Hydrocarbon vapor temperature treatment in the ATP process based on the DOCoR model

Accurate prediction of hydrocarbon vapor temperature is crucial to the operation of the Alberta Taciuk Processor (ATP) hydrocarbon vapor handling system. However, the problem of hydrocarbon vapor temperature prediction in ATP systems remains insufficiently studied. Existing models rarely distinguish variable-specific dynamic changes from system-level interactions. To address this limitation, this study proposes a novel deep learning model, termed DOCoR, which achieves effective disentanglement and fusion of individual characteristics and system-level coordination in multivariate time-series data through the synergistic integration of a private encoder, a shared encoder, and an orthogonality constraint mechanism. Compared with the traditional LSTM model, the proposed method improves RMSE, MAE, (R2), and MAPE by 41.60%, 46.93%, 8.72%, and 46.15%, respectively; compared with the GRU model, the corresponding improvements are 29.42%, 44.23%, 4.12%, and 43.24%. These results demonstrate the superior predictive accuracy of the proposed model. This study provides a reliable solution to the challenging problem of hydrocarbon vapor temperature prediction in ATP systems.