Grade prediction of rock burst based on PSO-RVM model

Abstract

A broad and accurate rock burst prediction model is crucial for preventing rock burst disasters in engineering. The relevance vector machine (RVM) algorithm based on the particle swarm optimization (PSO) is proposed for its prediction in this paper. The PSO is used to optimize the kernel parameter inside the RVM, while the RVM is applied to complete the prediction task. Firstly, according to a series of existing classification standards and theoretical research of rock burst, three impact indicators and four rock burst grades are summarized. Next, the PSO-RVM model is trained by the cross-validation method with main indicators as input and rock burst grades as output. Then, the universality and accuracy of the proposed model are verified by different engineering samples. Finally, a specific engineering case is used to verify the practicality of the model, and the prediction accuracy of rock burst grade is up to 100%.