Adaptive planning method for local destruction disassembly line considering feasibility and noise

Abstract

An adaptive planning method for a local destructive disassembly line considering feasibility and noise includes the following steps: obtaining basic information of the disassembly task; constructing a feasibility evaluation system for local destructive disassembly of the product to be disassembled, extracting corresponding coarse-grained evaluation indicators and fine-grained indicators from the product layer and the component layer respectively, and quantifying the indicators to determine whether the disassembly task adopts destructive disassembly or non-destructive disassembly; establishing a balance problem model for the local destructive disassembly line: with the goal of minimizing the number of open workstations, idle time balance indicators, disassembly costs and noise pollution; at the same time, establishing constraints for the balance problem of the local destructive disassembly line; based on the gray wolf algorithm, solving the above-mentioned balance problem model for the local destructive disassembly line, and outputting the optimal solution and the optimal plan. The present invention realizes the adaptive planning solution of the local destructive disassembly line considering feasibility and noise, and provides a feasible theoretical guidance for the layout optimization setting of the disassembly line.

Description

Technical Field

[0001] The invention relates to the technical field of disassembly line balance, in particular to a local destruction disassembly line adaptive planning method considering feasibility and noise. Background Art

[0002] Due to product replacement, a large number of products to be disassembled are generated, which has a huge impact on the environment and resources, and the related negative effects have actually increased exponentially. Remanufacturing plays an important role in the recycling and recovery of scrapped products, which can effectively improve resource utilization efficiency, reduce waste and promote clean production. Disassembly is the premise of remanufacturing, which directly affects the recovery and reuse of waste products and the recycling capacity of scrapped products. The disassembly line is an effective method for the recycling and reuse of remanufactured products. However, while pursuing profits, manufacturing companies should focus on reducing energy consumption. According to the depth of disassembly and disassembly operations, disassembly can be divided into selective disassembly, complete disassembly, parallel disassembly, partial destructive disassembly (PDD) and non-destructive disassembly. Good disassembly can reduce remanufacturing costs and improve the efficiency of large-scale disassembly.

[0003] In the actual disassembly process, the quality of the product to be disassembled is uncertain, and multiple faults such as wear, aging, corrosion and deformation may occur, resulting in uncertain changes in its external features and internal materials. In this case, traditional disassembly methods are either inefficient or ineffective. Therefore, the problem of partially destructive disassembly line balancing (PDDLB) came into being, which is between non-destructive disassembly and fully destructive disassembly. It aims to remove parts that cannot be removed due to serious faults by destroying some connectors or low-value parts. The feasibility of PDD is a prerequisite for evaluating whether PDD can be implemented. However, the existing PDDLB problem is mainly profit-oriented and requires random selection of parts for destruction and disassembly, ignoring the feasibility of PDD. In addition, the large amount of noise generated during the PDD process can have a negative impact on organisms and cause harm to human health, which may cause complications such as heart disease, diabetes, and increased blood pressure, and seriously affect mental health, leading to depression. Therefore, noise pollution has become a serious social problem and a major environmental risk in developing countries. To the best of our knowledge, there is no research on the feasibility evaluation and noise of PDD in the PDDLB problem.

[0004] In order to improve the level of PDDLB problem in industrial applications and achieve theoretical results, it is urgent to further study the PDDLB problem considering the feasibility of PDD and noise pollution. Summary of the invention

[0005] To solve the above problems, the main purpose of the present invention is to provide an adaptive planning method for local destruction disassembly lines based on the improved grey wolf optimization (IGWO) algorithm taking into account feasibility and noise, so as to obtain comprehensive and optimal solutions in many aspects and provide guidance and basis for the optimal setting of local destruction disassembly lines in various complex situations.

[0006] The technical solution of the present invention is an adaptive planning method for a local destruction disassembly line considering feasibility and noise, comprising the following steps:

[0007] Step 1: Obtain basic information of the disassembly task, including disassembly time, tools required for disassembly, disassembly noise, priority relationship and fault information;

[0008] Step 2: Construct a feasibility evaluation system for partial destructive disassembly of the product to be disassembled, extract the corresponding coarse-grained evaluation indicators and fine-grained indicators from the product layer and the component layer respectively, and quantify the indicators. According to the quantified value of the feasibility indicator, determine whether the disassembly task adopts destructive disassembly or non-destructive disassembly;

[0009] Step 3: Establish a partial destruction disassembly line balance problem model: with the goal of minimizing the number of open workstations, idle time balance index, disassembly cost and noise pollution; at the same time, establish the constraints of the partial destruction disassembly line balance problem (PDDLBP), including disassembly task allocation constraints, disassembly priority relationship constraints, time constraints, feasibility constraints, noise constraints and workstation allocation constraints;

[0010] Step 4: Based on the Grey Wolf Algorithm, solve the above-mentioned local destruction disassembly line balance problem model and output the optimal solution and optimal plan.

