﻿ 医疗废弃物多目标多周期可持续回收网络优化
 上海理工大学学报  2020, Vol. 42 Issue (5): 479-487 PDF

1. 上海理工大学 管理学院，上海 200093;
2. 上海理工大学 中德国际学院，上海 200093

Optimization of a multi-objective and multi-period sustainable recycling network for medical waste
HUO Qingqing, GUO Jianquan
1. Business School, University of Shanghai for Science and Technology, Shanghai 200093, China;
2. Sino-German College, University of Shanghai for Science and Technology, Shanghai 200093, China
Abstract: With the consideration of the uncertainty of medical waste recovery, and in view of the fact that multi-periodicity and improper treatment of some toxic and hazardous wastes not only pollute the environment, but also endanger the health of operators, a multi-objective multi-period sustainable recycling network model for medical waste under uncertain conditions was established aiming at minimizing economic cost, and environmental impact and maximizing social benefit. In order to reduce the influence of uncertain parameters, the fuzzy opportunity constraint method was adopted to convert the fuzzy constraints in the model into clear corresponding expressions. Taking a medical waste recycling enterprise in Shanghai as an example, a genetic algorithm (GA) was used to solve the model. The example results show that the multi-objective optimization is better than the single-objective optimization in general, and the multi-period recovery network planning is more flexible than the single-period planning.
Key words: medical waste     uncertainty     multi-objective     multi-period     sustainable recycling network     fuzzy     genetic algorithm

1 医疗废弃物可持续回收网络模型 1.1 医疗废弃物可持续回收网络结构

 图 1 医疗废弃物回收网络结构图 Fig. 1 Structure of the medical waste recycling network

1.2 模糊多目标模型 1.2.1 模型假设

a. 医疗机构、处理厂、回收加工厂的位置和数量已知；

b. 暂存处置中心、分类检测中心的候选位置与数量已知；

c. 各节点间的运输成本与运输量和运输距离成正比[22]

d. 运营成本包含运输成本；

e. 节点间的距离并非两点间的直线距离，而是货车行驶距离；

f. 各节点CO2排放量与医疗废弃物处理量成正比；

g. 运输过程中CO2排放量与运输量和运输距离成正比；

h. 员工经过培训而减少的工伤请假时数由节点历史数据预估得出；

i. 各运输车辆与节点设施均符合医疗废弃物运输与处置特殊要求。

1.2.2 符号

$c$ 代表医疗机构， $c$ ∈{1,2, $\cdots, C$ }； $s$ 代表暂存处置中心， $s$ ∈{1,2, $\cdots ,S$ }； $d$ 代表分类检测中心， $d$ ∈{1,2, $\cdots, D$ }； $p$ 代表处理厂， $p$ ∈{1,2, $\cdots, P$ }； $r$ 代表回收加工厂， $r$ ∈{1,2, $\cdots ,R$ }； $v$ 代表运输车辆， $v$ $\{1,2,\cdots,V\}$ $k$ 代表运输路线， $k$ ∈{1,2, $\cdots, K$ }。

1.2.3 参数

${F_s}$ 代表暂存处置中心的固定建设成本； ${F_d}$ 代表分类检测中心的固定建设成本； ${F_p}$ 代表处理厂的固定建设成本； ${F_r}$ 代表回收加工厂的固定建设成本； ${O_s}$ 代表暂存处置中心的运营成本； ${O_d}$ 代表分类检测中心的运营成本； ${O_p}$ 代表处理厂的运营成本； ${O_r}$ 代表回收加工厂的运营成本； $m{a_s}$ 代表暂存处置中心的设备维持成本； $m{a_d}$ 代表分类检测中心的设备维持成本； $m{a_p}$ 代表处理厂的设备维持成本； $m{a_r}$ 代表回收加工厂的设备维持成本； $di{s_{cs}}$ $di{s_{sd}}$ $di{s_{dp}}$ $di{s_{dr}}$ 分别代表两节点间的距离； ${Q_{cs}}$ ${Q_{sd}}$ ${Q_{dp}}$ ${Q_{dr}}$ 分别代表两节点间的运输量； ${q_{{\rm{CO}}_2}}$ 代表车辆载重单位重量的医疗废弃物行驶单位距离排放的 ${\rm{CO}}_2$ ${h_s}$ 代表暂存处置中心为员工提供的培训时数； ${h_d}$ 代表分类检测中心为员工提供的培训时数； ${h_p}$ 代表处理厂为员工提供的培训时数； ${h_r}$ 代表回收加工厂为员工提供的培训时数； $c{a_c}$ $c{a_s}$ $c{a_d}$ $c{a_p}$ $c{a_r}$ 分别代表各节点的处理能力； $r{e_{rt}}$ 代表t时期医疗废弃物回收量的模糊值。

