Longjun Dong

Prediction of rockburst classification using Random Forest

Abstract The method of Random Forest (RF) was used to classify whether rockburst will happen and the intensity of rockburst in the underground rock projects. Some main control factors of rockburst, such as the values of in-situ stresses, uniaxial compressive strength and tensile strength of rock, and the elastic energy index of rock, were selected in the analysis. The traditional indicators were summarized and divided into indexes I and II. Random Forest model and criterion were obtained through training 36 sets of rockburst samples which come from underground rock projects in domestic and abroad. Another 10 samples were tested and evaluated with the model. The evaluated results agree well with the practical records. Comparing the results of support vector machine (SVM) method, and artificial neural network (ANN) method with random forest method, the corresponding misjudgment ratios are 10%, 20%, and 0, respectively. The misjudgment ratio using index I is smaller than that using index II. It is suggested that using the index I and RF model can accurately classify rockburst grade.

A Microseismic/Acoustic Emission Source Location Method Using Arrival Times of PS Waves for Unknown Velocity System

To eliminate the location error of MS/AE (microseismic/acoustic emission) monitoring systems caused by the measurement deviations of the wave velocity, a MS/AE source location method using P-wave and S-wave arrivals for unknown velocity system (PSAFUVS) was developed. Arrival times of P-wave and S-wave were used to calculate and fit the MS/AE source location. The proposed method was validated by numerical experimentations. Results show that the proposed method without the need for a premeasured wave velocity has a reasonable and reliable precision. Effects of arrival errors on location accuracy were investigated, and it shows location errors enlarged rapidity with the increase of arrival errors. It is demonstrated the proposed method can not only locate the MS/AE source for unknown velocity system but also determine the real time PS waves velocities for each event in rockmass.

An efficient closed-form solution for acoustic emission source location in three-dimensional structures

This paper presents an efficient closed-form solution (ECS) for acoustic emission(AE) source location in three-dimensional structures using time difference of arrival (TDOA) measurements from N receivers, N ≥ 6. The nonlinear location equations of TDOA are simplified to linear equations. The unique analytical solution of AE sources for unknown velocity system is obtained by solving the linear equations. The proposed ECS method successfully solved the problems of location errors resulting from measured deviations of velocity as well as the existence and multiplicity of solutions induced by calculations of square roots in existed close-form methods.

Three-dimensional analytical solution of acoustic emission or microseismic source location under cube monitoring network

To find the analytical solution of the acoustic emission/microseismic(AE/MS) source location coordinates, the sensor location coordinates were optimized and simplified. A cube monitoring network of sensor location was selected, and the AE/MS source localization equations were established. A location method with P-wave velocity by analytical solutions (P-VAS) was obtained with these equations. The virtual location tests show that the relocation results of analytical method are fully consistent with the actual coordinates for events both inside and outside the monitoring network; whereas the location error of traditional time difference method is between 0.01 and 0.03 m for events inside the sensor array, and the location errors are larger, which is up to 1080986 m for events outside the sensor array. The broken pencil location tests were carried out in the cross section of 100 mm×98 mm, 350 mm-length granite rock specimen using five AE sensors. Five AE sources were relocated with the conventional method and the P-VAS method. For the four events outside monitoring network, the positioning accuracy by P-VAS method is higher than that by the traditional method, and the location accuracy of the larger one can be increased by 17.61 mm. The results of both virtual and broken pencil location tests show that the proposed analytical solution is effective to improve the positioning accuracy. It can locate the coordinates of AE/MS source only using simple four arithmetic operations, without determining the fitting initial value and iterative calculation, which can be solved by a conventional calculator or Microsoft Excel.

An Analytical Solution for Acoustic Emission Source Location for Known P Wave Velocity System

This paper presents a three-dimensional analytical solution for acoustic emission source location using time difference of arrival (TDOA) measurements from N receivers, . The nonlinear location equations for TDOA are simplified to linear equations, and the direct analytical solution is obtained by solving the linear equations. There are not calculations of square roots in solution equations. The method solved the problems of the existence and multiplicity of solutions induced by the calculations of square roots in existed close-form methods. Simulations are included to study the algorithms performance and compare with the existing technique.

