| Hematite resources are abundant in China, but they are all difficult to be separated for the nature of low grade, weak magnetic, complex mineral composition. Therefore, the roasting and magnetic separation technology is employed to produces the overall concentrate with qualified grade and tailings. The target of optimal operational control of a beneficiation process is to determine the targeted ranges of operational indices, which not only satisfy all kind of constraints but also maximize the overall concentrate grade and output. The mechanism of the relationship between the operational indices and the production indices is complex and unclear. As a result, it is difficult to be described using exact mechanism model and to avail classic optimal method to conduct the optimal decision-making of operational indices. So the decision-making of operational indices is usually carries out by engineers' experience, which often results in poor product quality, high energy consumption and resource consumption and other issues. Therefore, studies on how to determine the optimal decision-making of operational indices instead of engineers'experience is of great significant for improving the overall concentrate grade and output.Unlike classic optimization algorithm, the evolutionary algorithm for the optimization problem has the properties of wide applicability, such as non-differential, discontinuous, multimodal problems etc.. Taking advantage of intelligent evolutionary algorithm can realize the optimal decision-making of operational indices and choose the appropriate setpoints in terms of the actual production conditions and constraints. Therefore, it has important theoretical significance and application value to carry out the research on how to realize the optimal decision-making of operational indices under dynamic environment in order to achieve global optimization of the whole production process by applying intelligent optimization algorithms instead of engineers'experience.Recent years, the differential evolution algorithm has been successfully applied in many fields in terms of its characteristics, such as its simple structure, easy implementation, fast convergence, strong robustness and the easy ability to mix with other optimization techniques. Subject to the above problem, supported by the National Natural Science Foundation project "closed-loop optimal decision-making approach of technology index for complex industrial processes under dynamic environment", the research on the optimal decision-making of operational indices for beneficiation processes based on has been carried out. The detailed work has been summarized as follows:1) The mathematical formulation of the decision-making of operational indices for beneficiation process is presented. The performance indices of the optimal decision-making of operational indices, including overall concentrate grade and output, are described here as well as the constraints and decision variables. Moreover, the difficulties and importance in solving the above problem using the existing optimization methods are analyzed.2) An improved differential evolution algorithm based on orthogonal design and archive mechanism, so called OPEADE, is proposed in this paper, which take advantage of the orthogonal designs to improve the diversity of the population and the archive population to save the elite solutions to guide the evolution of the population. Meanwhile, self-adaptive ? dominance is applied to control the amount of solutions and markedly improve the speed of the algorithms. Meanwhile, the self-adaptive technique can reduce the number of undetermined parameter and improve the usability of the algorithm. In order to understand the mechanism of OPEADE in more detail,5 different algorithms, including classic differential evolution algorithm (DE), hybrid differential evolution algorithm combined with orthogonal experiment design technology (ODE), hybrid differential evolution algorithm based on Pareto dominance and archive mechanism (ArchiveDE), hybrid differential evolution algorithm based on ? dominance and archive mechanism (E-DE) and RPEADE are tested and compared roundly to judge the impact of different techniques. We use 3 different standard trial functions namely ZDT1 (High dimensional convex function), ZDT2 (High dimensional non convex function) and ZDT3 (High dimensional non convex function) to test 6 kinds of algorithms in 4 different evolutionary environments, and we utilize 3 evaluating indicators to test performance, including Convergence indicator ?, multifarious indicator ? and hyper_ratio which could consider convergence and distribution to analyze the optimization results are good or not. The analysis results show that adding every technology can really optimize algorithm in the aspects of convergence diversity and distribution, and the algorithm proposed in the paper have good performance of convergence and distribution, especially in the indicator named hyper_ratio which can evaluation the diversity and convergence of the algorithm comprehensively. For that indicator, OPEADE is the best every time except for a time of secondly which can verify that the algorithm proposed in paper have the advantage of strong convergence, good diversity and uniform distribution in the respect of multi-objective optimization.3) Using the OPEADE algorithm proposed in the paper, integrated with the problems that how decisions to be made in the process of the operation of extracting iron mineral, we put forward a method to be optimized based on OPEADE algorithm about that problem. Because of the complex mechanism of the exploded mathematic model of the indicators to be optimized in the process of extracting iron mineral, it is hard to determine the optimal solution, and we cannot use indicators named Y, ? and hyper_ratio proposed in the paper to judge the property of the result of the problems that how decisions to be made in the process of the operation of extracting iron mineral. So, this thesis will make comparison and judgment between the results of different algorithms by means of using solutions distribution picture and hyper_ratio of the results. It is worthy to mention that because NSGA-? has been applied to the indicators optimization in the process of the operation of extracting iron mineral, so, in this paper, we make a contrast of indicators optimization models in the process of the operation of extracting iron mineral between the two different algorithms based on NSGA-? and OPEADE. The contrast of the s.ne3olution via running OPEADE algorithm and NSGA-II algorithm can illustrate that the hybrid differential evolution algorithm based on orthogonal experimental design and archive technology can improve the model-optimization solution that already exist in the process of the operation of extracting iron mineral that already exist. |
PDF Full Text Request