【作者】 李慧；

【导师】 周建；

【作者基本信息】 上海大学 ， 管理科学与工程， 2022， 博士

【摘要】 城市生活垃圾收运系统优化问题是优化领域的热点问题之一,合理的系统构建能够大幅减少网络成本、改善环境质量以及提高居民满意度。随着新的垃圾分类标准试点工作在国内各大城市开展,一些亟待解决的问题逐渐显现:其一,垃圾清运企业需要调整组织架构以避免混装混投现象发生,现行清运组工作制度不仅费财耗时,且不利于收运车辆及员工的工作量平衡;其二,湿垃圾中的流质、易腐物较多,在存放和运输过程中可能发生“跑冒滴漏”引发环境污染和居民不满,部分地区尝试建立就地就近处理设施改善这一问题,但因成本高昂无法全面铺开。鉴于此,本文从系统优化的角度出发,采用数学建模、模糊集理论和案例分析的方法对城市生活垃圾收运网络优化问题展开研究。本文的主要内容包括:·不确定需求的识别及相关理论结果的证明。通过对不确定变量的识别、分析实际收运过程中每日的垃圾收运量存在的主观和客观不确定性,本文将顾客点的垃圾量考虑为区间二型模糊数。由于城市生活垃圾回收网络涉及的数学模型较为复杂,而现有研究关于区间二型模糊数的运算相对繁琐,为降低求解难度,本文基于区间二型模糊数的特性提出一系列定义,给出了规则区间二型模糊数的运算法则,进而得到包含区间二型模糊数的机会约束规划模型的清晰等价转化形式。·基于分类运输策略的城市生活垃圾回收网络优化问题建模。新的垃圾分类标准实行以来,为避免混投混运,清运企业采取清运组工作制度对干湿垃圾进行分类收运。在分析其特点后,本文首先对分类运输策略下的城市生活垃圾回收网络优化问题进行研究,将其拆分成两个独立的子问题:针对湿垃圾收运,将湿垃圾处理点选址纳入其中并设计以总成本最小化、工作量平衡以及运输途中引起的二次污染风险最小化为目标的三目标选址-路径优化模型;干垃圾收运则考虑以总成本最小化和工作量平衡的双目标驶进驶出路径优化问题。之后,证明构建的两个包含规则区间二型模糊参数的模型满足清晰转化的条件,并得到其清晰等价模型。·基于混运分装策略的城市生活垃圾回收网络优化问题建模。为了帮助企业提高收运效率,本文提出基于“多舱室”车辆的混运分装策略,并在分析该策略下城市生活垃圾收运网络与分类运输策略下的收运网络在成本结构及约束条件方面的异同后,以总成本最小化、工作量平衡及二次污染风险最小化为目标构建了分舱室同时运输干、湿垃圾的三目标选址-路径优化模型,并得到其清晰等价形式。·针对城市生活垃圾收运系统优化的混合智能算法设计及案例分析。由于城市生活垃圾回收网络优化问题本身的复杂性高,同时模型涉及多个目标,软件和现有算法求解难度很大。因此,本文基于非支配排序多目标遗传算法设计了适用于求解大规模实际问题的算法。首先,改进就近分配算法确定“车辆-路线-湿垃圾处理点”分配策略以求得每一个染色体对应的适应度值,使其在求解大规模问题时能较快输出非支配解。随后,从现有车辆路径规划数据集中找到较符合垃圾回收特性的Homeberger数据集并进行调整,将所提算法的实验结果与多目标粒子群算法和多目标差分进化算法的结果进行比较,对其效率、所得解的质量与稳定性以及多样性进行验证。最后,以垃圾分类回收第一批试点城市之一的上海市嘉定区为例进行实际案例分析,对两种收运策略模型的结果进行对比。结果表明,本文所提出的混运分装策略与现行的分类运输策略相比,在成本控制、工作量平衡以及减少“跑冒滴漏”引起的二次污染方面都表现更佳,其结果不仅有助于试点城市的收运网络优化,也能为其他城市收运策略选择、垃圾处理点选址、收运路径优化提供参考。总体来看,本文的研究成果既丰富了城市生活垃圾收运网络优化领域的研究,也在求解方法上提供了新思路,具有一定的实际价值。更多还原

