基于RVM-AFOSM模型的边坡可靠度分析

Slope Reliability Analysis Based on Relevance Vector Machine (RVM) and Advanced First Order Second Moment (AFOSM)

针对边坡稳定计算中的不确定性问题,提出基于相关向量机(RVM)的可靠度计算方法。模型利用RVM实现了对边坡安全系数的快速、准确估算,并推求模型导数,解决了传统计算的极限状态函数无法显式表达、难以求解导数的问题。随后,模型采用改进的一次二阶矩(AFOSM)进行可靠度计算,充分发挥其设计验算点位于失效面上的优势,使可靠度计算结果更加准确。对多层边坡实例的可靠度计算结果证明,RVM-AFOSM模型具有计算效率高、计算结果准确、简单易用等优势,因此基于该模型的边坡可靠度分析是可行的,具有良好的应用前景和工程价值。

For the uncertainty problem of slope stability analysis,the reliability calculation method was proposed based on Relevance Vector Machine(RVM).The RVM was used to calculate the slope safety factor in a fast and accurate way,avoiding the difficult that limit state function not expressed explicitly to solve derivatives in the model.Meanwhile,the design point located at the failure surface was calculated by using Advanced First Order Second Moment(AFOSM),so that the reliability determined by the RVM AFOSM was more accurate.Through reliability calculation of the multi layer slope examples,the results show that RVM AFOSM has high computational efficiency,accurate re sults and easy to use and other advantages.Therefore,slope reliability analysis based on the RVM is feasible and has good prospects.