Design and Multiobjective Optimization of Green Closed-Loop Manufacturing-Recycling Network Considering Raw Material Attribute

: Regarding decision planning in the electronic manufacturing industry, this paper designs a green closed-loop manufacturing-recycling network for multiperiod production planning for multiple products. The network considers the tradeoff between production costs and environmental pollution induced by production scraps. Therefore, a mixed integer programming model with a dual objective is designed to achieve environmental protection and reduce production costs through resource recovery and utilization. At the same time, the recycled materials are considered to be treated, not entirely new, which could affect the manufacturing qualiﬁed rate. Thus, material attributes are proposed to distinguish new raw materials from recycled (second-hand) ones through the closed-loop manufacturing-recycling process to enhance the manufacturing qualiﬁed rate. In order to solve the dual-objective optimization model and realize optimal decisions, an epsilon constraint is designed to generate a nonextreme solution set by changing the original feasible region. The results show its ability to obtain a more balanced solution in terms of cost and environmental factors compared with the fuzzy-weighted method. Meanwhile, the analysis proves that the dual-objective optimization model with distinguishing material attributes can improve the efﬁciency of the manufacturing qualiﬁed rate and achieve a win-win situation for production and environmental protection during enterprise production.


Introduction
The progress of mankind cannot be separated from industrial. With the proposal of industry 4.0, digital intelligent manufacturing becomes more meaningful to achieve realtime connectivity of people, equipment, and products through cyclic manufacturing and remanufacturing in a closed-loop supply chain [1]. However, although this has changed the manufacturing mode, the environmental impact of manufacturing cannot be ignored. Manufacturing with agglomeration exacerbates environmental pollution and affects people living in surrounding towns [2,3]. This has drawn attention to environmental management, energy conservation, and emission reduction. This has prompted decision-makers to take proactive measures to help companies to improve their productivity and reduce environmental pollution [4]. Thus, environmental performance is valued by enterprises and has become the driving force of production together with economic performance [5]. Recycling has become a key measure to improve environmental benefits [6]. From the perspective of an environmental protection system, for example, the recycling of batteries can ensure that the toxic substances in batteries do not cause secondary pollution to the environment [7]. Recycling is of great significance to reduce the burden on the environment. From the perspective of the manufacturing process, the waste of resources is caused by defective products and discarded parts in the production process. Recycling and remanufacturing can achieve sustainable development and improve resource utilization [8]. For example, in the automobile industry, cars and their parts can be recycled for repair, effectively capturing the residual value of waste [9]. Thus, a production scheduling system is the key to saving energy under considering environmental pollution. Regarding sustainable development [10], in this paper, the recycling process is incorporated into the manufacturing process to form a closed-loop manufacturing network to deal with the resource utilization problem. Further, a closed-loop manufacturing network structure, including production, inventory, and recovery and transportation units, is established to enable more accurate and efficient evaluation of manufacturing and recovery performance. The implementation of sustainable development in the manufacturing industry could drive the development of sustainable production via recycling worldwide [11]. It should be noted that although recycled materials (second-hand materials) can be reused, they still have some disadvantages, as they are not as perfect as newly purchased materials. Therefore, recycled and new materials should be distinguished from each other. To improve production efficiency, reduce waste discharge, and protect the environment, a production model that distinguishes material attributes (purchased new materials and second-hand materials) in the closed-loop manufacturing process should be developed. When the recycled materials are not distinguished from new materials, the result is a lower yield of qualified production in the manufacturing process, because of the combination of new and second-hand materials. This problem can be dealt with by distinguishing material attributes to allow a high-quality manufacturing qualified rate in a closed-loop manufacturing network.
On the contrary, if the byproducts of manufacturing are directly discarded, there is the potential to harm the environment. According to an investigation of the contamination of electronic waste in soil, the large amount of heavy metals in soil poses a risk to the environment, and the metal concentration is related to the electronic component waste [12]. Heavy metals pollute the environment through toxicity, persistence, and accumulation and are the most stubborn pollutants [13,14]. As people are focusing more on the environment, heavy metal pollution in soil has become a worldwide environmental problem, and this type of pollution has become the biggest threat to human beings [15]. Thus, a recycling manufacturing network is required to effectively reduce heavy metal pollution. The integration of forward logistics and reverse logistics can reduce environmental pollution by heavy metal waste through the secondary utilization of resources [16]. Driven by the environment, reverse logistics is becoming more and more important [17]. The multiobjective function is formed when multiple factors are considered in a model to effectively balance the problems faced in production. There are some solutions to the dual-objective optimization problem considering environmental and production cost factors. However, in closed-loop manufacturing, there have been few studies on dual-objective optimization. The genetic algorithm, fuzzy method, and weighting method can all be used to solve dual-objective optimization. In the weighting method, the scale of the objective function greatly affects the optimal solution. In the genetic algorithm, the optimal value of the same object represented by two endpoints in the Pareto solution set varies greatly between different optimal solutions. Thus, in dual-objective optimization, which is difficult to balance, the epsilon constraint is used to solve the dual-objective optimization decision problem, because the epsilon constraint produces nonextreme solutions and can control the number of effective solutions according to the range of objective functions, providing more choices for decision makers, which can effectively avoid these problems and achieve the most satisfactory solution to two objectives.
Therefore, in this work, a green closed-loop manufacturing-recycling network was designed to improve resource utilization by adding recycling to the manufacturing process. In addition, distinguishing materials into new and second-hand attributes can highlight the value of recycling and amplify the advantages of low-cost manufacturing and high-level environmental production decisions. In other words, this process can effectively reduce the output of production costs and enhance the use of materials during production by distinguishing material attributes. Additionally, incorporating a tradeoff between closedloop manufacturing decisions and heavy metal pollution into the design of manufacturing networks can achieve environmental protection and reduce the impact of heavy metal waste on the environment. An epsilon constraint is used to solve the tradeoff between environmental pollution and production costs to change the feasible region of the original problem to obtain a richer Pareto optimal solution set and achieve the double advantage purpose, whereby the lowest cost and least pollution are realized at the same time, achieving the maximum production benefit.
In this study, environmental protection is regarded as a long-term development strategy and the economic benefits and environmental benefits of production are balanced. The rest of the article is structured as follows: Section 2 discusses the relationship between manufacturing and the environment. Section 3 describes manufacturing-recycling problems. In Section 4, the formula of the model of electronic assembly is introduced, and in Section 5, a multiobjective optimization method is proposed to solve the dual-objective problem considering environmental pollution and production costs. Sections 6 and 7 analyze the results and discuss the conclusions of the study.

