Multi-Objective Optimization Method With Uncertain Parameters For Grinding-Classification Process

Grinding-classification process is one of the most important links in beneficiation production and the basis for ensuring the quality of subsequent flotation links.Stabilizing the product quality of grinding-classification circuit and improving the production efficiency of the circuit have always been the focus of concentrators and engineers.Optimizing the set points of operation is an important breakthrough to meet the requirements of production quality and improve production efficiency.Due to the complex process mechanism of grinding-classification process,it is very difficult to model and optimize it.In addition,there are many uncertain factors in grinding-classification process due to changeable environmental factors.Therefore,this paper studies the optimal setting of operating variables in grinding-classification process based on multi-objective optimization.The main research contents and innovations are as follows:(1)Aiming at the difficulty of modeling the grinding-classification process,a mill power model based on response surface method is established,which solves the problem that the traditional power modeling depends on some unmeasured variables.Furthermore,the grinding and classification models based on the population balance method are calibrated.The verification results of actual production data show that the model has high accuracy,which provides a basis for building a multi-objective optimization model.(2)A certain multi-objective optimization model for the grinding-classification process is developed to maximize the quality of primary overflow products and the power of the mill.An improved NSGA-Ⅱ algorithm is then proposed to solve the model,and a more evenly distributed solution set is obtained.The multi-attribute decision-making method is used to obtain the optimal setting value of operation variables.The simulation results show that the fineness of overflow products,the mill feed and the mill power are increased by 2.53%,3.83t/h and 40.46k W on average compared to manual operation.(3)Aiming at the uncertainty caused by the fluctuation of grinding stone particle size distribution,combined with fuzzy theory,a grinding-classification process fuzzy chance constrained programming(FCCP)model based on credibility measure is proposed.The actual production data are used for simulation.The results show that the fineness of overflow product,mill feed and mill power are increased by 0.99%,3.73t/h and5.76k W on average compared to deterministic optimization.Aiming at the problem that the credibility measure ignores the preference information of decision makers,the traditional method of dealing with fuzzy events is improved,and a FCCP model for grinding-classification process based on m_λmeasure is proposed.The simulation results show that on the basis of determining the optimization,the optimistic decision-maker further increases the fineness of overflow products by 3.81%,and the pessimistic decision-maker further increases the mill feed by 6.19t/h.