Method, device and equipment for disassembling retired electromechanical products with random process failure

Abstract

The present invention discloses a method, device and equipment for dismantling retired electromechanical products with random process failures, the method comprising: using a Bayesian network to analyze the failure causal relationship between each process in the dismantling process of retired electromechanical products, and determining the probability of failure state transfer; estimating the probability of random failure conditions of each dismantling process according to the failure sample data in the dismantling process of retired electromechanical products; combining a genetic algorithm, optimizing the dismantling process route according to the failure condition probability of the process, and searching for the global approximate optimal solution for the dismantling of retired electromechanical products; according to the designed failure repair strategy, taking corresponding repair operations for the process that has failed, and ensuring the continuity of the dismantling sequence during the dismantling process of retired electromechanical products. The present invention can deeply explore the failure behavior mechanism of the dismantling process, and decide the global optimal dismantling process route based on the failure causal relationship between the dismantling processes.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a method, device and equipment for disassembling retired electromechanical products with random process failures. Background Art

[0002] The end of the service life of electromechanical products and the acceleration of their updates and iterations have led to a large number of retired electromechanical products being produced every year. The rational and efficient recycling of retired electromechanical products can not only promote the recycling of resources, but also help reduce adverse consequences such as environmental pollution and safety hazards. In recent years, the recycling of retired electromechanical products has become one of the hot topics of global concern. Separating the valuable parts inside retired electromechanical products through disassembly operations is a prerequisite for their recycling and utilization, and is also the primary link in realizing circular economy and green manufacturing.

[0003] The key to improving recycling efficiency and resource utilization lies in implementing disassembly process planning, systematically analyzing, designing and optimizing the disassembly process, and deciding on the global optimal disassembly process route. However, the quality status of components in retired electromechanical products is often unknown and uncertain, which often leads to process failures during the disassembly process, which can easily lead to the failure of the established disassembly process route and the interruption of disassembly production operations, thereby increasing the complexity and difficulty of solving the disassembly process planning problem. In addition, the failure of the preceding process in the disassembly operation may cause cascading failures in the subsequent process, that is, the failure behaviors of the preceding and subsequent disassembly processes are not completely independent of each other, but have a certain causal correlation. Ignoring the causal correlation between the failures of the disassembly processes will compress the optimization space of the disassembly process route, making it difficult to decide on the global optimal disassembly process route. Therefore, disassembly process planning under the condition of random failure of the disassembly process has important research significance.

[0004] The uncertainty of disassembly information of retired electromechanical products is an important reason for the random failure behavior of the process during the disassembly process. Domestic and foreign scholars have conducted many studies on this direction. However, the existing research on the uncertainty of disassembly process planning mainly focuses on the quality status of retired electromechanical products and the characteristic parameters of the disassembly process. Although some studies have considered the random failure characteristics of the disassembly process, they often assume that the probability or probability distribution of random failure of the disassembly process is known, and do not consider the causal relationship between the previous and subsequent disassembly processes. There is still a lack of effective methods that can accurately estimate the probability of random failure of the disassembly process, and it is difficult to accurately characterize the failure behavior mechanism of the disassembly process, resulting in the inability to timely and effectively correct and decide on the global optimal disassembly process route under the conditions of the failure behavior of the disassembly process. Summary of the invention

[0005] The embodiments of the present invention provide a method, device and equipment for disassembling retired electromechanical products with random process failures, which can deeply explore the failure behavior mechanism of the disassembly process and decide the global optimal disassembly process route based on the failure causal relationship between the disassembly processes.

[0006] An embodiment of the present invention provides a method for disassembling retired electromechanical products with random process failures, comprising:

[0007] The Bayesian network is used to analyze the causal relationship between failures in the disassembly process of retired electromechanical products and determine the probability of failure state transfer.

[0008] Based on the failure sample data during the disassembly of retired electromechanical products, the conditional probability of random failure in each disassembly process is estimated;

[0009] Combined with genetic algorithm, the disassembly process route is optimized according to the failure condition probability of the process, and the global approximate optimal solution for the disassembly of retired electromechanical products is searched;

[0010] According to the designed failure repair strategy, corresponding repair operations are taken for the process where failure has occurred to ensure the continuity of the disassembly sequence during the disassembly of retired electromechanical products.

[0011] As an improvement of the above scheme, the method of estimating the random failure condition probability of each disassembly process according to the failure sample data in the disassembly process of retired electromechanical products includes:

[0012] Collect and organize failure sample data during disassembly;

[0013] Based on the failure sample data and applying the Bayesian network structure learning algorithm, the network topology of the failure causal relationship between processes is constructed;

[0014] The maximum a posteriori probability estimation method is used to calculate the node state parameters of the Bayesian network and obtain the random conditional failure probability of each disassembly process.

[0015] As an improvement of the above solution, the failure repair strategy includes:

[0016] Define a set of repair operations, each corresponding to a different failure state;

[0017] Select the most appropriate repair action based on the failure type and severity, which may include replacing tools, recalibrating equipment, or increasing manual supervision.

[0018] After performing the repair operation, re-evaluate the status of the repaired process to ensure that it meets the conditions for continued disassembly.

[0019] As an improvement of the above scheme, the genetic algorithm is combined to optimize the disassembly process route according to the failure condition probability of the process, and the global approximate optimal solution for the disassembly of retired electromechanical products is searched, including:

[0020] The genetic algorithm initializes the population, and each individual represents a disassembly sequence. The sequence takes into account the dependencies and failure probabilities of the processes.

[0021] The reliability weight of the disassembly sequence is adjusted by combining the recycling profit, cost and repair fee through the fitness function and using the failure probability estimated by the Bayesian network;

[0022] The priority crossover operation is adopted to maintain the logical relationship of the process, combined with the single-point mutation operation to ensure the feasibility and efficiency of the sequence;

[0023] Through multiple iterations, the genetic algorithm is used to search for the global optimal solution and find the best disassembly process route under failure and repair conditions.

