Multivariate Prediction Model of Strength and Acoustic Emission Energy considering Parameter Correlation of Coal or Rock

Abstract EN

Due to the heterogeneity of the internal structure and the different external loading conditions, the mechanical and acoustic emission (AE) characteristic parameters of coal and rock are discrete in the process of loading until failure, and many repeated and destructive tests need to be completed to obtain the performance parameters.

It is of theoretical significance to explore the correlation of various parameters and to establish multiparameter regression models of coal rock strength and AE characteristics for predicting the strength and acoustic emission characteristic parameters of coal rock and reducing the repeated tests.

For the coal sample from a coal seam of Longde Coal Mine in China, the mass density of coal samples and the acoustic velocity in the samples before loading are measured at first, and their respective coefficient of variation is analyzed.

Then, the stress-strain curve and the time history curve of AE characteristic parameters are obtained by the uniaxial compression AE test of each coal sample according to the different loading rates.

The influence of loading rate, mass density, and acoustic velocity on the mechanical and AE energy parameters of coal sample is analyzed by the section morphology of the coal sample after failure, the three-dimensional location map of AE, and the scanning micrograph of the electron microscope.

Based on the least-square method, the multiple regression models of compressive strength, elastic modulus, and the maximum AE energy are established by mass density, acoustic velocity, and loading rate of coal samples.

The results indicate that, for the coal samples from the same geological source, the obtained regression models can, respectively, predict the uniaxial compressive strength, elastic modulus, and the maximum AE energy according to the predesigned loading rate, the acoustic velocity, and mass density of coal samples measured before loading, without too many repeated loading failure tests.