[0011] The technical effects of the present invention are:

[0012] (1) Taking into account the feasibility of destructive disassembly and noise pollution, an adaptive planning method for local destructive disassembly is proposed, and a mathematical model of a local destructive disassembly line is established with the optimization objectives of minimizing the number of workstations opened, idle time balance index, disassembly cost and noise pollution. This model provides a feasible theoretical guidance for the layout optimization setting of the disassembly line.

[0013] (2) By using the IGWO algorithm to solve the P25 medium-scale and P52 large-scale examples and comparing the results with those in the existing literature, the proposed algorithm is significantly better than other classical algorithms. Combining the established multi-objective mathematical model with the proposed IGWO algorithm to solve the engine disassembly example, 10 groups of solution solutions are analyzed, which verifies the effectiveness of the IGWO algorithm in solving PDDLB and the diversity of its solution results. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below.

[0015] Figure 1 It is the local destruction disassembly feasibility evaluation system of the present invention;

[0016] Figure 2 It is a flow chart of the multi-granularity evaluation method for the feasibility of local destructive disassembly in the present invention;

[0017] Figure 3 is the fine-grained index membership function based on failure characteristics in the present invention;

[0018] Figure 4 It is a schematic diagram of coding in the present invention;

[0019] Figure 5 It is a schematic diagram of two-point intersection in the present invention;

[0020] Figure 6 Schematic diagram of the mutation process in the present invention;

[0021] Figure 7 Solve P52 results for different algorithms in the present invention;

[0022] FIG8 is a disassembly scheme according to the present invention that takes into account the negative impact of noise and feasibility on local destructive disassembly. DETAILED DESCRIPTION

[0023] The present invention will be further described in detail below in conjunction with the embodiments and drawings.

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] Description of symbols and decision variables:

[0026] Table 1 Symbols and explanation of decision variables

[0027]

[0028] In order to facilitate understanding of the meaning of the embodiments and other formulas in the subsequent parts of the application documents, a table of symbols and decision variables in Table 1 is provided hereby. Unless otherwise specified, the meanings of the symbols and decision variables described below that are the same as those in Table 1 are the same as those explained in Table 1.

[0029] In the following embodiments, unless otherwise specified, the operations described are routine operations in the art.

[0030] A local destruction disassembly adaptive planning method considering feasibility and noise, the method comprising the following steps:

[0031] Step 1: Obtain basic information of the disassembly task, including disassembly time, tools required for disassembly, disassembly noise, priority relationship and fault information;

[0032] Step 2: Construct a feasibility evaluation system for partial destructive disassembly of the product to be disassembled, extract the corresponding coarse-grained evaluation indicators and fine-grained indicators from the product layer and the component layer respectively, and quantify the indicators. According to the quantified value of the feasibility indicator, determine whether the disassembly task adopts destructive disassembly or non-destructive disassembly;

[0033] Specifically, partial destructive disassembly aims to improve disassembly efficiency by partially destroying the components while ensuring their integrity. Different disassembly methods have different disassembly directions, disassembly times, and disassembly tools for the products to be disassembled, but these factors cannot fully reflect the feasibility of partial destructive disassembly of the products. Based on the hierarchical characteristics of products, a feasibility evaluation system for partial destructive disassembly of products to be disassembled is constructed, and the corresponding coarse-grained evaluation indicators and fine-grained indicators are extracted from the target group and component layers, respectively. See the detailed evaluation system for partial destructive disassembly of products to be disassembled. Figure 1 .

[0034] Three indicators, namely failure rate, number of connections and cost of local destruction and disassembly, are introduced for coarse-grained evaluation to reflect the feasibility of overall local destruction and disassembly of the product.

[0035] (1) Failure rate: refers to the proportion of components with high failure probability in the target group. The difficulty of local destructive disassembly is related to the failure probability of the components. The higher the failure rate, the more difficult non-destructive disassembly is and the greater the feasibility of local destructive disassembly is. The failure rate entropy is defined as:

[0036] S1=log2(N i / N s )

[0037] Where N i is the total number of product components, N s is the number of components with high failure probability in the product;

[0038] Among them, N s Obtained by expert scoring method: In view of the uncertainty of component failure characteristics and the difficulty in accurate quantification, the expert evaluation method is used to quantify the fault type. The component failure status comment set is h = [h1, h2, h3, h4, h5], where [h1, h2, h3, h4, h5] represent basically no failure, slight failure, general failure, moderate failure, and severe failure respectively.

[0039] Invite D experts to evaluate the status of failure characteristics, and the comprehensive result is A = [a h ], (h = 1 to 5), where D is the number of experts involved in the evaluation, ah is the number of experts who select the hth evaluation value. At this time, the characteristic value of component i is defined as follows:

[0040] N h =h·A T

[0041] According to the above expert evaluation method, determine the number of components with high failure probability in the product, N s .

