摘要: |
为提高压裂水平井试井多参数自动拟合的计算精度、速度和稳定性,将传统方法、智能算法和并行算法相结 合,提出并行分群式粒子群优化算法,并将高斯-牛顿法与粒子群算法相结合,同时采用OpenMP并行算法求解。结 果表明:在粒子群优化算法中.通过粒子分群使粒子搜索方向趋近于线性,避免了粒子群算法易陷人局部最优的问 题,加快了搜索速度;与高斯-牛顿法相结合保证了计算的稳定性;采用OpenMP并行算法求解降低了模型的复杂度, 提高了计算效率;分群式粒子群优化算法比其他优化算法计算速度更快,计算精度更高,并可在一定程度上为多裂 缝水平并试并解释划分流动阶段。 |
关键词: 压裂水平井 分群式 粒子群算法 试井 自动拟合 |
DOI:10.3969/j.issn.1673-5005.2012.02.023 |
分类号: |
基金项目:高等学校学科创新引智计划项引B08028) |
|
Well testing curve automatic matching of fractured horizontal well based on group particle swarm optimization |
WANG Pei-xi1,ZHANG Jing2
|
(1.School of Petroleum Engineering in China University of Petroleum, Qingdao 266580,China ;2. SINOPEC International Petroleum Exploration and Production Corporation, Beijing 100083, China)
|
Abstract: |
In order to improve the accuracy, speed and stability of multi-parameters automatic matching for fractured horizontal well test analysis, a new parallel group particle swarm optimization algorithm combined with traditional method, intelligence algorithm and parallel algorithm wa$ proposed. "Hie proposed algorithm was combined with Gauss-Newton algorithm and solved by OpenMP parallel algorithm. The results show that in the particle swarm optimization algorithm, the searching direction of particle 叩proaches to linear by clusterir^ particles. The shortcoming of the local optimization is avoided and searching speed is increased. Tlie algorithm combines Gauss-Newton algorithm with particle swarm optimization, which ensures its stability. Meanwhile, OpenMP parallel optimization algorithm can reduce the compiexity of the model and improve the calcuiation efficiency. Compared with other optimization calculation algorithms, the algorithm shows high accuracy, high efficiency and the capability of dividing flow stage to some extent in well teat interpretation for multi-fractures horizontal well, |
Key words: fractured horizontal well group swarm particle swarm optimization well test automatic matching |