改进蚁群聚类算法在火山岩岩性识别中的应用

Application of Improved Ant Colony Clustering Algorithm to Volcanic Rock Lithology Identification

提出了一种基于蚁群算法和模糊聚类算法的改进蚁群聚类算法对火山岩岩性进行识别。介绍了蚁群算法的原理、K-均值聚类算法的实现过程及改进蚁群聚类算法的实现过程。用该方法对火山岩样本数据点进行训练和学习,获得最佳的岩性聚类中心,根据加权信息素浓度和的大小,识别实际测井数据点的岩性。对松辽盆地430个火山岩薄片的实际处理表明,与自组织神经网络及K-均值聚类算法相比,该方法识别准确率高、运算速度快,是一种有效的岩性识别手段。

Put forward is an improved ant colony clustering algorithm based on ant colony algorithm and fuzzy clustering algorithm to identify the volcanic rock lithology accurately. Introduced are the principle of ant colony algorithm, realization process of K-means clustering algorithm and improved ant colony clustering algorithm. After training and learning of the volcanic rock sample-data points, the best cluster centers are obtained. Then the lithology of actual logging data points can be identified by comparing the sum of weighted pheromone concentration values. Practical applications of 430 volcanic chips in Songliao basin show that, compared with SOM as well as K-means clustering algorithm, the improved ant colony clustering algorithm is more accurate, faster calculation and practical in lithology identification.