| A thorough understanding of the quality and structural characteristics of rock formations is a prerequisite for the safe implementation of various rock engineering projects.While traditional methods for evaluating rock properties,such as core drilling and geophysical exploration,not only consume time and effort but also difficult to implement on a large scale.Therefore,seeking more scientific methods for evaluating rock characteristics has always been a research goal for relevant scientific researchers.This thesis focuses on the essential drilling operations in geotechnical engineering as the research object and fully explored and compared the correlation between different in-drilling response indicators and rock properties through theoretical analysis,laboratory experiments,and simulation studies.Based on the drilling response,a rock property evaluation method was established,providing a reference and basis for intelligent drilling research.The main work and conclusions are as follows:(1)Research has been conducted to predict rock mechanics parameters based on indirect indicators,aiming to develop more reasonable methods for evaluating rock characteristics.A theoretical method was used to establish a formula for predicting rock strength parameters based on the "velocity-rebound-density" fusion indicator,which was then validated using laboratory experimental data.Subsequently,different intelligent algorithms and statistical regression methods were compared.The results showed that the fusion indicator of "velocity-rebounddensity" can improve the prediction accuracy of rock mechanics parameters.Intelligent computing techniques have significant advantages over statistical regression methods in predicting non-linear problems.In rock mechanics parameter prediction,the SVM model outperformed BP,SVM,RF,and KNN,and all four models can simultaneously predict the complex rock mechanics parameters.The GWO-SVM combined model,using the grey wolf optimization algorithm,further improved the prediction accuracy and can be prioritized for in-drilling rock property evaluation research.(2)By decoding and analyzing the actions of the drilling rig,a drilling process identification method based on the in-drilling geological measurement system was developed.This thesis also established a drilling quality evaluation model for rock masses by using the extracted pure drilling parameters.The results showed that the prediction effect of a single drilling rig parameter was ranked as:drilling speed>axial pressure>torque>rotational speed>flushing pressure;the multivariate nonlinear regression methods can improve the prediction effect,but determining the equation function form can be challenging;the drilling specific energy has the strongest correlation with rock mass quality and can also be used to identify interlayer structures.With the assistance of the in-drilling testing system,the classification and positioning of in-drilling rock mass can be achieved.(3)By analyzing the rock breaking process of PDC bit,the response relationship between drilling parameters,bit structure and rock mechanical properties is obtained.The theoretical calculation formula of rock mechanics parameters(compressive strength,shear strength,elastic modulus,Poisson’s ratio,internal friction Angle and cohesion)based on drilling parameters was established by using contact mechanics,elastic mechanics,limit equilibrium principle and energy principle,etc.,and verified by field investigation and laboratory test data.The results show that the rock mechanics parameters acquisition while drilling model based on the theoretical method can effectively reflect the changes of rock mechanics parameters,and has good guiding significance for engineering field.The prediction effect of uniaxial compressive strength is the best,but the prediction effect of cohesion and internal friction Angle is relatively poor.(4)An interface model drilling experiment was conducted using the finite element software Abaqus to analyze the changing characteristics of the stress and vibration signal responses of the drilling rod in different rock layers,and an automatic identification method for rock layer interfaces was established based on drilling rod vibration response,drilling speed,and rotation speed.The conclusions are as follows:the sensitivity ranking of the drilling rod vibration signal with respect to changes in lithology is:longitudinal vibration>transverse vibration>torsional vibration,which aligns with the ranking of the interface identification effect of the three types of vibration signals;the drilling rod vibration response proved to be a more effective method for identifying the rock layer interface compared to drilling speed and rotation speed,with greater data stability in the same rock layer and a lower risk of misjudgment;the change-point detection method was more effective in identifying the position of the rock layer interface.Naive Bayes interface identification was slightly better than the K-means method,but the difference in cumulative error between the two methods was not significant.(5)Using the Hilbert-Huang time-frequency transformation method,8 drilling rod vibration characteristic indicators suitable for lithology identification were selected,and a lithology identification model based on the principal components of vibration response was established.The results showed that as the rock strength increases,the number of main components obtained by empirical mode decomposition showed an upward trend,and the marginal spectrum thus showed a multi-peak feature;dimensionality reduction analysis(PCA)compressed the 8 vibration characteristic indicators into 3 principal components,which contained 93.68%of the original information;the recognition accuracy of the PCA-GWOSVM with axial pressure and 3 principal components as input matrix was 93.33%,and the accuracy was 86.25%when three principal components were used as input matrix.(6)Based on a self-developed drilling simulation test rig,the impact drilling friction and acoustic characteristics of granite and sandstone were studied,and a rock mechanics parameter prediction model using the drilling friction coefficient and the sound level was established.The results showed that the friction coefficient was negatively correlated with the rotational speed and rock strength in a power function but had no significant correlation with the impact velocity;the sound level was positively correlated with the impact velocity and lithology,and was not affected by the rotational speed;the GWO-SVM rock mechanics parameter prediction model outperformed regression analysis,,achieving the best results when using friction coefficient,sound level,impact velocity,and rotational speed as input parameters,with a determination coefficient of 0.958 to 0.985.(7)Through cyclic rotation-impact drilling experiments to explore the evolution laws of sound level,friction coefficient,and drill rod stress wave with rock damage.,and a lithology identification and rock damage evaluation method using impact drilling response was established.The results showed that under the radial displacement restriction,the sample exhibited a progressive damage process,with the axial splitting failure occurred with the loaded area as the center;the GWOSVM model,based on sound level,friction coefficient,and stress wave energy dissipation rate as input parameters,achieving a lithology identification accuracy of 96.51%;when data collection was difficult,effective lithology identification could still be achieved with sound level and energy dissipation rate as input parameters,with an accuracy of 92.64%;the change rates of sound level and friction coefficient with rock damage were not significant,while the stress wave energy dissipation rate increased synchronously with the aggravation of rock damage;there was a significant positive correlation between cumulative rock damage and impact intensity..(8)The thesis conducted cavity structure identification and depth prediction based on the changes in-drilling response indicators for structural specimens containing cavities at varying depths.The results showed that as the cavity depth increased,the specimen exhibited forms of breakthrough,axial splitting+layer splitting,and axial penetrating splitting;the correlation between the friction coefficient and peak stress ratio of drilling rod stress wave and cavity structure was not significant;the baseline sound level,stress wave energy dissipation rate,and reflection energy ratio were identified as indicators for cavity rock structure;The change rate of drilling rod stress wave reflection energy ratio was found to be the optimal indicator for identifying cavity structure depth identification;the cavity structure depth prediction model was established by using statistical methods showed a negative exponential correlation with a determination coefficient of 0.922,and had a good fitting effect. |
PDF Full Text Request