Closed-Loop Dynamic Blending Optimization Based on Variational Bayesian and its Application in Industry

There exist uncertainties in raw material components and distribution parameters blending process. Considering the dynamic uncertainty of change inventory, a general closed-loop optimization (CDBO) is proposed. The proposed method consists both data-driven system model-based system. In system, feedback problem reconstructed into linear regression problem, variational Bayesian to obtain mean variance distribution. Then, according incoming materials, value each parameter adjusted by expert rules. Last, ratio obtained chance-constrained programming. To verify effectiveness CDBO, detailed derivation process variable Bayesian, rule programming are established for zinc smelting Numerical studies an industrial application presented demonstrate advantages method. Compared with manual blending, volatility index chemical tests data greatly reduced compliance rates lead increased 6.7% 3.3%, respectively.