1.2.4 决策变量

${X_{st}}$ 为0−1变量，若在t时期选择候选暂存处置中心，则 ${X_{st}}$ =1，否则为0； ${X_{dt}}$ 为0−1变量，若在t时期选择候选分类检测中心d，则 ${X_{dt}}$ =1，否则为0； ${X_{pt}}$ 为0−1变量，若在t时期选择处理厂p，则 ${X_{pt}}$ =1，否则为0； ${X_{rt}}$ 为0−1变量，若在t时期选择回收加工厂r，则 ${X_{rt}}$ =1，否则为0； ${k}_{cst}$ ${k}_{sdt}$ ${k}_{dpt}$ ${k}_{drt}$ 为0−1变量，若t时期在两节点间运输时选择路线k ${k}_{cst}$ ${k}_{sdt}$ ${k}_{dpt}$ ${k}_{drt}$ 为1，否则为0。

1.2.5 数学模型的建立

 $\begin{split}\displaystyle {E_{1({\rm s})}} = &\sum\limits_{t = 1}^T {\sum\limits_{s = 1}^S {{F_s}} } {X_{st}} + \sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {{F_d}} } {X_{dt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{p = 1}^P {{F_p}} } {X_{pt}} + \sum\limits_{t = 1}^T {\sum\limits_{r = 1}^R {{F_r}} } {X_{rt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{s = 1}^S {{O_s}} } {X_{st}} + \sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {{O_d}} } {X_{dt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{p = 1}^P {{O_p}} } {X_{pt}} + \sum\limits_{t = 1}^T {\sum\limits_{r = 1}^R {{O_r}} } {X_{rt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{s = 1}^S {m{a_s}} } {X_{st}} + \displaystyle\sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {m{a_d}} } {X_{dt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{p = 1}^P {m{a_p}} } {X_{pt}} + \sum\limits_{t = 1}^T {\sum\limits_{r = 1}^R {m{a_r}} } {X_{rt}} \end{split}$ (1)

 $\begin{split}\displaystyle {E_{1({\rm m})}} = &\sum\limits_{t = 1}^T {\sum\limits_{s = 1}^S {{F_s}} } {X_{st}} + \sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {{F_d}} } {X_{dt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{p = 1}^P {{F_p}} } {X_{pt}} + \sum\limits_{t = 1}^T {\sum\limits_{r = 1}^R {{F_r}} } {X_{rt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{s = 1}^S {{O_s}} } {X_{st}} + \sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {{O_d}} } {X_{dt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{p = 1}^P {{O_p}} } {X_{pt}} + \sum\limits_{t = 1}^T {\sum\limits_{r = 1}^R {{O_r}} } {X_{rt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{s = 1}^S {m{a_s}} } {X_{s(t{\rm{ - }}1)}} + \sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {m{a_d}} } {X_{d(t{\rm{ - }}1)}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{p = 1}^P {m{a_p}} } {X_{p(t{\rm{ - }}1)}} + \sum\limits_{t = 1}^T {\sum\limits_{r = 1}^R {m{a_r}} } {X_{r(t{\rm{ - }}1)}} \end{split}$ (2)

