Multi-level resource optimal configuration method for cluster manufacturing

Abstract

The invention relates to a cluster manufacturing-oriented multi-level resource optimization configuration method, which comprises the following steps of: decomposing responsible product manufacturing into a component-level task, a part-level task and a process-level task, and matching an optimal equipment resource for completing the process-level task and an optimal equipment resource combination for completing the part-level task by adopting a grey correlation analysis method; a Louvain community detection algorithm and a data envelope analysis method are adopted to simplify matching of part-level tasks, a product manufacturing service configuration process is simplified, data value-added reference is provided for further decision making of managers, multi-level cooperation and collaboration among manufacturing resources are finally established, resource sharing and manufacturing cooperation are achieved, and product quality is improved. Compared with the prior art, the method has the advantages that the requirement for computing resources is lowered, high application expansion is achieved, different types of manufacturing resources and task requirements can be involved, and multiple complex scenes with changeable market requirements or unstable equipment resource attributes are met.

Description

Technical Field

[0001] The present invention relates to the technical field of resource scheduling optimization in manufacturing industry, and in particular to a multi-level resource optimization configuration method for cluster manufacturing. Background Art

[0002] With the rapid development of the global economy, the industrial and manufacturing markets are booming, but there is also a new situation of increasingly intensified competition and increasingly urgent transformation. User needs are gradually changing from unified to personalized, diversified and customized, which puts higher requirements on the existing manufacturing industry, especially in the rapid and accurate collection of manufacturing information, dynamic response to resource demand, and resource service packaging and sharing. The transformation and upgrading are urgent. Although demand has promoted the realization and promotion of various advanced manufacturing models, the current research focus is mainly on the research of production service optimization models and methods for fixed levels or a few nodes in traditional manufacturing systems, and there is less research on the production relationship from the perspective of process level, part level, component level, product level and networking. Therefore, in order to realize manufacturing interconnection and industrial informatization, it is necessary to deeply explore the use constraints and collaborative linkage relationships within and between levels, study the socialized and networked interaction of multiple subjects and multiple clusters, and then achieve the purpose of intelligent resource optimization and self-organization configuration of multi-level manufacturing tasks under resource clustering.

[0003] The Chinese application number is: CN114860435A, and the patent name is: Big data job scheduling method based on reinforcement learning of task selection process, which involves a job scheduling method based on machine learning and deep learning; the method first vectorizes the jobs in the cluster, and then for the selected task node, the intelligent agent determines how many task instances to schedule; in the third step, for the selected task node, the resource matching module selects the type of task executor with the best resource matching, and then selects an idle task executor from the task executor type to prepare to execute the currently scheduled task instance; finally, the scheduled task instance is deployed and executed on the big data task execution platform; however, since this method uses deep learning and reinforcement learning technology, it has high performance requirements for computing resources (such as processors and memory), and is suitable for mid- and high-level control centers in the industrial sector. In industrial manufacturing edge nodes, edge manufacturing resources that are mainly processing and manufacturing businesses do not have high computing capabilities, and the accuracy of the results depends on specific scenarios, and the flexibility is insufficient, so the use of this method is limited. Summary of the invention

[0004] The technical problem to be solved by the present invention is how to solve the resource allocation problem of multi-level complex manufacturing tasks based on the mutual induction and interconnection of manufacturing resources. In order to overcome the defects of the above-mentioned prior art, the present invention provides a multi-level resource optimization configuration method for cluster manufacturing.

[0005] The present invention provides a multi-level resource optimization configuration method for cluster manufacturing, comprising:

[0006] Step 1. Decompose the product tasks that need to be processed into component-level tasks, part-level tasks, and process-level tasks step by step; also include the equipment resources used to complete various manufacturing processes;

[0007] Step 2. For each process-level task, based on the task requirements, a functional classification method is used to obtain multiple equipment resources that can complete the process-level task, forming a primary manufacturing cluster; process-level evaluation indicators are formulated, and the gray correlation analysis method is used to evaluate and sort the equipment resources in the primary manufacturing cluster to select the optimal equipment resources;

[0008] Step 3. For each part-level task, according to the process manufacturing sequence of the part-level task, determine whether there is a resource cluster with the same process manufacturing sequence in the database. If yes, proceed to step 6; if not, proceed to step 4;

[0009] Step 4. According to the process manufacturing sequence of the part-level task, obtain the equipment resource combination that can complete each process in the part-level task to form a basic resource cluster; establish an evaluation matrix for the basic resource cluster in combination with the process-level evaluation index, use the grey correlation analysis method to evaluate each equipment resource combination in the basic resource cluster, select the optimal equipment resource combination, and save the process manufacturing sequence corresponding to the part-level task and the corresponding basic resource cluster in the database;

[0010] Step 5. Call N basic resource clusters corresponding to the same process manufacturing sequence from the database to merge them to form an equipment resource network, and use the Louvain community detection algorithm to divide and sort the equipment resource network to obtain resource clusters composed of multiple communities, which are saved in the database;

[0011] Step 6. Call the resource cluster from the database to process the part-level task of the same process manufacturing sequence, formulate the part-level task evaluation index, use the data envelopment analysis method to evaluate each community in the resource cluster, and select the best community to undertake the part-level task of the same process manufacturing sequence;

[0012] Step 7. For multiple part-level tasks with different process manufacturing sequences, repeat steps 4 to 5 to obtain multiple resource clusters with different process manufacturing sequences, and return to step 6 to obtain the optimal community corresponding to each part-level task;

[0013] Step 8. For multiple component-level tasks, decompose each component-level task into multiple part-level tasks, and then return to step 7.

