基于改进粒子群算法的阵列侧向测井快速反演

电阻率是评价储层含烃饱和度的重要物理参数，电阻率参数的精确度对测井储层参数评价及油气资源勘探开发方案有重要影响。在电法测井中，由于视电阻率影响复杂，如围岩—层厚、井斜、泥浆侵入、井眼、仪器分辨率等因素都会使测量的电阻率值产生较大的误差，测量资料准确度不够高，因而对测量的视电阻率数值进行校正很有必要。通过对电阻率资料反演的方法来进行视电阻率的校正，合适的反演算法一般都是要满足反演效率和精度的诉求。由于反演过程中伴随着大量的正演计算，所以实时反演对反演算法收敛速度及正演计算效率要求较高。提出了一种高速有效的实时反演算法——自适应混沌粒子群反演算法，与基本粒子群算法、马奎特算法相比，该算法在寻优成功率、收敛速度、抗噪性等方面具备一定优势。在反演的过程中，将严格的正演计算用预先制作的阵列侧向测井伪几何因子表格进行替代，采用线性插值法直接调用，在保证计算精度的前提下进一步提高了反演速度。构建三参数反演模型，将反演结果与基于渗流—对流扩散二场耦合的动态泥浆侵入模拟结果进行对比，证明了快速反演算法的适用性及准确性，可用于井场阵列侧向测井资料的实时反演处理。

Fast inversion of array laterolog based on improved particle swarm optimization algorithm

Resistivity is an important physical parameter to evaluate hydrocarbon saturation of reservoir.The accuracy of resistivity parameters has an important influence on well logging reservoir parameter evaluation and oil and gas exploration and development plan.In electric logging，due to the complicated influence of apparent resistivity，such as surrounding rock-layer thickness，well inclination，mud intrusion，well hole，instrument resolution and other factors，the resistivity value measured will produce a large error，and the measured data accuracy is not high enough.Therefore，it is necessary to correct the measured apparent resistivity value.Apparent resistivity is often corrected by the method of resistivity data inversion，and the appropriate inversion algorithm is generally to meet the demands of inversion efficiency and accuracy.Due to the large amount of forward calculation in the inversion process，real-time inversion requires higher convergence speed and forward calculation efficiency of the inversion algorithm.In this paper，a high-speed and effective real-time inversion algorithm adaptive chaotic particle swarm inversion algorithm is proposed.Compared with basic particle swarm optimization algorithm and Marquette algorithm，this algorithm has certain advantages in terms of optimization success rate，convergence speed，anti-noise，etc.In the process of inversion，the strict forward calculation is replaced by the pre-made array lateral log pseudo-geometric factor table，and the linear interpolation method is used directly to improve the inversion speed while ensuring the calculation accuracy.A three-parameter inversion model is established，and the inversion results are compared with the dynamic mud intrusion simulation results based on the two coupled fields of seepage and convection diffusion.The applicability and accuracy of the fast inversion algorithm are proved，and it can be used for real-time inversion processing of wellsite array lateral logging data.