摘要
为提高露天矿边坡变形预测精度,利用协同进化粒子群(CEPSO)优化多核相关向量机(MK-RVM)的参数,构建协同进化多核相关向量机(CEPSO-MK-RVM),并将此模型应用于露天矿边坡变形预测。将CEPSO-MK-RVM的结果与协同进化多项式核函数相关向量机(CEPSO-PolyRVM)、协同进化高斯核函数相关向量机(CEPSO-Gauss-RVM)及修正果蝇优化下的支持向量回归(MFOA-SVR)的结果进行对比,并分析CEPSO对MK-RVM参数的优化效果。结果表明,CEPSO比标准粒子群优化(PSO)算法具有更好的优化效率及最优解;用CEPSO-MK-RVM模型得到的结果,4个精度指标均优于其余3种方法,边坡变形预测的精度得到有效提高。
For the sake of improving prediction accuracy of slope deformation in open pit mines, a CEPSO- MK-RVM model was built by way of using the CEPSO to optimize the parameters of MK-RVM.And the new model was applied to prediction of slope deformation in open pit mines. An accuracy comparison was made between the results of CEPSO-MK-RVM and the results of optimized Polynomial kernel function (CEPSO-Poly-RVM), optimized Gauss kernel function RVM (CEPSO-Gauss-RVM) and the support vector regression based on modified fruit fly optimization algorithm(MFOA-SVR).The effect of MK-RVM parameters optimization based on CEPSO was analyzed. The experimental results show that the 4 accuracy indexes of CEPSO-MK-RVM model are better than other 3 methods, and that the accuracy of the prediction of slope deformation is effectively improved.
出处
《中国安全科学学报》
CAS
CSCD
北大核心
2016年第11期110-114,共5页
China Safety Science Journal
基金
国家自然科学基金资助(41374007)
江西省自然科学基金资助(20151BAB213031)
江西省教育厅科学技术研究项目(GJJ150592)