中文摘要：

概率反演中,马尔科夫链蒙特卡罗是一类重要的后验概率抽样方法,但由于该算法的搜索往往会陷入局部最优解,因而限制了其在具有非唯一解反问题中的应用。鉴于此,本文对基于Metropolis-Hastings算法的多链搜索的方法进行了改进,改进后的方法可以根据搜索结果实时调整链的个数,因而可以在搜索到尽可能多的解的同时节省了多链搜索的时间。最后将该算法应用于一个地下水污染源反问题的求解,计算结果表明改进后的算法对求解非唯一性反问题具有较好的效果。

英文摘要：

A multi-chain sampling method based on Metropolis-Hastings algorithm was used to improve the Markov Chain Monte Carlo(MCMC) method in order to prevent from trapped into the local optimal solutions that often occur to probability inversion by using current MCMC algorithm.The improved MCMC method can adjust the number of chains according to the feedback results from sampling process in real time,so that it can search out the non-unique solutions as much as possible while saving the time of multi-chain search....