[0042] (2) Entropy of the number of connectors: Since connectors are generally of low value, local destruction and disassembly mainly adopts destructive means to destroy the connectors. The more connectors there are, the more selective the components for destruction and disassembly are, and the higher the feasibility of local destruction and disassembly is. The entropy of the number of connectors is defined as

[0043] S2=log2[(N i +N k ) / N k ]

[0044] Where N k The number of connections formed by product components using methods k (screws, bolts, welding, riveting, non-detachable connections, etc.), N i The number of product components.

[0045] (3) Destructive disassembly cost entropy: refers to the ratio of the sum of non-destructive disassembly cost and destructive disassembly cost to non-destructive disassembly cost. The smaller the disassembly benefit entropy, the lower the local destructive disassembly cost and the higher the feasibility of local destructive disassembly. The local destructive disassembly cost entropy is defined as

[0046] S3=log2[(C n +C m ) / C n ]

[0047] Where: C n is the non-destructive disassembly cost; C m To destroy the demolition cost.

[0048] For complex products, the target group is first evaluated at a coarse-grained level, and then the total disassembly entropy of the target group is calculated according to the following formula:

[0049] S=0.33S1+0.33S2+0.34S3

[0050] According to the above formula, the product is evaluated at an overall coarse-grained level. When the product evaluation value exceeds the first threshold, it is proved that the product can be disassembled in a destructive manner. Next, it is necessary to determine which components are to be disassembled in a destructive manner, i.e., component fine-grained evaluation. The entire evaluation process is as follows: Figure 2 The first threshold is usually specified by the user according to actual conditions, and in this embodiment, the first threshold can be selected as 1.

[0051] The object of fine-grained evaluation is the component (i.e., disassembly task). In the actual disassembly process, some components cannot be disassembled due to their serious failure. Generally, some components are destroyed to improve the disassembly efficiency. Therefore, the feasibility of partial destructive disassembly is related to the failure degree of the component and the disassembly process.

[0052] The local destruction and disassembly process is as follows:

[0053] (1) Identify the locations of locally destructive disassembly components and non-destructive disassembly components. The identification of the locations of locally destructive disassembly components and non-destructive disassembly components can be evaluated using destructive component identification indicators, including identification of component connection types, identification of component failure characteristics, and identification of tools for different disassembly methods. The more severe the failure characteristics, the simpler the identification of destructive disassembly components, and the stronger the feasibility of local destructive disassembly.

[0054] (2) The replacement of tools for different disassembly methods and their alignment with the connecting parts. The conversion of tools for different disassembly methods and their relative positioning accuracy with the connecting parts seriously affect the efficiency of local destructive disassembly. If different disassembly tools are replaced many times and have high positioning accuracy, the cost of local destructive disassembly will increase and energy will be wasted. Therefore, the disassembly tools and positioning accuracy are used to evaluate their impact on the feasibility of local destructive disassembly.

[0055] (3) Destroy or remove the corresponding connectors. Destroying connectors involves disassembly direction, disassembly time, disassembly force, and accessibility. When the failure degree of the component is serious, destroying the connector will change the previous disassembly direction. Frequent changes in the disassembly direction will reduce the disassembly efficiency and bring inconvenience to local destructive disassembly. Severe component failure will increase the required disassembly force and make tool accessibility more difficult, which will seriously affect the efficiency of local destructive disassembly. Therefore, the disassembly direction, disassembly time, disassembly force, and accessibility are used to evaluate their impact on the feasibility of local destructive disassembly; removing the corresponding connector involves the size of the component structure. If the component failure degree is serious and the structure is large, it is difficult for the tool to grasp, and it is difficult to remove the corresponding connector and the connected components, which affects the efficiency of local destructive disassembly.

[0056] In summary, six fine-grained indicators, including damaged component identification, disassembly tools, disassembly direction, disassembly time, disassembly force, and accessibility, are established based on failure characteristics.

[0057] The failure characteristics of components are quantified by expert judgment, and the influence of component failure characteristics on fine-grained evaluation indicators is fuzzy. Fuzzy comprehensive evaluation comprehensively evaluates the results through a fuzzy set. Therefore, fuzzy trigonometric function is selected as the membership function of the influence of failure characteristics on fine-grained indicators. The evaluation level set is used to classify the influence of component failure degree on fine-grained indicators. The evaluation level and level value are shown in Table 1.

[0058] Table 1 The influence of part failure characteristics on fine-grained indicators

[0059]

[0060]

[0061] Among them, the evaluation level of each fine-grained indicator in Table 1 is set based on the expert scoring method. Since the specific operation steps of the expert scoring method have been proposed in the aforementioned process, its specific operation process will not be repeated here.

