CN-121997437-A - Intelligent design method and system for pre-supporting side wall of underground factory building with deep buried multi-structure surface

CN121997437ACN 121997437 ACN121997437 ACN 121997437ACN-121997437-A

Abstract

The invention relates to the technical field of artificial intelligence, and discloses an intelligent design method and system for pre-supporting side walls of a deeply buried multi-structure-face underground factory building, which aim to solve the problems of supporting lag and poor precision in the existing method, and the scheme mainly comprises the steps of setting the value ranges of parameters of a structural face and parameters of a pre-supporting direction, and constructing data sets under different working conditions by adopting discrete element numerical simulation; the method comprises the steps of constructing a multi-structure surface surrounding rock deformation prediction model for predicting the maximum deformation of a side wall, constructing an optimizing frame by taking the prediction model as an adaptability function of a particle swarm optimization algorithm, obtaining actual structure surface parameters of a region to be supported, iteratively searching by the particle swarm optimization algorithm to enable the adaptability function to output minimum optimal pre-supporting direction parameters, and carrying out inclined pre-supporting on a next layer of rock body which is not excavated according to the optimal parameters before the side wall of an underground plant is excavated. The invention realizes the advance of the pre-supporting time and improves the supporting precision, and is suitable for underground workshops in geology complex areas.

Inventors

ZHANG SHISHU
WANG MENG
ZHAO XIAOPING
RAN CONGYAN
CHEN WEITAO
SHAN SHIHAN
CHENG LIJUAN

Assignees

中国电建集团成都勘测设计研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260403

Claims (10)