[0024] Another embodiment of the present invention provides a dismantling device for retired electromechanical products with random process failures, including:

[0025] The analysis module is used to analyze the causal relationship between failures in various processes during the disassembly of retired electromechanical products using a Bayesian network to determine the probability of failure state transfer;

[0026] An estimation module is used to estimate the conditional probability of random failure in each disassembly process based on the failure sample data during the disassembly of retired electromechanical products;

[0027] A search module is used to combine genetic algorithms to optimize the disassembly process route according to the failure condition probability of the process and search for the global approximate optimal solution for the disassembly of retired electromechanical products;

[0028] The repair module is used to take corresponding repair operations for the failed processes according to the designed failure repair strategy, so as to ensure the continuity of the disassembly sequence during the disassembly of retired electromechanical products.

[0029] As an improvement of the above solution, the estimation module is specifically used for:

[0030] Collect and organize failure sample data during disassembly;

[0031] Based on the failure sample data and applying the Bayesian network structure learning algorithm, the network topology of the failure causal relationship between processes is constructed;

[0032] The maximum a posteriori probability estimation method is used to calculate the node state parameters of the Bayesian network and obtain the random conditional failure probability of each disassembly process.

[0033] As an improvement of the above solution, the failure repair strategy includes:

[0034] Define a set of repair operations, each corresponding to a different failure state;

[0035] Select the most appropriate repair action based on the failure type and severity, which may include replacing tools, recalibrating equipment, or increasing manual supervision.

[0036] After performing the repair operation, re-evaluate the status of the repaired process to ensure that it meets the conditions for continued disassembly.

[0037] As an improvement of the above scheme, the genetic algorithm is combined to optimize the disassembly process route according to the failure condition probability of the process, and the global approximate optimal solution for the disassembly of retired electromechanical products is searched, including:

[0038] The genetic algorithm initializes the population, and each individual represents a disassembly sequence. The sequence takes into account the dependencies and failure probabilities of the processes.

[0039] The reliability weight of the disassembly sequence is adjusted by combining the recycling profit, cost and repair fee through the fitness function and using the failure probability estimated by the Bayesian network;

[0040] The priority crossover operation is adopted to maintain the logical relationship of the process, combined with the single-point mutation operation to ensure the feasibility and efficiency of the sequence;

[0041] Through multiple iterations, the genetic algorithm is used to search for the global optimal solution and find the best disassembly process route under failure and repair conditions.

[0042] Another embodiment of the present invention provides a dismantling device for retired electromechanical products with random process failures, comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the dismantling method for retired electromechanical products with random process failures described in the above-mentioned embodiment of the invention when executing the computer program.

[0043] Another embodiment of the present invention provides a storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method for dismantling retired electromechanical products with random process failures described in the above-mentioned embodiment of the invention.

[0044] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0045] The embodiment of the present invention aims at the process failure behavior that often occurs during the disassembly of retired electromechanical products. From the perspective of internal parts failure and the execution level of the disassembly process, the random failure behavior mechanism of the disassembly process is systematically studied, including the accurate analysis of the failure causal relationship mechanism between multiple processes and the precise estimation of the probability of random failure (condition) of a single process, and the establishment of a disassembly process route decision model under the conditions of random failure of the process and its repair. A Bayesian network-genetic hybrid algorithm is proposed, and the failure causal relationship between the disassembly processes and the random failure conditional probability of each disassembly process are respectively obtained through the Bayesian network. On this basis, the genetic algorithm is used to efficiently solve the global approximate optimal disassembly process route. From the above analysis, it can be seen that the embodiment of the present invention focuses on the random failure problem of the disassembly process of retired electromechanical products. From the perspective of internal parts failure and the execution level of the disassembly process, the failure behavior mechanism of the disassembly process is deeply explored, and the global optimal disassembly process route is decided based on the failure causal relationship between the disassembly processes. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of a method for dismantling retired electromechanical products with random process failures provided by one embodiment of the present invention;

[0047] Figure 2 is a failure causal relationship diagram provided by an embodiment of the present invention;

[0048] Figure 3 is a topological structure diagram provided by an embodiment of the present invention;

[0049] Figure 4 is a BN-GHA calculation flow chart provided by an embodiment of the present invention;

[0050] Figure 5 It is an iterative curve diagram of different failure states of the initial process during general random failure provided by an embodiment of the present invention;

[0051] Figure 6 It is an iterative curve diagram provided by an embodiment of the present invention when the initial process failure state after repair is S1;

[0052] Figure 7 It is a schematic diagram of a process of dismantling a retired electromechanical product with random process failure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0054] See also Figure 1 , is a flow chart of a method for dismantling retired electromechanical products with random process failures provided by an embodiment of the present invention. The method for dismantling retired electromechanical products with random process failures comprises steps S10 to S13:

[0055] S10, using Bayesian networks to analyze the causal relationship between failures in various processes during the disassembly of retired electromechanical products and determine the probability of failure state transfer;

[0056] S11, based on the failure sample data during the disassembly of retired electromechanical products, estimate the conditional probability of random failure in each disassembly process;

[0057] S12, combined with genetic algorithm, optimizes the disassembly process route according to the failure condition probability of the process, and searches for the global approximate optimal solution for the disassembly of retired electromechanical products;

[0058] S13, according to the designed failure repair strategy, take corresponding repair operations for the process where failure has occurred to ensure the continuity of the disassembly sequence during the disassembly of retired electromechanical products.

