Hamiltonian Monte Carlo-Based D-Vine Copula Regression Model for Soft Sensor Modeling of Complex Chemical Processes

Multi-Objective Nonlinear Programming Model for Reducing Octane Number Loss in Gasoline Refining Process Based on Data Mining Technology

To simultaneously reduce automobile exhaust pollution to the environment and satisfy the demand for high-quality gasoline, the treatment of fluid catalytic cracking (FCC) gasoline is urgently needed to minimize octane number (RON) loss. We presented a new systematic method for determining an optimal operation scheme for minimising RON loss and operational risks. Firstly, many data were collected and preprocessed. Then, grey correlative degree analysis and Pearson correlation analysis were used to reduce the dimensionality, and the major variables with representativeness and independence were selected from the 367 variables. Then, the RON and sulfur (S) content were predicted by multiple nonlinear regression. A multi-objective nonlinear optimization model was established with the maximum reduction in RON loss and minimum operational risk as the objective function. Finally, the optimal operation scheme of the operating variable corresponding to the sample with a RON loss reduction greater than 30% in 325 samples was solved in Python.