1. The intelligent design method for the pre-supporting of the side wall of the underground factory building with the deep buried multi-structure surface is characterized by comprising the following steps: Step 1, setting a value range of structural surface parameters and pre-supporting direction parameters, wherein the structural surface parameters comprise structural surface trends, structural surface dip angles and structural surface distances, and the pre-supporting direction parameters comprise anchor rod trends and anchor rod dip angles; Step 2, constructing three-dimensional numerical models under different working conditions according to the value ranges, adopting discrete element numerical simulation programs to simulate the excavation and the inclined pre-supporting process of the side wall of the underground factory building, obtaining different structural surface parameters and the maximum deformation of the side wall under different pre-supporting direction parameters, and constructing a data set comprising the structural surface parameters, the pre-supporting direction parameters and the maximum deformation of the side wall; Step 3, constructing an improved multi-layer perceptron model, taking the structural surface parameters and the pre-supporting direction parameters in the data set as input parameters, taking the corresponding maximum deformation of the side wall as an output target, and training the improved multi-layer perceptron model to obtain a multi-structural surface surrounding rock deformation prediction model for predicting the maximum deformation of the side wall; step 4, constructing an optimizing frame based on a particle swarm optimization algorithm, and taking the multi-structure surface surrounding rock deformation prediction model as an adaptability function of the particle swarm optimization algorithm, wherein the output of the adaptability function is a maximum deformation prediction value of a side wall; Step 5, acquiring structural plane parameters of the rock mass in the area to be supported, performing iterative search in the range of values of the pre-support direction parameters based on the structural plane parameters of the rock mass in the area to be supported by the particle swarm optimization algorithm, and solving the anchor rod trend and anchor rod inclination angle combination which enable the output of the fitness function to be minimum as the optimal pre-support direction parameters; and 6, before the side wall of the underground powerhouse is excavated, performing oblique pre-support on the rock mass of the next excavated layer which is not excavated according to the optimal pre-support direction parameters.
2. The intelligent design method for the pre-supporting of the side wall of the underground powerhouse with the deeply buried multi-structure surface is characterized in that the range of the inclination of the structural surface is 0-360 degrees, and the range of the inclination of the structural surface is 0-90 degrees.
3. The intelligent design method for pre-supporting the side wall of the underground powerhouse with the deep buried multi-structure surface according to claim 1, wherein in the step2, the adopted discrete element numerical simulation program is a 3DEC discrete element program; Before the step 3, the method further comprises a process of preprocessing the data set, and specifically comprises the following steps: removing abnormal data in the data set, carrying out unified dimension processing on the data, dividing the preprocessed data set into a training set, a verification set and a test set according to the ratio of 7:1:2, and converting the data into a matrix form containing a feature matrix and a target vector.
4. The intelligent design method for pre-supporting the side wall of the underground powerhouse with the deep buried multi-structure surface according to claim 3, wherein in the step 3, the improved multi-layer perceptron model adopts a shallow network structure comprising an input layer, 2 to 3 hidden layers and an output layer; And training the improved multi-layer perceptron model by using the training set, and monitoring the mean square error of the training set and the verification set in real time in the training process to ensure that the trends of the training set and the verification set are consistent.
5. The intelligent design method for pre-supporting the side wall of the underground powerhouse with the deeply buried multi-structural surface according to claim 4, wherein in the step 3, after the prediction model for deformation of the surrounding rock with the multi-structural surface is obtained, the method further comprises the step of verifying and optimizing the model by adopting the test set: and evaluating the model prediction capability through at least one index of average absolute error, decision coefficient or root mean square error, and adjusting network structure or super-parameters of the multi-structure surface surrounding rock deformation prediction model when the error is larger than a preset threshold value.
6. The intelligent design method for pre-supporting side walls of a deeply buried multi-structure face underground powerhouse according to claim 1, wherein in the step 4, the fitness function is defined as: ; Wherein, the Is a structural surface trend, Is a structural plane inclination angle, For the distance between the structural faces, Is an anchor rod trend, For anchor rod inclination angle, MLP is a deformation prediction model of surrounding rock with multiple structural surfaces, and for a group of fixed structural surface parameters The output value of the fitness function is along with the parameters of the pre-supporting direction A wall maximum deformation prediction value which changes with the change; the particle quantity of the particle swarm optimization algorithm is 20-50, and the iteration times are 100-200.
7. The intelligent design method for pre-supporting the side wall of the underground powerhouse with the deep buried multi-structure surface according to claim 6, wherein in the step 5, the automated process of iterative search includes: And obtaining the structural plane trend, the structural plane inclination angle and the structural plane spacing of the rock mass in the area to be supported, iteratively searching different anchor rod trends and anchor rod inclination angles through the particle swarm optimization algorithm, and iteratively predicting the maximum deformation predicted value of the side wall by using the multi-structural-plane surrounding rock deformation predicted model until the anchor rod trend and the anchor rod inclination angle which enable the maximum deformation predicted value of the side wall to be minimum are found.
8. The intelligent design method for pre-supporting the side wall of the underground powerhouse with the deep buried multi-structure surface according to claim 1, further comprising the step of verifying the result of the optimal pre-supporting direction parameter after the step 5 and before the step 6: Selecting a plurality of groups of structural plane parameters of the rock mass in the area to be supported, inputting the corresponding structural plane parameters and the optimal pre-support direction parameters obtained by solving into the discrete element numerical simulation program for verification calculation, and comparing and verifying whether the maximum deformation value of the side wall obtained by numerical simulation calculation is smaller than the maximum deformation value of the side wall when other pre-support direction parameters are adopted under the group of optimal pre-support direction parameters.
9. The intelligent design method for pre-supporting the side wall of the underground powerhouse with the deep buried multi-structure surface according to claim 1, wherein in the step 6, the side wall of the underground powerhouse is a high side wall with the height of more than 30m, and the height is the vertical distance from the lower pit position to the upper arch line; The next excavated layer which is not excavated refers to a rock mass with the single excavated layer height of 5-13 m; The oblique pre-support refers to that in the layered downward excavation process of the side wall, before the current excavation layer is excavated, the oblique support operation of applying anchor rods or anchor ropes to the rock mass of the next excavation layer which is not excavated is carried out in advance according to the optimal pre-support direction parameters.
10. An intelligent design system for pre-supporting side walls of a deeply buried multi-structure face underground powerhouse, which is characterized by being used for executing the intelligent design method for pre-supporting side walls of a deeply buried multi-structure face underground powerhouse according to any one of claims 1 to 9, and comprising: The parameter range setting module is used for setting the value ranges of structural surface parameters and pre-supporting direction parameters, wherein the structural surface parameters comprise structural surface trends, structural surface dip angles and structural surface distances, and the pre-supporting direction parameters comprise anchor rod trends and anchor rod dip angles; The data set construction module is used for constructing three-dimensional numerical models under different working conditions according to the value range, simulating the underground factory building side wall excavation and the inclined pre-support process by adopting a discrete element numerical simulation program, acquiring the side wall maximum deformation under different structural surface parameters and different pre-support direction parameters, and constructing a data set comprising the structural surface parameters, the pre-support direction parameters and the side wall maximum deformation; The prediction model training module is used for constructing an improved multi-layer perceptron model, taking the structural surface parameters and the pre-supporting direction parameters in the data set as input parameters, taking the corresponding maximum deformation of the side wall as an output target, and training the improved multi-layer perceptron model to obtain a multi-structure surface surrounding rock deformation prediction model for predicting the maximum deformation of the side wall; The optimizing framework building module is used for building an optimizing framework based on a particle swarm optimization algorithm, the multi-structure surface surrounding rock deformation prediction model is used as an adaptability function of the particle swarm optimization algorithm, and the output of the adaptability function is a maximum deformation prediction value of the side wall; The optimal parameter solving module is used for acquiring structural plane parameters of the rock mass in the area to be supported, the particle swarm optimization algorithm is based on the structural plane parameters of the rock mass in the area to be supported, iterative search is carried out in the range of values of the pre-support direction parameters, and the anchor rod trend and anchor rod inclination angle combination which enables the output of the fitness function to be minimum is solved and serves as the optimal pre-support direction parameters; and the pre-supporting execution guidance module is used for carrying out inclined pre-supporting on the rock mass of the next excavation layer which is not excavated according to the optimal pre-supporting direction parameter before the side wall of the underground powerhouse is excavated.