[0059] The embodiment of the present invention aims at the process failure behavior that often occurs during the disassembly of retired electromechanical products. From the perspective of internal parts failure and the execution level of the disassembly process, the random failure behavior mechanism of the disassembly process is systematically studied, including the accurate analysis of the failure causal relationship mechanism between multiple processes and the precise estimation of the probability of random failure (condition) of a single process, and the establishment of a disassembly process route decision model under the conditions of random failure of the process and its repair. A Bayesian network-genetic hybrid algorithm is proposed, and the failure causal relationship between the disassembly processes and the random failure conditional probability of each disassembly process are respectively obtained through the Bayesian network. On this basis, the genetic algorithm is used to efficiently solve the global approximate optimal disassembly process route. From the above analysis, it can be seen that the embodiment of the present invention focuses on the random failure problem of the disassembly process of retired electromechanical products. From the perspective of internal parts failure and the execution level of the disassembly process, the failure behavior mechanism of the disassembly process is deeply explored, and the global optimal disassembly process route is decided based on the failure causal relationship between the disassembly processes.

[0060] As an improvement of the above scheme, the method of estimating the random failure condition probability of each disassembly process according to the failure sample data in the disassembly process of retired electromechanical products includes:

[0061] Collect and organize failure sample data during disassembly;

[0062] Based on the failure sample data and applying the Bayesian network structure learning algorithm, the network topology of the failure causal relationship between processes is constructed;

[0063] The maximum a posteriori probability estimation method is used to calculate the node state parameters of the Bayesian network and obtain the random conditional failure probability of each disassembly process.

[0064] In this embodiment, specifically:

[0065] Failure sample data includes relevant field data and historical records during the disassembly of retired electromechanical products. The failure behavior of the disassembly process may come from the damage of internal parts of retired electromechanical products, manual operation errors in the disassembly process, and disassembly equipment failures. These failure inducements will cause different degrees of failure impact on the disassembly process, thus forming different types of failure states. Si is used to describe the failure state of the disassembly process, which consists of failure inducements and failure degrees, and is expressed in the form of an array [u, v], where u = 1, 2, 3 respectively represent internal damage to parts, manual operation errors, and disassembly equipment failures; v = 1, 2, 3 respectively represent minor, general, and serious, then Si has 9 types as shown in Table 1. The effects of different failure inducements and failure degrees of the pre-disassembly process on the failure behavior of the post-disassembly process may vary. The failure causal relationship diagram that may exist between any two processes is shown in the following figure. Figure 2 shown.

[0066] Table 1 Failure status types

[0067]

[0068] In order to accurately describe the causal relationship between the failures of the two processes, the disassembly priority matrix P = [pij] is used to represent the priority relationship between the processes. Assuming that there are 10 processes in the process to be disassembled, each process corresponds to a component. The rows and columns in Table 2 correspond to processes i and j, respectively, where the value of pij can be calculated by formula (1). For example, p12 means that process 1 is the immediate predecessor of process 2, that is, process 2 can be executed immediately after process 1 is completed, and process 2 cannot be executed before process 1.

[0069]

[0070] Table 2 Priority Matrix

[0071]

[0072] In order to improve the disassembly efficiency under the condition of random failure of the process, it is usually necessary to consider whether to take corresponding repair measures for the failed process. Therefore, the DDPR-RFR model established in this embodiment includes two decision situations: general random failure and failure repair operation. The latter mainly considers the following three repair methods: (1) Direction change operation: keep the original disassembly operation unchanged and only change the disassembly direction; (2) Auxiliary recovery operation: under the condition that the existing disassembly situation remains unchanged, an additional operation is performed, such as applying vibration to the original operation; (3) Tool change operation: replace the disassembly tool with one that can successfully remove the disassembly target, such as using a hot air gun, cutter, electric drill, etc.

[0073] In order to facilitate the description of the problem model, the following symbols are defined first:

[0074] Subscript number:

[0075] i,j,o failure state number, i,j,o=1,2,…,H, where o is the initial process failure state and H is the total number of failure states;

[0076] g,k,w are the disassembly process numbers, a,k,w=1,2,…,J, where g is the initial process and J is the total number of processes;

[0077] l,u is the sequence number of the process in the disassembly sequence, l,u=1,2,…,J;

[0078] λ Repair method number, λ=1 (direction change), 2 (assisted recovery), 3 (tool change).

[0079] parameter:

[0080] c k The recovery cost per unit time of process k;

[0081] eP ojk When the initial process failure state type is o, the expected probability that the failure state type of process k is j;

[0082] f. Profits from recycling retired electromechanical products;

[0083] n k The number of parts corresponding to process k in a single retired electromechanical product

[0084] N ijk The number of samples with failure status type j of process k when the failure status of the parent process is i; P ijk The probability that the failure state type of process k is j when the failure state of the parent process is i;

[0085] rc λRepair method λ repair cost per unit time;

[0086] rP ojk After the repair operation is adopted, when the initial process failure state type is o, the expected probability that the failure state type of process k is j;

[0087] R Recycling income of retired electromechanical products;

[0088] S k Failure state of process k;

[0089] The cost of restoration incurred by the TC’s choice of restoration;

[0090] t k Disassembly time of process k;

[0091] t kλ The repair time of process k using repair method λ;

[0092] η jλ The success rate of repair process k using repair method λ;

[0093] φ k Decision variable y kl =1, the value of the sequence number of process k in the disassembly sequence;

[0094] δ k The recycling income of process k;

[0095] δP j Compared with the case where no failure occurs, the proportion of recovery income when the failure state is j;

[0096] Decision variables:

[0097] y kl If process k is the lth process to be executed, then y kl =1,φ k = l, otherwise y kl =0;

[0098] z kλ Use repair method λ to repair process k, then z kλ =1, otherwise z kλ =0.