[0014] Compared with the prior art, the present application has the following advantages: the responsible product manufacturing is decomposed into component-level tasks, part-level tasks, and process-level tasks, and the grey correlation analysis method is used to match the optimal equipment resources for completing process-level tasks and the optimal equipment resource combination for completing part-level tasks; at the same time, the Louvain community detection algorithm and the data envelopment analysis method are used to simplify the equipment resource combination matching of part-level tasks with the same process manufacturing sequence, and by simplifying the product manufacturing service configuration process, data value-added reference is provided for managers to make further decisions, and finally multi-level cooperation and collaboration between manufacturing resources are established to achieve resource sharing and manufacturing cooperation. Compared with the neural network model based on machine learning, deep learning, and reinforcement learning, the resource optimization configuration steps adopted by the present invention are consistent with the actual task allocation logic of the manufacturing enterprise, and the output results of each execution step have a clear physical meaning. The decision-making method is highly reliable, explainable, and controllable, which reduces the requirements for computing resources, and has strong application expansion, and can intervene in different types of manufacturing resources and task requirements to meet multiple complex scenarios with changing market demand or unstable equipment resource attributes.

[0015] In a possible implementation, the process-level evaluation index formulated in step 2 involves five evaluation factors: time, cost, reliability, satisfaction, and similarity, specifically including:

[0016]

[0017]

[0018]

[0019]

[0020]

[0021] In the formula, time represents time, cost represents cost, real represents reliability, sf represents satisfaction, sl represents similarity; n represents the number of device resources;

[0022] The calculation steps for evaluating the equipment resources in the primary manufacturing cluster using the grey correlation analysis method in step 2 are as follows:

[0023] Step 201. Combine the process-level evaluation indicators to establish an evaluation matrix for equipment resources:

[0024]

[0025] In the formula, represents the evaluation matrix of the i-th device resource; represents the mth evaluation factor of the i-th device resource;

[0026] Step 202. Extract the optimal evaluation index from the evaluation matrix of equipment resources as the evaluation benchmark. Specifically, for the time and cost evaluation factors, take the minimum value in the equipment resource evaluation matrix as the optimal index; for the reliability, satisfaction and similarity evaluation factors, take the maximum value in the evaluation matrix as the optimal index:

[0027]

[0028] Step 203. Evaluation index b for each device resource i (k) Perform the tempering treatment without any amount of heat treatment, the treatment formula is:

[0029]

[0030] Step 204. Evaluation index a of each equipment resource after the infinite tempering process i (k) Calculate the maximum difference Δ(max) and the minimum difference Δ(min), the calculation formulas are:

[0031] Δb0(k)=|b0(k)-b i (k)|,(i=0,1,2,…n,k=1,2,3,…m)

[0032] Δ(min)=min|b0(k)-b i (k)|,1≤i≤n,1≤k≤m

[0033] Δ(max)=max|b0(k)-b i (k)|,1≤i≤n,1≤k≤m

[0034] Step 205. Calculate the correlation coefficient of each device resource evaluation factor combination based on the maximum difference Δ(max) and the minimum difference Δ(min) The calculation formula is:

[0035]

[0036] Where ρ is the adjustment coefficient;

[0037] Step 206. According to the correlation coefficient Calculate the associated parameters of the evaluation factors of each equipment resource, and calculate the associated data of the evaluation factor combination and the evaluation benchmark in all equipment resources. The calculation formula is:

[0038]

[0039] According to the different weights of the evaluation factors of each equipment resource, the AHP method is used to calculate the weighted average of the correlation coefficient to obtain the correlation data. The calculation formula is:

[0040]

[0041] Where W k Indicates the weight of the evaluation factor of equipment resources;

[0042] Step 207: By comparing the correlation data of the evaluation factor combinations in each equipment resource, the optimal equipment resource is screened to complete the process-level task.

[0043] In a possible implementation, in step 4, an evaluation matrix of a basic resource cluster is established in combination with process-level evaluation indicators, and a grey correlation analysis method is used to evaluate and rank each equipment resource combination in the basic resource cluster to select the optimal equipment resource combination, specifically including:

[0044] Step 401. Combining the process manufacturing sequence and the process level evaluation index, the evaluation matrix expression of the basic resource cluster is:

[0045]

[0046] In the formula, represents the evaluation matrix of the i-th equipment resource combination; represents the evaluation index of the mth device resource in the i-th device resource combination;

[0047] Step 402: Extract the optimal evaluation index from the evaluation matrix of the basic resource cluster as the evaluation benchmark:

[0048]

[0049] Step 403. Evaluation index a for each device resource i (k) Perform the tempering treatment without any amount of heat treatment, the treatment formula is:

[0050]

[0051] Step 404. Evaluation index a of each equipment resource after the infinite tempering process i (k) Calculate the maximum difference Δ(max) and the minimum difference Δ(min), the calculation formulas are:

[0052] Δa0(k)=|a0(k)-a i (k)|,(i=0,1,2,…n,k=1,2,3,…m)

[0053] Δ(min)=min|a0(k)-a i (k)|,1≤i≤n,1≤k≤m

[0054] Δ(max)=max|a0(k)-a i(k),1≤i≤n,1≤k≤m

[0055] Step 405. Calculate the correlation coefficient of each device resource combination based on the maximum difference Δ(max) and the minimum difference Δ(min) The calculation formula is:

[0056]

[0057] Where ρ is the adjustment coefficient;

[0058] Step 406. According to the correlation coefficient Calculate the associated parameters of each equipment resource combination, and calculate the associated data of all equipment resource combinations and evaluation benchmarks. The calculation formula is:

[0059]

[0060] According to the different weights of the equipment resource combination according to the evaluation index of each equipment resource, the AHP method is used to analyze the relevant

[0061] The weighted average of the numbers is used to obtain the correlation data. The calculation formula is:

[0062]

[0063] Where W k Indicates the weight of the evaluation index of equipment resources to the equipment resource combination;

[0064] Step 407: By comparing the correlation data of each equipment resource combination, the optimal equipment resource combination is screened to complete the part-level task.

