CN-121765985-B - High-energy geological environment surrounding rock classification and decision method based on digital twin and multi-source feedback

CN121765985BCN 121765985 BCN121765985 BCN 121765985BCN-121765985-B

Abstract

The invention belongs to the technical field of intelligent construction of tunnels and underground engineering and geotechnical engineering informatization, discloses a high-energy geological environment surrounding rock classification and decision-making method based on digital twin and multi-source feedback, and solves the problems caused by difficulty in realizing surrounding rock state dynamic sensing, multi-source data fusion classification and construction decision-making closed-loop linkage in the prior art under the high-energy geological environment. The method comprises the steps of firstly constructing a tunnel three-dimensional geological-structural digital twin body based on initial investigation data, acquiring multi-source data such as geology, construction disturbance and surrounding rock response in real time by utilizing the Internet of things technology in the construction process, mapping the multi-source data into the digital twin body, and realizing virtual-real synchronous updating. Then, a deep learning-parameter inversion mixed model is constructed on the basis of multi-source fusion data, dynamic surrounding rock classification indexes DRCI and key mechanical parameters are output, and the dynamic surrounding rock classification indexes and the key mechanical parameters are input into a multi-target optimization module to give an adaptive drilling and blasting scheme. And finally, reversely correcting the model through a construction feedback result to realize closed-loop self-learning.

Inventors

ZHANG SHISHU
LI QINGCHUN
RAN CONGYAN
CHEN WEITAO
ZHAO XIAOPING

Assignees

中国电建集团成都勘测设计研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20260304

Claims (8)

