Study On Nonlinear Prediction Of Rockburst Of Kuocangshan Freeway Tunnel

Tunnel engineering and underground engineering will be highly developed in the 2lth century. With the unceasing development of the construction of our country, the depth of underground project has continued to increase. However, the engineering geological problem became more and more complex with the unceasing increase of buried depth. Being a kind of familiar geological hazard in deep and over-length highway tunnel, rockburst usually causes injury including death to workers, damage to equipment, and even substantial disruption and economic loss of underground space excavation.Since the first record of rockburst appeared at a tin mine in Britain in 1738,rockburst has occurred frequently in hydroelectric powerhouse,mining tunnel,road and railway tunnel,and nuclear power station etc. Because of the paroxysmal characteristic and greatly injury, the rockburst phenomenon arouses more and more concerns. Rockburst is considered as a dynamic instability phenomenon of surrounding rock mass of underground space in high geostatic stress and caused by the violent release of strain energy stored in rock mass.Rockburst occurs during excavating underground space in the form of stripe of rock slices or rock fall or throwing of rock fragments,sometimes accompanied by crack sound.The theoretical investigation of rockburst is developed from the 1920s～1930s. During the research process, many scholars have been analyzing the occurrence mechanism of rockburst from various angles. Moreover, the typical theory includes the strength theory, the energy theory, the rigidity theory, the defect theory, the energetic disturbances theory etc. The condition of the occurrence of rockburst is various. There is not a kind of theory to explain the occurrence mechanism of rockburst completely.The project region of Kuocangshan tunnel is in the southeast of Zhejiang province. This area is controlled by Cathaysian and neocathaysian structure. Kuocangshan tunnel occured many rockbrusts in left tunnel K156+300 and right tunnel K156+283 when the tunnel is in the actual construction process.The degree of rockburst is in the scope of feebleness to medium,and the rock of rockburst region is stiff and integrated, the joint is scarce, the tunnel is dying. The occurrence of rockburst is caused by the synthetic action of intrinsic factor and extrinsic factor. In addition, the intrinsic factor includes the rockburst orientation of the rock. The extrinsic factor includes terrestrial stress, explosion disturbance influence etc. In this article, the author appraises the orientation of Kuocangshan enclosing rock applying Wet andσc . As a result, the enclosing rock of Kuocangshan takes on the strong orientation of rockburst.The author evaluated the environment of the terrestrial stress about Kuocangshan tunnel through regional geological reconnaissance, focal mechanism solution and terrestrial stress test. As a result, the author got the following formulae:σH = 0.0373H + 4.4065,σh = 0.0288H + 2.4219,σv = 0.0258H (σH is the most horizon principal stress,σh is the least horizon principal stress,σh is the vertical stress, H is the depth of burial ). In the research of the terrestrial stress, the author can also calculated that the direction of the most horizon principal stress is N61°W. Based on the formulae, the terrestrial stress of Kuocangshan tunnel region is on the normal level. The author calculates the value of the terrestrial stress about the tunnel enclosing rock through founding finite element tunnel model. From the principal stress choropleth, we could descry that the maxima of bothσ1andσθappears at the ceiling dermal enclosing rock after the excavation of the tunnel. More over, the value of the stress is in the scope of feebleness to medium.In order to find out the rule of rockburst, and to classify and count the forecasting parameters of rockburst of Kuocangshan tunnel, several methods were adopted as following: investigating the Kuocangshan tunnel geology conditions, analyzing the influence factors of rockburst classifiably, synthetically studying on the rockburst prediction methods and how to assign a value of prediction parameter of rockburst. Meanwhile the rockburst intensity was predicted by the traditional methods such as stress judge criterion and rock character judge criterion. At last, two grade fuzzy comprehensive judgement was applied to the rockburst prediction according to the stress parameters and rock character parameters. However, study results indicated that the traditional prediction methods and two grade fuzzy comprehensive judgement could not objectively and correctly predict and evaluate the rockburst.Through compare and research on the rockburst problem, this paper brought forward that the rockburst was a synthesis result related to multitudinous factors, which was a puzzle with high nonlinear and indeterminacy. This kind of puzzle should be studied by nonlinear methods to insure the research achievements objective and correct, which was the essential cause that the traditional can not make the prediction objectively to the rockburst problem. A new way to research by nonlinear theory was put forward in this paper to make out the study on prediction for rockburst. The nonlinear methods have powerful capabilities in the data mining and knowledge discovery aspects. Based on them, a prediction model with well mapping ability established, the research achievements gained from which can closely approach the target to be studied.The nonlinear prediction of rockburst was studied for the first time based on the rough set theory (RS), support vector machine and extenics methodology in this paper. Firstly, the influencing factors of rockburst were disposed with by the rough set theory, which realized the attribute reduction to those influencing factors. Then the weight coefficients of the attributes were figured out by the concept of the significance of attributes. In the process, the significance was normalized to weight coefficients. Therefore, the most simplified and reasonable parameters model can be provided for the prediction studies for rockburst based on support vector machine and extenics methodology. Finally, the nonlinear theories based on support vector machine and extenics methodology was applied to predict the rockburst of Kuocangshan tunnel.Research achievements indicated that it was feasible to make out the prediction for rockburst based on the nonlinear theories, and the prediction results can reflect the degree of rockburst after the tunnle was excavated. Therefore, the new prediction methods by the nonlinear theories can solve the puzzles of practical projects. Furthermore, research achievements can provide scientific reference for the design of rockburst prevention. By means of the deep and system research on the rockburst of Kuocangshan tunnel, the primary results and conclusions were gained as following:1,The surrounding rock of Kuocangshan tunnel is integrity and stiffness. Through the test of single axle compressive strength and elastic deformation energy index and the combination of rockburst experience criterion, the result show that the surrounding rock of Kuocangshan takes on the strong orientation of rockburst.2,Based on the research of the terrestrial stress, the direction of the most horizon principal stress of Kuocangshan tunnel region is N61°W, which is on the normal level. But the stress differentiation caused by the tunnel excavation induced the value of the terrestrial stress to be higher from the original foundation, which offers the favourable conditon for rockbust. By analysis, we could descry that the maxima of bothσ1 andσθappears at the ceiling dermal enclosing rock and the value of the stress is in the scope of feebleness to medium rockburst.3,After getting secondary stress field caused by the tunnel excavation, the traditional rockburst prediction methods such as TaoZhenyu criterion, LuSen criterion, Rb/σ1 criterion are applied to rockburst prediction of Kuocangshan tunnel. The results show that the results of the traditional prediction methods are different from the practical situation greatly. The rockburst prediction is a puzzle with high nonlinear and is affected by many factors, so a prediction model should be established according to the nonlinear theories and the rockburst influencing factors so that the objective and reasonalbe result can be got. 4,Based on the analyse of the rockburst influencing factors, the rough set theory is adopted to go to attributes reduction to the rockburst influencing factors. The reduction results are show that the five factors reduced can include all the rockburst influencing factors of Kuocangshan tunnel, which offer the better index set for the following rockburst prediction.5,The composition of the decision information table affects the analyzing result of rough set theory greatly. After confirmed the rockburst influencing factors, the method of orthogonal experimental design is adopted to construct the decision table of rockburst prediction. Owing to the samples produced by orthogonal experimenttal have the attributes of uniform scatter and regular comparability, few samples can include all the information of samples set, which induces that the final decision table produced by orthogonal experimental can include all the information of samples set.6,Aimed at the problem of the prodigious influence of the SVM kernel function parameters to model classification ability, genetic algorithm is adopted to search the best parameters of SVM kernel function. The result shows that this method not only decreases the work done, but also increases the confidence level of the parameters.7,There are the crucial action of the rockburst samples choice to the SVM training results. The learning samples of SVM include two parts. One part is obtained by the practical underground engineering, and another is obtained by orthogonal experimental. The samples obtained by orthogonal experimental is applied to train the SVM and the samples obtained by the practical underground engineering is applied to choose the the parameters of SVM kernel function. The result show that the SVM model obtained by this method have the high the confidence level.8,The examination method to training model used to the method of the individual sample examination, which is unilateral greatly. Aimed at this problem, the holistic examination method is provided, that is adopting the change tendency of single factor to rockburst degree to examine the SVV prediction model. This method can examine the extension ability of the SVV prediction model and avoid the limitation of the individual sample examination method.9,Aiming at the problem lied in extenics evaluation method of maximal correlation degree criterion fail and weight coefficient choice, the methods are provided that maximal correlation degree criterion replaced by asymmetry overlay degree and weight coefficient decided by RS theory. The result show that the improve extenics evaluation method have the high accuracy to rockburst prediction.10,Compared with the two nonlinear prediction methods for the rockburst prediction of Kuocangshan tunnel, it can be known that the prediction results by extenics and SVM are similar and accordant with the fact. It is validated that the nonlinear theories applied to rockburst prediction are feasible and reasonable.