Study On Rock Mass Classification Method And Intelligent Classification System Of Road Tunnel

This paper studies the rock mass classification problem of Niba mountain tunnel of the Yaxi highway.Using the ideology of combining site investigation and laboratory analysis,qualitative analysis and quantitative evaluation,modern mathematical theory and engineering practice,geological engineering and software engineering,this paper comprehensively studies the tunnel rock mass classification problem.The results of the study refine and complete the BQ rock mass classification method of road tunnel.Using qualitative parameters as input parameters,this paper builds intelligent classification methods based on support vector machine,neural network and fuzzy reasoning models.Finally,a rock mass classification website is designed to accomplish the functions of storing rock mass information,quantitatively and qualitative-intelligent classifying rock mass.The main contents are as following.(1)Considering the influence of the strike and dip of structural plane on the surrounding rock mass,this paper optimized the the weak structural plane influence parameter K2 of the road tunnel rock mass classification(BQ)method.Firstly,the angle between the strike of structural plane and the direction of tunnel axis influence the rock mass classification.When the structure plane strike is parallel to the direction of tunnel axis,it is unfavorable for the construction.K2 gets a larger value.When the structure plane strike is perpendicular to the direction of tunnel axis,it is favorable for the construction,especially,when the tunneling direction is the same with the structural plane inclination direction.K2 gets a smaller value.Secondly,the dip of structural plane has influence on the stabilization of surrounding rock mass.When the strike of the structural plane with large dip angle is parallel to the tunnel axis direction,it is unfavorable for rock mass stability.K2 gets a larger value.While,when the strike of the structural plane with large dip angle is perpendicular to the tunnel axis direction,it is favorable for rock mass stability.K2 gets a smaller value.Furthermore,the structure of rock mass has influence on K2.The more intact the rock mass is,the better for rock mass stability.So K2 gets a relatively smaller value.(2)This paper optimized the initial stress status influence parameter K3 of the BQ method,according to the intensity comprehensive classification system of rock burst that considering geomechanical patterns,which Li Tianbin proposed.Slight rock burst could incite the rifting and peeling of rock mass into flakes.This process has little influence on rock mass stability,so K3 gets a small value.Medium rock burst could burst rock mass into lenticular,lamellar or tabular pieces,during to rifting-slippage and bending-bulging failure.It has a medium influence on rock mass stability,so K3 gets larger value.Intense rock burst could burst rock mass into tabular,lump or wedge blocks,during to shearing or quaquaversal bursting failure.It has a big influence on rock mass stability,so K3 gets a larger value between 0.6 and 1.1.Super intense rock burst could burst rock mass into tabular,lump or even dispersed pieces.It has a huge influence on rock mass stability,so K3 gets a large value between 1.1 and 1.5.(3)After analysis the parameters that other primary rock mass classification methods used and the request that the parameters could be quickly obtained,this paper proposes using 7 qualitative parameters,such as the rigidity of rock,the intactness of rock mass,embedded condition of rock mass,rock mass structure,weathering condition of joints,underground weather condition and ground stress condition,as input parameters of intelligent classification method.(4)Intelligent Rock mass classification models have been established based on support vector machine,neural network and fuzzy logic models.282 site samples have been used as training samples of support vector machine and neural network models.And 60 samples have been classified using these three models.The results show that: support vector machine with polynomial kernel could classify rock mass perfectly,the correct rate is 85%;neural network performed normal with a correct rate of 66.7%;Fuzzy logic is the worst with a correct rate of 45%.(5)A web site system has been developed to classify the rock mass in tunnel.This web site system developed by Visual Studio.Using the technique of ASP.net and SQL database,this system can store and read the information of rock mass.Interface programming has been used to integrate MATLAB in this system,so that it's possible to classify rock mass using the intelligent models.Dependent web pages have been developed to accomplish the functions of cataloging,classifying,checking and printing the geological information of rock mass.(6)A workflow that based on the web site platform has been proposed to classify rock mass.Technical staffs from the construction unit catalog the rock mass information into the system.Then the system will store this information and classify the rock mass using multiple methods.Professional supervision engineers will review the information of rock mass and comment.Designers will change the design parameters based on the actual situation.Experts will review the design parameters and propose suggestions.Then the proprietor will consider all the suggestions,safety and cost factors,organize meeting to determine the rock mass classification.Finally,the construction unit will organize construction based on the classification of rock mass.