[0065] In a possible implementation, the N basic resource clusters corresponding to the same process manufacturing sequence in step 5 are merged to form an equipment resource network, specifically including:

[0066] Step 501A. Number all device resources;

[0067] Step 502A. Connect the equipment resources in each equipment resource combination in the basic resource cluster according to the process manufacturing sequence;

[0068] Step 503A. Combine multiple device resource combinations in the basic resource cluster, combine device resources with the same number into one, and retain the connection with other device resources to form a basic resource network;

[0069] Step 504A. Continue to merge N basic resource networks, merge the device resources with the same number into one, and retain the connection with other device resources to form a device resource network;

[0070] In step 5, the Louvain community detection algorithm is used to divide and sort the multi-resource cooperation clusters to obtain resource clusters, which specifically include:

[0071] Step 501B. Each resource device in the device resource network is regarded as a network node. First, each independent network node is regarded as a community; and each community is numbered;

[0072] Step 502B. Calculate the modularity of each community:

[0073]

[0074]

[0075] In the formula, w xy represents the total number of links between any community x and community y, s x =∑ y w xy Represents the total number of connections between community x and other communities; m = 1 / 2∑ xy w xy Indicates the total number of all connections in the device resource network; xy Used to determine whether community x and community y are the same community. If so, δ xy =1, if not, δ xy =0;

[0076] Step 503B. Randomly select a community i, and find multiple adjacent communities j1, j2…j based on the lines connected to community i. u , calculate the relationship between community i and each adjacent community j1, j2…j u The modularity difference ΔP1, ΔP2…ΔP u , where the calculation formula for the modularity difference ΔP is:

[0077]

[0078] In the formula, s i,in Represents the total number of connections of all network nodes in community i; ∑tots i represents the total number of connections between network nodes in community i and other communities;

[0079] Step 504B. From ΔP1, ΔP2 ... ΔP u The maximum value of the modularity difference ΔP is selected max , and determine ΔP max Is it greater than or equal to 0? If so, set ΔP max Corresponding community kMerge with community i into one community, update the community number, and reset the counter S to zero; if not, execute counter S=S+1 to determine whether S is less than or equal to the preset threshold W. If so, return to step 503B; if not, output the resource cluster.

[0080] In a possible implementation, the part-level task evaluation index formulated in step 6 involves a service efficiency index and a service quality index, wherein the service efficiency index includes four evaluation factors: time CT, cost CC, scale GM, and number of projects, specifically including:

[0081]

[0082]

[0083]

[0084]

[0085] In the formula, ct i represents the time it takes for the i-th community in the resource cluster to complete the part-level task; cc i represents the cost of completing the part-level task in the i-th community within the resource cluster, l represents the number of processes included in the process route of the part-level task, and gm i,g represents the number of resources in the i-th community within the resource cluster that can complete process g; n i,g represents all resources in the i-th community in the resource cluster that can complete process g, and the total number of tasks they have completed in history; m represents the total number of communities in the resource cluster;

[0086] The service quality indicators include four evaluation factors: completion degree P, satisfaction degree SL, time Time, and cost Cost, which specifically include:

[0087]

[0088]

[0089]

[0090]

[0091] In the formula, s i represents the completion degree of the part-level task of the i-th community in the resource cluster; sl i represents the satisfaction of the i-th community in the resource cluster in completing the part-level task; t i represents the time it takes for the ith community in the resource cluster to complete the part-level task; c i It represents the cost of completing the part-level task for the i-th community in the resource cluster.

[0092] In a possible implementation, step 6 uses a data envelopment analysis method to evaluate each community in the resource cluster and select the best community, which specifically includes:

[0093] Step 601. n given part-level tasks have n evaluation matrices in the manufacturing cluster, where the i-th evaluation matrix is ​​represented by EUV i =[a i ,b i ] T , i = 1, 2, ... n, each evaluation matrix has m input index vectors and s output index vectors. The m input index vectors are expressed as: a i =[a 1i ,a 2i …a mi ] T , the s-term output indicator vector is expressed as: b i =[b 1i ,b 2i …b si ] T , evaluation matrix EUV i Unit efficiency evaluation value U i The calculation formula is:

[0094]

[0095] Where wv i =[wv1,wv2,…wv m ],wu i =[wu1,wu2,…wu s ],wv i is a i The weight matrix of i Yes i Weighting matrix;

[0096] Step 602: Using the C2R model in the data envelopment analysis method to maximize the unit efficiency evaluation value U i To solve the goal, the C2R model is:

[0097]

[0098]

[0099] Solve for the optimal weight wu i and wv i , so that the unit efficiency evaluation value U i Highest;

[0100] Step 603. For a given part-level task, the unit efficiency evaluation values ​​U of multiple communities in the resource cluster are i The maximum value 1 is possible. In order to further judge the advantages and disadvantages of each community, the contradictory mutual evaluation method is used to expand the difference between community i (i = 1, 2, ... n) and other communities k (k = 1, 2, ... i-1, i+1, ... n) and minimize the unit efficiency evaluation value U of other communities k. k The optimization model is:

[0101]

[0102]

[0103] Solve for the optimal weight wu k and wv k , so that the unit efficiency evaluation value U k lowest;

[0104] Step 604. Construct community mutual evaluation matrix E:

[0105]

[0106] The community mutual evaluation matrix E is an n-order square matrix, including diagonal elements and off-diagonal elements. The diagonal elements are the highest unit efficiency evaluation values ​​U obtained in step 602. i (i=1, 2, ... n), the non-diagonal elements are the minimum unit efficiency evaluation values ​​U obtained in step 603 k The reciprocal of (k=1,2,…i-1,i+1,…n), that is, U' k =1 / U k ;

[0107] Step 605. Calculate the average value of each row element in the community mutual evaluation matrix E; the average value of the e-th row element U e The calculation formula is:

[0108]

[0109] For a given part-level task, the unit efficiency evaluation value U i The larger the value is, the higher the unit efficiency of the corresponding community i is, and the lower the unit efficiency of other communities k is. Therefore, community i is selected as the left and right community to undertake the given part-level task. BRIEF DESCRIPTION OF THE DRAWINGS

[0110] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0111] First, those skilled in the art should understand that these implementations are only used to explain the technical principles of the embodiments of the present application, and are not intended to limit the protection scope of the embodiments of the present application. Those skilled in the art can make adjustments to them as needed to adapt to specific application scenarios.

