A modified Fuzzy C-Means (FCM) Clustering algorithm and its application on carbonate fluid identification

作者：Liu, Lifeng^*; Sun, Sam Zandong; Yu, Hongyu; Yue, Xingtong; Zhang, Dong

来源：Journal of Applied Geophysics, 2016, 129: 28-35.

DOI：10.1016/j.jappgeo.2016.03.027

• Fuzzy C-Means Clustering
• Chaotic Quantum Particle Swarm Optimization
• Fluid factor
• Fluid identification
• Carbonate rock

Considering the fact that the fluid distribution in carbonate reservoir is very complicated and the existing fluid prediction methods are not able to produce ideal predicted results, this paper proposes a new fluid identification method in carbonate reservoir based on the modified Fuzzy C-Means (FCM) Clustering algorithm. Both initialization and globally optimum cluster center are produced by Chaotic Quantum Particle Swarm Optimization (CQPSO) algorithm, which can effectively avoid the disadvantage of sensitivity to initial values and easily falling into local convergence in the traditional FCM Clustering algorithm. Then, the modified algorithm is applied to fluid identification in the carbonate X area in Tarim Basin of China, and a mapping relation between fluid properties and pre-stack elastic parameters will be built in multi-dimensional space. It has been proven that this modified algorithm has a good ability of fuzzy cluster and its total coincidence rate of fluid prediction reaches 97.10%. Besides, the membership of different fluids can be accumulated to obtain respective probability, which can evaluate the uncertainty in fluid identification result.

• 出版日期2016-6
• 单位中国石油大学（北京）; 油气资源与探测国家重点实验室