PARTICLE SWARM OPTIMIZATION ALGORITHM BASED FUZZY C-MEANS CLUSTER ANALYSIS FOR DISCONTINUITIES OCCURRENCE IN ROCK MASS

The analysis of occurrence of discontinuities is a basic work for further study on mechanical analysis and stability of rock mass. Traditional analysis of the plot method is insufficient and inadequate. And they are unable to analyze occurrence data of discontinuities accurately. Although the fuzzy cluster method can achieve accurate dominant occurrences, the initial cluster centroids must be given in advance. And the method uses a local optimized algorithm. Different choices of initial guesses of cluster centroids can lead to different partitions of the same data. In order to obtain objective results of classification and the dominant occurrence, a fuzzy c-means cluster analysis method based on the Particle Swarm Optimization(PSO)algorithm is proposed. This hybrid method uses the PSO algorithm to solve the problem. So it can avoid the subjectivity of the initial cluster centroids specified manually, overcome the defects of the fuzzy c-means algorithm such as the local optima and sensitivity to initialization, and analyse the occurrence data efficiently, even if there are a large number of discontinuities. Meanwhile, the optimal cluster number can be determined automatically during the operational process of the algorithm. On the basis of the field measured data from the real rock mass, the proposed approach has been testified to be reliable and reasonable. And the classification and dominant occurrences are more realistic.