Description

Intelligent design method and system for pre-supporting side wall of underground factory building with deep buried multi-structure surface Technical Field The invention relates to the technical field of artificial intelligence, in particular to an intelligent design method and system for pre-supporting side walls of a deeply buried multi-structure-surface underground plant. Background The rock mass structural plane of geology complex areas (such as Jinsha river plate joint strips and east structure knots) is developed in a complex way, and the deep fracture is widely distributed, so that the rock mass has poor integrity and stability. And the underground cavern group excavation construction of the hydropower station is carried out under the environment, and the surrounding rock is easy to deform or collapse and unstably generate disaster problems due to strong unloading of the rock mass and open structural surface in the excavation process, so that the construction safety is threatened. The hydropower station underground powerhouse is provided with a high steep side wall, the height of the side wall exceeds 30m, layered excavation is generally adopted, and the height of a single excavation layer is generally 5-13 m. The high side wall has the advantages that due to complex excavation procedures and long construction period, the surrounding rock is continuously unloaded and subjected to stress adjustment, the structural surface is cracked and displaced, the surrounding rock of the side wall is easy to continuously deform, the problem of large deformation is finally caused, and the safety and stability of a grotto are threatened. Therefore, for the deep buried underground factory building under the complex condition of multiple structural surfaces, the stable control of surrounding rock deformation during the excavation of the high side wall is a key difficult problem for restricting engineering construction. The difficulty of the difficult problem is mainly two aspects, namely that the deformation control is difficult to be timely. Because the anchor spraying and supporting technology is generally carried out after the excavation of the side wall layer is finished, unloading of surrounding rock occurs at the moment, cracking of the structural surface occurs along with unloading, damage of the surrounding rock gradually expands from the outside to the inside, and supporting measures are applied at the moment and are later than unloading deformation. Secondly, deformation control is difficult to be accurate. Because the structural surface develops with a certain yield and forms a certain inclination angle with the excavation surface, the general anchor bolt support is applied in a single direction, the structural surface yield is not considered, and the targeting support effect aiming at the structural surface characteristics is difficult to realize. Aiming at the problem of deformation control of surrounding rocks of a high-side wall of a deeply buried multi-structure-surface underground plant, three related or similar supporting technologies mainly exist at present, namely, concrete spraying and common anchor rod supporting are adopted, namely, the supporting means which are most commonly used in the underground plant at present are that common mortar anchor rods are adopted for supporting the excavated surrounding rocks, and concrete spraying and sealing are carried out. The anchor cable or the prestressed anchor cable support is widely applied to underground workshops, and particularly the prestressed anchor cable or the opposite-penetrating anchor cable and the like are possibly adopted at the side wall part of the workshops according to actual demands. Thirdly, the anchor rod support is pressed. The method is a supporting means mainly aiming at extrusion large deformation, and has relatively good supporting effect in the soft rock tunnel based on the concepts of yielding supporting and yielding before carrying. However, the prior supporting technology has the following defects that firstly, the supporting time is delayed. The existing high side wall support technology starts to perform support after the excavation of the lower excavation layer is finished. The rock mass with multiple structural surfaces has specificity, under the strong unloading effect of high ground stress excavation, the densely distributed structural surfaces can crack along with the unloading, so that the rock mass cracks and damage are expanded, the excavation surface is continuously developed from the outside to the inside, surrounding rock deformation is also continuously developed and is not converged, and the rock mass is a special mechanical property which is not possessed by the complete rock mass. If the method is still applied to the excavation-first and the support-later of the complete rock mass, the support is obviously lagged, and the deformation development is difficult to control in time. Secondly, the supporting prec