Research And Application Of Filling Material Gradation Based On Fractal Theory And Bond Strength

In a filling system, the strength of the filling body directly affects the stability of surrounding rock and operation sequence, as a consequent, it plays a crucial role in mining safety. Paper is testified by a large number of experimental analysis and orthogonal method, for the cementing strength of the filling body, the main influencing factors are:fill gradation, water cement ratio (cementing agent content) and concentration. And the grading of filling material is the most significant impact. Filling grading refers to the distribution from the fine particle to medium particle and to coarse particle size. The maximum density curve theory commonly used to describe the particle size distribution of filling materials, but in coarse and fine grain size varies widely, the description of the theory is further form the truth.1. With the help of fractal mathematical, the author has established a formula which can describe particle size distribution curve of filling materials effectively and completely. The formula is P(x)=(1-a)(x3-d-xmax3-d)/(xmax3-D-x3min3-D)+1，P(x) refer to distribution curve of filling material in negative accumulation, a is the smallest sieve pore size particles of the corresponding frequency, D is the dimension of the filling material gradation. Results show that the formula can accurately describe particle size distribution curve of filling materials, and in the case of a variety of filling ratio of arbitrary can effectively distinguish between different ratio of fill materials, at the same time, the output parameters D is a dimensionless quantity.2. For cementing strength model of filling material although many scholars and experts have done a lot of analysis and research, and many predict model is put forward but its promotion and range of application is still very narrow. Artificial neural network’s ability of self-organization, parallel self-learning and distributed storage of information, which provides a favorable tool for the nonlinear prediction model. Therefore the author using the BP neural network, which optimized by genetic algorithm, and support vector machine (SVM) to establish a prediction model of cementing strength which involved dimensions of filling material, water cement ratio and concentration, and forecast data that does not participate in the study. The results show that the prediction effect of the two models are ideal that means the predicted values close to the actual value, the actually MSE of two models were 0.3 and 0.2 MPa. The above two model can predict the cementing strength accurately after measured the particle size distribution function of filling materials.3. Optimal gradation is a kind of gradation which can achieve maximum cementing strength and meet the pipeline conditions. With the help of experimental data and the prediction model of cementing strength, the author get the relationship between dimension and cementing strength which show the optimal gradation is D=2.75.The results in conformity with the actual engineering situation, it also can provide a basis to optimize the design of the cementing strength. At the same time it can provide the parameters of the actual filling engineering preliminary design with the convenience and security.