[0062] The membership function of the indicators and fault levels in Table 1 is established through expert experience and literature. Figure 3 As shown in Figure 2. The value of each indicator is determined by the membership function. Taking the damage component identification indicator as an example, Figure 3 It can be seen that according to the range of the failure characteristic value, it can be determined that it belongs to [Δa i , Δa j The membership degrees of the intervals are Calculate the damage component identification index in, Figure 3 (a) represents the relationship between fault and damage and disassembly identification indicators, Figure 3 (b) represents the relationship between the fault and the disassembly tool conversion index, Figure 3 (c) represents the relationship between the fault and the disassembly direction index, Figure 3 (d) represents the relationship between the failure and disassembly time indicators, Figure 3 (e) represents the relationship between the failure and the disassembly force index, Figure 3 (f) represents the relationship between fault and reachability index, Figure 3 (g) represents the relationship between fault and positioning accuracy index, Figure 3 (h) Represents the relationship between failure and component structure size index.

[0063] In order to avoid the distortion of the evaluation results caused by mutation values, this paper uses the structural entropy weight method to determine the index u yThe weights are as follows: y=1, 2, ```, 6 represent y=1 as the damage component identification index, y=2 as the disassembly tool index, y=3 as the disassembly direction index, y=4 as the disassembly time index, y=5 as the disassembly force, and y=6 as the accessibility index.

[0064] The specific steps are as follows:

[0065] Step 1. Invite experts to rank the importance of each fine-grained indicator.

[0066] Step 2. Build a fine-grained indicator set, collect expert opinions, and build an expert opinion matrix.

[0067]

[0068] In the formula, a xy (x=1,2,…,z;y=Δa,Δb,Δc,Δt,Δd,Δq) represents the importance ranking of the yth indicator by the xth expert, and the value range is (1,2,…,n). Calculate a xy is the membership degree b xj for

[0069]

[0070] In the formula, u is the conversion parameter, and u = n + 2 is usually assumed. xy The average degree of awareness y for:

[0071]

[0072] Then the blindness Q y for:

[0073]

[0074] Then z experts have different opinions on indicator u y The overall awareness of xy for

[0075] p y =(1-b y )(1-Q y ),p y >0

[0076] It is easy to get the evaluation vector of z experts on all indicators U:

[0077] P=[p1,p2,…,p y ]

[0078] Step 3. Normalize (9) to get the index u j The weight is

[0079]

[0080] For components, firstly, the component layer is fine-grained evaluated, and then the comprehensive evaluation of the fine-grained indicators of the component layer, that is, the feasibility value of local destruction and disassembly, e, is calculated according to the following formula:

[0081]

[0082] In the formula, α y (y=1,2,…,8) is the weight corresponding to each fine-grained indicator calculated by applying the structural entropy weight method.

[0083] Step 3: Establish a partial destruction disassembly line balance problem model: with the goal of minimizing the number of open workstations, idle time balance index, disassembly cost and noise pollution; at the same time, establish the constraints of the partial destruction disassembly line balance problem (PDDLBP), including disassembly task allocation constraints, disassembly priority relationship constraints, time constraints, feasibility constraints, noise constraints and workstation allocation constraints;

[0084] The objective function is as follows:

[0085] F=[NS,SI,DC,DN]

[0086]

[0087] Where NS is the number of open workstations; S g The variable for opening the workstation is a 0,1 variable. It takes 1 when the workstation is opened and takes 0 when it is closed. g is the workstation number; m is the maximum number of open workstations.

[0088]

[0089] SI is the idle time balance indicator; AT is the actual cycle time; i, j represent the disassembly task numbers; r is the disassembly mode, when r=1, it is destructive disassembly, when r=0, it is non-destructive disassembly; N is the set of disassembly task numbers.

[0090]

[0091] DC is the disassembly cost index; N is the disassembly task number set; c ir is the disassembly cost when task i is in disassembly mode r; c s is the unit time cost of opening a workstation.

[0092]

[0093] DN is the negative impact index of noise pollution; L lirWhen task i switches to disassembly mode r, the equivalent continuous sound pressure level of the noise caused by the disassembly operation is L lir (dB(A)); t ir is the disassembly time of task i when it is in disassembly mode r.

[0094] An implementation manner of the present invention is that in step 3, the constraints include cycle time constraints, disassembly task priority relationship constraints, workstation sequence start constraints, complete disassembly constraints, local damage disassembly feasibility constraints, and noise constraints:

[0095] Cycle time constraints:

[0096]

[0097]

[0098] Where N is the total number of disassembly tasks, i is the disassembly task number; r represents the disassembly task mode, when r = 0, it represents the non-destructive disassembly mode, when r = 1, it represents the destructive disassembly mode; t ir represents the disassembly time when task i is in disassembly mode r; x ig Indicates the allocation variable. When x ig = 1, indicating that disassembly task i is assigned to workstation g. ig =0, it means that disassembly task i is not assigned to workstation g; CT is the maximum allowed cycle time; AT is the actual completed cycle time;

[0099] Among them, the constraint states that all operations are completed within the cycle time.

[0100] Disassembly task precedence constraint:

[0101]

[0102] In the formula, h represents the workstation number; y ij represents the priority relationship between disassembly task i and disassembly task j. If disassembly task i is the predecessor task of disassembly task j, then y ij =1, otherwise y ij =0;

[0103] Workstation sequential startup constraints:

[0104]

[0105] Completely disassemble the constraint:

[0106]

[0107] This constraint condition indicates that the local destructive disassembly in this embodiment is complete disassembly.