[0099] The DDPR-RFR model aims to maximize the recycling profit of a retired electromechanical product during the dismantling process, which is:

[0100] max f=R-TC (2)

[0101] For formula (2), in the disassembly process flow with general random failure behavior, the recycling profit is related to the process failure probability. This embodiment only considers the impact of process failure on recycling income. The corresponding recycling income and cost of retired electromechanical products are expressed as follows:

[0102]

[0103] If there are random failure behaviors in the process of disassembly and failure repair operations are taken, the recycling income is related to the failure probability after the process is repaired, and the total recycling cost increases a repair cost:

[0104]

[0105] In summary, the objective function of expected recovery profit in two cases, when random failure of the disassembly process occurs and when repair is not adopted, can be expressed by formula (7):

[0106]

[0107] In addition, to ensure the feasibility of the disassembly process route, the model also needs to meet the following constraints:

[0108]

[0109]

[0110] Among them, formula (8) is the expression of the sequence number of process when process k is the lth process to be executed; formula (9) restricts that multiple processes cannot be executed simultaneously in the disassembly process flow; formula (10) is used to describe the priority constraint relationship between processes. When p lu =1, the disassembly sequence number l of process k must be smaller than the disassembly sequence number u of process w.

[0111] The failure probabilities under the two conditions of general random failure and failure repair are the key factors for solving the DDPR-RFR model. In order to obtain these two failure probabilities, it is necessary to accurately characterize the failure causal relationship between the disassembly processes and accurately estimate the random failure (conditional) probability of the processes. Bayesian network (BN), as a probabilistic graph model that describes the dependency relationship between variables, can be used to describe the causal relationship between the failure behaviors of the previous and subsequent disassembly processes. It includes a directed acyclic graph (DAG) composed of a node set composed of random variables and directed edges that describe the causal relationship between nodes, which can directionally describe the dependency relationship between variables, and a conditional probability table (CPT) composed of the probability of occurrence of each node state when its parent node set is in a certain state, which can quantify the interaction relationship between variables. Therefore, this embodiment designs a BN-GHA method that combines Bayesian network with genetic algorithm to solve the DDPR-RFR model.

[0112] Before using BN to analyze the random failure behavior mechanism of the disassembly process, it is necessary to define the failure causal relationship between multiple processes in the BN model in combination with the description of the failure state in the DDPR-RFR model: map each disassembly process to each node inside the BN one by one, so that each node corresponds to a disassembly process, the node state corresponds to the failure state of the disassembly process, and the failure causal relationship between the disassembly processes corresponds to the directed edges connecting the BN nodes. Use the causal relationship structure equation X k =Φ(X w ) defines this logical relationship mapping, let {d1,d2,…,d 10} is a set of network nodes, then the structural equation can be expressed as d k |S i =Φ(d w |S j ), where k, w = 1, 2, ..., 10; i, j = 1, 2, ..., 9; such as d5|S3 = Φ(d1|S5), which means that the general failure of the first process due to human error will lead to internal damage of parts in the fifth process and cause serious failure.

[0113] The basic idea of ​​the BN-GHA method is as follows: first, based on the K2 algorithm and the scoring search strategy, the Bayesian network topology structure is learned from the sample data set to accurately characterize the failure causal relationship between each disassembly process; then, based on this structure, a Bayesian network node state parameter estimation method for failure causal relationship between multiple processes is proposed to accurately represent the conditional probability of random failure of the subsequent process under different failure states of the preceding process; finally, with the goal of maximizing the recycling profit, the relevant parameters obtained by the Bayesian network are integrated into the genetic algorithm to obtain the optimal disassembly process route under two decision-making situations.

[0114] BN topology generation:

[0115] Structural learning refers to learning the corresponding directed acyclic graph structure from a training sample data set. Currently, it is mainly summarized into three types of methods: constraint-based, score search-based, and hybrid algorithm-based structural learning methods. The K2 algorithm has extremely high accuracy and effect, but it has a major challenge that the output quality is highly dependent on the given node ordering. If the variable order is inaccurate, a network that is not similar to the real network will be obtained. Therefore, this embodiment is designed to determine the node order based on the priority relationship of the disassembly process, and then use the K2 algorithm based on hybrid structural learning to generate the BN topological structure. The central idea is: call the topological sorting in the given priority constraint relationship to obtain a linear arrangement of the parent node in front of the child node as the initial node order, and then select the node candidate parent set with the best BIC score, and then learn to obtain the optimal network structure. The specific algorithm steps are as follows:

[0116] (1) Import the disassembled sample data set and the priority matrix P;

[0117] (2) Based on the priority matrix P, select the initial node and its corresponding child nodes for sorting. If the child node has no child nodes, proceed to the next child node. If there are child nodes, sort their child nodes first, and finally determine the node order [nd1, nd2, …, nd 10 ]. If the initial node is d2, its child nodes are d5 and d6, d5 has a child node d3 and d3 has a child node d9, and d9 and d6 have no child nodes, then the current order can be [d2, d5, d3, d9, d6];

[0118] (3) Arrange at node nd i The first (i-1) nodes (i>1) are taken as candidate parent nodes, and nd is calculated according to the scoring function formula. i The Bayesian Information Criterion (BIC) score of the local network structure formed with each candidate parent node;

[0119]

[0120] Where Nijk is the number of samples in the disassembly sample data set that satisfy the failure state type j of process k under the condition that the failure state of the parent process is i, N is the number of disassembly samples, the first term represents the log likelihood of the disassembly sample data and the BN topological structure, and the second term is used to measure the parameter complexity and the storage space of the disassembly sample data. The larger the BIC(G|D) value, the more accurately the current BN topological structure can describe the causal relationship between the disassembly processes.

[0121] (4) If the parent node with the highest score among the candidate parent nodes has a higher score than when the parent node is not added, then the parent node is added to node d. i The parent node is concentrated, otherwise the parent node is discarded;

[0122] (5) Loop through the best parent node of each node in turn, add edges from the parent node to the child node, and obtain the final network structure and adjacency matrix dag;

[0123] (6) Draw a topological structure diagram. The result is as follows Figure 3 shown.

