模拟回火马尔可夫链蒙特卡罗全波形分析方法  

A new full waveform analysis approach using simulated tempering Markov chain Monte Carlo method

作　　者：尹文也 何伟基[1] 顾国华[1] 陈钱[1] 

机构地区：[1]南京理工大学光电技术系,南京210094

出　　处：《物理学报》2014年第16期198-204,共7页Acta Physica Sinica

基　　金：国家自然科学基金青年科学基金(批准号:61101196);国家自然科学基金(批准号:61271332);中国博士后科学基金(批准号:2012M521085)资助的课题~~

摘　　要：针对传统的全波形分析方法不能快速自动处理全波形数据的缺点,提出了一种模拟回火马尔可夫链蒙特卡罗全波形分析法,用于求解全波形数据中的波峰数和峰值位置等参量.该方法采用Metropolis更新策略求解波峰数量和噪声两个参量,以达到快速求解的目的;而峰值位置和波峰幅值则采用改进的模拟回火策略求解,通过添加的主动干预回火步骤实现对参量更新过程的有效探测,以满足对速度或运算收敛性的要求.模拟回火马尔可夫链蒙特卡罗全波形分析方法以马尔可夫算法为基础,仍保持马氏链的收敛性,从而保证本方法具有良好的鲁棒性,实现对全波形数据的自动化处理.To reconstruct the target shape distribution in the distance, full waveform analysis algorithm is utilized by extracting and analyzing the number of the peaks, the time of the peak maximum and other parameters. A novel fast full waveform analysis algorithm （simulated tempering Markov chain Monte Carlo algorithm, STMCMC） is proposed, which is able to process the waveform data automatically. For the different types of the parameters, simulated tempering strategy and the Metropolis strategy are presented. In simulated tempering strategy, due to the demand of speed or accuracy, active intervention tempering is used to control the process of solving the vector parameters. On the other hand, the Metropolis strategy is adopted for non-vector parameters to reduce computation amount. Both the strategies are based on Markov chain algorithm, and meanwhile can hold the convergence of the Markov chain, which makes the STMCMC algorithm robust.

关 键 词：全波形分析 模拟回火马尔可夫链蒙特卡罗方法 主动干预回火 Metropolis策略 

分 类 号：O211.62[理学—概率论与数理统计] O441[理学—数学]