为了提高岩爆倾向性预测模型的精度，确保岩爆多指标综合评价方法中指标赋权方式和关联度函数的选用更加全面合理，建立了基于组合赋权的T-FME岩爆倾向性预测模型。该模型在选取岩石脆性系数、切向应力指数和弹性应变能指数作为评价指标的基础上，由序关系分析法和Vague熵确定指标主、客观权重，引入最小鉴别信息原理对指标组合赋权，最后采用理想点法计算贴近度复合模糊物元得到岩爆倾向性等级。运用国内外15组工程岩爆实例对该模型进行测试，与其他模型预测结果进行对比，并将该模型应用于国内若干实际工程。结果表明：该模型预测精度更高，预测等级更加安全，对国内几项实际工程岩爆倾向性的预测等级与实际情况相符，说明该模型具有较强的适用性。

Due to the limitations of its own operating conditions，many multi-index comprehensive evaluation methods of rockburst are liable to cause low accuracy，and there is currently no unified prediction standard.In order to improve the accuracy of rockburst tendency prediction model,we must ensure that the index weighting method and the selection of the correlation function are more comprehensive and reasonable then a prediction model of T-FME rockburst tendency was established.According to the mechanism and condition of rock ex-plosion，brittle coefficient，tangential stress index and elastic strain energy index were selected as the evaluation indexes from three aspects:Surrounding rock stress，lithological conditions and surrounding rock energy storage.On the one hand，the excessive subjectivity of subjective judgments will affect the objectivity of index weights，on the other hand，in the case of limitated information，entropy weight method excessively depends on the degree of index variation will lead to bias.In order to make up for these deficiencies，the principle of minimum discriminant information was introduced，and the indicators were combined and weighted by combining the subjective and objective weights that the subjective weight of the indicator is determined by the ordinal relationship analysis method and the objective weight of the indicator was determined by the conventional entropy weight method which has been modified by the vague entropy.The T-FME rockburst propensity pre-diction model is based on the fuzzy matter-element analysis method and combines the principles of the TOPSIS method to construct the ideal fuzzy matter-element.The concept of ideal difference-square compound fuzzy matter-element was proposed.Post progress calculation has been optimized，the closeness compound fuzzy matter element was calculated.Finally，the degree of rockburst tendency can be obtained through the closeness analysis.Using the data of 15 domestic and foreign engineering rockburst examples to test the T-FME model and other 4 rockburst propensity prediction models that use different weighting methods and correlation degree functions，and these rockburst propensity prediction models are the ideal fuzzy matter element method based on Vague entropy weight，the ideal fuzzy matter-element method based on expert experience method，the euclid approach degree fuzzy matter-element method based on combined weighting，and the gray favorably membership degree fuzzy matter-element method.By analyzing the results of this test of these models，it is known that the prediction accuracy of the T-FME rockburst tendency prediction model is as high as 93.3%.Compared with other models，the accuracy of the prediction is improved by 6.6%~10.0%，and the prediction of the rockburst propensity level which is biased is higher than actual，so the prediction result is safer.Finally，the model was applied to 5 domestic practical projects，and the prediction results are consistent with the actual rockburst propensity level，which proves that the model has strong feasibility and applicability.

http://www.goldsci.ac.cn/article/2020/1005-2518/1005-2518-2020-28-4-565.shtml