[0112] In the description of the embodiments of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "connected" and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.

[0113] In the embodiments of the present application, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may mean that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, a first feature being "above", "above" or "above" a second feature may mean that the first feature is directly above or obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" or "below" a second feature may mean that the first feature is directly below or obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0114] The present application is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0115] See also Figure 1 As shown, the embodiment of the present application discloses a multi-level resource optimization configuration method for cluster manufacturing. The application scenario of this specific embodiment is a product task processed and produced in a workshop environment, specifically including:

[0116] Step 1. Decompose the product tasks that need to be processed and produced step by step into component-level tasks, part-level tasks, and process-level tasks; also include the equipment resources used to complete various manufacturing processes.

[0117] Step 2. For each process-level task, based on the task requirements, a functional classification method is used to obtain multiple equipment resources that can complete the process-level task to form a primary manufacturing cluster; process-level evaluation indicators are formulated, and the gray correlation analysis method is used to evaluate and sort the equipment resources in the primary manufacturing cluster to select the optimal equipment resources; among them, the formulated process-level evaluation indicators involve five evaluation factors: time, cost, reliability, satisfaction, and similarity, specifically including:

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] In the formula, time represents time, cost represents cost, real represents reliability, sf represents satisfaction, sl represents similarity; n represents the number of device resources;

[0124] The calculation steps for evaluating the equipment resources in the primary manufacturing cluster using the grey correlation analysis method are as follows:

[0125] Step 201. Combine the process-level evaluation indicators to establish an evaluation matrix for equipment resources:

[0126]

[0127] In the formula, represents the evaluation matrix of the i-th device resource; represents the mth evaluation factor of the i-th device resource;

[0128] Step 202. Extract the optimal evaluation index from the evaluation matrix of equipment resources as the evaluation benchmark. The optimal evaluation index is conducive to the high efficiency and high quality completion of the production process. Specifically, for the time and cost evaluation factors, the minimum value in the equipment resource evaluation matrix is ​​taken as the optimal index; for the reliability, satisfaction and similarity evaluation factors, the maximum value in the evaluation matrix is ​​taken as the optimal index:

[0129]

[0130] Step 203. Evaluation index b for each device resource i (k) Perform the tempering treatment without any amount of heat treatment, the treatment formula is:

[0131]

[0132] Step 204. Evaluation index a of each equipment resource after the infinite tempering process i (k) Calculate the maximum difference Δ(max) and the minimum difference Δ(min), the calculation formulas are:

[0133] Δb0(k)=|b0(k)-b i (k)|,(i=0,1,2,…n,k=1,2,3,…m)

[0134] Δ(min)=min|b0(k)-b i(k)|,1≤i≤n,1≤k≤m

[0135] Δ(max)=max|b0(k)-b i (k)|,1≤i≤n,1≤k≤m

[0136] Step 205. Calculate the correlation coefficient of each device resource evaluation factor combination based on the maximum difference Δ(max) and the minimum difference Δ(min) The calculation formula is:

[0137]

[0138] Where ρ is the adjustment coefficient;

[0139] Step 206. According to the correlation coefficient Calculate the associated parameters of the evaluation factors of each equipment resource, and calculate the associated data of the evaluation factor combination and the evaluation benchmark in all equipment resources. The calculation formula is:

[0140]

[0141] According to the different weights of the evaluation factors of each equipment resource, the AHP method is used to calculate the weighted average of the correlation coefficient to obtain the correlation data. The calculation formula is:

[0142]

[0143] Where W k Indicates the weight of the evaluation factor of equipment resources;

[0144] Step 207: By comparing the correlation data of the evaluation factor combinations in each equipment resource, the optimal equipment resource is screened to complete the process-level task.

[0145] Step 3. For each part-level task, according to the process manufacturing sequence of the part-level task, determine whether there is a resource cluster with the same process manufacturing sequence in the database. If yes, proceed to step 6; if not, proceed to step 4; Step 4. According to the process manufacturing sequence of the part-level task, obtain the equipment resource combination that can complete each process in the part-level task to form a basic resource cluster; establish an evaluation matrix for the basic resource cluster in combination with the process-level evaluation index, use the grey correlation analysis method to evaluate each equipment resource combination in the basic resource cluster, select the optimal equipment resource combination, and save the process manufacturing sequence corresponding to the part-level task and the corresponding basic resource cluster to the database; wherein, using the grey correlation analysis method to evaluate each equipment resource combination in the basic resource cluster specifically includes: Step 401. Combined with the process manufacturing sequence and the process-level evaluation index, the evaluation matrix expression of the basic resource cluster is:

[0146]

[0147] In the formula, represents the evaluation matrix of the i-th equipment resource combination; represents the evaluation index of the mth device resource in the i-th device resource combination;

[0148] Step 402. Extract the optimal evaluation index from the evaluation matrix of the basic resource cluster as the evaluation benchmark. In this specific embodiment, the optimal index is conducive to the high efficiency and high quality completion of the production process. Specifically, for indicators such as time, cost, and energy consumption, the minimum value in the evaluation matrix of the basic resource cluster should be taken as the optimal evaluation index; for indicators such as reliability and satisfaction, the maximum value in the evaluation matrix should be taken as the optimal evaluation index:

[0149]

[0150] Step 403. Evaluation index a for each device resource i (k) Perform the tempering treatment without any amount of heat treatment, the treatment formula is:

[0151]

[0152] Step 404. Evaluation index a of each equipment resource after the infinite tempering process i (k) Calculate the maximum difference Δ(max) and the minimum difference Δ(min), the calculation formulas are:

[0153] Δa0(k)=|a0(k)-a i (k)|,(i=0,1,2,…n,k=1,2,3,…m)