 $\begin{split} \displaystyle{E_2} =& \sum\limits_{t = 1}^T {\sum\limits_{c = 1}^C {\sum\limits_{s = 1}^S {di{s_{cs}}} } } {Q_{cs}}{q_{{\rm{CO}}_2}}{K_{cst}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{s = 1}^S {\sum\limits_{d = 1}^D {di{s_{sd}}} } } {Q_{sd}}{q_{{\rm{CO}}_2}}{K_{sdt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {\sum\limits_{p = 1}^P {di{s_{dp}}} } } {Q_{dp}}{q_{{\rm{CO}}_2}}{K_{dpt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {\sum\limits_{r = 1}^R {di{s_{dr}}} } } {Q_{dr}}{q_{{\rm{CO}}_2}}{K_{drt}} \end{split}$ (3)

 $\begin{split} \displaystyle{E_3}= &\sum\limits_{t = 1}^T {\sum\limits_{s = 1}^S {{h_s}} } {X_{st}} + \sum\limits_{t = 1}^T {\sum\limits_{d = 1}^D {{h_d}} } {X_{dt}} + \\ &\displaystyle\sum\limits_{t = 1}^T {\sum\limits_{p = 1}^P {{h_p}} } {X_{pt}} + \sum\limits_{t = 1}^T {\sum\limits_{r = 1}^R {{h_r}} } {X_{rt}} \end{split}$ (4)

 $\begin{split} {\rm{s}}{\rm{.t}}{\rm{.}}\quad \displaystyle\sum\limits_{c = 1}^C {\sum\limits_{s = 1}^S {{Q_{cs}}} } = \sum\limits_{s = 1}^S {\sum\limits_{d = 1}^D {{Q_{sd}}} } \displaystyle{{ = r}}{{{e}}_{rt}},\quad \forall c,s,d \end{split}$ (5)
 $\begin{split} \quad\displaystyle\sum\limits_{s = 1}^S \sum\limits_{d = 1}^D {{Q_{sd}}} =& \sum\limits_{d = 1}^D {\sum\limits_{p = 1}^P {{Q_{dp}}} + } \displaystyle\sum\limits_{d = 1}^D {\sum\limits_{r = 1}^R {{Q_{dr}}} } ,\\ &\forall s,d,p,r \end{split}$ (6)
 $\sum\limits_{c = 1}^C {\sum\limits_{s = 1}^S {{Q_{cs}}} } \leqslant c{a_s},\quad \forall c,s$ (7)
 $\sum\limits_{s = 1}^S {\sum\limits_{d = 1}^D {{Q_{sd}}} } \leqslant c{a_d},\quad \forall s,d$ (8)
 $\sum\limits_{c = 1}^C {\sum\limits_{s = 1}^S {{Q_{cs}}} } \leqslant c{a_s},\quad\forall c,s$ (9)
 $\sum\limits_{d = 1}^D {\sum\limits_{r = 1}^R {{Q_{dr}}} } \leqslant c{a_r},\quad \forall d,r$ (10)
 $\sum\limits_{d = 1}^D {\sum\limits_{p = 1}^P {{Q_{dp}}} } \leqslant c{a_p},\quad \forall d,p$ (11)
 $\sum\limits_{s = 1}^S {{X_{st}}} \geqslant 1$ (12)
 $\sum\limits_{d = 1}^D {{X_{dt}}} \geqslant 1$ (13)
 $\sum\limits_{p = 1}^P {{X_{pt}}} \geqslant 1$ (14)
 $\sum\limits_{r = 1}^R {{X_{rt}}} \geqslant 1$ (15)
 $\sum\limits_{c = 1}^C {\sum\limits_{s = 1}^S {\sum\limits_{d = 1}^D {\sum\limits_{p = 1}^P {\sum\limits_{r = 1}^R {{K_{cst}}} } } } } {K_{sdt}}{K_{dpt}}{K_{drt}} \geqslant 0$ (16)
 ${Q_{cs}},\;{Q_{sd}},\;{Q_{dp}},\;{Q_{dr}} \geqslant 0。$ (17)

2 模型求解 2.1 多周期下模糊参数清晰化

 ${U_{\widetilde {re}_{r{t_i}}}} = \left\{ \begin{array}{l} \dfrac{{t - r{e_{rt_1}}}}{{r{e_{rt_2}} - r{e_{rt_1}}}},\quad t \in [r{e_{rt_1}},\; r{e_{rt_2}}]\\ \dfrac{{r{e_{rt_3}} - t}}{{r{e_{rt_3}} - r{e_{rt_2}}}},\quad t \in (r{e_{rt_2}},\; r{e_{rt_3}}]\\ 0,\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;\;{\text{其他}} \end{array} \right.$ (18)

