基於SVM 和K-means 之混合架構應用於辨識震測資料

Abstract

近代科技文明的進步，石油與天然氣扮演著舉足輕重的角色。無論是 提供我們生活中燃料的來源，或是生產我們日常用品的石化工業。近幾年中國大陸與印度的經濟崛起，使得全球石油需求量急遽上升，但供給量卻沒有增加，石油供需失衡的情形由近兩年原油價格節節攀升可以看出端倪。雖然現在世界各國致力於開發新能源，但近期還未出現有效且具經濟效益的技術能完全替代石油。 於是，在此論文中提出一個SVM+K-means 的混合架構，對震測資料 作訓練並辨識可能存在油氣的地層，以期能協助地質專家們判斷鑽井位置，進而提高石油探勘的成功率。

While advancing of the modern science and technologies, petroleum and natural gas are playing an essential role. The petrochemical industry provides fuel source and supplies for our daily lives. In recent years, the economic rise of Mainland China and India, it rapidly increases oil demand globally. However, the oil supply has not been matching up. The imbalance situation between oil supply and demand can be figured out according to the skyrocketed crude oil prices in the past two years. Although, countries all over the world dedicate to developing and deploying new energy resources, there is no effective or efficient technology currently to replace the economic benefits of the oil. Therefore, the thesis proposed a hybrid framework based upon SVM (Support Vector Machine) and K-means classifiers. The framework utilizes the trained seismic data to identify, classify potential oil and gas. It can help the geological experts to determine the well drilling locations and improve the successful rate of oil exploration.