基于海洋捕食者优化算法的电磁监测裂缝识别

Identification of fracture in electromagnetic monitoring based on improved marine predators algorithm

水力压裂是实现深层油气增产的关键技术,压裂储层裂缝扩展监测对于油气田勘探开发具有重要意义.电磁法在裂缝监测中优势明显,但传统反演手段依赖于初始模型和约束条件,且反演精度受限,难以准确表征裂缝参数.为此,本文提出了基于海洋捕食者优化算法(Improved Marine Predators Algorithm, IMPA)的电磁监测裂缝参数识别方法.首先,采用Sobol序列初始化种群,提出阶段差异划分的寻优方式,改进了寻优参数,增加了边界自适应约束条件;然后通过Rastrigin函数测试,验证了算法优化的有效性;最后构建基于面积观测和剖面观测下的水力压裂单缝缝长模型与方位模型,分析了IMPA算法在不同噪声影响下对裂缝参数识别的适用性.结果表明,面积观测裂缝参数反演效果更稳定,反演精度更高,缝长反演相对误差小于0.15%,方位反演绝对误差小于0.04°;剖面观测反演效果受噪声影响程度稍大,缝长反演相对误差小于0.2%,方位反演绝对误差小于0.1°.

Hydraulic fracturing is a key technology to increase deep oil and gas production. The monitoring of fracture propagation in fracturing reservoirs is of great significance in the exploration and development of oil and gas fields. Electromagnetic monitoring technology has certain advantages for fracture monitoring. However, traditional electromagnetic inversion methods rely on the initial model and constraint conditions with limited inversion accuracy, which make it difficult to identify fracture parameters accurately. In order to improve the accuracy of fracture identification, in this paper, a fracture parameters identification method based on Improved Marine Predator Algorithm(IMPA) using electromagnetic monitoring technology is proposed. Firstly, the IMPA population parameters are initialized by Sobol random sequence, then through a large number of tests, the stage of population optimization is divided by difference, which changed the traditional method of three equal stages. At the same time, combined with the actual application environment, the optimization parameters are improved, and the boundary adaptive constraints are added. Then, based on the multi-minmum characteristics of Rastrigin function, the algorithm is tested by comparative experimental research method, and the optimization effect of PSO, MPA and IMPA algorithm is analyzed, and the effectiveness of IMPA algorithm optimization is verified. Finally, the single fracture length model and azimuth model of hydraulic fracturing based on area observation and profile observation are constructed,and the applicability of IMPA algorithm for fracture parameter identification under different noise influences is analyzed. The results show that the two observation methods can realize the fine identification of fracture length and orientation. For the area observation,the inversion effect of fracture parameters is more stable with higher accuracy,the relative error of fracture length inversion is less than 0.15%, and the absolute error of azimuth inversion is less than 0. 04°. The inversion effect under section observation is affected by noise,while the relative error of slit length inversion is less than 0.2%,and the absolute error of azimuth inversion is less than 0.1°.