[0124] Under the condition that the BN topology structure facing the causal relationship of failure among multiple processes is known, a BN node state parameter estimation method for the causal relationship of failure among multiple processes is proposed based on the Bayesian maximum a posteriori probability estimation theory: Let the BN node state parameter P ijk Normal distribution N(μ,σ) subject to approximate equality constraints 2 ) describes the expert prior knowledge, by approximating P ijk The prior normal distribution parameters of the Beta distribution are obtained, and finally the hyperparameter ɑ of the Dirichlet distribution is obtained. ijk , the random failure CPT of each disassembly process is obtained according to the Bayesian maximum a posteriori probability estimation formula. The specific algorithm steps are as follows:

[0125]

[0126] (1) Import the disassembly sample data set, the adjacency matrix dag obtained by structure learning, and the pre-set failure state type;

[0127] (2) Determine the number of levels L of the topological structure division, and count the nodes dL and the number of samples N of nodes in each layer whose child node state is j when the parent node state is i dL ;

[0128] (3) According to dL and N dL Get the number of samples N where node k is in node state j when the parent node state combination is i ijk ;

[0129] (4) Calculate the expectation μ, EB and variance σ of normal distribution and Beta distribution respectively 2 ,DB; Taking the minimum distance between the two distribution variances as the objective function, the equal expectation, the Beta distribution as a convex function (α, β>1) and the expression of its expectation and variance as the constraint conditions to solve the Beta distribution parameters α and β;

[0130]

[0131] (5) The hyperparameter ɑ of the Dirichlet distribution is obtained from the parameters α and β ijk , combined with N ijk Get BN node status parameter P ijk , that is, the conditional probability table of 9 nodes excluding the starting node is obtained, namely the random failure CPT of the disassembly process. Table 3 shows the CPT2 result when k=d2.

[0132]

[0133] Table 3 Conditional probability table example (CPT2)

[0134]

[0135]

[0136] Table 3 reflects the probability distribution of each failure state of process 2 under different conditions of the failure state of the preceding process. For example, when the failure state of the preceding process is S1, the probability values ​​of the failure state of process 2 being S1, S2, and S3 are 0.11, 0.15, and 0.06, respectively. Similarly, there are random failures CPT3, CPT4, …, CPT5, and CPT6 of processes 3, 4, …, 10. 10 Furthermore, CPT n The most likely failure states and corresponding probability values ​​of each process under different failure states of the predecessor processes are deduced. For example, when the failure states of the predecessor processes are S1, S2, and S3 respectively, the most likely failure states of process 2 are S8, S8, and S9, and the corresponding probability values ​​are 0.20, 0.201, and 0.30 respectively. Similarly, there are other most likely failure states of other processes under different failure states of the predecessor processes.

[0137] To facilitate analysis and solution, this embodiment only considers the case of a single parent node. The CPT obtained by Bayesian network structure learning and parameter estimation is the probability P that the failure state type of process k is j under the condition that the failure state of the parent process is i. ijk , calculate the expected failure probability under two decision scenarios in the DDPR-RFR model.

[0138] In the case of general random failure, the expected probability of different failure state types in each process can be expressed by formula (15), where k, w represent different disassembly processes, g is the initial process, k is the sub-process of w; j, i, o represent the failure state types corresponding to processes k, w, g.

[0139]

[0140] When repair operations are taken, the expected probability of different failure state types in each process can be expressed by formula (16). jλ It represents the success rate of process repair for different failure state types obtained by training different repair operation combinations.

[0141]

[0142] As an improvement of the above solution, the failure repair strategy includes:

[0143] Define a set of repair operations, each corresponding to a different failure state;

[0144] Select the most appropriate repair action based on the failure type and severity, which may include replacing tools, recalibrating equipment, or increasing manual supervision.

[0145] After performing the repair operation, re-evaluate the status of the repaired process to ensure that it meets the conditions for continued disassembly.

[0146] As an improvement of the above scheme, the genetic algorithm is combined to optimize the disassembly process route according to the failure condition probability of the process, and the global approximate optimal solution for the disassembly of retired electromechanical products is searched, including:

[0147] The genetic algorithm initializes the population, and each individual represents a disassembly sequence. The sequence takes into account the dependencies and failure probabilities of the processes.

[0148] The reliability weight of the disassembly sequence is adjusted by combining the recycling profit, cost and repair fee through the fitness function and using the failure probability estimated by the Bayesian network;

[0149] The priority crossover operation is adopted to maintain the logical relationship of the process, combined with the single-point mutation operation to ensure the feasibility and efficiency of the sequence;

[0150] Through multiple iterations, the genetic algorithm is used to search for the global optimal solution and find the best disassembly process route under failure and repair conditions.

[0151] In this embodiment, specifically:

[0152] Genetic algorithm is a heuristic search algorithm used to solve optimization problems. In order to solve the DDPR-RFR model, the relevant parameters obtained by the Bayesian network are integrated into the genetic algorithm to obtain the optimal disassembly process route under two decision-making situations. The algorithm flow is as follows: Figure 4 shown.

[0153] The population is initialized as follows:

[0154] (1) Initialize the feasible disassembly sequence: Based on the priority matrix and the number of immediate predecessors of each process, a certain number of feasible disassembly sequences are randomly generated. The number of immediate predecessors of each process is updated after each process is completed.

[0155] (2) Initialize the repair selection behavior combination: Based on the initial feasible disassembly sequence, randomly generate a certain number of repair operation combinations corresponding to each process as the initial population, and use z kλ Indicates whether to use the repair operation λ to disassemble process k. When multiple repair operations are used, they are disassembled in sequence, such as {z k1 ,z k2 ,z k3}=010 means that only auxiliary recovery operation is adopted when failure behavior occurs; k1 ,z k2 ,z k3}=101 indicates that the first and third repair operations are selected, and the tool change operation will continue to be used when the direction change operation cannot completely repair the failed process.

