Prediction of Rockburst Based on Multidimensional Connection Cloud Model and Set Pair Analysis

Abstract

Prediction of rockburst involves numerous random and fuzzy indicators asymmetrically distributed in finite intervals. Herein, a novel multidimensional connection cloud model was introduced to depict uncertainties and distribution characteristics of indicators, and the fuzziness of the classification boundary. In the model, numerical characteristics of the connection cloud model were first determined on the basis of the set pair analysis (SPA) of measured indicators relative to the classification standard. Then a multidimensional connection cloud model was presented to express the interval-valued classification standard. Next, based on the combination weight specified by a distance function, the integrated connection degree for a grade was identified for the sample. Finally, a case study and comparison of the proposed model with the normal cloud model and extensible evaluation method were performed to confirm the validity and reliability of the proposed model. The results show that the proposed model, with a quicker and simpler calculation process than a normal cloud model, can describe the multiple types of uncertainties of interval-valued indicators and overcome the subjectivity when determining numerical characteristics of the cloud model.