An improved multi-view collaborative fuzzy C-means clustering algorithm and its application in overseas oil and gas exploration

Abstract

The existing classification methods for oil and gas exploration projects are imperfect for their subjectivity in classification and ranking, which hinders subsequent investment decision-making. This paper proposes a weighted multi-view collaborative fuzzy C-means clustering algorithm improved using the double-layer-nested particle swarm optimization (Double PSO-WCoFCM, or DPSO-WCoFCM in short). The cluster center of the weighted clustering algorithm is optimized by combining the standard particle swarm optimization algorithm and WCoFCM algorithm. The first PSO is placed in the outer layer to optimize the cluster center vector, and the second PSO is nested in the PSO-WCOFCM algorithm in the upper layer and optimizes the weight of each view. Firstly, through three dimensions of resources, performances and risks, evaluation system of overseas oil and gas exploration projects with 20 indexes is established, and the extreme value processing method is adopted to conduct standardized processing of data. Then DPSO-WCoFCM clustering algorithm is applied to classify exploration projects, and membership matrix is obtained. The classification sensitivity coefficient is used to analyze whether the absence of a single evaluation index influent on the classification results of exploration projects, and the weight of each index is determined. Finally, through the weighted sum of the score values of each index, the comprehensive score of each project is obtained, and then the ranking and the investment priority of all projects is carried out. The results show that 24 exploration projects can be divided into 4 classes, of which, 14 first-class exploration projects are preferentially invested, 6 s-class ones are maintained steadily, 3 third-class ones are retained, and 1 fourth-class one is abandoned. In the case of the same class, investment priority is given to the projects with high comprehensive ranking. Compared to the existing subjective classifications, the proposed method can more accurately and effectively classify oil and gas exploration projects, its application produced the direct economic benefit 220 million US dollars. Therefore, the method can be widely used in exploration investment decision-making.