CN113190929B - Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning - Google Patents

Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning Download PDF

Info

Publication number
CN113190929B
CN113190929B CN202110560654.XA CN202110560654A CN113190929B CN 113190929 B CN113190929 B CN 113190929B CN 202110560654 A CN202110560654 A CN 202110560654A CN 113190929 B CN113190929 B CN 113190929B
Authority
CN
China
Prior art keywords
workpieces
workpiece
longest common
common subsequence
manufacturing system
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202110560654.XA
Other languages
Chinese (zh)
Other versions
CN113190929A (en
Inventor
黄思翰
王国新
阎艳
聂世琪
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Institute of Technology BIT
Original Assignee
Beijing Institute of Technology BIT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Institute of Technology BIT filed Critical Beijing Institute of Technology BIT
Priority to CN202110560654.XA priority Critical patent/CN113190929B/en
Publication of CN113190929A publication Critical patent/CN113190929A/en
Application granted granted Critical
Publication of CN113190929B publication Critical patent/CN113190929B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/17—Mechanical parametric or variational design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Abstract

The workpiece family building method of the delay reconstruction manufacturing system based on the machine learning determines the longest common subsequence of any two workpieces based on the process route of the two workpieces of the delay reconstruction manufacturing system; extracting the longest common subsequence of the relative position between the two workpieces according to the longest common subsequence and the position of the longest common subsequence on the process route of the two workpieces; calculating similarity coefficients of the two workpieces according to the length of the process route of the workpieces and the characteristic values of the longest common subsequence of the relative positions; calculating similarity coefficients among all workpieces of the delay reconstruction manufacturing system to obtain a similarity matrix of the workpieces of the delay reconstruction manufacturing system; and clustering convergence is carried out on the similarity matrix of the workpiece by utilizing a machine learning clustering algorithm, and the construction of the workpiece family of the delay reconstruction manufacturing system is completed. The efficiency and effectiveness of the construction of the workpiece family can be improved, and the operating efficiency of the delayed reconstruction manufacturing system can be further improved.

