改进区间多目标进化优化方法及其在RFID阅读器布局中的应用

发布时间：2018-10-31 21:29

【摘要】：目标函数含有区间不确定性的多目标优化问题广泛存在,多目标进化优化算法是解决该类问题的可行方法。当目标函数较多时,如何有效比较区间目标的优劣,以获得分布性、延展性等较好的Pareto前沿成为该类问题研究的焦点。现有研究往往采用单一的区间数大小比较方法,不能全面反映区间的信息;此外,已有方法对于进化过程中的知识应用不足,针对上述这些问题,本文重点开展了如下研究工作:(1)提出融合两种区间数可能度排序比较策略的区间多目标进化优化算法(Interval Multi-objective Evolutionary Optimization Problems Algorithm,IMOP),并用于解决实际问题。首先,论文分析了μ比较和可能度P比较两种策略的特点,然后,提出了基于上述两种方法的μ"昉混合比较策略,并将该策略与NSGA-II算法框架进行融合,给出了基于混合比较策略的区间多目标进化优化算法。在数值函数中的应用验证了所提方法的有效性。(2)提出了基于有向图的改进区间多目标进化优化算法。在研究内容(1)的基础上,进一步研究进化过程中知识的利用,首先,基于有向图理论给出了邻占优的概念,并根据该概念以及个体之间的μ"昉支配关系构建了有向图模型;其次,依据有向图中形成的进化方向对路径进行延伸,以此根据种群的收敛性方向预测出优势个体;最终,本文将根据有向图模型所预测出的个体与原有个体执行交叉操作,从而起到引导种群进化趋势的作用。算法在数值函数中的应用验证了算法的有效性。(3)算法在煤矿井下射频识别(Radio Frequency Identification,RFID)阅读器布局中的应用。RFID技术被广泛地应用于井下人员的定位,并起到了令人喜悦的效果。而井下RFID系统阅读器的布局直接关系射频识别的可靠性,而煤矿井下由于环境等的影响、RFID的价格等,在考虑布局经济性和可靠性的前提下,使得RFID的布局具有不确定性,鉴于此,本文将上述所提方法应用于该实际问题中。给出了煤矿井下RFID布局的多目标不确定建模,并利用上述方法对该模型进行求解,最后对结果进行了对比分析。
[Abstract]:Multi-objective optimization problem with interval uncertainty exists widely in objective function, and multi-objective evolutionary optimization algorithm is a feasible method to solve this kind of problem. When there are more objective functions, how to compare the advantages and disadvantages of interval targets effectively in order to obtain better Pareto frontier such as distribution and ductility becomes the focus of this kind of research. The existing studies often use a single method to compare the number of intervals, which can not reflect the information of the interval. In addition, existing methods are inadequate for the application of knowledge in the course of evolution, and in response to these problems, The main work of this paper is as follows: (1) an interval multi-objective evolutionary optimization algorithm (Interval Multi-objective Evolutionary Optimization Problems Algorithm,IMOP) is proposed to solve the practical problems. Firstly, the paper analyzes the characteristics of 渭 comparison and possibility P comparison, then proposes a 渭 "Fang hybrid comparison strategy based on the two methods, and combines the strategy with the NSGA-II algorithm framework. An interval multiobjective evolutionary optimization algorithm based on hybrid comparison strategy is presented. The application in the numerical function proves the effectiveness of the proposed method. (2) an improved interval multi-objective evolutionary optimization algorithm based on directed graph is proposed. On the basis of research content (1), the use of knowledge in evolutionary process is further studied. Firstly, the concept of neighbor dominance is given based on directed graph theory. According to the concept and the 渭 "Fang dominating relation between individuals, a directed graph model is constructed. Secondly, the path is extended according to the evolutionary direction formed in the digraph, and the dominant individuals are predicted according to the convergence direction of the population. Finally, based on the digraph model, the individuals predicted by the digraph model will perform cross operations with the original individuals to guide the evolution trend of the population. The application of the algorithm in the numerical function verifies the validity of the algorithm. (3) the application of the algorithm in the layout of (Radio Frequency Identification,RFID reader in coal mine. RFID technology is widely used in the location of underground personnel. And it has a delightful effect. The layout of downhole RFID system reader is directly related to the reliability of RFID, while the layout of RFID is uncertain because of the influence of environment, the price of RFID and so on. In view of this, this paper applies the above method to the practical problem. The multi-objective uncertain modeling of RFID layout in coal mine is presented, and the model is solved by using the above method. Finally, the results are compared and analyzed.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP18;TP391.44

【参考文献】