基于反赋权与MBCT-SR多维云模型算法岩爆预测研究

基于岩爆事故的模糊性与随机性特征,为解决在当前岩爆烈度等级预测研究中,通过正向云发生器经验式计算多维云模型特征值进而导致预测结果主观性较强的问题.本文选取应力比Ts=σθ/σt、岩石脆性指数B=σc/σt以及弹性应变储能指数Wet作为评价指标,结合机器学习理论,采用多维逆向MBCT-SR云发生器算法对岩爆等级进行预测,...

Research on rockburst prediction based on anti-weighting and MBCT-SR multi-dimensional cloud model algorithm

Based on the fuzziness and randomness characteristics of rockburst accidents,in order to solve the problem of strong subjectivity of prediction results by calculating the eigenvalues of multi-dimensional cloud model through the empirical formula of forward cloud generator in the current research on rock burst intensity prediction.In this paper,the stress ratio Ts=σθ/σt,the rock brittleness index B=σc/σt,and the elastic strain energy storage index Wet were selected as the evaluation indexes,and combined with the machine learning theory,the multi-dimensional reverse cloud generator algorithm was used,and the MBCT-SR algorithm was also used to study the grade prediction of rockburst.The digital characteristics were calculated through the example data,then the multi-dimensional cloud model and the dynamic fitness function were established,and finally the optimization algorithm was selected to reversely solve the optimal weight.192 sets of rockburst case data at home and abroad were selected to verify the prediction results of the established model,and the prediction results were compared with those of the anti-weighting one-dimensional cloud model and the conventional reverse cloud generator multi-dimensional cloud model.The results showed that the accuracy of this model for the classification prediction of rockburst intensity could reach 89%,and the accuracy of rockburst tendency prediction could reach 100%.Compared with other models,it had higher accuracy and provides a more scientific,effective and realistic evaluation model for the rockburst prediction.