Soft sensing method for octane number of catalytic reforming product based on IEO-MKELM model

Objective  Aiming at the poor real-time performance of octane number measurement of catalytic reforming product, this study proposes a soft sensing model for octane number of reforming product based on the improved equilibrium optimizer-multi-kernel extreme learning machine (IEO-MKELM).

Method The IEO algorithm used the tent chaotic mapping, opposition-based learning strategy to improve the initial population diversity randomness, introduced the optimized nonlinear factor control strategy, Lévy flight and greedy-choice strategy to improve the accuracy and optimization effect. Then, the IEO algorithm was utilized to optimize the weighted coefficients and other parameters of the MKELM and a soft sensing model for octane number of catalytic reforming product was established based on the data from a real petrochemical plant.

Result The accuracy of the proposed IEO-MKELM model was validated using actual measurement data from a refining enterprise. The results showed that the error between the predicted values obtained by the IEO-MKELM model and the measured values was below the order of magnitudes of 10⁻³, indicating higher prediction accuracy compared to other similar algorithm models.

Conclusion This study is of great significance for improving the automation level of catalytic reforming production process.