Literature Review
Production planning and scheduling is the focus of decision makers in manufacturing. Challenges such as the manufacturing complexity, product lead times, and customer satisfaction need to be addressed [18]. According to a previous survey, enterprises focus more on production costs, operational efficiency, and compliance with environmental regulations, but do not make long-term plans for environmental protection [19]. Thus, environmental factors are rarely a consideration for decision makers. Environmental protection is the social responsibility of manufacturers, and the adoption of sustainable manufacturing technology helps to achieve environmental protection and the economic benefits of enterprises [20]. Remanufacturing has attracted worldwide attention because of its great potential in environmental and economic aspects [21]. Remanufacturing can not only save the cost of procurement but can also allow the repeated utilization of resources and relieve environmental pressure. In a multiproduct, multicycle manufacturing process, products involve components, parts, and other different materials. Remanufacturing processes and reuses these waste materials. In the field of auto parts, remanufacturing can reduce the waste of resources and alleviate the energy crisis [22]. Similarly, the conversion of end-of-life products into new ones is an important factor for small-and medium-sized enterprises to allow them to remain competitive in the global market [23]. Thus, a new model that can be used to carry out production and solve the problem of recycling has become the focus. A closed-loop manufacturing-recycling network is designed in this paper. It combines manufacturing and recycling processes and achieves the recycling of waste through a collaborative operation involving forward logistics and reverse logistics. The closed-loop manufacturing-recycling network focuses on the manufacturing process and recycling process and drives production by enterprises through production demand while the recycling process is carried out. The closed-loop manufacturing-recycling network can clearly identify the waste of resources brought by the production process and the utilization of resources achieved by the recycling process. However, current closed-loop manufacturing-recycling networks do not distinguish between recycled and purchased materials [24]. In other words, the recycled materials can be reused, but second-hand recycled materials have certain defects compared with newly purchased materials, which is often one of the reasons for a reduction in the qualified rate of production. Therefore, the best idea is proposed. The paper defines x and y attributes by distinguishing the differences between these two materials in the closed-loop manufacturing-recycling network. Distinguishing second-hand recycled materials from newly purchased materials and manufacturing them accordingly is conducive to consolidating the advantages of product manufacturing and improving production efficiency.
Although the closed-loop manufacturing-recycling model alleviates the environmental pollution caused by some waste through the recycling process, the environmental pollution caused by heavy metal discharge in the soil from the manufacturing industry cannot be ignored. The pollution risk of heavy metals is so high that the environment cannot degrade them. Copper, tin, lead, nickel, and other heavy metals are extremely harmful to ecology. Assessing the ecological risks of heavy metals in soil can help to provide decision-makers with insights into reducing and managing pollution risks [25]. It is very important to pay attention to the heavy metal pollution content in soil. Through the study of heavy metal pollution index, evaluation index and pollution degree [26], the pollution degree of products is quantified as the content of heavy metals, which greatly solves the problem where pollution cannot be measured. Thus, design of the closed-loop manufacturing-recycling network under the multiproduct, multicycle, multilink scenario considering influence of heavy metal pollution, should be investigated. By quantifying the heavy metal pollution in the environment, the impact of heavy metals on the environment is replaced by the equivalent quantity of products. A objective optimization problem is formed, which takes into account both production costs and environmental pollution, but some research has not produced a good balance between them [24]. Dual-objective optimization can balance environmental pollution and production costs. Regarding closed-loop manufacturing networks, few studies have solved dual-objective optimization problems. Membership functions in a fuzzy method are used to deal with the dual-objective problem. However, these are chosen by experts subjectively and lack the self-study ability, so convincing results cannot be achieved [27,28] using the enhanced global criterion approach to dealing with the dual-objective problem of production costs and pollution. This method can only improve the effectiveness of non-convex effects in dealing with Pareto optimal problems, but cannot improve the search range of feasible region. However, an epsilon-constraint method can produce nonextreme, efficient solutions to the dual-objective problem regarding production costs and pollution. In the epsilon-constraint method, the optimal value of one objective function is selected as a constraint and added to the solution of another objective function to improve the search range of the feasible region. In this way, environmental protection is also double-layered under recycling and manufacturing with x and y attributes. The maximum balance between production costs and environmental pollution is maintained. In this closed-loop manufacturing system with a recycling and remanufacturing structure, a win-win situation in terms of production costs and environmental protection can be achieved by integrating the degree of heavy metal pollution into production decisions through green and sustainable production technology.