[0156] Setting of fitness function:

[0157] Adding repair behavior to the general random failure situation involves changes in the expected failure probability, increases in repair costs and time, so it is necessary to set different fitness functions based on the general random failure and the recovery profit after adding failure repair behavior, represented by f1 and f2 respectively.

[0158] From equations (3), (4), (7), and (15), we can get that the fitness function under general random failure behavior is:

[0159]

[0160] Expected probability of failure states in the same process in different disassembly sequences eP ojk The differences lead to different final fitness function values, thus forming the differences between the disassembly process routes.

[0161] From equations (5), (6), (7), and (16), we can get the fitness function considering the failure repair behavior:

[0162]

[0163] Similarly, rP ojk The difference in the disassembly process routes leads to the differences. In order to select the appropriate repair operation and obtain the best disassembly sequence after incomplete repair, it is necessary to train the success rate η of the repair operation for each failure state. jλ , using ρ kλ It represents the repair situation of process k by repair operation λ:

[0164]

[0165] Based on the sample data set, the ratio of the number of successful repairs of each failure state by the three repair operations to the total number of each failure state is taken as η jλ , which is used to update the expected failure probability after adding the repair operation when each process is in different failure states.

[0166] 2.3.2 Update and Iteration

[0167] In order to obtain chromosomes with better performance while ensuring the feasibility of the offspring during the solution process, this embodiment adopts the method of preferentially retaining crossover

[23] (Precedence preservative crossover, PPX) and single-point mutation operations update the disassembly sequence, as shown in the following table.

[0168]

[0169] Based on a certain crossover probability, the individuals to be crossed are paired in pairs, and two arrays containing only 1 and 2 and equal in length to the parent sequence are randomly generated to determine the choice of the offspring individuals for the current parent gene. After each selection, the selected genes are removed from the parent before the next selection. In order to improve the efficiency of single-point mutation and ensure that the updated sequence still satisfies the priority relationship between processes, the logical relationship of the BN topological structure is selected: first, the mutated individuals are selected based on a certain mutation probability, and then two different processes in the same layer of the topological structure are randomly selected for exchange.

[0170] For the case of general random failure, the optimal fitness value under the condition of H initial node failure state is selected and updated, and the corresponding optimal disassembly sequence is retained; for the case of adding failure repair behavior, the optimal fitness value under different repair selection behavior conditions is selected and updated, and the corresponding optimal disassembly sequence is retained. By comparing the optimal fitness values ​​corresponding to different repair selection combinations, the optimal repair selection behavior combination z is obtained. kλ and the optimal disassembly sequence.

[0171] In order to verify the feasibility and effectiveness of the DDPR-RFR model established in this example and the proposed BN-GHA method, this chapter uses power batteries as the disassembly object for case study. All experiments were programmed on MATLAB 2016a software and on an Intel(R) Core(TM) i7-8550U processor. CPU@1.80GHz The computer has a processor and 8GB of memory and runs in the win10 system environment.

[0172] For ease of understanding, the following test examples are provided here:

[0173] Power batteries are one of the core components of electric vehicles and directly determine the performance and cost of the vehicle.

[24] In this example, the power battery of Audi A3 is used as the disassembly object. Some case data are from the literature

[25] . Other DDPR-RFR model input data that cannot be directly collected and obtained are randomly generated by referring to the situation of batteries of the same type using an evaluation method, including the disassembly time t of each process. k , the average repair cost per unit time of different repair methods rc λ Table 4 shows the 17 important components of the power battery, Table 5 shows the disassembly priority of the power battery, and Table 6 shows the repair status of each component by three different repair operations based on previous disassembly experience.

[0174] Table 4 Disassembly information of important components of power battery

[0175]

[0176]

[0177] Table 5 Disassembly priority matrix of power batteries

[0178]

[0179]

[0180] Table 6 Repair status of power battery disassembly process

[0181]

[0182] The integrated genetic algorithm is used to perform disassembly process planning for general random failure behavior and failure repair behavior. The population size is set to 20 and the maximum number of iterations is 30. The iteration curves of the power battery in the two cases are as follows: Figure 5 , 6 shown.

[0183] In the case of general random failure behavior, the effects of different failure states in the initial process on the solution are as follows: Figure 5 As shown. It can be seen that the recycling profits obtained from different initial process failure states show a slight stratification phenomenon, that is, the recycling profits of S1, S4, and S7 are in the range of 459-461, the recycling profits of S2, S5, and S8 are between 457-459, and the recycling profits of S3, S6, and S9 are between 454-456, indicating that the failure degree of the initial process is an important factor affecting the recycling profit; and through the same-layer comparison of the fitness values ​​of the final convergence, it can be seen that S4>S7=S1, S5>S8>S2, S6>S9>S3, that is, when the failure degree of the initial process is certain, the impact of human operation errors in the failure inducement on the recycling profit will be greater than the impact caused by disassembly equipment failure and internal damage to parts.

[0184] When failure repair behavior is added, since genetic algorithms often require the system to run continuously for multiple times to obtain the optimal solution, the population size and the maximum number of iterations are kept unchanged, and the system is run continuously for 10 times to obtain the change curve of the objective function value with the number of iterations when the initial process failure state is S1. Figure 6 It can be seen that when the number of iterations is greater than 11, the system calculation results converge stably to 828.5181.

[0185] 3.2.1 Comparison of three scenarios

[0186] The DDPR-RFR model of this embodiment considers two decision-making situations under the condition of disassembly process failure, namely general random failure and adding failure repair operation. The comparison results of the two and the disassembly process planning without disassembly process failure behavior are shown in Table 10. When disassembly process failure occurs and repair operation is taken, the optimal repair operation combination is {101, 101, 100, 110, 111, 010, 101, 110, 001, 000, 011, 010, 000, 000, 100, 100}. The recycling profit when process failure occurs will be lower than the recycling profit when process failure does not occur, among which general random failure behavior reduces by nearly 56.69%, and adding repair behavior reduces by nearly 21.98%, indicating that the recycling profit of waste product disassembly will be significantly affected by process failure, and taking repair operation can alleviate this impact to a certain extent.

