Application of Data Mining in Development Index Forecast of an Oilfield during the Middle and Later Stage

When the development of an oil field comes into the middle and later stage, it is difficult to grasp the dynamic law due to continuous rising of water cut. A key to realize the optimal control of the whole process of oil development is to realize the accurate forecast of development indexes. This paper mainly investigates the application of data mining methods in development indexes forecast of an oil field during the middle and later stage. At first, the impact factor system of a development index which we want to forecast is established based on the static-dynamic historical data of the oil field by integrating of two data mining methods: grey relational analysis and fuzzy clustering. Then, a support vector regression (SVR) model is developed to establish the input-output nonlinear relationship between the development index and its influence factors. In the end, a practical development indexes forecast problem of an oil field in China is given, successfully solved, and the computational results are presented.