1. The high-energy geological environment surrounding rock classification and decision-making method based on digital twin and multi-source feedback is characterized in that, The method comprises the following steps: S1, constructing a tunnel-surrounding rock digital twin body in a high-energy geological environment, wherein the digital twin body integrates a three-dimensional geological model, tunnel geometric information and high-energy environmental field variables; s2, acquiring geological and reconnaissance data, construction disturbance data and surrounding rock response data in real time, and mapping the geological and reconnaissance data, the construction disturbance data and the surrounding rock response data to the digital twin body after space-time alignment; S3, preprocessing and feature extraction are carried out on the collected multi-source data, and feature vectors are formed by integration; S4, inputting the feature vector into a deep learning-inversion mixed model, calculating a dynamic surrounding rock classification index DRCI, inverting the key mechanical parameters of the surrounding rock, and updating a digital twin body; the deep learning-inversion mixed model comprises a CNN-LSTM deep learning model and a PSO-NN inversion model; The CNN-LSTM deep learning model takes a feature vector as input, outputs a dynamic surrounding rock classification index DRCI and a surrounding rock class grade corresponding to the current excavation cycle, and the objective function of the training process is as follows: ; Wherein N is the number of training samples; And Respectively a predicted value and a true value of a kth sample; And The predicted surrounding rock category and the real category of the kth sample are respectively; is a cross entropy loss function; Is a weight coefficient; the PSO-NN inversion model has the objective function that: ; Wherein, the Is the objective function value; for the set of surrounding rock constitutive parameters to be inverted, In order for the cohesive force to be high, Is the internal friction angle of the steel plate, In order to achieve a modulus of deformation, Is poisson's ratio; the actual measurement response vector of the kth working condition; To be at the parameters of A predicted response vector is obtained through numerical simulation of the digital twin body; The calculation formula of the dynamic surrounding rock classification index DRCI is as follows: ; Wherein, the Is the surrounding rock stability index; Is the explosiveness index of surrounding rock; Is a dynamic risk index; 、 And Is a weight coefficient and satisfies ; ; Wherein U c is peripheral convergence of surrounding rock, U c,lim is limit convergence allowed by corresponding surrounding rock grade or specification, U v is vault subsidence, U v,lim is limit vault subsidence allowed, N s is internal force of a supporting structure, N s,lim is internal force allowed by supporting structure design, alpha 1 、α 2 、α 3 is weight coefficient, and alpha 1 +α 2 +α 3 =1 is satisfied; ; Wherein F p is drilling average thrust or equivalent drilling resistance, F p,ref is a reference thrust value, d 50 is slag tapping block median particle size, d 50,ref is target or reference block particle size, V p is blasting vibration speed peak value, V p , ref is allowable or designed vibration reference value, and beta 1 、β 2 、β 3 is a weight coefficient to satisfy beta 1 +β 2 +β 3 =1; ; Wherein E m is the accumulated energy of the microseismic event in unit time or unit mileage, E m,ref is a reference energy threshold, n m is the occurrence frequency of the microseismic event, n m , ref is a reference frequency threshold, H e is a high-energy geological environment factor, gamma 1 、γ 2 、γ 3 is a weight coefficient, and gamma 1 +γ 2 +γ 3 =1 is satisfied; ; Wherein sigma H is the maximum main stress, T is the surrounding rock temperature, p is the pore water pressure, and D is the construction disturbance intensity index; 、、、 corresponding reference values of indexes of maximum main stress, surrounding rock temperature, pore water pressure and construction disturbance intensity are respectively shown, wherein delta 1 、δ 2 、δ 3 、δ 3 is a weight coefficient, and delta 1 +δ 2 +δ 3 +δ 4 =1 is satisfied; S5, determining surrounding rock types and corresponding constraint thresholds based on the dynamic surrounding rock classification indexes DRCI, based on the surrounding rock key mechanical parameters, constructing a multi-objective optimization model and solving the multi-objective optimization model through numerical simulation quantification drilling and blasting parameters and association relations of super-undermining, surrounding rock damage and blasting vibration of a digital twin body, and obtaining an optimal drilling and blasting scheme adapting to the current surrounding rock state through solving the multi-objective optimization model through a multi-objective optimization algorithm; S6, after blasting is implemented according to an optimal drilling and blasting scheme, actual feedback data of construction is collected, error indexes are calculated by comparing the actual feedback data with the prediction results outputted by the digital twin body through numerical simulation, and the error indexes are used for updating parameters of the deep learning-inversion mixed model.
2. The method for classifying and deciding surrounding rock in high-energy geological environment based on digital twinning and multi-source feedback according to claim 1, wherein in step S1, the manner of constructing the tunnel-surrounding rock digital twins in high-energy geological environment comprises: constructing a three-dimensional geological model along the tunnel based on the exploration data; Embedding the designed tunnel geometric information into a three-dimensional geological model to form a tunnel-surrounding rock integrated initial digital twin body; Presetting a high-energy environment field variable in the tunnel-surrounding rock integrated initial digital twin body, and recording the spatial distribution of the high-energy characteristic region to obtain the tunnel-surrounding rock digital twin body.
3. The method for classifying and deciding surrounding rock of high-energy geological environment based on digital twin and multi-source feedback as set forth in claim 1, wherein in step S2, the geological and reconnaissance data comprises face geological images, front abnormal body information recognized by advanced geological prediction system, slag block degree and morphology data; the construction disturbance data comprise drilling parameters such as drilling speed, thrust and torque of a drilling machine, and blasting parameters such as blasting hole spacing, hole depth, explosive loading quantity and detonation network structure; the surrounding rock response data comprise surrounding rock displacement monitoring quantity, blasting vibration parameters and magnitude, energy and position coordinate data of a microseismic monitoring event.
4. The method for classifying and deciding surrounding rock of high-energy geological environment based on digital twinning and multi-source feedback according to claim 1, wherein in step S2, mapping to the digital twinning body after space-time alignment comprises: in the data acquisition process, a construction mileage pile number, a time stamp and a three-dimensional coordinate in a hole are added for each type of data, space-time alignment of multi-source data is realized through a time axis standardization and space coordinate matching technology, and then aligned data are mapped to the digital twin body in real time.
5. The method for classifying and deciding surrounding rocks in high-energy geological environment based on digital twin and multi-source feedback according to claim 1, wherein in step S3, the preprocessing comprises the steps of carrying out time axis standardization on data with different sampling frequencies, realizing time alignment through linear or spline interpolation, projecting point location monitoring data onto tunnel axes and surrounding rock units, and carrying out spatial interpolation through a Kriging or inverse distance weighting method.
6. The method for classifying and deciding surrounding rock in high-energy geological environment based on digital twin and multi-source feedback according to claim 1, wherein in step S3, the feature extraction comprises extracting texture features, fracture trend and blockiness distribution from image data, extracting average propulsion resistance and fluctuation coefficient from drilling parameters, and extracting energy spectrum, dominant frequency and duration from blasting vibration parameters and microseismic monitoring event data.
7. The method for classifying and deciding surrounding rock in high-energy geological environment based on digital twin and multi-source feedback according to claim 1, wherein in step S5, the constraint threshold includes a minimum threshold of surrounding rock stability and a permissible threshold of blasting vibration; The objective function of the multi-objective optimization model is as follows: ; The constraint conditions include: Surrounding rock stability constraints: ; Vibration control constraints: ; Engineering constraints of safe distance and charge limitation; Wherein, the In order to drill and burst the parameter vector, , In order to provide a hole spacing of the holes, In order to achieve a row spacing of the rows, In order for the hole to be deep, Is the drug loading quantity; Is the lowest threshold of surrounding rock stability; For drilling and blasting parameters The maximum vibration speed predicted by numerical simulation of the digital twin body is obtained; a threshold value is allowed for blasting vibration; is the over-run or the profile deviation; is an index of damage or breakage degree of surrounding rock induced by blasting; is an index of the vibration speed or energy of blasting; Is the cost per cycle.
8. The method for classifying and deciding surrounding rock in high-energy geological environment based on digital twinning and multi-source feedback as claimed in claim 7, wherein in step S6, the error index is calculated by the following method: ; Wherein, the Is an error index; And The actual measurement and the numerical simulation prediction of the digital twin phantom are respectively carried out on the super-underexcavated volume; And Respectively predicting the damage index of the surrounding rock through actual measurement and numerical simulation of the digital twin phantom; And Respectively predicting microseismic energy or vibration energy through actual measurement and numerical simulation of a digital twin phantom; 、 And Is a weight coefficient and satisfies + + =1。

