value. Finally, the MEA-BP model has been well applied to predicting the mechanical parameter for the surrounding rock in the Pingdingshan mine area, which proves the accuracy and reliability of the optimized model.

A novel in situ stress monitoring method, based on rheological stress recovery (RSR) theory, was proposed to monitor the stress of rock mass in deep underground engineering. The RSR theory indicates that the tiny hole in the rock can close gradually after it was drilled due to the rheology characteristic, during which process the stress that existed in the rock can be monitored in real-time. Then, a three-dimensional stress monitoring sensor, based on the vibrating wire technique, was developed for in field measurement. Furthermore, the in-field monitoring procedures for the proposed technique are introduced, including hole drilling, sensor installation, grouting, and data acquisition. Finally, two in situ tests were carried out on deep roadways at the Pingdingshan (PDS) No. 1 and No. 11 coal mines to verify the feasibility and reliability of the proposed technique. The relationship between the recovery stress and the time for the six sensor faces are discussed and the final stable values are calculated. The in situ stress components of rock masses under geodetic coordinates were calculated via the coordinate transformation equation and the results are consistent with the in situ stress data by different methods, which verified the effectiveness of the proposed method.

The rheological stress recovery (RSR) method was proposed to obtain measurements of in-situ stress. Rock stress can be evaluated by monitoring the recovery process of stress sensors embedded in rock mass. In order to achieve this application, a novel stress sensor employing the fiber Bragg grating (FBG) technique was designed and manufactured. This stress sensor consisted of three parts: A sensing spherical head, connecting rod, and coupler box. In the sensing spherical head, six independent pressure sensing units were assembled together with a temperature compensation unit. In addition, wavelength division multiplexing (WDM) technology was adopted to ensure that only one fiber splice for each stress sensor is output. The fiber splicing of the sensing units was assembled in the coupler box. The transformation equations from the six pressure sensing units to the stress sensor were established. Furthermore, a calibrating device for the stress sensor was designed, and the general calibration and long-term stability tests were carried out to investigate the characteristic indexes (maximum range, full-range output, and sensitivity) and measurement error (zero drift index, hysteresis index, and repeatability index). Measurement errors showed that the degrees of linearity, zero drift, hysteresis, and repeatability were all below 1.5%. The stability test indicated that the creep of the stress sensor can gradually stabilize in 24 days, and the errors were less than 1.5%. As a result, the stress sensor developed here satisfies the requirements for the RSR method and can be used in field.

In tunnel construction in western China, a vast amount of carbonaceous slate is encountered. High in situ stress and foliation structures cause the rock mass to exhibit pronounced anisotropic creep, readily inducing a series of engineering disasters like collapses and lining cracks. Investigating the anisotropic time-dependent characteristics of carbonaceous slate is beneficial to the long-term stability of tunnel construction and operation. In view of this, carbonaceous slate specimens with different angles, β, between the foliation plane and loading direction were studied using a graded loading method through uniaxial compression creep tests. The results show that the instantaneous axial strain, ε_i, the axial creep strain, ε_c, the duration time of decelerating creep stage, t_d, and the steady creep strain rate, ${\dot{\epsilon }}_s$, increased with the rise in the loading ratio, k. Their variations followed a power law relationship, with the R² (Coefficient of Determination) values all exceeding 0.95. The value of ${\dot{\epsilon }}_s$ was observed to be less than 1.5 × 10⁻⁴/h when β < 45°, while it was found to exceed 1.5 × 10⁻⁴/h in the cases of $\beta \ge 45°$. The long-term strength, σ_L, of carbonaceous slate showed a U-shaped pattern with the variation in β. The maximum σ_L occurred at β = 90° and the minimum was observed at β = 15°. A fractional nonlinear creep model (FNC model) was developed. The sensitivity analysis reveals that the larger the fractional order n is, the t_d and ${\dot{\epsilon }}_s$ increase. η₂ and E₂ primarily affect the decelerated creep stage, while the ${\dot{\epsilon }}_s$ exhibits a rapid increase with the rise of η₁. To further validate the FNC model, a comparison is made with the traditional Nishihara model. The R² of the FNC model is larger than 0.965, which is higher than that of the Nishihara model (R² ≤ 0.911). The FNC model can effectively cope with the impact of the sudden increase in strain and well describe the characteristics of the decelerating, steady-state, and accelerating creep stages at any stress level and any angle. The results provide a reference for the study of the creep mechanism of layered rocks.