Problem Description
In production scheduling, the production process needs warehouses and manufacturing and other units to provide production services for enterprises. Due to the existence of production defects, some materials may not be fully utilized, and some waste products are produced. At this point, the recycling process through the unified recovery of waste and through the use of some reprocessing units qualifies the waste for reuse. Therefore, the design of a closed-loop manufacturing-recycling network can clearly show the specific process and different links used by decision makers. In particular, when the model involves multiple production variables, it is convenient to conduct effective planning for complex processes.
As shown in Figure 1, the solid line represents forward logistics, and the dotted line represents reverse logistics. In forward logistics, the processing of materials with two attributes, namely x and y attributes, is included. The material with attribute x is brand new, because it comes from the supplier and has not been processed in the recycling process. On the contrary, the material with attribute y has gone through the recycling process. It is a "second-hand" material that can be reused after being treated by the recycling process, rather than being brand new. In reverse logistics, waste products from forward logistics are transported to the collection unit of the recycling process. They will have the y attribute. These waste products with the y attribute are processed into materials qualified for reproduction and utilization and then transported to the production process. When they are transported to the recycling process, there is a uniform attribute, y. It is worth noting that no matter whether the material with has attribute x or y, the collection unit in the recycling process is attributed to attribute y. By distinguishing the two attributes of materials, it is possible to accurately distinguish between the two different production qualified rates during manufacturing. Thus, there are more manufacturing possibilities.
tributes, namely x and y attributes, is included. The material with attribute x is brand new, because it comes from the supplier and has not been processed in the recycling process. On the contrary, the material with attribute y has gone through the recycling process. It is a "second-hand" material that can be reused after being treated by the recycling process, rather than being brand new. In reverse logistics, waste products from forward logistics are transported to the collection unit of the recycling process. They will have the y attribute. These waste products with the y attribute are processed into materials qualified for reproduction and utilization and then transported to the production process. When they are transported to the recycling process, there is a uniform attribute, y . It is worth noting that no matter whether the material with has attribute x or y , the collection unit in the recycling process is attributed to attribute y . By distinguishing the two attributes of materials, it is possible to accurately distinguish between the two different production qualified rates during manufacturing. Thus, there are more manufacturing possibilities.  In the recycling process, waste units collect materials that can no longer be used. Different kinds of heavy metal pollution in the environment come from waste units. Considering the pollution degree of different heavy metals to the environment, these different pollution degrees are equivalent to the same benchmark. A comparison of environmental pollution caused by different heavy metals was carried out. It is convenient to make production decisions based on environmental pollution.