Description

Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning
Technical Field
The invention belongs to the technical field of advanced manufacturing, and particularly relates to a workpiece family construction method of a delayed reconstruction manufacturing system based on machine learning.
Background
In order to alleviate the impact of reconstruction on the conventional Reconstruction Manufacturing System (RMS), the concept of delaying reconstruction is proposed, taking into full account the complexity of reconstruction, and trying to create conditions to delay the reconstruction activity to the process in the back end of the manufacturing system. Combining the Delayed reconstruction with the RMS forms a Delayed reconstruction manufacturing system (D-RMS). D-RMS can reduce reconstruction difficulty and downtime loss due to reconstruction, thereby improving the practical capabilities of conventional RMS. D-RMS is a sub-class of RMS, inherits the characteristics of RMS, and needs to consider reconfigurability at the beginning of design and further consider delay reconfiguration characteristics. D-RMS is a manufacturing system for a specific workpiece family, and in practice, the first link is the construction of the workpiece family.
The essence of the building of a family of workpieces is the grouping technique. The RMS-oriented workpiece family requires more attention to reconstruction factors than conventional considerations. Due to the flexibility of the Similarity coefficient approach (Similarity coefficient) in combination with different production factors, it has been largely adopted in the work family construction studies of RMS. And obtaining a similarity relation between the workpieces based on a similarity coefficient method, and performing post-processing on the similarity coefficient method, namely clustering the workpieces into groups by adopting a clustering algorithm.
At present, galan and the like comprehensively consider production factors and reconstruction factors such as modularization, compatibility reusability, demand and the like when researching RMS workpiece family construction methods, and clustering is carried out by adopting an analytic hierarchy process. Ma Limei and the like take modularization and reusability with RMS characteristics, universality of conventional production factors and the like as evaluation indexes of similarity, obtain a similarity matrix taking each index as a row and column, and divide a workpiece family by adopting an improved analytic hierarchy process. Modularity is a measure of the individual modules of a product; reusability is the use of the same resources (e.g., modules) in different configurations and does not require any changes; compatibility refers to the ease of installation between modules. Zhang Xuehua and the like adopt a similarity coefficient method, also consider the weight relation among indexes of an evaluation similarity system, distribute index weight by using an analytic hierarchy process, and introduce a uniform link clustering algorithm to cluster workpieces into a group. Askin and Zhou put forward the concept of Longest Common Subsequence (LCS), and further consider the sequence among functions to solve the similarity coefficient among workpieces on the basis of only considering the processing function in the prior art, so that the construction result of the workpiece family is closer to the actual production. Abdi and Labib attempt to comprehensively analyze market demands and existing manufacturing system conditions and propose a concept of reconfiguration link to reduce the gap between the market and the manufacturing enterprise. In the subsequent research, based on RMS reconfigurability, the method of network analysis (Analytical network process) is used to construct a workpiece family in consideration of factors such as capacity demand and cost. Ashraf and Hasan propose a multi-product similarity RMS workpiece family construction method, and deeply analyze the characteristics of a modular product. Irani and Huang, in the layout research of the customized facilities, in consideration of the diversity of product lines, propose a concept of "layout module" to specify the special material flow mode of a single module, and introduce a systematic method to assist the implementation of layout design based on the workpiece family. Rajesh et al propose an RMS workpiece family construction method that mixes the similarity coefficient method. Goyal et al consider factors causing reduction in similarity between workpieces, including idle machine tools and workpiece detours, and complete construction of workpiece families by adopting a uniform link clustering algorithm. Wang et al comprehensively consider factors such as LCS, idle machine tools, workpiece detours and the like to study the RMS workpiece family construction method. Besides the similarity coefficient, the clustering algorithm is also a key link for building the RMS workpiece family, and most of the algorithms adopt a hierarchical clustering method, including two clustering ideas of top-down clustering and bottom-up clustering. The top-down clustering process is to regard all the workpieces as the same workpiece family and then decompose the workpieces step by step; the clustering process from bottom to top is reversed. In practice, fully linked Clustering (CLC), single Linked Clustering (SLC) and Average Linked Clustering (ALC) are the most common. Wherein ACL is the most common clustering algorithm in RMS workpiece family construction studies. In addition, gupta et al employ the K-means method as a clustering algorithm for the RMS workpiece family. Kashkoush and ElMaraghy propose a Consensus tree based Clustering Algorithm (Consensus tree-based Clustering Algorithm). Khanna and Kumar were clustered using the boundary Energy method (Bond Energy Algorithm).
Based on the existing workpiece family construction method and the combination of the D-RMS characteristics, the method has the following technical defects: ideally, the D-RMS can handle all families of workpieces built for conventional RMS's. However, the existing RMS workpiece family construction method does not consider the delayed reconstruction characteristic, so that the D-RMS operation efficiency is low; in addition, the grouping technology is a key technology for constructing a workpiece family, and a hierarchical clustering method is commonly used. At the present stage, requirements are personalized, diversified and uncertainty is aggravated, new requirements are provided for clustering effect and efficiency, and the existing method cannot meet the requirements.
Disclosure of Invention
The invention overcomes one of the defects of the prior art, provides the workpiece family construction method of the delayed reconstruction manufacturing system based on the machine learning, and can improve the efficiency and effectiveness of the construction of the workpiece family by combining the flexibility of the similarity coefficient method, the delayed reconstruction characteristic and the clustering algorithm of the machine learning, thereby improving the operating efficiency of the delayed reconstruction manufacturing system.
