Abstract: In order to improve the accuracy and computational efficiency of slope stability estimation, an MK-TCRVM based on HIOA (HIOA-MK-TCRVM)was constructed. Then the HIOA-MK-TCRVM was applied to the field of rock slope and soil slope stability estimation. At the same time, other slope stability estimation models (they were based on single kernel two classification relevance vector machine, and support vector machine (SVM) among other algorithms) were established, and a comparison was made between the models and the HIOA-MK-TCRVM for the accuracy and the sparsity. Finally, the effect of optimizing MK-TCRVM parameters based on HIOA was analyzed. The experimental results show that the accuracy rates of HIOA-MK-TCRVM for the training set and the test set are 100%, which are better than those of the other methods. The proportion of the correlation vector of HIOA-MK-TCRVM to the number of training samples is less than 25%,which confirms that HIOA-MK-TCRVM has an obvious sparse effect. After the genetic operation is added to the HIOA algorithm, the evolution speed and the optimal solution of HIOA are improved.

Key words: multi-kernel two-class relevance vector machine(MK-TCRVM) , slope stability estimation, hybrid intelligent optimization algorithm (HIOA), particle swarm optimization(PSO), two-class relevance vector machine(TCRVM)