模糊模式识别法及LVQ神经网络在岩体质量分级中的应用研究

Application of Fuzzy Pattern Recognition and LVQ Neural Network in the Rock Mass Classification

针对地下工程的围岩质量分级问题,以某水电站窑洞式安装间为实例,选取岩石质量指标RQD、岩石湿抗压强度RW、岩体完整性系数kv、结构面强度系数kf和地下水渗流量W 5个指标作为评价分级的基本因素,分别采用模糊模式识别直接法和LVQ神经网络模式识别法对其岩体质量进行分级,并对分级结果进行了对比。结果表明:模糊模式识别直接法和LVQ神经网络模式识别法的分类结果基本一致,只在单个岩层略微有所出入;基于LVQ神经网络的岩体分级有较好的分类识别性能,该分级方法比较简单、易于掌握,对类似工程设计和施工具有参考价值或指导意义。

Taking the cave-type installation room of a hydropower station as an example, this paper selected the rock quality index （ RQD ）, rock wet compressive strength （ R W ）, rock mass integrity factor （ k v ）, structural plane strength factor（ k f ） and groundwater percolation （ W ） as the basic factors for the evaluation grading to study the quality classification of surrounding rock in underground engineering.The fuzzy method and the LVQ neural network method were used to classify the rock mass quality, respectively.The grading results showed that, firstly, the classification results of the fuzzy method and the LVQ neural network method were basically the same, only slightly different in the single rock formation.Secondly, rock mass classification based on LVQ neural network had better recognition performance.This method was relatively simple and easy to master. It had reference value or guiding significance to similar engineering design and construction.