AUTHOR=Hu Yating , Li Ouyi , Song Lianteng , Liu Zhonghua , Zhang Qiong , Wu Huilin , Wang Yan , Zhang Yanru TITLE=Acoustic Prediction of a Multilateral-Well Unconventional Reservoir Based on a Hybrid Feature-Enhancement Long Short-Term Memory Neural Network JOURNAL=Frontiers in Energy Research VOLUME=Volume 10 - 2022 YEAR=2022 URL=https://www.frontiersin.org/journals/energy-research/articles/10.3389/fenrg.2022.888554 DOI=10.3389/fenrg.2022.888554 ISSN=2296-598X ABSTRACT=Due to the complexity of unconventional reservoirs, logs obtained are often incomplete or expensive, especially acoustic data, which can provide parameters for the necessary fracturing operations. To address this challenge, a novel hybrid Long Short-Term Memory (LSTM) neural network for effective acoustic log prediction is proposed. The network combines Convolutional Neural Network (CNN) and LSTM into a multi-channel prediction model, where CNN is used to extract spatial features of the logs and LSTM network extracts temporal features with the assistance of an adaptive attention mechanism implemented for key feature recognition. These spatial and time features are consequently feed into a multi-channel network whose output is the predicted log. In addition, the strong heterogeneity of unconventional reservoir also increases the difficulty of data prediction. Therefore, according to the characteristics of unconventional reservoir, we designed three feature enhancement method to mine the hidden information of logs. To prove the performance of the proposed network, a case study is presented with data acquired from Jimusar Oilfield, one of the largest unconventional reservoirs in China. The results show that because of its advanced key-feature extraction capability, the proposed network improves the predication accuracy significantly when compared to conventional networks such as LSTM, CNN and ensemble learning methods such as random forest. It provides efficient and accurate logging data for geological research and reduces exploration costs.