Prediction of reservoir rock brittleness index based on PSO- BiLSTM

Brittleness index is considered as an important rock parameter， for which many scholars have proposed some different calculation methods， however， the measurement methods are more complex or costly. In order to solve the above problems， in the report， the deep learning method was used to predict the brittleness index， which can effectively fuse multiple data and make full use of data to mine the relationship between independent variables and dependent variables. Therefore， based on conventional logging curve data， a bidirectional long short-term memory （BiLSTM） model optimized by particle swarm optimization （PSO） was established. The results showed that the MAE and RMSE of PSO-BiLSTM were 0.050 1 and 0.054 8， respectively， which was more accurate， more practical， and more applicable than other traditional methods.