Optimization Method Based On Sampling Analog Technology, Non-ferrous Metallurgy Blending Process

The blending process is the first procedure in nonferrous metallurgical production. Due to a great many blending indexes that coupling seriously, frequently fluctuating raw materials ingredients of which online real time detection cannot be obtained, and the uncertainty of process parameters, it is difficult to conduct the blending optimization calculation. Therefore, simulation technique based on random sampling is studied to address the issue of multi-objective and uncertainty in nonferrous metallurgical blending process, in order to improve raw materials utilization rate, save production cost and optimize blending production.The distribution uniformity of Monte Carlo, Latin Square and Hamersley Sequence is analysed in one-dimensional and multi-dimensional space, and test functions are employed to compare the convergence of these sampling methods. A hybrid sequence sampling method HST approach based on orthogonal Latin square and Hamersley Sequence sampling sequence is proposed. Aiming at the multi-objective optimization issue in copper flash smelting blending process, multi-objective optimization issue is transformed to a series of single objective issues by hybrid sequence sampling method to getε-constraint, to resolve Pareto sets of multi-objective optimization issue, providing guidance with copper flash smelting blending optimization. With respect to the nonlinear interval optimization problem, a nonlinear interval transformation equivalence class model is constructed on the basis of the hybrid sequence sampling, and a hybrid sequence enhanced genetic algorithm is raised to resolve it. Parameter confidence interval alumina blending uncertain model is established for the alumina blending process, and is resolved by the proposed uncertainty optimization approach based on hybrid sequence sampling and genetic algorithm. The validity of the model and the approach is verified in practical industrial application, and the results manifest that the uncertainty optimization method shows more robust, and is less vulnerable to the raw material composition fluctuation.