 $\left\{ \begin{split} {Q_{cs}} \geqslant (1 - {\alpha ^{rt}})r{e_{rt_1}} + {\alpha ^{rt}}r{e_{rt_2}} \\ {Q_{cs}} \leqslant (1 - {\alpha ^{rt}})r{e_{rt_3}} + {\alpha ^{rt}}r{e_{rt_2}} \end{split} \right.$ (19)

 ${\rm{Pos}}\left\{ {{Q_{cs}}} \right. = \left. {r{e_{rti}}} \right\} \geqslant {\alpha ^{rt}}$ (20)
2.2 多周期下多目标模糊化

 ${\mu _{g(x)}} \!=\! \left\{\!\!\!\! \begin{array}{l} 0,\qquad\qquad\qquad\;\,{g_v} \!\leqslant\! {g_{v\min }} \vee {g_s} \!\leqslant\! {g_{s\min }}\\ \left( \dfrac{{{g_v} - {g_{v\min }}}}{{{g_{v\max }} \!-\! {g_{v\min }}}}\right)^\lambda , \;{g_{v\min }} \!\leqslant\! {g_v} \!\leqslant\! {g_{v\max }}\\ 1,\qquad\qquad\qquad\;\,{g_v} \!\geqslant\! {g_{v\max }} \vee {g_s}\! \leqslant \!{g_{s\min }}\\ \left(\dfrac{{{g_{s\max }} - {g_s}}}{{{g_{s\max }} \!-\! {g_{s\min }}}}\right)^\gamma ,\;{g_{s\min }}\! \leqslant\! {g_s} \!\leqslant\! {g_{s\max }} \end{array} \right.\!\!\!\!\!\!$ (21)

 ${\left( {\frac{{{E_1} - {E_{1{\rm{min}}}}}}{{{E_{1{\rm{max}}}} - {E_{1{\rm{min}}}}}}} \right)^\lambda }\geqslant\xi$ (22)
 ${\left( {\frac{{{E_2} - {E_{2\min }}}}{{{E_{2\max }} - {E_{2\min }}}}} \right)^\lambda }\geqslant\xi$ (23)
 ${\left( {\frac{{{E_{3\max }} - {E_3}}}{{{E_{3\max }} - {E_{3\min }}}}} \right)^\gamma } \geqslant\xi$ (24)

2.3 遗传算法编写过程

a. 染色体编码与初始化。根据企业制定的置信水平在所有多目标组合中选择初始种群本体，计算机中每个种群本体都有与其对应的染色体基因编码。多目标集合由n个多目标组合组成，表示为 ${ E}=\{{{ E}}^{1}, {{ E}}^{2}, {{ E}}^{3}, \cdots, {{ E}}^{n}\}$ ，每个多目标组合又包括3个子目标，子目标矩阵表示为 ${ E}= \{{{ E}}_{1}^{1}, {{ E}}_{2}^{1}, {{ E}}_{3}^{1}, {{ E}}_{1}^{2}, {{ E}}_{2}^{2}, \cdots, {{ E}}_{3}^{n}\}$ ，每个子目标矩阵对应唯一的二进制编码，如图2所示。

 图 2 染色体组 Fig. 2 Genome

b. 适应度评估与选择。对各子目标进行综合评价，计算初始本体各项性能的适应度值，然后根据优化准则进一步进行优化，保留适应度值高于上一代的本体，并将其作为新一代本体，同时淘汰适应度值低的个体。本文选取的适应度函数

 $P{\rm{ = }}\displaystyle\sum\limits_{n = 1}^N {\sum\limits_{i = 1}^3 {{E_{ni}}} }$

c. 交叉与变异。按照一定的交叉和变异方法生成新的子代种群，并作为新的一代融入多目标组合集。交叉操作决定遗传算法的全局搜索能力，变异操作增强算法的局部搜索能力。

d. 终止条件。在优化过程中，种群个体相同且连续20代个体都无任何改进或者达到预先设定的最大迭代次数，即可终止迭代。设置最大迭代次数为400，交叉概率为0.9，变异概率为0.05。