[0187] Table 7 Disassembly process planning results under different failure considerations

[0188]

[0189] In summary, the technical objectives of the embodiments of the present invention are as follows:

[0190] (1) Starting from the uncertainty of retired electromechanical products, considering the random failure behavior of the process in the disassembly process and the causal relationship of failure between the previous and next processes, a DDPR-RFR model including general random failure and failure repair is constructed, which provides a model basis for the disassembly process planning problem involving process failure.

[0191] (2) A BN-GHA method is proposed based on the DDPR-RFR model: the Bayesian network topology structure is learned based on the K2 algorithm, the parameters of the Bayesian network node state are estimated based on the Bayesian maximum a posteriori probability estimation theory, and the relevant result parameters are integrated into the genetic algorithm to obtain the optimal disassembly process route and improve the efficiency of resource recovery.

[0192] (3) The model established in this embodiment and the proposed method are applied to the power battery case. The calculation results show that adding a repair operation can reduce the impact of process failure on the disassembly target, which has a certain guiding significance for the disassembly process decision under the condition of process failure behavior.

[0193] It is worth noting that the method proposed in this embodiment requires the support of factors such as failure causes, failure degrees, repair operations and related knowledge, and assumes that each process has only a single parent process. In reality, there are often some disassembly processes with multiple parent processes. Therefore, the next step can focus on the failure problem of multiple parent processes. At the same time, consider how to accurately obtain information related to process failure to provide a scientific and reasonable solution for disassembly process decision-making.

[0194] See also Figure 7 , is a schematic diagram of the structure of a dismantling device for retired electromechanical products with random process failures provided by an embodiment of the present invention. The dismantling device for retired electromechanical products with random process failures comprises:

[0195] The analysis module 10 is used to analyze the failure causal relationship between each process in the disassembly process of retired electromechanical products by using a Bayesian network to determine the probability of failure state transfer;

[0196] An estimation module 11 is used to estimate the conditional probability of random failure of each disassembly process based on the failure sample data during the disassembly of retired electromechanical products;

[0197] A search module 12 is used to optimize the disassembly process route according to the failure condition probability of the process in combination with a genetic algorithm, and search for a global approximate optimal solution for disassembly of retired electromechanical products;

[0198] The repair module 13 is used to take corresponding repair operations for the process where failure has occurred according to the designed failure repair strategy, so as to ensure the continuity of the disassembly sequence during the disassembly of retired electromechanical products.

[0199] The embodiment of the present invention aims at the process failure behavior that often occurs during the disassembly of retired electromechanical products. From the perspective of internal parts failure and the execution level of the disassembly process, the random failure behavior mechanism of the disassembly process is systematically studied, including the accurate analysis of the failure causal relationship mechanism between multiple processes and the precise estimation of the probability of random failure (condition) of a single process, and the establishment of a disassembly process route decision model under the conditions of random failure of the process and its repair. A Bayesian network-genetic hybrid algorithm is proposed, and the failure causal relationship between the disassembly processes and the random failure conditional probability of each disassembly process are respectively obtained through the Bayesian network. On this basis, the genetic algorithm is used to efficiently solve the global approximate optimal disassembly process route. From the above analysis, it can be seen that the embodiment of the present invention focuses on the random failure problem of the disassembly process of retired electromechanical products. From the perspective of internal parts failure and the execution level of the disassembly process, the failure behavior mechanism of the disassembly process is deeply explored, and the global optimal disassembly process route is decided based on the failure causal relationship between the disassembly processes.

[0200] As an improvement of the above solution, the estimation module is specifically used for:

[0201] Collect and organize failure sample data during disassembly;

[0202] Based on the failure sample data and applying the Bayesian network structure learning algorithm, the network topology of the failure causal relationship between processes is constructed;

[0203] The maximum a posteriori probability estimation method is used to calculate the node state parameters of the Bayesian network and obtain the random conditional failure probability of each disassembly process.

[0204] As an improvement of the above solution, the failure repair strategy includes:

[0205] Define a set of repair operations, each corresponding to a different failure state;

[0206] Select the most appropriate repair action based on the failure type and severity, which may include replacing tools, recalibrating equipment, or increasing manual supervision.

[0207] After performing the repair operation, re-evaluate the status of the repaired process to ensure that it meets the conditions for continued disassembly.

[0208] As an improvement of the above scheme, the genetic algorithm is combined to optimize the disassembly process route according to the failure condition probability of the process, and the global approximate optimal solution for the disassembly of retired electromechanical products is searched, including:

[0209] The genetic algorithm initializes the population, and each individual represents a disassembly sequence. The sequence takes into account the dependencies and failure probabilities of the processes.

[0210] The reliability weight of the disassembly sequence is adjusted by combining the recycling profit, cost and repair fee through the fitness function and using the failure probability estimated by the Bayesian network;

[0211] The priority crossover operation is adopted to maintain the logical relationship of the process, combined with the single-point mutation operation to ensure the feasibility and efficiency of the sequence;

[0212] Through multiple iterations, the genetic algorithm is used to search for the global optimal solution and find the best disassembly process route under failure and repair conditions.

[0213] Another embodiment of the present invention provides a dismantling device for retired electromechanical products with random process failures. The dismantling device for retired electromechanical products with random process failures of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a dismantling program for retired electromechanical products with random process failures. When the processor executes the computer program, the steps in the above-mentioned embodiments of the method for dismantling retired electromechanical products with random process failures are implemented. Alternatively, when the processor executes the computer program, the functions of each module / unit in the above-mentioned device embodiments are implemented.