[0108] Feasibility constraints for local destruction and disassembly:

[0109]

[0110]

[0111] Where R(i) is the disassembly mode assignment variable. If task i is assigned to destructive disassembly, R(i) = 1; otherwise, R(i) = 0. i represents the local destruction feasibility value of disassembly task i, and E represents the second threshold value, which is usually the average value of the local destruction feasibility values ​​obtained by all disassembly tasks in the product to be disassembled.

[0112] The constraint condition indicates that when the local destructive feasibility value of the disassembly task exceeds the second threshold, the disassembly mode of the disassembly task is pre-set to destructive disassembly; otherwise, the disassembly mode is set to non-destructive disassembly.

[0113] Noise constraints:

[0114]

[0115] Where, L eqi is the noise pollution function, L eqs is the noise threshold; L i is the instantaneous noise value when component i is removed.

[0116] This constraint indicates that when the noise of destructive disassembly exceeds the noise threshold, it is necessary to change to non-destructive disassembly.

[0117] Step 4: Based on the Grey Wolf Algorithm, solve the above-mentioned local destruction disassembly line balance problem model and output the optimal solution and optimal plan;

[0118] Among them, the Grey Wolf Algorithm is a common algorithm in this field, which generally includes the following steps:

[0119] S1. Generate the initial population and calculate its objective function value;

[0120] S2. Based on the hunting behavior of gray wolves, crossover and mutation operations are performed on non-inferior solutions;

[0121] S3, using the Pareto elite strategy and NSGA-Ⅱ crowding distance mechanism to select the non-inferior solution;

[0122] S4. Repeat the above operations until the iteration limit is reached, and output the optimal solution and optimal disassembly plan.

[0123] However, compared with the conventional gray wolf algorithm, this embodiment has the following differences:

[0124] 1. A double-layer coding method is used to generate the initial population. Compared with conventional single-layer coding, the advantages of double-layer coding are: 1. It can clearly represent the priority relationship and disassembly mode of tasks during the coding process; 2. It uses heuristic rules to adaptively change the disassembly mode to fit the non-disassembly components and high-value components encountered in the actual process, and select a more appropriate disassembly mode. Figure 4 Describes the generation process of the initial population, where X1 is the disassembly sequence and X2 is the disassembly pattern. <e max When e>e, the random selection of disassembly mode for tasks 6, 8, and 3 is represented as 0-1 random in X2; but when e>e max In engineering practice, only the destruction and disassembly mode can be selected for tasks 6 and 8.

[0125] 2. Adopt adaptive planning method for decoding; judge the noise pollution L of task i according to the noise constraint ieq Whether it exceeds the threshold. If it does, the disassembly mode should be changed to NDD; if it does not, the original disassembly mode will continue to be used, so as to achieve the effect of real-time updating of the disassembly mode.

[0126] 3. Use a two-point crossover method to simulate gray wolf hunting behavior;

[0127] The traditional crossover operation is random and may miss some solutions and eventually fall into the local optimum. To this end, a convergence factor is introduced to determine the crossover process:

[0128]

[0129] A=2a·r1-a

[0130] Where: T is the maximum number of iterations, t is the current number of iterations, ζ is the modulation factor that can effectively change the rate of change after multiple experiments, and ζ=20; r1 is a random number between [0,1]; A is the coefficient vector;

[0131] When |A|>1, the gray wolf moves away from the current prey and enters a larger space for global search; when |A|≤1, the gray wolf approaches the prey and conducts a local search.

[0132] by Figure 5 Y α For example, Y1 represents the disassembly sequence, and Y2 represents the disassembly pattern generated according to the encoding rules. αTwo points P1 and P2 are randomly generated on the Y2 matrix. The disassembly task sequence between P1 and P2 is used as the crossover segment, and the corresponding tasks 1 and 8 are used as the crossover points. The sequence {6, 9, 10} before Y2 sequence 1 and its corresponding disassembly pattern {0, 1, 0} are kept unchanged, and the sequence {3, 2, 11} after task 8 and its corresponding disassembly pattern {1, 0, 1} are kept unchanged. β Map to {1,5,7,4}, and transform Y β Y2 in is mapped to {1,0,0,1}, and the sequence between the two intersection points is {1,4,5,7,8} and its corresponding disassembly pattern {0,1,1,0,1}, thus obtaining Y α The new individual {0,1,0,1,0,0,1,1,0,1} also produces Y β New individual. α During the crossover process, the sequences before and after the crossover satisfy the priority constraint, and the sequence between the two crossover points is obtained by mapping the wolf pack Y β Got it, Y β The priority relation constraint is also satisfied in α The priority relationship must be satisfied.