Model Assumption and Notations
In this paper, a closed-loop manufacturing-recycling network for electronic assembly is designed. In this network, several manufacturing units and warehouse units are considered in the forward production process, while collection units, waste units, and several recycling units are considered in the reverse process. Waste units cause pollution to the environment. Therefore, we evaluated them and determined the degree of pollution to the environment to facilitate production planning with the comprehensive consideration of production costs and the environment. Some of the assumptions in the model are as follows: 1. The machining process involves the purchase of components and parts. Components, modules, parts, and semifinished products are processed, and finished products are In the recycling process, waste units collect materials that can no longer be used. Different kinds of heavy metal pollution in the environment come from waste units. Considering the pollution degree of different heavy metals to the environment, these different pollution degrees are equivalent to the same benchmark. A comparison of environmental pollution caused by different heavy metals was carried out. It is convenient to make production decisions based on environmental pollution.

Model Assumption and Notations
In this paper, a closed-loop manufacturing-recycling network for electronic assembly is designed. In this network, several manufacturing units and warehouse units are considered in the forward production process, while collection units, waste units, and several recycling units are considered in the reverse process. Waste units cause pollution to the environment. Therefore, we evaluated them and determined the degree of pollution to the environment to facilitate production planning with the comprehensive consideration of production costs and the environment. Some of the assumptions in the model are as follows: 1.
The machining process involves the purchase of components and parts. Components, modules, parts, and semifinished products are processed, and finished products are inspected. The recycling process includes collection, dismantling, remanufacturing, and waste. Additionally, components, modules, parts, and semifinished products are processed, and finished products can be recycled to output second-hand components, modules, and parts for use.

2.
Each unit in the model is connected through the transport process. By default, there is no loss during transportation.