[0214] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the dismantling equipment for retired electromechanical products with random failure in the process.

[0215] The dismantling equipment for retired electromechanical products with random process failures may be computing devices such as desktop computers, notebooks, PDAs, and cloud servers. The dismantling equipment for retired electromechanical products with random process failures may include, but are not limited to, processors and memories. Those skilled in the art will appreciate that the schematic diagram is merely an example of the dismantling equipment for retired electromechanical products with random process failures, and does not constitute a limitation on the dismantling equipment for retired electromechanical products with random process failures. It may include more or fewer components than shown in the diagram, or a combination of certain components, or different components. For example, the dismantling equipment for retired electromechanical products with random process failures may also include input and output devices, network access devices, buses, and the like.

[0216] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the dismantling equipment for retired electromechanical products with random failures in the process, and uses various interfaces and lines to connect various parts of the dismantling equipment for retired electromechanical products with random failures in the entire process.

[0217] The memory can be used to store the computer program and / or module, and the processor realizes various functions of the dismantling equipment of retired electromechanical products with random failure in the process by running or executing the computer program and / or module stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0218] Wherein, if the module / unit integrated in the disassembly equipment of retired electromechanical products with random failure in the process is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0219] It should be noted that the device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the accompanying drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art may understand and implement it without paying any creative effort.

[0220] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for dismantling retired electromechanical products with random process failures, characterized in that: include: The Bayesian network is used to analyze the causal relationship between failures in the disassembly process of retired electromechanical products and determine the probability of failure state transfer. Based on the failure sample data during the disassembly of retired electromechanical products, the conditional probability of random failure in each disassembly process is estimated; Combined with genetic algorithm, the disassembly process route is optimized according to the failure condition probability of the process, and the global approximate optimal solution for the disassembly of retired electromechanical products is searched; According to the designed failure repair strategy, take corresponding repair operations for the failed process to ensure the continuity of the disassembly sequence during the disassembly of retired electromechanical products; The method of estimating the random failure condition probability of each disassembly process based on the failure sample data during the disassembly of retired electromechanical products includes: Collect and organize failure sample data during disassembly; Based on the failure sample data and applying the Bayesian network structure learning algorithm, the network topology of the failure causal relationship between processes is constructed; The maximum a posteriori probability estimation method is used to calculate the state parameters of the Bayesian network nodes and obtain the random conditional failure probability of each disassembly process. The combined genetic algorithm optimizes the disassembly process route according to the failure condition probability of the process, and searches for the global approximate optimal solution for the disassembly of retired electromechanical products, including: The genetic algorithm initializes the population, and each individual represents a disassembly sequence. The sequence takes into account the dependencies and failure probabilities of the processes. The reliability weight of the disassembly sequence is adjusted by combining the recycling profit, cost and repair fee through the fitness function and using the failure probability estimated by the Bayesian network; The priority crossover operation is adopted to maintain the logical relationship of the process, combined with the single-point mutation operation to ensure the feasibility and efficiency of the sequence; Through multiple iterations, the genetic algorithm is used to search for the global optimal solution and find the best disassembly process route under failure and repair conditions.

2. The method for dismantling retired electromechanical products with random process failures as claimed in claim 1, characterized in that: The failure recovery strategy includes: Define a set of repair operations, each corresponding to a different failure state; Select the most appropriate repair action based on the failure type and severity, which may include replacing tools, recalibrating equipment, or increasing manual supervision. After performing the repair operation, re-evaluate the status of the repaired process to ensure that it meets the conditions for continued disassembly.

3. A dismantling device for retired electromechanical products with random process failure, characterized in that: include: The analysis module is used to analyze the causal relationship between failures in various processes during the disassembly of retired electromechanical products using a Bayesian network to determine the probability of failure state transfer; An estimation module is used to estimate the conditional probability of random failure in each disassembly process based on the failure sample data during the disassembly of retired electromechanical products; A search module is used to combine genetic algorithms to optimize the disassembly process route according to the failure condition probability of the process and search for the global approximate optimal solution for the disassembly of retired electromechanical products; The repair module is used to take corresponding repair operations for the failed processes according to the designed failure repair strategy to ensure the continuity of the disassembly sequence during the disassembly of retired electromechanical products; Wherein, the estimation module is specifically used for: Collect and organize failure sample data during disassembly; Based on the failure sample data and applying the Bayesian network structure learning algorithm, the network topology of the failure causal relationship between processes is constructed; The maximum a posteriori probability estimation method is used to calculate the state parameters of the Bayesian network nodes and obtain the random conditional failure probability of each disassembly process. The search module is specifically used for: The genetic algorithm initializes the population, and each individual represents a disassembly sequence. The sequence takes into account the dependencies and failure probabilities of the processes. The reliability weight of the disassembly sequence is adjusted by combining the recycling profit, cost and repair fee through the fitness function and using the failure probability estimated by the Bayesian network; The priority crossover operation is adopted to maintain the logical relationship of the process, combined with the single-point mutation operation to ensure the feasibility and efficiency of the sequence; Through multiple iterations, the genetic algorithm is used to search for the global optimal solution and find the best disassembly process route under failure and repair conditions.

4. The dismantling device for retired electromechanical products with random process failures according to claim 3, characterized in that: The failure recovery strategy includes: Define a set of repair operations, each corresponding to a different failure state; Select the most appropriate repair action based on the failure type and severity, which may include replacing tools, recalibrating equipment, or increasing manual supervision. After performing the repair operation, re-evaluate the status of the repaired process to ensure that it meets the conditions for continued disassembly.

5. A dismantling device for retired electromechanical products with random process failure, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and when the processor executes the computer program, it implements the method for dismantling retired electromechanical products with random process failures as described in any one of claims 1 to 2.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for dismantling retired electromechanical products with random process failures as described in any one of claims 1 to 2.

Images