[0133] 4. In order to increase the diversity of the solution set and reduce the possibility of premature convergence and falling into the local optimum, we added a perturbation factor as a local change operation. The specific method of the local mutation operation is as follows: as the convergence factor decreases, the gray wolf gradually approaches the prey. When the convergence factor is 0, the coefficient vector A=0, and the gray wolf begins to attack. In the later stage of this process, the traditional gray wolf algorithm will fall into a local optimal state. In order to increase the diversity of the solution set and reduce the possibility of premature convergence and falling into the local optimum, we improved the gray wolf algorithm and added a perturbation factor as a local change operation:

[0134] The perturbation factors are as follows:

[0135]

[0136]

[0137] In the formula, For the gray wolf individual, is the individual after mutation, d is a random number between [0,1], P d is the disturbance factor, y ipr for y i The position of the task's immediate predecessor, y isu for y i The task immediately following the task at the task, '[]' is the rounding function, λ is a random number between [0,1]; p∈[1,2,…,n]. When d <P dWhen the gray wolf individuals are mutated.

[0138] 5. Use the Pareto elite strategy to screen the results of each iterative calculation, select non-inferior individuals that meet the objective function and do not dominate each other, and store them in an external archive set as the next iterative population or non-inferior solution. After the number of iterative calculations meets the termination condition, use the NSGA-Ⅱ crowding distance evaluation standard to screen out the non-inferior solution of the objective function in the external archive set and output it as the final result.

[0139] 6. Finally, the optimal solution is selected as the final output through the hypervolume value (HV). HV (Hypervolume) refers to the spatial volume formed by the preset reference points distributed in the target space as the boundary and the Pareto solution set obtained by the algorithm. The HV indicator is usually used to evaluate the convergence and diversity of the algorithm. During the operation of the algorithm, when the HV value tends to be stable, the algorithm is considered to be in a convergence state. The larger the HV value, the larger the target space volume formed by the obtained solution, which proves that the diversity and quality of the Pareto solution set solved by the algorithm are better. The calculation formula of the HV indicator is: where λ represents the Lebesgue constant to measure the HV value; |S| refers to the number of solutions in the Pareto solution set; v i represents the HV value of the i-th non-inferior solution and the reference point.

[0140] In order to further illustrate the effect of the present invention, specific examples are used to illustrate it below.

[0141] The following is the specific program test environment of this embodiment:

[0142] The simulation computing environment is Intel(R) Core(TM) i5-12400 CPU@2.5GHz, 16GB RAM, Windows10Pro operating system, and MATLAB R2016a is used for programming and running.

[0143] 1. Comparison of similar algorithms:

[0144] The method of the present invention and some conventional methods were used to solve 55 disassembly cases of waste printers of a certain enterprise.

[0145] Table 2 Results of solving P52 using different algorithms

[0146]

[0147]

[0148] The P52 solving algorithms include ant colony optimization (ACO), multi-objective bacterial foraging optimization (MBFO), ant colony and genetic algorithm (ACGA), artificial fish swarm algorithm (AFSA), improved cat swarm optimization (ICSO), genetic algorithm / simulated annealing (GASA), improved whale optimization (IWOA), ant colony and genetic algorithm (ACGA), improved cat swarm optimization (ICSO), genetic algorithm / simulated annealing (GASA), improved whale optimization (IWOA) and improved moth flame optimization (IMFO).

[0149] As can be seen from Table 1, the proposed method obtains 10 non-inferior solutions, among which the idle rate is 0.0579, the equilibrium index range is 69.66% to 100%, and the disassembly cost range is 124.71 to 147.324 yuan. Compared with the results obtained by other algorithms, the IGWO algorithm not only obtains the maximum smoothing rate Fsmooth = 1 and the minimum disassembly cost Fcost = 124.71, achieving the optimization purpose, but also each solution dominates the results obtained by other algorithms. The comparison results are shown in Figure 7 As shown, it shows that the results obtained by the IGWO algorithm are closer to the Pareto frontier, the quality and distribution of the solution are better, and it has a good optimization effect in solving the DLBP problem of large-scale P52. The figure shows the comparison between the algorithm of the present invention and other algorithms.

[0150] 2. Description of the actual disassembly task effect:

[0151] In order to further verify the superiority of the proposed IGWO algorithm in solving the problem, the present invention takes the retired engine disassembly line of a disassembly company in southwest China as an example, and uses the proposed model and method to optimize the performance of the disassembly line to verify the proposed method.

[0152] There are 37 disassembly tasks in the retired engine, as shown in Figure 8. By evaluating and recording the noise value of each task on site, the disassembly time t of two different disassembly modes is calculated. ir / second), cost c ir / element) and noise L ir / dB(A)). Table 3 lists the specific statistical data.