3.
Forward logistics involves the processing and production of materials with two attributes. Because the qualified rate of the production of materials with two attributes differs λ i x > λ mix > λ i y , materials with different attributes cannot be mixed for processing. There is no difference in the final product after the processing of materials with different attributes. The forward process has a qualified rate; the reverse process has a conversion rate.
Based on the above assumptions, a mixed integer model of electronic assembly with a closed-loop structure was established. The notations used in the model are as follows: Indices: x: New attribute of materials; y: Secondary attribute of materials; MP rs i x ,s,t : Quantity of purchased of materials i x from S; MM i x ,j m ,t , MM i y , j m ,t : Quantity of manufactured materials i x and i y in manufacturing units; MY i y , j r ,t : Quantity of treated materials i y in recycling units; f p i y ,t : Quantity of warehouse materials i x and i y (components, parts, modules, semifinished products, and finished products) in warehouse units; T i x ,j w ,j m ,t , T i y ,j w ,j m ,t : Quantity of transported materials i x and i y from M(j w ) to M(j m ); T i x ,j m ,j w ,t , T i y ,j m ,j w ,t : Quantity of transported materials i x and i y from M(j m ) to M(j w ); T i x ,j m ,j r ,t , T i y ,j m ,j r ,t : Quantity of transported materials i x and i y from M(j m ) to Y(j r ); T i y ,j r ,j r ,t : Quantity of transported materials i y from Y(j r ) to Y(j r ); T i y ,j r ,j w ,t : Quantity of transported materials i y from Y(j r ) to M(j w ).

Objective Function
In order to achieve the lowest production cost at the same time as achieving low pollution and environmental protection, the proposed model considers the manufacturing process comprehensively and optimizes production costs and heavy metal pollution.

Minimizing Production Costs
In manufacturing, taking into account the depreciation of manpower, material resources, and facilities, the production process involves various kinds of expenditure, as shown in Equations (1)-(6).
Here, Equation (1) gives the total cost of five types of expenditure, including material purchases (CRM), equipment damage (CMF), recycling (CRC), inventory use (CIV), and transportation costs (CTP). CT j r ,j r = ∑ j r ∈J r ∑ j r ∈J r ∑ i y ∈Y(j r ) ∑ t∈T trr T i y ,j r ,j r ,t (10) Equations (7)- (11) show the details of the transportation costs. In other words, it is all the connected processes between units in the production process, including between the warehouse and manufacturing units, the manufacturing and recycling units, the recycling units and warehouse units, and between recycling units.

Minimizing Pollution by Heavy Metals
In this model, environmental pollution is actually caused by different kinds of heavy metals (such as copper, tin, lead, and nickel) in discarded items in the recycling process, and their sources are uniformly summarized as waste units. The objective function of minimizing heavy metal pollution is shown in Equation (12).
where PD is the sum of pollution equivalents of different heavy metals d. The calculation of the heavy metal pollution equivalent is shown in Equations (13) and (14).
In Equation (13), P m is obtained by calculating the content of heavy metal pollutants in waste units, and the pollution equivalent of heavy metal pollution is calculated through division by the equivalent quantity factor w using Equation (14).

Constraints
Because the production equipment has a capacity limit, the manufacturing unit has a production limit that cannot be exceeded, as shown in Equations (15) and (16).
Equations (17) and (18) describe the logistics balance of the manufacturing units. The quantity of materials transported from the warehouse to the manufacturing units is equal to the sum of the logistics of materials transported from the manufacturing units to the next warehouse and those transported to the collection unit during the recycling process. Here, λ i x , λ i y is the qualified rate of production, and its value varies with different attributes, 2T i y ,j w ,j m = 1 1−λ iy T i y ,j m ,j r + 1 λ iy T i y ,j m ,j w ∀i y ∈ M(j m ), ∀j m ∈ J m , ∀j w , j w ∈ J w , ∀j r ∈ J r , t ∈ T Equations (19)- (23) are different types of warehouse balances, including components, parts, modules, and semifinished products. Because the material is the same, but the two attributes are different, it has an independent inventory balance. In particular, Equation (19) covers procurement, and Equations (20) and (22) cover second-hand materials from the previous period. The difference is that Equation (25) shows the inventory balance of the finished products. Since there is no difference between the output products, the finished products of the two attributes are represented together.
Equations (26)-(28) describe the details of the recycling process. As shown in Equation (26), materials with attributes x and y are uniformly planned as y attributes in the collecting unit. Materials in the recycling units are eventually transported to the warehouse or waste unit, and there is a certain proportional relationship, as shown in Equations (26) and (28).
T i x ,j m ,j r ,t + T i y ,j m ,j r ,t = MY i y ,j r ,t ∀i x , i y ∈ M(j m ), ∀j r ∈ J r , ∀t ∈ T (26) MY i y ,j r ,t = T i y ,j r ,j r ,t + T i y ,j r ,j w ,t ∀i y ∈ Y(j r ), ∀j r , j r ∈ J r , ∀j w ∈ J w , ∀t ∈ T (27) (1 − δ i y ) · T i y ,j r ,j r ,t = δ i y · T i y ,j r ,j w ,t ∀i y ∈ Y(j r ), ∀j r , j r ∈ J r , ∀j w ∈ J w , ∀t ∈ T (28)

