Application of the artificial bee colony algorithm-based projection pursuit method in statistical rock mass stability estimation

Because of the complexity of factors that affect rock mass stability, the design and decision-making in related engineering cannot rely solely on theoretical analysis and numerical calculation, but depend on the comprehensive judgment of experts. In pursuit of a statistical approach that may improve this disparity, an artificial bee colony algorithm-based projection pursuit (ABC-PP) method is presented for rock mass stability determination. The ABC-PP method is a powerful tool to deal with high-dimension problems, which characterize rock mass stability assessment practice. Two experiments are employed to demonstrate the efficiency of the ABC-PP method. In the first case, the state of stability is classified at two levels: stable and failed, whereas in the second case stability is classified at five levels (1–5) to test the capability of multi-level prediction of the ABC-PP method. Results show that the ABC-PP method could predict the rock mass stability accurately and may also provide the relative importance of specific controls on stability.