[0154] Δ(min)=min|a0(k)-a i (k)|,1≤i≤n,1≤k≤m

[0155] Δ(max)=max|a0(k)-a i (k)|,1≤i≤n,1≤k≤m

[0156] Step 405. Calculate the correlation coefficient of each device resource combination based on the maximum difference Δ(max) and the minimum difference Δ(min0). The calculation formula is:

[0157]

[0158] Where ρ is the adjustment coefficient;

[0159] Step 406. According to the correlation coefficient Calculate the associated parameters of each equipment resource combination, and calculate the associated data of all equipment resource combinations and evaluation benchmarks. The calculation formula is:

[0160]

[0161] According to the different weights of the equipment resource combination given by the evaluation index of each equipment resource, the AHP method is used to calculate the weighted average of the correlation coefficient to obtain the correlation data. The calculation formula is:

[0162]

[0163] Where W k Indicates the weight of the evaluation index of equipment resources to the equipment resource combination;

[0164] Step 407: By comparing the correlation data of each equipment resource combination, the optimal equipment resource combination is screened to complete the part-level task.

[0165] Step 5. Call N basic resource clusters corresponding to the same process manufacturing sequence from the database to merge to form an equipment resource network, and use the Louvain community detection algorithm to divide and sort the equipment resource network to obtain resource clusters consisting of multiple communities, which are saved in the database; wherein, the N basic resource clusters corresponding to the same process manufacturing sequence are merged to form an equipment resource network, specifically including:

[0166] Step 501A. Number all device resources;

[0167] Step 502A. Connect the equipment resources in each equipment resource combination in the basic resource cluster according to the process manufacturing sequence;

[0168] Step 503A. Combine multiple device resource combinations in the basic resource cluster, combine device resources with the same number into one, and retain the connection with other device resources to form a basic resource network;

[0169] Step 504A. Continue to merge N basic resource networks, merge the device resources with the same number into one, and retain the connection with other device resources to form a device resource network;

[0170] The Louvain community detection algorithm is used to divide and sort out the multi-resource cooperation clusters to obtain resource clusters composed of multiple communities, including:

[0171] Step 501B. Each resource device in the device resource network is regarded as a network node. First, each independent network node is regarded as a community; and each community is numbered;

[0172] Step 502B. Calculate the modularity of each community:

[0173]

[0174]

[0175] In the formula, w xy represents the total number of links between any community x and community y, s x =∑ y w xy Represents the total number of connections between community X and other communities; m = 1 / 2∑ xy w xy表 Indicates the total number of all connections in the device resource network; xy Used to determine whether community x and community y are the same community. If so, δ xy =1, if not, δ xy =0;

[0176] Step 503B. Randomly select a community i, and find multiple adjacent communities j1, j2…j based on the lines connected to community i. u , calculate the relationship between community i and each adjacent community j1, j2…j u The modularity difference ΔP1, ΔP2…ΔP u , where the calculation formula for the modularity difference ΔP is:

[0177]

[0178] In the formula, s i,in Represents the total number of connections of all network nodes in community i; ∑tots i represents the total number of connections between network nodes in community i and other communities;

[0179] Step 504B. From ΔP1, ΔP2 ... ΔP u The maximum value of the modularity difference ΔP is selected max , and determine ΔP max Is it greater than or equal to 0? If so, set ΔP max Corresponding community k Merge with community i into one community, update the community number, and reset the counter S to zero; if not, execute counter S=S+1 to determine whether S is less than or equal to the preset threshold W. If so, return to step 503B; if not, output the resource cluster.

[0180] Step 6. Call the resource cluster from the database to process the part-level task of the same process manufacturing sequence, formulate the part-level task evaluation index, use the data envelopment analysis method to evaluate each community in the resource cluster, and select the best community to undertake the part-level task of the same process manufacturing sequence; wherein, formulating the part-level task evaluation index involves service efficiency index and service quality index, wherein the service efficiency index includes four evaluation factors: time CT, cost CC, scale GM and number of projects, specifically including:

[0181]

[0182]

[0183]

[0184]

[0185] In the formula, ct i represents the time it takes for the i-th community in the resource cluster to complete the part-level task; cc i represents the cost of completing the part-level task in the i-th community within the resource cluster, l represents the number of processes included in the process route of the part-level task, and gm i,g represents the number of resources in the i-th community within the resource cluster that can complete process g; n i,g represents all resources in the i-th community in the resource cluster that can complete process g, and the total number of tasks they have completed in history; m represents the total number of communities in the resource cluster;

[0186] The service quality indicators include four evaluation factors: completion degree P, satisfaction degree SL, time Time, and cost Cost, which specifically include:

[0187]

[0188]

[0189]

[0190]

[0191] In the formula, s i represents the completion degree of the part-level task of the i-th community in the resource cluster; sl i represents the satisfaction of the i-th community in the resource cluster in completing the part-level task; t i represents the time it takes for the ith community in the resource cluster to complete the part-level task; c i represents the cost of completing the part-level task for the i-th community in the resource cluster;

[0192] Data envelopment analysis is used to evaluate each community in the resource cluster and select the best community, including:

[0193] Step 601. n given part-level tasks have n evaluation matrices in the manufacturing cluster, where the i-th evaluation matrix is ​​represented by EUV i =[a i ,b i ] T , i = 1, 2, ... n, each evaluation matrix has m input index vectors and s output index vectors. The m input index vectors are expressed as: a i =[a 1i ,a 2i …a mi ] T , the s-term output indicator vector is expressed as: b i =[b 1i ,b 2i …b si ] T , evaluation matrix EUV i Unit efficiency evaluation value U i The calculation formula is:

[0194]

[0195] Where wv i =[wv1,wv2,…wv m ],wu i =[wu1,wu2,…wu s ],wv i is a i The weight matrix of i Yes i Weighting matrix;

[0196] Step 602: Using the C2R model in the data envelopment analysis method to maximize the unit efficiency evaluation value U i To solve the goal, the C2R model is:

[0197]

[0198]

[0199] Solve for the optimal weight wu i and wv i , so that the unit efficiency evaluation value U i Highest;