Multiobjective Optimization Approach
In this section, two different methods for solving multiobjective optimization problems are proposed: an epsilon-constraint method and a fuzzy-weight method. They have different characteristics in the process of solving.

Epsilon-Constraint Method Based on a Closed-Loop Manufacturing-Recycling Network
An epsilon-constraint method is introduced to solve dual-objective optimization problems, as shown in Equation (29) There are two objectives when finding the optimal solution f 1 (x), f 2 (x). S is the feasible region of the decision variable x. One objective function is selected to solve for the minimum and the result of another objective function is add as a constraint.
In Equation (30), where f 2 (x) ≤ e 2 represents the constraint that is added to the constraints. e 2 is the value of the objective function f 2 (x * ) when solving another objective function f 1 (x). In this way, the constraint is added, so that no feasible solution is lower than the corresponding solution to the constraint, which satisfies all requirements of the Pareto optimal.

A Fuzzy-Weight Method Based on a Closed-Loop Manufacturing-Recycling Network
In dual-objective programming, all Pareto optimal solutions have a fuzzy weight to satisfy the optimal solutions of single objective programming problems. Thus, in order to solve the production decision-making problem with the two objectives of minimizing production costs and minimizing heavy metal pollution, a fuzzy membership function method is adopted in this section to simplify objective functions.
In Equation (31), a linear function is used to construct the membership function of the objective function, where f k (x) is the target function, k ∈ N. f * k = min f k (x) and f 1 k = max f k (x) represent the minimum and maximum values of an objective function.
The membership function method is used to establish the following new membership function, as shown in Equation (32). The new multiobjective function obtained by fuzzy processing is linearly weighted to transform the dual-objective optimization problem. The existence of α 1 and α 2 makes the Pareto optimal solution of problem Equation (33) the optimal solution of the corresponding single-objective problem.

Analytical Data
This section presents a case study of the electronic factory production process. The production process is optimized with dual objectives to save resources, reduce production costs and achieve environmental protection. Through the two attributes, x and y, the production has two sets of independent assembly modes; in other words, production is processed together with the same attribute. Table 1 shows the qualified rate of materials in the manufacturing unit. Because different enterprises have different abilities to process materials, the qualified rate has a certain range. It is worth noting that the qualified rate is also distinguished when the second-hand materials are distinguished from brand-new purchased materials. When second-hand materials are processed by the recycling unit for secondary utilization, the qualified rate of second-hand materials will be lower than that of purchased materials. Therefore, when the source of the material is not distinguished, the unique fixed value is assumed λ mix , and the qualified rate is in the middle λ i x > λ mix > λ i y . As shown in the topology of Figure 1, there are two types of purchased materials involved in the electronics factory (nine components and three parts). Five different kinds of warehouse units and three different kinds of manufacturing units are used to deal with components, parts, modules, semifinished products, and finished products. The relationships among these materials are shown in Tables 2 and 3. In addition to the collecting unit and waste unit, there are three different types of remanufacturing units in the recycling process. Pollution impact by heavy metals is ranked using pollution equivalent numbers.
Nine components were eventually assembled into three finished products. Table 2 describes the quantitative relationships of components (1 to 3, 4 to 6, 7 to 9) to the material during the production process. Table 3 introduces the quantitative relationship of materials assembled between modules and parts.  Table 3. Quantitative relationships between modules and parts.