According to an aspect of the present disclosure, the present invention provides a workpiece family building method for a machine learning-based delayed reconstruction manufacturing system, the method comprising:
determining a longest common subsequence of any two workpieces of the delayed reconstruction manufacturing system based on process routes of the two workpieces;
extracting the longest common subsequence of the relative position between the two workpieces according to the longest common subsequence and the position of the longest common subsequence in the process route of the two workpieces;
calculating similarity coefficients of the two workpieces according to the length of the process route of the workpieces and the characteristic values of the longest common subsequence of the relative positions;
calculating similarity coefficients among all workpieces of the delay reconstruction manufacturing system to obtain a similarity matrix of the workpieces of the delay reconstruction manufacturing system;
and clustering convergence is carried out on the similar matrix of the workpiece by utilizing a machine learning clustering algorithm, and the construction of the workpiece family of the delay reconstruction manufacturing system is completed.
In one possible implementation, the reciprocal of the position of the process route element of the workpiece is a characteristic value of the longest common subsequence of relative positions.
In one possible implementation, the longest common subsequence of the two workpieces is a longest process route subsequence having the same processing function and processing order between process routes of the two workpieces;
the longest common subsequence of relative positions between the two workpieces is a position in the process route of the two workpieces where the longest common subsequence is added on the basis of the longest common subsequence of the two workpieces.
In one possible implementation, the calculating the similarity coefficient of the two workpieces according to the characteristic values of the longest common subsequence of the relative positions and the length of the process route of the workpieces includes:
calculating a feature value of the longest common subsequence of relative positions of the workpiece and the workpiece with the element position of the process route of the workpiece as the feature value;
determining a penalty factor of the similarity coefficient of the two workpieces according to the position difference of the elements of the longest common subsequence of the relative positions of the workpieces;
calculating the similarity coefficient of the two workpieces according to the length of the process route of the workpieces, the characteristic value of the longest common subsequence of the relative positions and the penalty factor of the similarity coefficient of the two workpieces.
In one possible implementation, the clustering algorithm is K-medoids.
The workpiece family building method of the delay reconstruction manufacturing system based on machine learning determines the longest common subsequence of any two workpieces based on the delay reconstruction manufacturing system through the process route of the two workpieces; extracting the longest common subsequence of the relative position between the two workpieces according to the longest common subsequence and the position of the longest common subsequence in the process route of the two workpieces; calculating similarity coefficients of the two workpieces according to the length of the process route of the workpieces and the characteristic value of the longest common subsequence of the relative positions; calculating similarity coefficients among all workpieces of the delay reconstruction manufacturing system to obtain a similarity matrix of the workpieces of the delay reconstruction manufacturing system; and clustering convergence is carried out on the similarity matrix of the workpiece by utilizing a machine learning clustering algorithm, and the construction of the workpiece family of the delay reconstruction manufacturing system is completed. The efficiency and effectiveness of the construction of the workpiece family can be improved, and the operating efficiency of the delayed reconstruction manufacturing system can be further improved.
Drawings
The accompanying drawings are included to provide a further understanding of the technology or prior art of the present application and are incorporated in and constitute a part of this specification. The drawings for illustrating the embodiments of the present application together with the embodiments of the present application serve to explain the technical solutions of the present application, but do not limit the technical solutions of the present application.
FIG. 1 illustrates a flow diagram of a method of workpiece family building for a machine learning-based delayed reconstruction manufacturing system according to an embodiment of the present disclosure;
FIG. 2 illustrates a longest process route sub-sequence diagram between process routes of a workpiece according to one embodiment of the present disclosure;
FIG. 3 illustrates a longest process route sub-sequence diagram of relative positions between process routes of a workpiece according to an embodiment of the present disclosure;
FIG. 4 is a schematic diagram illustrating element discontinuities of a longest process route sub-sequence of relative positions between process routes of a workpiece according to an embodiment of the present disclosure;
FIG. 5 is a schematic diagram illustrating a deviation in the position of an element of a longest process route sub-sequence of relative positions between process routes of a workpiece according to one embodiment of the present disclosure;
FIG. 6 shows a schematic diagram of a workpiece similarity matrix according to an embodiment of the present disclosure;
FIG. 7 is a schematic diagram illustrating workpiece family build results for a machine learning-based lag reconstruction manufacturing system according to an embodiment of the present disclosure.
Detailed Description
The following detailed description of the embodiments of the present invention will be provided with reference to the accompanying drawings and examples, so that how to apply the technical means to solve the technical problems and achieve the corresponding technical effects can be fully understood and implemented. The embodiments and the features of the embodiments can be combined without conflict, and the technical solutions formed are all within the scope of the present invention.
Additionally, the steps illustrated in the flow charts of the figures may be performed in a computer such as a set of computer-executable instructions. Also, while a logical order is shown in the flow diagrams, in some cases, the steps shown or described may be performed in an order different than here.
FIG. 1 shows a flow diagram of a method for building a family of workpieces for a machine learning based delayed reconstruction manufacturing system according to an embodiment of the present disclosure. As shown in fig. 1, the method may include:
step S1: determining a longest common subsequence of any two workpieces of the delayed reconstruction manufacturing system based on process routes of the two workpieces.
