Analyzing Feature Selection of Chromatographic Fingerprints for Oil Production Allocation

摘要

Commingling is employed in the petroleum industry to enhance oil recovery and reduce costs. It is of great importance to monitor the production of each oil well oilfields. Nowadays, more and more oilfields use chromatographic fingerprint to estimate single-zone production allocation. However, how to select the features of chromatographic fingerprint remains an unresolved problem. So far, the features of chromatographic fingerprint are still selected by the professional experts. This leads to a certain degree of subjectivity, which easily results in a poor performance of estimation the single-zone production. To our knowledge, there are few researches exploiting the selection of the features of chromatographic fingerprints. In order to select the features of chromatographic fingerprint, principal component analysis (PCA) method, linear correlation method and the variable importance method used in random forest are exploited in this paper. Meanwhile, a joint feature selection method, which combines the linear correlation method and the variable importance method, is proposed. Experimental results with oil samples from an oil field in Hainan offshore basin show that the proposed method can achieve good results.