Abstract: A model of rockburst prediction was established based on cluster analysis method. Some of the rockburst data in China and abroad were collected by statistics and were chosen as known samples. Three major factors which affect the prediction and evaluation of rockburst were selected as indexes. These factors are ratio of cavern’s maximum tangential stress to rock’s uniaxial compressive strength,ratio of rock’s uniaxial compressive strength to uniaxial tensile strength,and elastic energy index. According to the importance of these factors,combination weight method was adopted to give weight to the three factors so as to make the data more scientific and reasonable. The rockburst data were processed and analyzed by using cluster analysis method,and then by using this method,the intensity levels of 5 rockburst examples were predicted. Research findings show that methods in this research could well classify rockburst grades and predict rockburst accurately. It provides a basis for rockburst prediction.

Key words: rockburst, prediction, cluster analysis, combination weight method