Fig. 2 illustrates a longest process route sub-sequence diagram between process routes of a workpiece according to an embodiment of the present disclosure.
Wherein, the Longest Common Subsequence (LCS) is the Longest process route subsequence having the same processing function and processing sequence among process routes. The longest common subsequence of two workpieces may be the longest process route subsequence having the same processing function and processing order between the process routes of the two workpieces.
The goal of workpiece family building is to divide workpieces with more identical processes into the same workpiece family. In consideration of the order between the processes, the LCS is usually found in the same process of two workpieces to calculate the similarity of the two workpieces. As shown in fig. 2, the process routes of workpiece a and workpiece c are OPa {1,2,3,4,5} and OPc {2,6,7,4,5}, respectively, and the longest process route subsequence LCSac of workpiece a and workpiece c is {2,4,5}.
Step S2: and extracting the longest common subsequence of the relative position between the two workpieces according to the longest common subsequence and the position of the longest common subsequence in the process route of the two workpieces.
Wherein the longest common subsequence of relative positions (LPCS) may be a position of adding the longest common subsequence in the process route of two workpieces on the basis of the longest common subsequence of the two workpieces.
Fig. 3 illustrates a longest process route sub-sequence diagram of relative positions between process routes of a workpiece according to an embodiment of the present disclosure.
The LCS ignores the position in the process route of each workpiece, the position of the LCS being closely related to the length of the manufacturing system, which can be greatly reduced if the LCS is also located in the same way in the various process routes. On the basis of LCS, the relative position of LCS in the process route of each workpiece is combined, and the longest common subsequence (LPCS) with the most relative position of the process elements is screened out. As shown in FIG. 3, the process routes of the workpiece a and the workpiece c are OPa {1,2,3,4,5} and OPc {2,6,7,4,5}, respectively, the longest process route subsequence LCSCA of the workpiece a and the workpiece c is {2,4,5}, but the relative positions of the processes {2}, {4}, and {5} in the respective process routes are different, and the process {2} is removed from LCSCA {2,4,5} to obtain LPCSCA {4,5} of the workpiece a and the workpiece c. A similarity coefficient method based on the longest common subsequence (LPCS) of the relative position is adopted to fuse delay reconstruction, process route characteristic values and the like, and an important foundation is laid for the construction of a D-RMS exclusive workpiece family.
And step S3: calculating similarity coefficients of the two workpieces according to the length of the process route of the workpieces and the characteristic values of the longest common subsequence of the relative positions.
The similarity coefficient constructed by the workpiece family is derived from the Jaccard coefficient and is combined with factors such as LCS, idle machine tool, detour action, processing time, capacity requirement and the like. Similarly, the D-RMS workpiece family construction method is designed by using a similarity coefficient derived from the Jaccard coefficient. The similarity coefficients between workpieces can generally be characterized by the same number of process steps, such as the LPCS.
FIG. 4 is a schematic diagram illustrating element discontinuities of a longest process route sub-sequence of relative positions between process routes of a workpiece according to an embodiment of the present disclosure; FIG. 5 is a schematic diagram illustrating a deviation in the element positions of the longest process route sub-sequence of relative positions between the process routes of a workpiece according to an embodiment of the present disclosure.
D-RMS implements a functional redundancy strategy in the system front-end process, and the common subsequence should appear as far as possible in the front-end considering the investment cost of the extra functions. However, there may be a discontinuity in the LPCS, as shown in fig. 4, where the LPCSac of workpiece a and workpiece b is {1,2,4}, but there is an element {3} between element {1,2} and element {4} of workpiece a, and an element {6} between element {1,2} and element {4} of workpiece b, that is, there are different elements {3} and {6} between element {1,2} and element {4} of the LPCSac. The LPCS may also have a position deviation, as shown in fig. 5, the LPCSbc of the workpiece b and the workpiece c is {2,6,7}, where the element {2,6,7} is located at the 2 nd, 3 rd, and 4 th positions of the process route of the workpiece b, and is located at the 1 st, 2 nd, and 3 th positions of the process route of the workpiece c, respectively. This results in additional idle, discrete situations and positional deviation situations that require the presence of the LPCS during the similarity coefficient calculation.
For D-RMS, the similarity of two workpieces is larger the more the same process occurs at the front end of the original process route, and the importance degree of the process in the LPCS in calculating the similarity is decreased. In the design process of the similarity coefficient, the positions of the process routes are considered to be positive integers, such as 1,2,3, … and the like, and a monotone increasing sequence is formed. The reciprocal of the position of the process route element of the workpiece can be used as the characteristic value of the LPCS (longest common subsequence of relative positions), for example {1,1/2,1/3, … }, which just form a monotone decreasing sequence, and the sequence is consistent with the decreasing of the importance of the same process from the front end to the rear end, and the reciprocal of the position of the process element in the original process route in the LPCS can be used as the characteristic value of the process.
In one example, calculating the similarity coefficient of two workpieces according to the feature values of the longest common subsequence of the length and relative position of the process route of the workpieces may include:
calculating the characteristic value of the longest common subsequence of the relative positions of the workpiece and the workpiece by taking the element position of the process route of the workpiece as the characteristic value;
determining penalty factors of similarity coefficients of the two workpieces according to the position difference of elements of the longest common subsequence of relative positions of the workpieces;
and calculating the similarity coefficient of the two workpieces according to the length of the process route of the workpieces, the characteristic value of the longest common subsequence of the relative positions and the penalty factor of the similarity coefficient of the two workpieces.
For example, as shown in equation (1), a similarity coefficient for a D-RMS workpiece family is calculated based on the process line length and the characteristic values of the LPCS,
Figure BDA0003078830590000081
