Fuzzy C-Means Cluster Analysis Based on Variable Length String Genetic Algorithm for the Grouping of Rock Discontinuity Sets

Discontinuities have huge impact on civil and mining engineering. To understand the spatial features of discontinuities, it is common to group them into different sets based on orientation. In this paper, a new algorithm is introduced for the identification of discontinuity sets. The new algorithm is developed by combined fuzzy C-means algorithm with variable length string genetic algorithm. In the new method, the number of discontinuity sets is not the necessary input parameter any more. This method is robust, global optimal and totally automatic. To verify its validity, the new method was firstly applied to an artificial data as well as a published data. For artificial data set, the assignment error rate is only 7.4%. For published data set, only 2 discontinuities are assigned to wrong sets. The results indicate that the new algorithm is better than fuzzy C-means algorithm and comparable with other common methods. Afterwards, the new method was utilized to analyze the orientation data sampled at an underground storage cavern site. The new method determines that the ideal number of sets is 3. The new method provided satisfactory results, which confirm its effectiveness and convenience.