自适应混合反演剩余静校正

Residual static correction by self-adaptive mixed inversion.

本文针对剩余静校正问题的非线性特点 ,分析了已有的线性反演及非线性反演方法的优缺点 ,在此基础上提出了一种计算效率较高、搜寻全局最优解的混合反演方法 ;即在求解过程中每次扰动采用线性寻优方法搜寻局部最优解 ,在迭代过程中采用模拟退火法随机搜寻 ,就可以用较少的迭代次数搜寻全局最优解。此文中还提出了反演参数的选取方法 ,使得该法具有自适应的特点。理论分析和数值计算表明 ,该法不仅解决了剩余静校正较强的非线性全局优化问题 ,而且提高了计算效率 。

In view of the non-linear character of residual static correction,the paper analyzed the advantages and disadvantages of currently-used linear and non-linear inversion methods and presented a new mixed inversion method having high computational efficiency and searching global optimum solution on that basis,that is:in the process of solving the problem the linear optimization is used to look for local optimum solution for each disturbance and simulated annealing method is used for random searching in each iteration,resulted in global optimum solution by lesser iterative number.The paper also presented the selective method for inversion parameters,making the method have self-adaptive character.The theoretical analysis and numeric calculation showed that the method not only solved the non-linear global optimum problem of larger residual static corrections,but also improved the efficiency,being a feasible self-adaptive residual static correction method.