Study On Classification Method Of Road Tunnel Surrounding Rock According To The Hierarchical Cluster Analysis

With China’s rapid economic development, road construction has madesubstantial progress. As an important part of it, the construction of roadtunnel has also got a good opportunity to make a considerable advance.In the process of tunnel building, firstly, the quality of engineeringrock should be judged, and the tunnel surrounding rock classificationmethod is a commonly used evaluation means. A variety of classificationmethods have been produced by researchers both in domestic and abroad,they grade surrounding rock from different points of view. The nationalstandards recommends the classification through the value of BQ(basicquality of rock mass), which calculated by Rc、 Kv and other threeInfluencing factors. It divides the surrounding rock into five levels intwo steps, the level Ⅰis the best, and the level Ⅴ is the worst.In practice, due to the large span of level Ⅲ、Ⅳ and Ⅴ, the judgmentof the quality of surrounding rock is too sweeping, which leads to theproperties of the rock masses that are classified as the same level appearquiet different. However, the use of construction methods and designtheory is based on the divided levels. It can not make effective selectionaccording to the individual characteristics of the rock section.Therefore more detailed description is need for rock mass of the abovethree levels, it means the rock classification should be refined, thusimproving the pertinence of construction methods and design theory.In this paper, the hierarchical cluster analysis method was used todivide the level Ⅲ、Ⅳ and Ⅴ into sub-levels, that highlighted the mainengineering geological properties of the tunnel surrounding rock mass. The hierarchical cluster is a multivariate analysis method. It directlycompares the properties of various things, and classifies the similar onesas one class, so that the properties between any two classes are different.If define the rock sections as different samples, the hierarchical clusteranalysis could classify those who get the similar properties into oneclass. In this way, it clearly show the engineering properties of thedifferent rock sections. In order to make rock section be a sample, someindicators who are able to describe its engineering geologicalcharacteristics are needed. Therefore, the rock uniaxial saturatedcompressive strength(Rc)、 coefficient of rock mass integrity(Kv) andunderground water(Unit inflow) were chosen as the indicators. They werealso used in sub-level classification.The methods to calculating the distance between samples and thedistance between classes should be selected before hierarchical clusteranalysis. Through comparative analysis, the Euclidean distance was chosenas the calculation method of distance between samples, and the Furthestneighbor was selected as the calculation of distance between classes.According to the result of hierarchical cluster analysis of13groups ofsamples provided by the engineering example, and the labeled samples,three major classes with dominant factor and five subclasses withsub-dominant factor were divided.18sub-levels were made by integratingthe clustering results with the BQ classification results. After thedivision of the sample s sub-levels, some advices of targeted measureswere given to improve the construction and design methods.