PT 1 PT 2 PT 3
In Tables 2 and 3, material CP can be first processed into MD and then combined with PT to form SP and PF. MD and PT are combined in a one-to-one relationship. As shown in Table 4, the relationship between materials and units is introduced into the closed-loop manufacturing-recycling network. Specifically, the components are processed in unit A, the parts and the modules are processed in unit B together, and the semi-finished products are inspected in unit C. Finally, the planning horizon contains seven time periods. The production requirements for seven time periods are shown in Table 5. Finished products 1, 2, and 3 have different demands at each time. Considering that the four heavy metals in components and modules have different environmental pollution levels, w d represents the equivalent factor of heavy metal d(w d = {0.1, 0.08, 0.07, 0.04}, d = {1, 2, 3, 4}), referring to copper, tin, lead, and nickel, respectively, and the corresponding heavy metal pollution quantities of materials are shown in Table 6.  Table 5. Production demand in seven periods.   Table 6 shows the environmental heavy metal pollution quantities of copper, tin, lead, and nickel among wasted materials CP 1 to 9 and MD 1 to 3 in waste units. The different heavy metals come from nine kinds of wasted components and three kinds of wasted modules, respectively.

Result Analysis
In order to solve the dual-objective optimization problem with the lowest production cost and the least environmental pollution, the epsilon-constraint and fuzzy with weight methods were used, as shown in Table 7. In order to more clearly compare the performance of the two methods, the solution of the dual-objective model is given without distinguishing x and y attributes, where λ = λ mix = 0.7. Table 7. Cost and pollution results of the two methods.

