利用阻尼型高斯牛顿法的激发极化数据聚焦反演

摘要

利用阻尼型高斯牛顿法求解反演目标函数，并将粗糙度矩阵约束和最小支撑泛函约束以及参考模型和模型参数界限约束引入到目标函数中，同时结合预条件共轭梯度法进行反演迭代，实现了激发极化数据三维聚焦反演快速计算。通过对几个典型模型的试算，并与传统光滑模型约束的反演结果进行比较，表明该反演算法的可靠性和稳定性较好，并且算法速度快，占用内存低且易于并行化。%In this paper,roughness,minimum support functional constraints and reference model,model parameter range constraints are incorporated into the objective function of induced polarization (IP)data inversion.Then we use the damped Gauss-Newton method to find the optimization of the objective function,with the model update being calculated using a pre-conditioned conjugate gradient algorithm.The method is demonstrated on two synthetic models by comparisons with the results obtained by the traditional inversion method based on smoothness stabilizing functional constraint.The model tests show that the proposed inversion algorithm has good reliability,stability,and high speed.The comparisons show that the program is more convergent and this approach helps to generate more focused and resolved images of blocky geoelectrical structures than traditional inversion approach.