Lithology identification from well-log curves via neural networks with additional geologic constraint

• DOI: 10.1190/geo2020-0676.1
• Publication Date: 2021-06-24T16:43:00Z
• Authors: Chunbi Jiang; Dongxiao Zhang; Shifeng Chen
• Keywords: Lithology; Sequence (biology); Identification; Feature (linguistics)

Lithology identification is of great importance in reservoir characterization. Recently, many researchers have applied machine-learning techniques to solve lithology problems from well-log curves, and their works indicate three main characteristics. First, most predict lithofacies using features measured during logging, whereas very few consider adding stratigraphic sequence information that available prior drilling this problem. Second, studies properties one depth point, take the influence neighboring formation into account. Third, due a lack publicly interpreted data, previous research has concentrated on applying different algorithms private data set, making it impossible perform comparison. We developed framework classification problem curves by incorporating an additional geologic constraint. The constraint unit, we use as feature. evaluate types recurrent neural networks (RNNs), bidirectional long short-term memory, gated unit (Bi-GRU), GRU-based encoder-decoder architecture with attention, well two 1D convolutional (1D CNNs), temporal network multiscale residual network, set North Sea. RNN-based CNN-based can process sequential enabling model access formations when performing prediction at particular depth. Our experiments improves performance models significantly, overall better more consistent.