Cost of Production Pollution Equivalent Numbers
Epsilon-constraint method 1.8130176 × 10 7 1.1515220 × 10 7 Fuzzy with weighted method 1.8661670 × 10 7 1.1675352 × 10 7 Optimization was achieved by two different methods without distinguishing between x and y attributes, in other words, with only one set of material attributes.
Using the fuzzy with weighted method, it is difficult for the dual-objective optimization problem to achieve the relative optimality of two objectives at the same time, because it deviates from the solution of the other objective when it meets the relative optimality of one objective. For linear problems, the weighted value is applied in the original feasible region, while the epsilon-constraint method changes the original feasible region to produce nonextreme solutions. Thus, the epsilon-constraint method can obtain a better solution than the fuzzy with weighted method in terms of both production costs and environmental pollution equivalents, as shown in Table 7. The epsilon-constraint method gives a 2.85% lower cost of production and a 1.38% lower pollution equivalent than the fuzzy with weighted method. Therefore, the epsilon-constraint method has more advantages for the closed-loop manufacturing-recycling network.
Next, the model that distinguishes x and y attributes (λ i x ∈ [0.75, 0.9], λ i y ∈ [0.5, 0.65]), and those the model that does not distinguish (λ mix = 0.7) are solved by the epsilonconstraint method, where λ i x and λ i y take the extreme value of the boundary, and the production cost and environmental pollution equivalent are shown in Figures 2 and 3. Figures 2 and 3 compare cases with no attributes distinguished (λ mix ) with those with attributes distinguished (λ i x , λ i y ) in the model. The percentages of the distinguished attribute and no distinguished attribute cases were calculated, and the production cost was 15.78% to 33.22% higher than the production cost where the attributes were not distinguished, as shown in Figure 2. Similarly, the pollution equivalents were 25.28% to 75% higher than the pollution equivalents where the attributes were not distinguished, as shown in Figure 3. Thus, distinguishing between new materials purchased in the production process and second-hand materials can effectively reduce production costs and greatly reduce environmental pollution equivalents. The model that distinguishes material attributes has obvious advantages in terms of saving costs and protecting the environment.  tribute and no distinguished attribute cases were calculated, and the production cost was 15.78% to 33.22% higher than the production cost where the attributes were not distinguished, as shown in Figure 2. Similarly, the pollution equivalents were 25.28% to 75%  tribute and no distinguished attribute cases were calculated, and the production cost was 15.78% to 33.22% higher than the production cost where the attributes were not distinguished, as shown in Figure 2. Similarly, the pollution equivalents were 25.28% to 75% Distinguishing material attributes is of great significance in the production process. Table 8 presents a further analysis of this case in the solution set for the two objectives, 1 and 2. When λ i x = 0.7λ i y ∈ [0.5, 0.65], the optimal solution sets of the two objectives under different qualified rates (λ i y = 0.5, 0.55, 0.6, 0.65) with the y attribute are displayed. Under the same λ i y , it can be found that the optimal solutions are almost the same, whether object 1 or object 2 is considered. Object 1 and object 2 represent production costs and pollution equivalents, respectively. However, when λ i y is different, the higher the qualified rate is, the lower the production cost is and the lower the environmental pollution equivalent is.
To be precise, when the fixed λ i x , λ i y changes, the pollution caused by the waste material basically does not change due to the limited amount of second-hand materials. When λ i y is the largest, the utilization of materials is improved, and there is a solution that best meets the production cost requirements in the solution set of the objective function. On the contrary, dual-objective optimization Pareto solution sets with different qualified rates and the x attribute are shown in Figure 4. Because λ i x ∈ [0.75, 0.9], four kinds of λ i x (λ i x = 0.9, 0.85, 0.8, 0.75) were selected for analysis. Due to the small amount of recycled material and the large proportion of new material, the production qualification rate of new materials has a greater impact on the environment. The production cost fluctuates greatly with the reduction of the low qualified rate. Thus, under the four different λ i x values with λ i y = 0.65, the pollution environmental pollution equivalent gradually increases with the reduction of production costs, forming a Pareto distribution. It is worth noting that under the condition of a high qualified rate, the relationship between the production cost and pollution equivalent is relatively stable. When achieving the lowest production cost, the least environmental pollution is also achieved. The higher the value of λ i x , the stronger the tradeoff between the two objectives, the smaller the impact on the objectives, and the higher the overall superiority of the dual-objective optimization compared with cases where the value of λ i x is low. Therefore, in the case of distinguishing material attributes, achieving a high qualified rate of new materials can greatly reduce production costs and improve the environmental pollution decision optimization quality, which is conducive to the production and environmental protection of enterprises.
Processes 2022, 10, x FOR PEER REVIEW 16 of 18 production costs and improve the environmental pollution decision optimization quality, which is conducive to the production and environmental protection of enterprises.

Conclusions
In this paper, the impact of electronic production on the environment was considered, and a dual-objective optimization decision problem for a green closed-loop manu-

Conclusions
In this paper, the impact of electronic production on the environment was considered, and a dual-objective optimization decision problem for a green closed-loop manufacturing recycling network for use in electronic assembly was studied. A new idea was put forward where the network divides the sources of materials into two attributes, new materials and second-hand materials, to integrate high-quality raw materials to improve the overall production qualified rate. Here, only materials with the same attribute can be assembled in the same product. Compared with the traditional mode, regardless of the material attributes, the proposed network results in a higher qualified rate of materials. In order to enhance the excellence of decision optimization, the epsilon-constraint method is used to solve the dual-objective optimization problem. The epsilon-constraint method can change the feasible region of the original problem, produce effective solutions with nonextreme values, and strengthen the superiority of the solution while balancing the weights of the two objectives. Compared with the fuzzy with weight method, the epsilon-constraint method can obtain a relatively superior solution that satisfies economic and environmental objectives at the same time. Finally, through a case study analysis, on the premise of distinguishing material attributes, improving the qualified rate of production for new materials can once again improve the decision optimization effect in terms of costs and the environment. This closed-loop manufacturing-recycling network considering raw material attributes can provide more economic benefits for enterprises in this era of focus on the environment.