基于模拟退火的M稳健估计精确算法  

An Exact Robust M-Estimation Algorithm Based on Simulated Annealing

作　　者：胡宇 康雄华[1] 曾文宪[1] HU Yu;KANG Xionghua;ZENG Wenxian(School of Geodesy and Geomatics,Wuhan University,Wuhan 430079,China)

机构地区：[1]武汉大学测绘学院,湖北武汉430079

出　　处：《测绘地理信息》2022年第5期54-57,共4页Journal of Geomatics

基　　金：国家自然科学基金(41674002,41774009)

摘　　要：经典选权迭代算法在实现M抗差估计时,通过迭代的方法对参数进行解算,所以仅能求得目标函数最优参数解的近似值。基于M抗差估计的目标函数,将模拟退火算法应用在目标函数最小值的直接求解中,并给出了相应的抗差算法流程。实验结果表明,基于模拟退火算法的M估计的参数解优于经典选权迭代算法的近似参数解;并且在测量数据处理中具有实用意义。When M-robust estimation is achieved by the classical weight-selection iterative algorithm,only the approximate value of optimal parameter solution of the objective function can be obtained.Based on the objective function of M-robust estimation,the simulated annealing algorithm is applied to the solution of the minimum value of the objective function,and the corresponding robust algorithm is given.The experimental results show that,compared with the classical weightselection iterative algorithm,the M-estimation algorithm based on simulated annealing gives a better parametric solution of the objective function,and it has practical significance in measuring data processing.

关 键 词：启发式优化 抗差估计 选权迭代算法 模拟退火算法 M估计 

分 类 号：P207[天文地球—测绘科学与技术]