[0153] Table 3 Data information of various disassembly tasks of retired engines

[0154]

[0155]

[0156] Test results:

[0157] According to the actual disassembly of the engine, the predetermined cycle time CT=580s was set, and the average HV value was calculated after running each parameter combination of the IGWO algorithm ten times. The experimental results show that the optimal parameter settings are MI=150, WN=500 and Pd=0.85. After determining the optimal parameters of the algorithm, the IGWO algorithm was run ten times with the optimal parameter combination. When the IGWO algorithm reached the maximum HV value, 265 non-dominated solutions were obtained. In order to provide decision makers with a more comprehensive disassembly solution, we recorded the optimal disassembly solutions under the four objectives and marked them as S1, S2, S3 and S4, respectively, as shown in Fig. 8A , Figure 8B , Figure 8C and Fig.8D As shown, Fig. 8A , Figure 8B , Figure 8C and Fig.8D They are the disassembly plans when the negative impact index of noise pollution is optimal, the idle time balance index is optimal, the disassembly cost is optimal, and the number of open workstations is optimal. Fig. 8A NS=9, SI=186822, DC=6301.514, DN=80.38, Figure 8B NS=8, SI=18821, DC=6118.31, DN=81.37, Figure 8C NS=8, SI=125885, DC=5693.47, DN=84.76, Fig.8D NS = 8, SI = 38123, DC = 7695.118, DN = 83.38. In Figure 8, the darker disassembly tasks use destructive disassembly, and the rest use non-destructive disassembly.

[0158] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the embodiments of the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An adaptive planning method for local destruction disassembly lines considering feasibility and noise, characterized in that: The following steps are involved: Step 1: Obtain basic information of the disassembly task, including disassembly time, tools required for disassembly, disassembly noise, priority relationship and fault information; Step 2: Construct a feasibility evaluation system for partial destructive disassembly of the product to be disassembled, extract the corresponding coarse-grained evaluation indicators and fine-grained indicators from the product layer and the component layer respectively, and quantify the indicators. According to the quantified value of the feasibility indicator, determine whether the disassembly task adopts destructive disassembly or non-destructive disassembly; Step 3: Establish a balance problem model for the local destruction disassembly line: minimize the number of open workstations, idle time balance index, disassembly cost and noise pollution as the objective function; at the same time, establish the constraint conditions of the balance problem of the local destruction disassembly line, including disassembly task allocation constraints, disassembly priority relationship constraints, time constraints, feasibility constraints, noise constraints and workstation allocation constraints; Step 4: Based on the Grey Wolf Algorithm, solve the above-mentioned local destruction disassembly line balance problem model and output the optimal solution and optimal plan; The objective function is: F=[NS,SI,DC,DN] Where NS is the number of open workstations; S g is the workstation enable variable, which is a 0,1 variable. It takes 1 when the workstation is enabled and 0 when it is disabled. g is the workstation number. m is the maximum number of enabled workstations. SI is the idle time balance index; AT is the actual cycle time; i, j represent the disassembly task numbers; r is the disassembly mode, when r=1, it is destructive disassembly, when r=0, it is non-destructive disassembly; N is the set of disassembly task numbers; DC is the disassembly cost index; c ir is the disassembly cost when task i is in disassembly mode r; c s is the unit time cost of opening the workstation; DN is the negative impact index of noise pollution; L lir When disassembly task i is switched to disassembly mode r, the equivalent continuous sound pressure level of the noise caused by the disassembly operation is L lir (dB(A)); t ir is the disassembly time when disassembly task i is in disassembly mode r; The constraints include part removal constraints, workstation task allocation constraints, takt time constraints, real takt time constraints, workstation sequential opening constraints, disassembly task priority relationship constraints, local destruction disassembly feasibility constraints, and noise constraints; Cycle time limit: In the formula, x ig Indicates the allocation variable. When x ig = 1, indicating that disassembly task i is assigned to workstation g. ig =0, it means that disassembly task i is not assigned to workstation g; CT is the maximum allowed cycle time; Disassembly task precedence constraint: In the formula, h represents the workstation number; y ij represents the priority relationship between disassembly task i and disassembly task j. If disassembly task i is the predecessor task of disassembly task j, then y ij =1, otherwise y ij =0; Workstation sequential startup constraints: Completely disassemble the constraint: Feasibility constraints for local destruction and disassembly: Where R(i) is the disassembly mode assignment variable. If task i is assigned to destructive disassembly, R(i) = 1; otherwise, R(i) = 0. i represents the local destruction feasibility value of disassembly task i, and E represents the local destruction feasibility threshold; Noise constraints: Where, L eqi is the noise pollution function, L eqs is the noise threshold; L i is the instantaneous noise value when component i is removed.

2. The method according to claim 1, characterized in that The operation of step 2 is: Coarse-grained evaluation: The failure rate, number of connecting parts, and destructive disassembly cost of the product to be disassembled are used as coarse-grained evaluation indicators. The disassembly entropy is used to quantify the coarse-grained indicators. Based on the evaluation value of the coarse-grained evaluation indicator and the first threshold, it is determined whether the product to be disassembled can be disassembled by destructive disassembly. If destructive disassembly is possible, proceed to the next step; Fine-grained evaluation: Based on fine-grained indicators: destructive component identification, disassembly tools, disassembly direction, disassembly time, disassembly force, accessibility, positioning accuracy, and component structure size, fuzzy trigonometric functions are selected as the membership functions of the failure characteristics on fine-grained indicators. The evaluation level set is used to classify the results of the impact of component failure degrees on fine-grained indicators. Based on expert experience and experience, the membership functions of each indicator and the failure level are established and the fine-grained indicator values ​​are calculated; the weights of each indicator are determined by the structural entropy weight method, and finally the evaluation value of the fine-grained evaluation indicator and the second threshold are used to determine whether the disassembly task is destructive disassembly or non-destructive disassembly.

