论文部分内容阅读
针对现有岩爆预测方法权重的确定带有随意性和主观性问题,提出了一种岩爆烈度分级预测的离散Hopfield神经网络(DHNN)模型。该模型选取应力系数、岩石脆性系数及弹性能量指数作为评价指标,将岩爆等级分为强岩爆、中等岩爆、弱岩爆及无岩爆4级,然后进行编码,不需要对样本数据进行归一化处理,只需转换成“1”和“-1”的二值型模式,编码简单,网络迭代次数少,具有很好的联想记忆能力,使岩爆烈度分级预测更加科学合理,可为深部地下工程岩爆烈度分级预测提供一种新途径。典型岩爆工程实例预测结果证明了该模型的正确性。
Aiming at the randomness and subjectivity of determining the weights of the existing rockburst prediction methods, a discrete Hopfield neural network (DHNN) model for hierarchical prediction of rockburst is proposed. The model selects the stress coefficient, the rock brittleness coefficient and the elastic energy index as the evaluation index, and classifies the rock burst into strong rock burst, medium rock burst, weak rock burst and no rock burst, and then encodes it without the need of sample data For normalization, we only need to convert into binary patterns of “1” and “-1”, which has the advantages of simple coding, fewer network iterations, good associative memory ability, more rigorous prediction of rockburst intensity classification, It can provide a new way for grading the rockburst intensity of deep underground engineering. The prediction of the typical rockburst project proves the correctness of the model.