3 算　例 3.1 数据来源

3.2 算例结果分析

 图 3 单周期多目标最优 Fig. 3 Multi-objective optimization in a single period

 图 4 第一周期多目标最优 Fig. 4 Multi-objective optimization in the first period

 图 5 第二周期多目标最优 Fig. 5 Multi-objective optimization in the second period

 图 6 第三周期多目标最优 Fig. 6 Multi-objective optimization in the third period

 图 7 CPLEX收敛性 Fig. 7 CPLEX convergence

 图 8 GA收敛性 Fig. 8 GA convergence

4 总结与展望

 [1] ZAMPARAS M, KAPSALIS V C, KYRIAKOPOULOS G L, et al. Medical waste management and environmental assessment in the Rio University Hospital, Western Greece[J]. Sustainable Chemistry and Pharmacy, 2019, 13: 100163. DOI:10.1016/j.scp.2019.100163 [2] KORKUT E N. Estimations and analysis of medical waste amounts in the city of Istanbul and proposing a new approach for the estimation of future medical waste amounts[J]. Waste Management, 2018, 81: 168-176. DOI:10.1016/j.wasman.2018.10.004 [3] ANSARI M, EHRAMPOUSH M H, FARZADKIA M, et al. Dynamic assessment of economic and environmental performance index and generation, composition, environmental and human health risks of hospital solid waste in developing countries; a state of the art of review[J]. Environment International, 2019, 132: 105073. DOI:10.1016/j.envint.2019.105073 [4] HONG J M, ZHAN S, YU Z H, et al. Life-cycle environmental and economic assessment of medical waste treatment[J]. Journal of Cleaner Production, 2018, 174: 65-73. DOI:10.1016/j.jclepro.2017.10.206 [5] MAKAJIC-NIKOLIC D, PETROVIC N, BELIC A, et al. The fault tree analysis of infectious medical waste management[J]. Journal of Cleaner Production, 2016, 113: 365-373. DOI:10.1016/j.jclepro.2015.11.022 [6] ROLEWICZ-KALIŃSKA A. Logistic constraints as a part of a sustainable medical waste management system[J]. Transportation Research Procedia, 2016, 16: 473-482. DOI:10.1016/j.trpro.2016.11.044 [7] GRAIKOS A, VOUDRIAS E, PAPAZACHARIOU A, et al. Composition and production rate of medical waste from a small producer in Greece[J]. Waste Management, 2010, 30(8/9): 1683-1689. [8] MEYER R, CAMPANELLA S, CORSANO G, et al. Optimal design of a forest supply chain in Argentina considering economic and social aspects[J]. Journal of Cleaner Production, 2019, 231: 224-239. DOI:10.1016/j.jclepro.2019.05.090 [9] 曹锋, 郭健全, 刘欣欣. 考虑碳排放的多周期医药逆向物流网络联建研究[J]. 华东师范大学学报(自然科学版), 2017(2): 52-60. [10] GUO J Q, FANG J, GEN M. Dynamic joint construction and optimal strategy of multi-objective multi-period multi-stage reverse logistics network: a case study of lead battery in Shanghai[J]. Procedia Manufacturing, 2018, 17: 1171-1178. DOI:10.1016/j.promfg.2018.10.005 [11] JOHN S T, SRIDHARAN R, KUMAR P N R, et al. Multi-period reverse logistics network design for used refrigerators[J]. Applied Mathematical Modelling, 2018, 54: 311-331. DOI:10.1016/j.apm.2017.09.053 [12] KUMAR V N S A, KUMAR V, BRADY M, et al. Resolving forward-reverse logistics multi-period model using evolutionary algorithms[J]. International Journal of Production Economics, 2017, 183: 458-469. DOI:10.1016/j.ijpe.2016.04.026 [13] 周向红, 成思婕, 成鹏飞. 