[0200] Step 603. For a given part-level task, the unit efficiency evaluation values ​​U of multiple communities in the resource cluster are iThe maximum value 1 is possible. In order to further judge the advantages and disadvantages of each community, the contradictory mutual evaluation method is used to expand the difference between community i (i = 1, 2, ... n) and other communities k (k = 1, 2, ... i-1, i+1, ... n) and minimize the unit efficiency evaluation value U of other communities k. k The optimization model is:

[0201]

[0202]

[0203] Solve for the optimal weight wu k and wv k , so that the unit efficiency evaluation value U k lowest;

[0204] Step 604. Construct community mutual evaluation matrix E:

[0205]

[0206] The community mutual evaluation matrix E is an n-order square matrix, including diagonal elements and off-diagonal elements. The diagonal elements are the highest unit efficiency evaluation values ​​U obtained in step 602. i (i=1, 2, ... n), the non-diagonal elements are the minimum unit efficiency evaluation values ​​U obtained in step 603 k The reciprocal of (k=1,2,…i-1,i+1,…n), that is, U' k =1 / U k ;

[0207] Step 605. Calculate the average value of each row element in the community mutual evaluation matrix E; the average value of the e-th row element U e The calculation formula is:

[0208]

[0209] For a given part-level task, the unit efficiency evaluation value U i The larger the value is, the higher the unit efficiency of the corresponding community i is, and the lower the unit efficiency of other communities k is. Therefore, community i is selected as the left and right community to undertake the given part-level task.

[0210] Step 7. For multiple part-level tasks with different process manufacturing sequences, repeat steps 4 to 5 to obtain multiple resource clusters with different process manufacturing sequences, and return to step 6 to obtain the optimal community corresponding to each part-level task;

[0211] Step 8. For multiple component-level tasks, decompose each component-level task into multiple part-level tasks, and then return to step 7.

[0212] The method of the present invention aims to reconstruct, model and dynamically match the dynamic manufacturing clusters and manufacturing needs on the basis of realizing the mutual induction and interconnection of manufacturing resources, so as to improve productivity and the efficiency of equipment resource utilization. The first step is to simply classify the corresponding equipment resources according to the inherent characteristics of a single process to form a primary manufacturing cluster, and formulate evaluation criteria and service sets according to the optimization configuration method, and obtain the selection of equipment resources to complete the process-level tasks according to the method of grey correlation analysis; the second step is to analyze the upstream and downstream cooperation and coordination constraints between resource nodes and specific processes from a network perspective based on the multi-process cooperation requirements of part-level tasks, and expand the primary manufacturing cluster for a single process in the first step to a dynamic multi-process cooperation resource cluster for the production of specific parts; the third step is to formulate evaluation indicators and standards as well as evaluation algorithms to comprehensively evaluate the part-level tasks formed in the previous step, and finally select the most reasonable manufacturing cluster to process the part-level tasks to complete the optimal configuration of resources.

[0213] Compared with the prior art, the advantages of the present invention are:

[0214] Applying intelligent sensing and industrial Internet technologies to the production and manufacturing process of parts and components to realize the automatic collection and communication of manufacturing information and management strategies, improving the shortcomings of traditional manual recording, which is inefficient and prone to errors;

[0215] By combing and modeling manufacturing resources and manufacturing relationships, the concept of cluster manufacturing is applied to workshop production, forming iterative multi-level resource cluster dynamic division and resource allocation standards for different tasks, which are suitable for various processing tasks such as process-level tasks, part-level tasks, and component-level tasks;

[0216] By simplifying the manufacturing service configuration process, it provides managers with data value-added references for further decision-making, and ultimately establishes multi-level cooperation and collaboration among manufacturing resources, realizes resource sharing and manufacturing cooperation, and changes the traditional production process of fixed resources and fixed tasks. It improves the self-organization and adaptability of the manufacturing system and the interpretability of the algorithm, thereby improving production efficiency and resource utilization.

[0217] In addition, in addition to using the Louvain community detection algorithm in step 5 of the present invention, other algorithms with similar effects can also be selected according to the characteristics of the manufacturing resource network. When the network scale is large, that is, the number of resources and tasks is large, the Louvain algorithm in the steps of the present invention is used preferentially to ensure a faster computational convergence speed under the premise of good community detection effect; when there are a large number of public resources in the resource network, that is, when the divided communities have many overlapping structures, the Clique Percolation algorithm can be used to identify overlapping community structures; when different manufacturing resources need to be assigned different weights in the process of matching task requirements, the Infomap algorithm can be used to make full use of weight information; when there may be large-scale communities and small communities with significant scale differences in resources, the Girvan-Newman algorithm can be used to improve the recognition effect of community structures of different scales. In short, the Louvain algorithm has strong versatility and computing performance, and other algorithms can be flexibly selected according to the needs of actual scenarios.

[0218] In the description of the embodiments of the present application, it should be noted that in the description of the present application, terms such as "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is only for the convenience of description, and does not indicate or imply that the device or component must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present application.