wherein S is xy Representing a similarity coefficient between the workpiece x and the workpiece y;
Figure BDA0003078830590000082
representation of a sequence of characteristic values LPCS xy The sum of (1);
Figure BDA0003078830590000083
a sum of a series of characteristic values representing workpiece x and workpiece y; alpha is a penalty factor based on the situation of the position difference of the LPCS element in the workpiece x and the workpiece y; beta is a penalty factor based on the position discontinuity of the LPCS element.
Figure BDA0003078830590000084
Can be calculated according to equations (2) - (4):
Figure BDA0003078830590000085
Figure BDA0003078830590000086
Figure BDA0003078830590000087
wherein, the first and the second end of the pipe are connected with each other,
Figure BDA0003078830590000088
represents the sum of the series of eigenvalues of the workpiece x;
Figure BDA0003078830590000089
represents the sum of the series of characteristic values of the workpiece y; OP (optical fiber) x Represents the process path length of the workpiece x; OP (optical fiber) y Indicating the process path length of the workpiece y.
Figure BDA0003078830590000091
Can be calculated by following equations (5) - (9):
Figure BDA0003078830590000092
Figure BDA0003078830590000093
Figure BDA0003078830590000094
Figure BDA0003078830590000095
Figure BDA0003078830590000096
wherein the content of the first and second substances,
Figure BDA0003078830590000097
represents the LPCS xy Sum of the characteristic values of the elements in the workpiece x;
Figure BDA0003078830590000098
represents the LPCS xy Sum of the feature values of the elements in the workpiece y.
The penalty factor α can be calculated according to equations (10) - (12):
Figure BDA0003078830590000099
Figure BDA00030788305900000910
Figure BDA00030788305900000911
the penalty factor β can be calculated according to equations (13) - (19):
Figure BDA00030788305900000912
Figure BDA0003078830590000101
Figure BDA0003078830590000102
Figure BDA0003078830590000103
Figure BDA0003078830590000104
Figure BDA0003078830590000105
Figure BDA0003078830590000106
wherein, the first and the second end of the pipe are connected with each other,
Figure BDA0003078830590000107
and
Figure BDA0003078830590000108
represents the LPCS xy The upper and lower limits of the position of the element in the workpiece x;
Figure BDA0003078830590000109
and
Figure BDA00030788305900001010
represents the LPCS xy The position of the element in the workpiece y is limited.
And step S4: and calculating the similarity coefficient among all the workpieces of the delay reconstruction manufacturing system to obtain a similarity matrix of the workpieces of the delay reconstruction manufacturing system. And (3) calculating a similarity coefficient between any two workpieces according to the step 3, and further obtaining a workpiece similarity matrix of D-RMS. For example, the process route of the workpieces to be grouped is shown in table 1, and the similarity matrix of the calculated workpieces is shown in fig. 6.
TABLE 1
Figure BDA00030788305900001011
Figure BDA0003078830590000111
Step S5: and clustering convergence is carried out on the similarity matrix of the workpiece by utilizing a machine learning clustering algorithm, and the construction of the workpiece family of the delay reconstruction manufacturing system is completed.
After the similarity coefficient calculation among the workpieces is completed in the steps S1-S4, the specific grouping of the workpieces is realized by using a clustering algorithm, and the construction of a workpiece family is realized. The hierarchical clustering algorithm includes a top-down and a bottom-up approach, such as a link-by-link clustering algorithm. However, the method has low calculation efficiency, and randomness exists when the similarity between the workpieces is the same, so that the clustering result cannot meet the requirement.
In recent years, the machine learning algorithm is applied to the engineering field, is very hot, and is also applied to the clustering aspect of workpiece family construction. Wherein, the machine learning clustering algorithm can be K-medoids. K-means is similar to K-means, where K-means can select non-input data points when selecting a center point, a phenomenon in which the center point is difficult to interpret. The K-medoids clustering algorithm specifies that data points must be input at the center point of the K-medoids clustering algorithm, and a similarity matrix between workpieces can be used as input data of the K-medoids clustering algorithm.
For example, a machine learning clustering algorithm K-medoids is adopted to perform clustering convergence on the similarity matrix of the D-RMS workpiece, and the specific process is as follows:
(1) Randomly selecting k workpieces from the n workpieces as cluster centers (k is less than or equal to n), wherein C = { C is shown as formula (20) 1 ,c 2 ,...c k Formula (20);
(2) Connecting other workpieces to the nearest cluster centerline point;
(3) Each cluster is updated according to equation (21):
Figure BDA0003078830590000112
wherein m is j Representing the jth cluster center point; d (p) i ,m j ) Representing a distance matrix, which can be calculated by taking the difference between 1 and the similarity matrix;
(4) And (5) repeatedly executing the steps (2) to (4) until convergence.
FIG. 7 is a schematic diagram illustrating workpiece family build results for a machine learning-based lag reconstruction manufacturing system according to an embodiment of the present disclosure.
As shown in fig. 7, the similarity matrix of the workpiece in table 1 and the workpiece in fig. 6 is clustered by using a machine learning clustering k-medoids algorithm. The clustering results of fig. 7 are collated, and as shown in table 2, the clustering results are classified into 5 workpiece groups, for example, workpiece group 1{ workpiece 7, workpiece 12}, workpiece group 2{ workpiece 5, workpiece 11}, workpiece group 3{ workpiece 15, workpiece 19}, workpiece group 4{ workpiece 8, workpiece 3, workpiece 18, workpiece 14, workpiece 4, workpiece 1}, and workpiece group 5{ workpiece 9, workpiece 10, workpiece 13, workpiece 16, workpiece 17, workpiece 20, workpiece 6, workpiece 2}.
According to the workpiece family construction method of the machine learning-based delay reconstruction manufacturing system, the delay reconstruction characteristic is considered at the beginning of the construction of the workpiece family, the defects of the existing research method are overcome, the exclusive D-RMS workpiece family is constructed, and the operation efficiency of the D-RMS is improved; by combining the flexibility of the similarity coefficient method, the delayed reconstruction characteristic and the machine learning K-medoids clustering algorithm, the accuracy and the effectiveness of the construction of the workpiece family can be improved, and the operation efficiency of the delayed reconstruction manufacturing system is further improved.
Although the embodiments of the present invention have been described above, the above descriptions are only for the convenience of understanding the present invention, and are not intended to limit the present invention. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (3)

