The Applied Research Of Algorithm Of Fuzzy C-mens In Electrofacies Analysis

Currently, the work about oil and gas is more and more difficulty, Dealing with electrofacies data increasingly dependent on software and hardware, in the face of a large number of data collection, work of data processing is a new challenge.In data analysis and data visualization, Clustering analysis is a highly effective tool. Because of cluster analysis is widely used, a variety of clustering algorithms are endless increasingly and improved. Fuzzy C means (FCM) algorithm is one of the most widely used cluster analysis methods. However, the FCM algorithm has its limitations.When we use this clustering algorithm, the number of clusters about the data sets must be specified, but the number of clusters is difficult to know in advance. This paper detailed description of the theoretical basis of the traditional FCM clustering algorithm, and carry out a detailed analysis of the specific steps of the algorithm, and recognizing that the two inadequacies of the algorithm.In order to overcome the limitations that the algorithm must know the number of clusters in advance, this article give a new algorithm based on the average information entropy to find the number of clusters. Combined with previous results, this article improved the target clustering function of the FCM algorithm. Using the Lagrange multiplier method to solve the cluster center matrix of the iterative algorithm, when using the IRIS data set to validation the improved algorithms get a better results.In the process of explaining Logging data, identification of lithology and stratigraphic division are an important foundation for content about the formation evaluation and reservoir characterization. According to the different response of the different lithologies in the conventional logging data, based on the improved C-means to Classification of the lithology of the wells. The biggest advantage of using this method is that does not require to know or to set the specific physical characteristics of a variety of lithologic in advance, just give the set of wells rock types, the method can be drilled in the borehole lithology identification and divided, so as to achieve the purpose of identifying different lithologies.In order to study and improved of clustering effect in well logs, select the Xinchang gas field about six wells, using hierarchical clustering algorithm improved these well logs. Take xin5Well for example, get four categories of15small. After get the category of the logging data, refer to the correspondence between well logs and lithology, draw xin5Well relative to the type of division results. Then take the same approach to other well logging data for cluster analysis, and draw a profile of the phase diagram. Cutting logging compared with the known goodness of fit better to prove that the method is practical. When Logging data is abundant, the improved practicality of the clustering algorithm will be more significant.