[0219] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "in the present embodiment", "specific example", or "some examples" etc. means that the specific features, mechanisms, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, mechanisms, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0220] The above is only a specific implementation of the present application, but the protection scope of the present application 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 present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A multi-level resource optimization configuration method for cluster manufacturing, characterized in that: include: Step 1. Decompose the product tasks that need to be processed into component-level tasks, part-level tasks, and process-level tasks step by step; also include the equipment resources used to complete various manufacturing processes; Step 2. For each process-level task, based on the task requirements, a functional classification method is used to obtain multiple equipment resources that can complete the process-level task, forming a primary manufacturing cluster; process-level evaluation indicators are formulated, and the gray correlation analysis method is used to evaluate and sort the equipment resources in the primary manufacturing cluster to select the optimal equipment resources; Step 3. For each part-level task, according to the process manufacturing sequence of the part-level task, determine whether there is a resource cluster with the same process manufacturing sequence in the database. If yes, proceed to step 6; if not, proceed to step 4; Step 4. According to the process manufacturing sequence of the part-level task, obtain the equipment resource combination that can complete each process in the part-level task to form a basic resource cluster; establish an evaluation matrix for the basic resource cluster in combination with the process-level evaluation index, use the grey correlation analysis method to evaluate each equipment resource combination in the basic resource cluster, select the optimal equipment resource combination, and save the process manufacturing sequence corresponding to the part-level task and the corresponding basic resource cluster in the database; Step 5. Call N basic resource clusters corresponding to the same process manufacturing sequence from the database to merge them to form an equipment resource network, and use the Louvain community detection algorithm to divide and sort the equipment resource network to obtain resource clusters composed of multiple communities, which are saved in the database; Step 6. Call the resource cluster from the database to process the part-level task of the same process manufacturing sequence, formulate the part-level task evaluation index, use the data envelopment analysis method to evaluate each community in the resource cluster, and select the best community to undertake the part-level task of the same process manufacturing sequence; Step 7. For multiple part-level tasks with different process manufacturing sequences, repeat steps 4 to 5 to obtain multiple resource clusters with different process manufacturing sequences, and return to step 6 to obtain the optimal community corresponding to each part-level task; Step 8. For multiple component-level tasks, decompose each component-level task into multiple part-level tasks, and then return to step 7.

2. The multi-level resource optimization configuration method for cluster manufacturing according to claim 1 is characterized in that: The process-level evaluation indicators formulated in step 2 involve five evaluation factors: time, cost, reliability, satisfaction, and similarity, which specifically include: In the formula, time represents time, cost represents cost, real represents reliability, sf represents satisfaction, sl represents similarity; n represents the number of device resources; The calculation steps for evaluating the equipment resources in the primary manufacturing cluster using the grey correlation analysis method in step 2 are as follows: Step 201. Combine the process-level evaluation indicators to establish an evaluation matrix for equipment resources: (B’1,B’2,B’3…B’ n ),B’ i =(b’ i (1),b’ i (2),b’ i (3)…b’ i (m)) In the formula, B' i represents the evaluation matrix of the i-th device resource; b' i (m) represents the mth evaluation factor of the i-th device resource; Step 202. Extract the optimal evaluation index from the evaluation matrix of equipment resources as the evaluation benchmark. Specifically, for the time and cost evaluation factors, take the minimum value in the equipment resource evaluation matrix as the optimal index; for the reliability, satisfaction and similarity evaluation factors, take the maximum value in the evaluation matrix as the optimal index: B'0=(b'0(1),b'0(2),b'0(3)...b'0(m)) Step 203. Evaluation index b for each device resource i (k) Perform the tempering treatment without any amount of heat treatment, the treatment formula is: Step 204. Evaluation index a of each equipment resource after the infinite tempering process i (k) Calculate the maximum difference Δ(max) and the minimum difference Δ(min), the calculation formulas are: Δb0(k)=|b0(k)-b i (k)|,(i=0,1,2,…n,k=1,2,3,…m) Δ(min)=min|b0(k)-b i (k),1≤i≤n,1≤k≤m Δ(max)=max|b0(k)-b i (k)|,1≤i≤n,1≤k≤m Step 205. Calculate the correlation coefficient of each device resource evaluation factor combination based on the maximum difference Δ(max) and the minimum difference Δ(min) The calculation formula is: Where ρ is the adjustment coefficient; Step 206. According to the correlation coefficient Calculate the associated parameters of the evaluation factors of each equipment resource, and calculate the associated data of the evaluation factor combination and the evaluation benchmark in all equipment resources. The calculation formula is: According to the different weights of the evaluation factors of each equipment resource, the AHP method is used to calculate the weighted average of the correlation coefficient to obtain the correlation data. The calculation formula is: Where W k Indicates the weight of the evaluation factor of equipment resources; Step 207: By comparing the correlation data of the evaluation factor combinations in each equipment resource, the optimal equipment resource is screened to complete the process-level task.

3. The multi-level resource optimization configuration method for cluster manufacturing according to claim 2 is characterized in that: In step 4, an evaluation matrix of a basic resource cluster is established in combination with process-level evaluation indicators, and a grey correlation analysis method is used to evaluate and rank each equipment resource combination in the basic resource cluster to select the optimal equipment resource combination, specifically including: Step 401. Combining the process manufacturing sequence and the process level evaluation index, the evaluation matrix expression of the basic resource cluster is: (A’1,A’2,A’3…A’ n ),A’ i =(a’ i (1),a’ i (2),a’ i (3)…a’ i (m)) In the formula, A' i represents the evaluation matrix of the i-th equipment resource combination; a' i (m) represents the evaluation index of the mth device resource in the i-th device resource combination; Step 402: Extract the optimal evaluation index from the evaluation matrix of the basic resource cluster as the evaluation benchmark: A'0=(a'0(1),a'0(2),a'0(3)...a'0(m)) Step 403. Evaluation index a for each device resource i (k) Perform the tempering treatment without any amount of heat treatment, the treatment formula is: Step 404. Evaluation index a of each equipment resource after the infinite tempering process i (k) Calculate the maximum difference Δ(max) and the minimum difference Δ(min), the calculation formulas are: Δa0(k)=|a0(k)-a i (k)|,(i=0,1,2,…n,k=1,2,3,…m) Δ(min)=min|a0(k)-a i (k)|,1≤i≤n,1≤k≤m Δ(max)=max|a0(k)-a i (k),1≤i≤n,1≤k≤m Step 405. Calculate the correlation coefficient of each device resource combination based on the maximum difference Δ(max) and the minimum difference Δ(min) The calculation formula is: Where ρ is the adjustment coefficient; Step 406. According to the correlation coefficient Calculate the associated parameters of each equipment resource combination, and calculate the associated data of all equipment resource combinations and evaluation benchmarks. The calculation formula is: According to the different weights of the equipment resource combination given by the evaluation index of each equipment resource, the AHP method is used to calculate the weighted average of the correlation coefficient to obtain the correlation data. The calculation formula is: Where W k Indicates the weight of the evaluation index of equipment resources to the equipment resource combination; Step 407: By comparing the correlation data of each equipment resource combination, the optimal equipment resource combination is screened to complete the part-level task.

