TIME SERIES PREDICTION OF THE TREND OF HYDRATE RISK USING PRINCIPAL COMPONENT ANALYSIS AND DEEP LEARNING

Hydrate risk strategy is a critical concern in offshore gas and oil production systems. Several empirical models were employed to predict the hydrate formation behaviors related to plugging risk. However, these empirical approaches have limitations in becoming universally used due to their dependency on geometries and fluid characteristics. Also, hydrate kinetic behaviors are statistical and nonlinear relationships on the dependent variables, which means it is difficult to develop the model to describe its behavior. In this work, time series prediction using data-driven methods is applied rather than these model-based methods to analyze the kinetic experimental data during the hydrate formation. Deep learning models, specifically LSTM(Long Short-Term Memory) were used to be trained based on the lab-scale experiment data to make the real-time prediction. The transition trend of hydrate formation from homogenous to heterogeneous particles was predicted by using the model. The prediction was made on hydrate risk indicator, which is PCA(Principal Component Analysis) treated sensor data including pressure, temperature, relative torque, and others. The results suggested that the deep learning techniques incorporated with time series prediction could be a promising method for hydrate risk management.