自营回收模式下再制造逆向物流网络多周期多目标选址规划[J]. 系统工程, 2018, 36(9): 146-153. [14] MANTZARAS G, VOUDRIAS E A. An optimization model for collection, haul, transfer, treatment and disposal of infectious medical waste: application to a Greek region[J]. Waste Management, 2017, 69(9): 518-534. DOI:10.1016/j.wasman.2017.08.037 [15] CUI Y Y, GUAN Z L, SAIF U, et al. Close loop supply chain network problem with uncertainty in demand and returned products: genetic artificial bee colony algorithm approach[J]. Journal of Cleaner Production, 2017, 162: 717-742. DOI:10.1016/j.jclepro.2017.06.079 [16] 许闯来, 胡坚堃, 黄有方. 不确定需求下快递配送网络鲁棒优化[J]. 计算机工程与应用, 2020, 56(3): 272-278. DOI:10.3778/j.issn.1002-8331.1811-0198 [17] 李进, 朱道立. 模糊环境下低碳闭环供应链网络设计多目标规划模型与算法[J]. 计算机集成制造系统, 2018, 24(2): 494-504. [18] 杨晓华, 郭健全. 新零售下生鲜产品闭环物流网络模糊规划[J]. 计算机工程与应用, 2019, 55(2): 198-205. DOI:10.3778/j.issn.1002-8331.1809-0042 [19] 张鑫, 赵刚, 李伯棠. 可持续闭环供应链网络设计的多目标模糊规划问题[J]. 控制理论与应用, 2020, 37(3): 513-527. DOI:10.7641/CTA.2019.80955 [20] 李进. 低碳环境下闭环供应链网络设计多目标鲁棒模糊优化问题[J]. 控制与决策, 2018, 33(2): 293-300. [21] 李伯棠, 赵刚, 张鑫. 低碳闭环物流网络模糊目标规划模型及算法[J]. 计算机集成制造系统, 2019, 25(8): 2087-2100. [22] 姜芳, 郭健全. 多目标多周期多阶段逆向物流网络动态联建[J]. 上海理工大学学报, 2018, 40(3): 274-281. [23] GUO J Q, HE L, GEN M. Optimal strategies for the closed-loop supply chain with the consideration of supply disruption and subsidy policy[J]. Computers & Industrial Engineering, 2019, 128: 886-893. [24] RAMOS T R P, GOMES M I, BARBOSA-PÓVOA A P. Planning a sustainable reverse logistics system: balancing costs with environmental and social concerns[J]. Omega, 2014, 48: 60-74. DOI:10.1016/j.omega.2013.11.006 [25] 狄卫民, 马祖军, 代颖. 制造/再制造集成物流网络模糊优化设计方法[J]. 计算机集成制造系统, 2008, 14(8): 1472-1480. [26] KHISHTANDAR S. Simulation based evolutionary algorithms for fuzzy chance-constrained biogas supply chain design[J]. Applied Energy, 2019, 236: 183-195. DOI:10.1016/j.apenergy.2018.11.092 [27] RIBEIRO V H A, REYNOSO-MEZA G. A holistic multi-objective optimization design procedure for ensemble member generation and selection[J]. Applied Soft Computing, 2019, 83: 105664. DOI:10.1016/j.asoc.2019.105664 [28] CHOI A E S, PARK H S. Fuzzy multi-objective optimization case study based on an anaerobic co-digestion process of food waste leachate and piggery wastewater[J]. Journal of Environmental Management, 2018, 223: 314-323. DOI:10.1016/j.jenvman.2018.06.009 [29] 丁丹军, 戴康, 张新松, 等. 基于模糊多目标优化的电动汽车充电网络规划[J]. 电力系统保护与控制, 2018, 46(3): 43-50. DOI:10.7667/PSPC170006 [30] 胡东方, 吴盘龙. 基于遗传算法的个性化服务型产品设计[J]. 计算机集成制造系统, 2019, 25(8): 2036-2044. [31] HE Z G, LI Q, FANG J. The solutions and recommendations for logistics problems in the collection of medical waste in China[J]. Procedia Environmental Sciences, 2016, 31: 447-456. DOI:10.1016/j.proenv.2016.02.099