3. The method according to claim 2, characterized in that The specific method for determining the disassembly method of the product to be disassembled is as follows: Get the quantitative entropy of failure rate, number of joints, and cost of damage removal: S1=log2(N i / N s ) S2=log2[(N i +N k ) / N k ] S3=log2[(C n +C m ) / C n ] In the formula, S1 represents the quantitative entropy of failure rate; S2 represents the quantitative entropy of the number of joints; S3 represents the quantitative entropy of damage and disassembly cost; N i Represents the number of disassembly tasks in the product to be disassembled, N s Represents the number of disassembly tasks with high failure probability in the product to be disassembled, and its specific data is determined by expert evaluation method; N k Indicates the number of connecting parts in a disassembled product; C n is the non-destructive disassembly cost; C m for the destruction of demolition costs; Based on the quantitative entropy of failure rate, number of joints and damage and disassembly cost, the coarse-grained evaluation index is obtained: S=0.33S1+0.33S2+0.34S3 When the coarse-grained evaluation index S of the product to be disassembled is greater than a first threshold value given by the user, the product to be disassembled adopts a destructive disassembly mode; The disassembly task of the product to be disassembled with a high failure probability is obtained based on an expert scoring method.

4. The method according to claim 2, characterized in that: The specific method for determining the disassembly method of a disassembly task is as follows: Based on the fault characteristics, eight fine-grained indicators are established, including damaged component identification, disassembly tools, disassembly direction, disassembly time, disassembly force, accessibility, positioning accuracy, and component structure size; The fuzzy trigonometric function is selected as the membership function of the failure characteristics on the fine-grained index, and its membership [Δa x , Δa y The membership degrees of the intervals are Calculate the damage component identification index value According to the above method, the index values ​​of the damaged component identification, disassembly tools, disassembly direction, disassembly time, disassembly force, accessibility, positioning accuracy, and component structure size of the disassembly task are obtained, u y , y=1,2,```,6, where y=1 represents the damage component identification index, y=2 represents the disassembly tool index, y=3 represents the disassembly direction index, y=4 represents the disassembly time index, y=5 represents the disassembly force, and y=6 represents the accessibility index; Based on the structural entropy weight method, the weight α of the fine-grained index is obtained y , y=1,2,```,6; Get the local destructive disassembly feasibility value of the disassembly task: In the formula, e represents the feasibility value of local destructive disassembly of the disassembly task; The local destructive disassembly feasibility value of the disassembly task is compared with a second threshold. When the local destructive disassembly feasibility value of the disassembly task is greater than the second threshold, destructive disassembly is performed on the disassembly task, otherwise non-destructive disassembly is performed. The second threshold is the average value of the local destructive disassembly feasibility values ​​of all disassembly tasks in the product to be disassembled.

5. The disassembly line adaptive planning method according to claim 1, characterized in that: In step 4, the specific steps of the gray wolf algorithm are: S1. Generate the initial population based on the double-layer encoding method and calculate its objective function value; S2. Based on the hunting behavior of gray wolves, crossover and mutation operations are performed on non-inferior solutions; S3, using the Pareto elite strategy and NSGA-Ⅱ crowding distance mechanism to select the non-inferior solution; S4. Repeat the above operations until the iteration limit is reached, and output the optimal solution and optimal disassembly plan.

6. The method for self-adaptive planning of a disassembly line according to claim 5, characterized in that: In S2, during the crossover operation, a convergence factor is introduced to determine the crossover process, and the convergence factor is as follows: A=2a·r1-a Where: T is the maximum number of iterations, t is the current number of iterations, ζ is the modulation factor that can effectively change the rate of change after multiple experiments, and ζ=20; r1 is a random number between [0,1]; A is the coefficient vector; When |A|>1, the gray wolf moves away from the current prey and enters a larger space for global search; when |A|≤1, the gray wolf approaches the prey and conducts a local search.

7. The method for adaptive planning of a disassembly line according to claim 5, characterized in that: In S2, during the mutation operation, a disturbance factor is introduced as a local change operation: In the formula, For the gray wolf individual, is the individual after mutation, d is a random number between [0,1], P d is the disturbance factor, y ipr for y i The position of the task's immediate predecessor, y isu for y i The position of the task immediately following the task, '[]' is the rounding function, and λ is a random number between [0,1]; p∈[1,2,…,n], when d <P d When the gray wolf individuals are mutated.

Images