1. A method for building a family of workpieces for a machine learning-based delayed reconstruction manufacturing system, the method comprising:
determining a longest common subsequence of any two workpieces of the delayed reconstruction manufacturing system based on process routes of the two workpieces;
extracting the longest common subsequence of the relative position between the two workpieces according to the longest common subsequence and the position of the longest common subsequence in the process route of the two workpieces;
calculating similarity coefficients of the two workpieces according to the length of the process route of the workpieces and the characteristic values of the longest common subsequence of the relative positions;
calculating similarity coefficients among all workpieces of the delay reconstruction manufacturing system to obtain a similarity matrix of the workpieces of the delay reconstruction manufacturing system;
clustering convergence is carried out on the similar matrix of the workpiece by utilizing a machine learning clustering algorithm, and the construction of the workpiece family of the delay reconstruction manufacturing system is completed;
the longest common subsequence of the two workpieces is the longest process route subsequence with the same processing function and processing sequence between the process routes of the two workpieces;
the longest common subsequence of relative positions between the two workpieces is the position of the longest common subsequence added to the process route of the two workpieces on the basis of the longest common subsequence of the two workpieces;
the calculating the similarity coefficient of the two workpieces according to the length of the process route of the workpieces and the characteristic value of the longest common subsequence of the relative positions comprises:
calculating a feature value of the longest common subsequence of relative positions of the workpiece and the workpiece with the element position of the process route of the workpiece as the feature value;
determining a penalty factor of the similarity coefficient of the two workpieces according to the position difference of the elements of the longest common subsequence of the relative positions of the workpieces;
calculating the similarity coefficient of the two workpieces according to the length of the process route of the workpieces, the characteristic value of the longest common subsequence of the relative positions and the penalty factor of the similarity coefficient of the two workpieces.
2. The workpiece family building method of claim 1, wherein the reciprocal of the position of the process route element of the workpiece is the characteristic value of the longest common subsequence of relative positions.
3. The workpiece family construction method of claim 1, wherein the clustering algorithm is K-medoids.
CN202110560654.XA 2021-05-21 2021-05-21 Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning Active CN113190929B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110560654.XA CN113190929B (en) 2021-05-21 2021-05-21 Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110560654.XA CN113190929B (en) 2021-05-21 2021-05-21 Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning

Publications (2)

Publication Number Publication Date
CN113190929A CN113190929A (en) 2021-07-30
CN113190929B true CN113190929B (en) 2022-12-30

Family

ID=76984638

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110560654.XA Active CN113190929B (en) 2021-05-21 2021-05-21 Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning

Country Status (1)

Country Link
CN (1) CN113190929B (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103996080A (en) * 2014-05-30 2014-08-20 北京理工大学 Manufacturing system configuration optimization method for achieving the highest connectedness
CN103995978A (en) * 2014-05-30 2014-08-20 北京理工大学 Reconfigurable manufacturing system part family construction method with comprehensive production factors considered

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7574409B2 (en) * 2004-11-04 2009-08-11 Vericept Corporation Method, apparatus, and system for clustering and classification
CN104317244B (en) * 2014-09-28 2018-06-12 北京理工大学 A kind of reconfigurable manufacturing system part family construction method
CN111915153A (en) * 2020-07-11 2020-11-10 天津大学 Method for dividing reconfigurable manufacturing system workpiece family by considering multiple indexes

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103996080A (en) * 2014-05-30 2014-08-20 北京理工大学 Manufacturing system configuration optimization method for achieving the highest connectedness
CN103995978A (en) * 2014-05-30 2014-08-20 北京理工大学 Reconfigurable manufacturing system part family construction method with comprehensive production factors considered

Also Published As

Publication number Publication date
CN113190929A (en) 2021-07-30

Similar Documents

Publication Publication Date Title
AlGeddawy et al. Optimum granularity level of modular product design architecture
CN103927346B (en) Query connection method on basis of data volumes
TW514804B (en) Structural regularity extraction and floorplanning in datapath circuits using vectors
CN102073700B (en) Discovery method of complex network community
CN101751499B (en) Automatic assembly method for three-dimensional ship modeling part
CN104063559A (en) Layout legalization method and system for distributed computing of large-scale integrated circuit
EP2759952B1 (en) Efficient genomic read alignment in an in-memory database
CN107656978B (en) Function dependence-based diverse data restoration method
CN111899882A (en) Method and system for predicting cancer
CN111444220A (en) Cross-platform SQ L query optimization method combining rule driving and data driving
CN109902142A (en) A kind of character string fuzzy matching and querying method based on editing distance
CN111915153A (en) Method for dividing reconfigurable manufacturing system workpiece family by considering multiple indexes
CN107977504A (en) A kind of asymmetric in-core fuel management computational methods, device and terminal device
CN110335641B (en) Four-body combination genetic relationship identification method and device
CN113190929B (en) Workpiece family construction method of delayed reconstruction manufacturing system based on machine learning
CN102622334A (en) Parallel XSLT (Extensible Style-sheet Language Transformation) conversion method and device for use in multi-thread environment
CN104317244A (en) Reconfigurable manufacturing system part family construction method
CN107657050A (en) One kind is based on " with the one-to-one join of conflation algorithm calculating, one-to-many join " contraposition segmentation parallel method
CN112001635A (en) Process flow determination method, device, server and storage medium
CN103793597B (en) Model similarity measuring method based on complete backbone subsystems
CN104794010B (en) Information exchange optimization method in a kind of system
CN113986990B (en) Data resource acquisition and labeling method and device based on block chain data mining
CN113672615B (en) Data analysis method and system for automatically generating SQL based on relationships among tree tables
Jemal et al. What if mixing technologies for Big Data mining and queries optimization
CN109685453B (en) Method for intelligently identifying effective paths of workflow

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant