Rockburst is a difficult problem in underground resource development and infrastructure construction. With scientific and technological progress, more research technologies and methods to prevent rockburst have been proposed and applied. Research content has become more comprehensive, and positive progress and achievements have been made. However, due to the complexity of rockburst control factors and the sudden occurrence of rockburst, the occurrence mechanism and accurate prediction of occurrence intensity and time are still difficult worldwide problems. In this paper, the research development and achievements of rockburst are reviewed. Firstly, various definitions and types of rockburst are briefly summarized. Then, research progress and achievements in four aspects are comprehensively reviewed, including methods and techniques of rockburst research (theoretical research, numerical simulation, physical model tests and in situ monitoring), mechanism of rockburst, classification and prediction of rockburst (empirical criteria, in situ monitoring, mathematical model approaches and rockburst chart) and prevention and control of rockburst. Of particular significance is that the classification and prediction with prevention and control are summarized in detail. Finally, limitations, deficiencies and some promising directions for future research are listed.

By applying the Wavelet Relevance Vector Machine (WRVM) method, this research proposes the loose zone of roadway surrounding rock prediction. Based on the theory of relevance vector machine (RVM), the wavelet function is introduced to replace the original Gauss function as the model kernel function to form the WRVM. Five factors affecting the loose zone of roadway surrounding rock are selected as the model input, and the prediction model of the loose zone of roadway surrounding rock based on WRVM is established. By using cross-validation method, the kernel parameters of three kinds of wavelet relevance vector machines (RVMs) are calculated. By comparing and analyzing the root mean square (RMS) error of the test results of each predictive model, the advantages and accuracy of the model are verified. In practical engineering applications, the average relative prediction errors of the Mexican relevance vector machine, the Morlet relevance vector machine and the difference of Gaussian (DOG) relevance vector machine models are accordingly 4.581%, 4.586% and 4.575%. The square correlation coefficient of the predicted samples is 0.95 > 0.9, which further verifies the accuracy and reliability of the proposed method.

To combat the uncertainty of the multiple factors affecting roadway surrounding rock stability, five initial indexes are selected for reduction according to concept lattice theory: rock quality designation (RQD), uniaxial compressive strength (Rc), the integrity coefficient of rock mass, groundwater seepage, and joint condition. The aim of this study is to compute correlation coefficients among various indexes and verify the effectiveness of lattice reduction. Alpha stable distribution is used to replace the commonly used Gauss distribution in probabilistic neural networks. A prediction model for the stability of roadway surrounding rock is then established based on a concept lattice and improved probabilistic neural network. 100 groups of training sample data are plugged into this model one by one to examine its rationality. The established model is employed for engineering application prediction with ten indiscriminate sample groups from the Jianlinshan mining area of the Daye iron mine, revealing accuracy of up to 90%. This demonstrates that our prediction model based on a concept lattice and improved probabilistic neural network has high reliability and applicability.