4. The multi-level resource optimization configuration method for cluster manufacturing according to claim 1 is characterized in that: In step 5, the N basic resource clusters corresponding to the same process manufacturing sequence are merged to form an equipment resource network, specifically including: Step 501A. Number all device resources; Step 502A. Connect the equipment resources in each equipment resource combination in the basic resource cluster according to the process manufacturing sequence; Step 503A. Combine multiple device resource combinations in the basic resource cluster, combine device resources with the same number into one, and retain the connection with other device resources to form a basic resource network; Step 504A. Continue to merge N basic resource networks, merge the device resources with the same number into one, and retain the connection with other device resources to form a device resource network; In step 5, the Louvain community detection algorithm is used to divide and sort out the multi-resource cooperation clusters to obtain resource clusters, which specifically include: Step 501B. Each resource device in the device resource network is regarded as a network node. First, each independent network node is regarded as a community; and each community is numbered; Step 502B. Calculate the modularity of each community: In the formula, w xy represents the total number of links between any community x and community y, s x =∑ y w xy Represents the total number of connections between community x and other communities; m = 1 / 2∑ xy w xy Indicates the total number of all connections in the device resource network; xy Used to determine whether community x and community y are the same community. If so, δ xy =1, if not, δ xy =0; Step 503B. Randomly select a community i, and find multiple adjacent communities j1, j2…j based on the lines connected to community i. u , calculate the relationship between community i and each adjacent community j1, j2…j u The modularity difference ΔP1, ΔP2…ΔP u , where the calculation formula for the modularity difference ΔP is: In the formula, s i,in Represents the total number of connections of all network nodes in community i; ∑tots i represents the total number of connections between network nodes in community i and other communities; Step 504B. From ΔP1, ΔP2 ... ΔP u The maximum value of the modularity difference ΔP is selected max , and determine ΔP max Is it greater than or equal to 0? If so, set ΔP max Corresponding community k Merge with community i into one community, update the community number, and reset the counter S to zero; if not, execute counter S=S+1 to determine whether S is less than or equal to the preset threshold W. If so, return to step 503B; if not, output the resource cluster.

5. The multi-level resource optimization configuration method for cluster manufacturing according to claim 1 is characterized in that: The part-level task evaluation index formulated in step 6 involves service efficiency index and service quality index, wherein the service efficiency index includes four evaluation factors: time CT, cost CC, scale GM and number of projects, specifically including: In the formula, ct i represents the time it takes for the i-th community in the resource cluster to complete the part-level task; cc i represents the cost of completing the part-level task in the i-th community within the resource cluster, l represents the number of processes included in the process route of the part-level task, and gm i,g represents the number of resources in the i-th community within the resource cluster that can complete process g; n i,g represents all resources in the i-th community in the resource cluster that can complete process g, and the total number of tasks they have completed in history; m represents the total number of communities in the resource cluster; The service quality indicators include four evaluation factors: completion degree P, satisfaction degree SL, time Time, and cost Cost, which specifically include: In the formula, s i represents the completion degree of the part-level task of the i-th community in the resource cluster; sl i represents the satisfaction of the i-th community in the resource cluster in completing the part-level task; t i represents the time it takes for the ith community in the resource cluster to complete the part-level task; c i It represents the cost of completing the part-level task for the i-th community in the resource cluster.

6. The multi-level resource optimization configuration method for cluster manufacturing according to claim 5 is characterized in that: The step 6 uses a data envelopment analysis method to evaluate each community in the resource cluster and select the best community, which specifically includes: Step 601. n given part-level tasks have n evaluation matrices in the manufacturing cluster, where the i-th evaluation matrix is ​​represented by EUV i =[a i ,b i ] T , i = 1, 2, ... n, each evaluation matrix has m input index vectors and s output index vectors. The m input index vectors are expressed as: a i =[a 1i ,a 2i …a mi ] T , the s-term output indicator vector is expressed as: b i =[b 1i ,b 2i …b si ] T , evaluation matrix EUV i Unit efficiency evaluation value U i The calculation formula is: Where wv i =[wv1,wv2,…wv m ],wu i =[wu1,wu2,…wu s ],wv i is a i The weight matrix of i Yes i Weighting matrix; Step 602: Using the C2R model in the data envelopment analysis method to maximize the unit efficiency evaluation value U i To solve the goal, the C2R model is: Solve the optimal weight wu i and wv i , so that the unit efficiency evaluation value U i Highest; Step 603. For a given part-level task, the unit efficiency evaluation values ​​U of multiple communities in the resource cluster are i The maximum value 1 is possible. In order to further judge the advantages and disadvantages of each community, the contradictory mutual evaluation method is used to expand the difference between community i (i = 1, 2, ... n) and other communities k (k = 1, 2, ... i-1, i+1, ... n) and minimize the unit efficiency evaluation value U of other communities k. k The optimization model is: Solve the optimal weight wu k and wv k , so that the unit efficiency evaluation value U k lowest; Step 604. Construct community mutual evaluation matrix E: The community mutual evaluation matrix E is an n-order square matrix, including diagonal elements and off-diagonal elements. The diagonal elements are the highest unit efficiency evaluation values ​​U obtained in step 602. i (i=1, 2, ... n), the non-diagonal elements are the minimum unit efficiency evaluation values ​​U obtained in step 603 k The reciprocal of (k=1,2,…i-1,i+1,…n), that is, U' k =1 / U k ; Step 605. Calculate the average value of each row element in the community mutual evaluation matrix E; the average value of the e-th row element U e The calculation formula is: For a given part-level task, the unit efficiency evaluation value U i The larger the value is, the higher the unit efficiency of the corresponding community i is, and the lower the unit efficiency of other communities k is. Therefore, community i is selected as the left and right community to undertake the given part-level task.

Images