【Abstract】 Municipal solid waste network design is one of the hot spots in the field of optimization,a reasonable network can significantly reduce costs,improve environmental quality as well as enhance residents’ satisfaction.With the pilot projects of the new waste classification progressively promoted in some cities in China,some urgent problems have gradually emerged.First,the waste collection enterprises are forced to make personnel rearrangement to avoid mixed transportation,since the current classified working system is not only costly and time-consuming,but also not conducive to the workload balance.Second,the wet waste is easy to be rotten and inevitably brings about environmental pollution and residents’ dissatisfaction,whereas the way that some pilot areas adopted to improve this hidden danger by establishing the nearby facilities is too expensive to be fully implemented.Additionally,in the daily operation,the waste amounts are of high objective and subjective uncertainty since they are affected by numerous factors and can only be subjectively estimate by the managers according to the long-term contracts and some inaccurate historical data.Based on the above-mentioned issues,this thesis studies the problem of solid waste collection and transportation network design from the perspective of the system optimization by employing mathematical modeling,fuzzy sets theory and case study.The main contents of this thesis include four parts:First,identification of uncertain demands and relevant theoretical analyses.After identifying the uncertain variables,the waste amounts at the customer points are considered as symmetric trapezoidal interval type-2 fuzzy numbers.Since the mathematical modeling of municipal solid waste recycling network is of high complexity,and the operations of interval type-2 fuzzy numbers in the existing research are relatively cumbersome,this thesis proposes a series of definitions based on the characteristics of interval type-2 fuzzy numbers,and provides the operational law for the regular interval type-2 fuzzy numbers,in order to reduce the solving difficulty.Accordingly,the explicit equivalent forms of the programming model containing interval type-2 fuzzy numbers based on fuzzy chance constrained programming are obtained.Second,modeling of municipal solid waste classified collection network considering uncertain demands and secondary pollution.By analyzing the features of the waste collection process,this thesis firstly focuses on the classified transportation network currently adopted by the collection enterprises and divide the problem into two independent subproblems: the triple-objective location-routing problem for the wet waste aiming at system costs and secondary pollution minimization as well as workload balancing with consideration of disposal site selection;and the bi-objective rollon-rolloff vehicle routing problem for the dry waste considering transportation related costs minimization and workload balancing.Following that,the explicit equivalent models are obtained after proving that the proposed models satisfy the condition of equivalent transformation.Third,modeling of municipal solid waste collection network considering combined collection and transportation strategy.For the purpose of improving the collection and transportation efficiency,the combined collection and transportation strategy on the basis of multi-compart vehicles is put forward,and the corresponding triple-objective locationrouting model aiming at system costs and secondary pollution minimization as well as workload balancing is formulated after analyzing the similarities and differences between the network of this strategy and the classified collection and transportation strategy.Analogously,the model is also explicitly transformed by the proposed theory.Forth,hybrid intelligent algorithm design and case study for the municipal solid waste collection network.Due to the high complexity of the optimization problem of the municipal solid waste collection network as well as the multiple objectives considered,the model is too difficult for software and existing solutions to solve.Therefore,this thesis designs the algorithm suitable for the large-scale practical problems based on the multiobjective non-dominated sorting genetic algorithm.By adjusting the nearest allocation algorithm,the allocation strategy of ”vehicles-routes-wet waste treatments” is determined before the fitness values of the chromosomes are calculated,so that it can efficiently output non-dominated solutions.Then,the Homberger dataset is found from the existing vehicle routing datasets and adjusted according to the characteristics of the problem,and the experimental results of the proposed algorithm are compared with those of the multi-objective particle swarm optimization algorithm and multi-objective differential evolution algorithm from the perspectives of efficiency,stability and diversity.Finally,the case study of Jiading Distric,Shanghai,one of the first batch of pilot cities for waste classification,is taken as an example for comparing the two collection and transportation strategies and provide decision supports for the operation of municipal management departments and enterprises.The results indicate that the combined collection and transportation strategy proposed in this thesis performs better in cost control,workload balance and reducing the secondary pollution compared with the current classified collection and transportation strategy,which not only help to optimize the network of the pilot cities,but also provide references for other cities to establish new networks.Overall,the research results of this thesis can not only enrich the research in the field of urban domestic waste collection and transportation network optimization,but also provide new ideas for related problems in solving methods,which have certain practical values.更多还原