Geostress Survey And Intelligent Prediction Of Surrounding Rock Stability

Ground stress is the basic primary data of rock masses and the source material to confirm the attributes of engineering rock masses, identify surrounding rock stability, and design for soil and rock engineering excavation.The survey principle, process and key problem of strain gauge of improved hollow inclusion cell were expounded in this paper. It was discussed how to carry out in-situ stress survey and analysis from the aspects of geostress station selection, survey theory and the analysis of geostress survey results etc. Based on the geostress survey results of Juji Mine, the distribution law of ground stress field has been gained by theoretical anlysis and linear regression analysis in this region. That is, firstly, it is near linear relationship between primary stress and burying depth, and primary stress increases along with the growth of burying depth; secondly, the dominant ground stress is the horizontal tectonic stress, and the maximum stress lies in the level; thirdly, the mine is merely influenced by press stress; at last, the abrupt change of ground stress has never occurred, and the direction of area tectonic stress field is at the direction of south to north.Based on the survey station stress ellipsoid, it was analyzed the changing relationship of ground stress components in different cross section with roadway inclination angle and direction.There are many influencing factors of roadway surrounding rock stability, and which have complexly nonlinear relationship with surrounding rock stability. On account of the indeterminacy, inconsistency and imperfection, the key problem affecting the surrounding rock stability has not been solved, that is, the selection and the quantification presentation and explanation of principal influence factors.Based on traditional identification method of surrounding rock stability, it was brought forward comprehensive and systemic research of RS-NN combination classification model which was successfully applied to identifying the stability of wall rock for the first time. Rough set theory of knowledge discovery field is adopted to deal with the indeterminacy, inconsistency and imperfection between influence factors on the foundation of practical geological data. Decisive factors could be mined from survey information field by attribute reduction, core reduction and significance calculation, which can better condense network dimensions. Specific identification rules could be acquired by rule acquisition from information field, which can be utilized to verifying and reconstructing samples. It is indicated by the application that RS-NN combination classification model could deal with not only quantification presentation and explanation of classification indexes but qualitative analysis problems and find out the internal relationship between classification indexes and surrounding rock stability. Furthermore, the relationship would be adopted to predict the surrounding rock stability on the similar project condition.Relevant combination method of information field was adopted to establish RS-NN combination classification model in this paper, and in which the research results of rough set were successfully introduced to the innovation of neural network. It is dominance complementation, to exert each advantage, prevent each shortcoming, and promote each development.