基于云模型及粗糙集理论的围岩稳定性分级方法研究

Stability evaluation for the surrounding rock structure based on the normal cloud model and the rough set theory

围岩分级是隧道及地下工程岩体稳定性评价和安全风险控制的重要依据。针对围岩分级评价过程中评价指标自身的随机模糊性问题,结合云模型相关理论实现了围岩分级定性概念与各待评价指标隶属不同分级的确定度之间的自然转化。同时,充分考虑各个评价指标相对于最终围岩分级结果的重要度不同,利用粗糙集知识发现相关算法,从已有工程数据中挖掘规律,确定各个评价指标相对权重。在计算得到各评价因子隶属不同分级的确定度后,与其对应的权重相乘,并将所有评价因子计算结果进行代数叠加,得到待评价样本隶属于各个特定围岩等级的综合确定度,以其中最大值对应的等级为最终围岩评价结果。工程实例结果表明,该方法评价所得结果与围岩实际等级基本相符,表明该方法有效可行。

The paper is aimed at making an exploration of the tunnel surrounding rock stability based on the Normal Cloud Model and Rough Sets.As regard to the randomness and fuzziness of the evaluation indicators of the surrounding rock categories,qualitative concept has to be converted into quantitative one by means of a cloud model.Particularly speaking,all the continuous attribute data evaluation indicators should be discretized according to the surrounding rock levels.Thus,it would be possible for us to gain some certain degrees of the indicator attributes corresponding to the different rock levels.At the same time,it is possible to identify and determine the significance weight of each evaluation indicator in correspondence with the evaluation result through the knowledge discovery algorithm of the rough sets.In so doing,the underlying rules on how to make the evaluation indicators differentiated on the eventual surrounding rock levels can be found based on the currently existing engineering methods,which can actually be more objective than the traditional ones,such as the Experts Mark and Analytical Hierarchy Process（AHP）.Furthermore,the contribution of each evaluation indicator to the final surrounding rock level can be calculated with the product of the certainty degree and significance weight.In this way,the final integrated certainty degree（ICD）can be determined by the sum of all the evaluation indicators whereas the maximum ICD can indicate the proper categories of the testing samples.The said method can be validated through a group of samples quoted from the corresponding technical literatures.For instance,it is possible to establish an initial decision table and 5 evaluation indicators by using at least 20 existing samples including the uniaxial compressive strength,the RQD,the structural surface condition,the underground water condition and the angle between the tunnel axis and major structural surface.Then,a normal cloud model of the5 evaluation indicators can be worked out with their significance weights gained through discernible matrix algorithm（the rough set）.And,finally,the other set of 7 testing surrounding rock samples can be evaluated via the ICD and the evaluating results generally gained in accordance with the actual level.Thus,the method proposed in this paper can thus be proven both valid and feasible.