Three Dimensional Comprehensive Analytical Solutions for Locating Sources of Sensor Networks in Unknown Velocity Mining System

The accuracy of localization methods based on the arrival time difference is usually affected by the iterative algorithm, the initial value, and the pre-measured wave velocity. The analytical solutions are non-unique because of the square root operations in the calculating process for source coordinates. To solve these significant problems, the nonlinear equations were simplified to the linear equations. An analytical localization method without the pre-measured velocity or the square root operations was developed. The explicit formulas for analytical solutions were resolved for the six sensors network. The source coordinates can be solved for real time by substituting the arrival times and coordinates of sensors. Focusing on the practical engineering where the number of sensors is greater than six, the comprehensive analytical solutions were proposed on account of sensor networks, which formed through the combination of different sensors, and the logistic probability density function. The blasting tests in two mines verified its effectiveness and accuracy. Results show that the locating accuracy of three dimensional comprehensive analytical solutions is superior to the traditional methods. The assumed examples proved that the proposed method performs well under different scales of arrival time errors. This proposed method highlights four advantages: without iterative algorithm, without pre-measured velocity, without initial value, and without square root operations.

Comprehensive Models for Evaluating Rockmass Stability Based on Statistical Comparisons of Multiple Classifiers

The relationships between geological features and rockmass behaviors under complex geological environments were investigated based on multiple intelligence classifiers. Random forest, support vector machine, bayes classifier, fishers classifier, logistic regression, and neural networks were used to establish models for evaluating the rockmass stability of slope. Samples of both circular failure mechanism and wedge failure mechanism were considered to establish and calibrate the comprehensive models. The classification performances of different modeling approaches were analyzed and compared by receiver operating characteristic (ROC) curves systematically. Results show that the proposed random forest model has the highest accuracy for evaluating slope stability of circular failure mechanism, while the support vector Machine model has the highest accuracy for evaluating slope stability of wedge failure mechanism. It is demonstrated that the established random forest and the support vector machine nmodels are effective and efficient approaches to evaluate the rockmass stability of slope.

Effects of sonic speed on location accuracy of acoustic emission source in rocks

To quantitatively study the location errors induced by deviation of sonic speed, the line and plane location tests were carried out. A broken pencil was simulated as acoustic emission source in the rocks. The line and plane location tests were carried out in the granite rod using two sensors and the cube of marble using four sensors, respectively. To compare the position accuracy between line and plane positions, the line poison test was also carried out on the marble surface. The results show that for line positioning, the maximum error of absolute distance is about 0.8 cm. With the speed difference of 200 m/s, the average value of absolute difference from the position error is about 0.4 cm. For the plane positioning, in the case of the sensor array of 30 cm, the absolute positioning distance is up to 8.7 cm. It can be seen that the sonic speed seriously impacts on the plane positioning accuracy. The plane positioning error is lager than the line positioning error, which means that when the line position can satisfy the need in practical engineering, it is better to use the line position instead of the plane location. The plane positioning error with the diagonal speed is the minimum one.

A Comparison of Mine Seismic Discriminators Based on Features of Source Parameters to Waveform Characteristics

To find efficient methods for classifying mine seismic events, two features extraction approaches were proposed. Features of source parameters including the seismic moment, the seismic energy, the energy ratio of S- to P-wave, the static stress drop, time of occurrence, and the number of triggers were selected, counted, and analyzed in approach I. Waveform characteristics consisting of two slope values and the coordinates of the first peak and the maximum peak were extracted as the discriminating parameters in approach II. The discriminating performance of the two approaches was compared and discussed by applying the Bayes discriminant analysis to the characteristic parameters extracted. Classification results show that 83.5% of the original grouped cases are correctly classified by approach I, and 97.1% of original grouped cases are correctly classified by approach II. The advantages and limitations pertaining to each classifier were discussed by plotting the event magnitude versus sample number. Comparative analysis shows that the proposed method of approach II not only has a low misjudgment rate but also displays relative constancy when the testing samples fluctuate with seismic magnitude and energy.

Energy conversion of rocks in process of unloading confining pressure under different unloading paths

Based on energy theory and tests of rocks with initial confining pressures of 10, 20 and 30 MPa under different unloading paths, the processes of strain energy conversion were investigated. The absorbing strain energy for axial compression, the dissipating strain energy for plastic deformation and cracks propagation, the expending strain energy for circumferential deformation, and the storing and releasing elastic strain energy were considered. Unloading paths included the condition of fixing axial pressure and unloading axial pressure, increasing axial pressure and unloading confining pressure, as well as unloading axial pressure and confining pressure simultaneously. Results show that expending strain energy for circumferential deformation has mainly evolved from absorbing strain energy for axial compression in three unloading paths during unloading processes. Dissipating strain energy is significantly increased only near the peak point. The effect of initial confining pressure on strain energy is significantly higher than that of unloading path. The strain energy is linearly increased with increasing initial confining pressure. The unloading path and initial confining pressure also have great influence on the energy dissipation. The conversion rate of strain energy in three paths is increased with increasing initial confining pressure, and the effect of initial confining pressure on conversion rate of strain energy is related with the unloading paths.