Abstract: In order to improve the accuracy and reliability of surface subsidence prediction of underground mining, a surface subsidence prediction method of underground mining based on HIOA and MK-RVM is proposed. First, the HIOA and MK-RVM algorithms are constructed, and the parameters of MK-RVM are optimized by using HIOA. Then, the prediction model of the surface subsidence geometric parameters and the dynamic subsidence prediction model are constructed based on the optimized MK-RVM. Finally, based on the above model, the rise moving angle, dip moving angle, central moving angle, maximum subsidence and the dynamic subsidence are predicted. The accuracy and reliability of the prediction results are analyzed in order to verify the effectiveness of the proposed method. Experimental results show that the accuracy and reliability of this method are better than single kernel correlation vector machine and support vector machine, and the accuracy and reliability of the new method are excellent. The above analysis confirms the effectiveness of the new method.