需求不确定条件下的应急物资储备库选址模型构建研究

发布时间：2017-12-28 21:21

本文关键词：需求不确定条件下的应急物资储备库选址模型构建研究　出处：《合肥工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：近几年各种突发事件频频发生,对人们的生命和生活造成了严重的威胁,为了能够提高应急突发事件的预防和处置能力,加强应急物资储备工作建设,保障突发事件发生时对物资的需求,本文在前人的基础之上探讨了需求不确定条件下的应急物资储备库选址问题,主要重点解决三个问题:需求量的预测、目标函数的确定、求解方法的确定。本文针对这三个问题进行了深入的研究,构建了多目标决策的库存选址模型,并以合肥市为例进行实证分析。对突发事件发生前应该如何对应急物资储备库布局选址的问题研究属于前期预防阶段研究,本文综合运用管理学、统计学和选址的建模仿真等理论知识,通过定性分析和定量分析相结合,多角度考虑选址问题。首先,在考虑应急物资储备库选址问题之前,要明确应急物资储备库所涉及范围内的应急物资的需求量,即需求量的预测。通常应急物资的需求量会因为影响因素的变动而表现出不确定性,因此在考虑需求不确定的情况下对选址模型进行研究提高了模型的严谨性。通过阅读文献和查找相关的资料,发现对需求的不确定性的处理方式主要有以下三种:一是针对运用鲁棒性对模型的不确定性进行优化;二是运用模糊数表示不确定性因素;三是运用联系数处理不确定性因素。其中,运用联系数处理不确定因素的方法的优势是可以借助较少的样本数据,达到较精确的预测效果,较适用于突发事件数据短缺的情景,所以本文选择采用联系数的方法对需求进行预测。预测方法是先将需求区间数转化为联系数形式,再对基于联系数的区间数进行灰数预测,最后得到基于联系数的区间灰数预测结果。然后,对传统的选址模型进行了总结、对比和研究,构建包含应急时间响应程度,应急物资覆盖程度和应急物资储备库使用成本的多目标模型,从三个方面对选址模型进行了考虑。随后对比多目标函数的求解方法,运用改进的NSGA II对构建的模型进行求解,在传统的NSGA II算法中结合差分法,改善了解的分布情况,保证了解的多样性和准确性。最后进行算例分析,以整个合肥市为研究对象,根据人口、经济水平和交通通达情况确定合肥市各个区域的需求权数,之后采用灰数预测法对需求量进行预测,然后运用NSGA II和DE-NSGA II方法对模型求解并分析。求解结果表明:一是基于联系数的灰数预测模型具有较高的准确度;二是多目标决策函数的解呈现离散型分布,三个目标函数之间是相互联系和制约的,DE-NSGA II方法在时间上优于NSGA II,并且改善了可行解的分布情况。
[Abstract]:In recent years, all kinds of unexpected events occur frequently, causing a serious threat to people's life and life, in order to improve the prevention and disposal of emergency ability, strengthen the construction of emergency materials storage, security requirements on materials of unexpected incidents, this paper based on the previous study under demand uncertainty emergency material storage location problem, mainly focused on solving the three problems: demand forecasting, the objective function is determined, the solving method of determining. This paper makes an in-depth study on these three problems, and constructs a multi objective decision-making model of inventory location, and takes Hefei as an example to carry out an empirical analysis. On the occurrence of unexpected events should be how to study early corresponding prevention stage belongs to the study of problem base location emergency material reserves, science, statistics and the location of the modeling and Simulation of knowledge based management, through the combination of qualitative analysis and quantitative analysis, considering the location problem of multi angle. First, before considering the location problem of emergency material storage, we should make clear the demand for emergency materials in the scope of emergency supplies repository, that is, the prediction of demand. Usually, the demand for emergency materials will show uncertainty due to the change of influencing factors. Therefore, considering the uncertainty of demand, the location model is studied, which improves the rigour of the model. By reading the literature and find relevant information, there are mainly three kinds of demand uncertainty treatment: one is for the use of robustness to model uncertainty is optimized; two is the use of fuzzy uncertainty factors; three is the use of connection number to deal with uncertainty. Among them, the advantage of using connection number to deal with uncertain factors is that we can use fewer sample data to achieve a more accurate prediction effect, which is suitable for scenarios of shortage of emergency data. So this paper chooses connection number method to predict demand. The prediction method first transforms the demand interval number into the connection number form, and then forecasts the grey number based on the connection number interval number, and finally obtains the interval grey number prediction result based on connection number. Then, the traditional location models are summarized, compared and studied, and a multi-objective model including emergency response time, emergency material coverage and emergency material storage cost is built. The location model is considered from three aspects. Then we compare the solution of multi-objective function, and use the improved NSGA II to solve the model. In the traditional NSGA II algorithm, we combine the difference method to improve the distribution of knowledge, and ensure the diversity and accuracy of the understanding. Finally, the example analysis, the whole city of Hefei as the research object, according to the situation of population, economic level and traffic demand in various regions of the Hefei city to determine the weights, using grey prediction method to forecast the demand, and then use the NSGA II and DE-NSGA II method and analysis of model solution. The results show that: one is the gray number based on connection number prediction model has higher accuracy; two is a discrete distribution solution of multiobjective decision function, the objective function is between the three mutual connection and restriction, DE-NSGA II method is better than that of NSGA II in time, and improve the distribution of feasible solution.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F251

【相似文献】