Description

High-energy geological environment surrounding rock classification and decision method based on digital twin and multi-source feedback Technical Field The invention belongs to the technical field of intelligent construction of tunnels and underground engineering and geotechnical engineering informatization, and particularly relates to a high-energy geological environment surrounding rock classification and decision-making method based on digital twin and multi-source feedback. Background Along with expansion of infrastructure construction such as traffic, water conservancy and energy to deep underground space, engineering construction of deep buried long tunnels, high-temperature tunnels, rich water broken zones and mining earthquake frequent areas is increased gradually, geological environment where engineering surrounding rocks are located gradually evolves from conventional to high-energy, and a typical high-energy geological environment is formed. The environment has the core characteristics of high ground stress, high ground temperature, high osmotic pressure and strong disturbance (frequent blasting impact and mechanical vibration), and under the condition, the surrounding rock often shows strong nonlinearity, strong anisotropy and strong space-time effect, and the engineering requirements are difficult to meet by the traditional surrounding rock classification and construction design method based on static investigation data. In a high-energy geological environment, surrounding rock is easy to generate stress concentration, local fracture, rapid expansion or shearing sliding of a structural surface and other phenomena under the coupling action of excavation unloading, blasting disturbance and underground water-stress, so that the problems of serious super-underexcavation, large deformation of the surrounding rock, instability of a surrounding rock-supporting system, exceeding of blasting vibration and the like are caused, and the construction safety and the construction period control are seriously influenced. Therefore, how to realize real-time, dynamic and visual sensing and classification of surrounding rock states in a high-energy environment and adaptively optimize drilling and blasting schemes according to the real-time, dynamic and visual sensing and classification of surrounding rock states becomes an important scientific and engineering problem in the fields of tunnels and underground engineering. The classical surrounding rock classification methods such as Q system, RMR system, BQS and the like widely applied in the current engineering are mostly based on initial investigation data and limited field information, and have the following defects: (1) Strong static property and lack of time sequence evolution capability The traditional classification method mainly determines surrounding rock classification at one time before construction or in the early stage of excavation, and only carries out a small amount of correction when obvious abnormality is found in the follow-up process, so that the dynamic evolution process of the surrounding rock under the continuous blasting disturbance, unloading relaxation and supporting effects cannot be reflected, and the rapid deterioration or abrupt change response of the surrounding rock property in a high-energy environment is delayed. (2) Multisource data fracturing and underutilization of information During construction, a large amount of heterogeneous data can be generated, including drilling parameters (thrust, torque and footage speed), blasting parameters (hole net and charging structure), monitoring data (surrounding rock displacement, convergence, blasting vibration, ultra-short laser scanning, microseismic events and the like), but the data are often collected and stored by different systems unification and independence, so that space-time alignment and fusion are difficult to perform on a unified platform, and a large amount of potential information is wasted. (3) Surrounding rock classification and construction decision-making disjoint The traditional surrounding rock classification result is only used as a reference of a design stage, an automatic association mechanism is absent between the traditional surrounding rock classification result and blasting parameter design, charging structure optimization and supporting parameter adjustment, the surrounding rock classification cannot directly drive construction decision optimization, and the traditional surrounding rock classification result still depends on experience judgment of field technicians seriously, so that the objectivity and stability of decisions are insufficient. (4) Neglecting high-energy characteristics and space-time coupling effect The existing method generally assumes that a ground temperature field, a water pressure field and a stress field are approximately stable in a short time, and is difficult to describe the thermal-seepage-force-disturbance coupling effect of surroundin