基于粒子群和支持向量机的裂缝识别

裂缝识别是裂缝性储层勘探和开发研究中所面临的关键问题和难点之一。基于常规测井资料对裂缝的响应特征,提出粒子群优化算法(PSO)和支持向量机(SVM)相结合的裂缝识别方法(PSO-SVM)。以麻黄山西区块延安组和延长组储层为例,应用交会图技术分析能较好响应裂缝的常规测井参数,用粒子群优化算法对模型参数进行全局优化选取,从而建立起研究区裂缝识别模型。用建立起的模型对研究区单井进行裂缝识别研究,将识别结果与取心照片和测井曲线进行对比,绘制出综合柱状图。实际分析表明,基于粒子群优化算法和支持向量机的裂缝识别方法的识别结果与实际地质情况相符,能较好地反应裂缝发育情况。

PSO-SVM-based fracture identification method

Fracture identification is one of the key issues and difficulties in exploration and development of fractured reservoirs.Based on response characteristics of conventional logging data to fractures,we proposed an identification method called PSO-SVM that integrates the Particle Swarm Optimization(PSO) with the Support Vector Machine(SVM).By taking the reservoirs in the Yan'an and Yanchang formations in the western Mahuang-shan block for examples,we identified conventional logging parameters that can respond well to fractures through a crossplot analysis,and performed an overall optimal selection of model parameters by using the PSO.Based on these works,we built a model for identification of fractures in the study area.The model was applied to single well in the study area for identification of fractures and a synthetic column map was produced by comparing the model outputs with core photos and logging curves.The application of the method shows that the modeling result matches well with the geological reality and can truthfully reflect the growth of fractures.