基于自适应差分进化算法的阵列侧向测井快速反演

严格意义上的电测井反演在每一次反演算法迭代过程中都要进行多次正演计算，耗时长，在复杂三维地层模型中电测井数据的实时反演不太可能实现，对现场实时高效处理电测井数据提出了挑战.本文基于径向阶跃介质模型，预先计算正演响应，在反演过程中直接进行线性插值调用，大大节省反演时间.提出一种自适应差分进化算法，将阵列侧向反演问题转化为一种非线性的全局优化问题，通过与经典差分进化算法、马奎特算法进行比较，自适应差分进化算法具备寻优成功率高（90%）、平均迭代次数少（21）、抗噪性好的特点.利用不同侵入条件下的层状介质模型进行算法验证，结果表明改进算法反演的电阻率较视电阻率更加接近储层真实电阻率，可以满足油气藏评价的需要.

Fast Inversion of Array Laterolog Based on Adaptive Differential Evolution Algorithm

In strict sense of electric log inversion, multiple forward calculations are performed many times during each iterative procedure of inversion algorithm, which takes a long time. In particularly, real-time inversion of electrolog data is unlikely to be achieved in complex three-dimensional stratigraphic models. There is a great challenge to process efficiently electrolog data in real time. In this paper, based on radial step medium model, forward response is pre computed and linear interpolation is called directly in inversion process, which greatly saves inversion time. An adaptive differential evolution algorithm is proposed to transform array lateral inversion problem into a nonlinear global optimization problem. Compared with classical differential evolution algorithm and Marquette algorithm, the adaptive differential evolution algorithm has characteristics of high search success rate (90%), low average number of iterations (21), and good noise immunity. The algorithm is tested under layered media model with different invasive conditions. Resistivity using the algorithm is closer to the true resistivity than apparent resistivity, which meets evaluation needs of oil and gas reservoirs.