Abstract: Rockburst is one of the major geological hazards in underground engineering.It has strong damage，uncertaintyand suddenness.In order to accurately and effectively predict the tendency of rockburst in underground engineering，a combination weighting-improving set pair analysis prediction model was established.Firstly，predictive indicators were established from three aspects including lithology，stress and surrounding rocks.Secondly，the subjective weight，objective weightand combination weight were calculated by using the C-OWA operator weighting method，CRITIC weighting method and game theory，respectively.Finally，the ternary connection degree in the traditional set pair analysis theory was improved to a quaternary connection degree， and the improved set pair analysis theory was used to calculate the comprehensive connection degree to predict the rockburst level.The model was applied to the prediction of rockburst tendency of Jiama copper polymetallic deposit in Tibet. The calculated comprehensive connection degree （[μ1=-0.302 9] ） suggested the rockburst level of the mine was weak， which was consistent with the reality.The study results show that the established model is suitable for prediction of rockburst propensity.Compared with interactive multicritia decision makeing（TODIM）model and technique for order preference by similarity to an ideal solution（AHP-TOPSIS）model，it has better accuracy and reliability，which provides a new method for rockburst tendency prediction.

Key words: rockburst, C-OWA operator, CRITIC weighting method, game theory, improved set pair analysis