CN109543899B - Two-dimensional contour layout sequencing method based on deep learning - Google Patents
Two-dimensional contour layout sequencing method based on deep learning Download PDFInfo
- Publication number
- CN109543899B CN109543899B CN201811384644.XA CN201811384644A CN109543899B CN 109543899 B CN109543899 B CN 109543899B CN 201811384644 A CN201811384644 A CN 201811384644A CN 109543899 B CN109543899 B CN 109543899B
- Authority
- CN
- China
- Prior art keywords
- layout
- matrix
- stock
- stock layout
- geometric
- 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
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/043—Optimisation of two dimensional placement, e.g. cutting of clothes or wood
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Strategic Management (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Game Theory and Decision Science (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Document Processing Apparatus (AREA)
Abstract
The invention discloses a two-dimensional contour layout sequencing method based on deep learning, which specifically comprises the following steps: obtaining and preprocessing stock layout history big data, and calibrating all stock layout parts in the history data to obtain a stock layout sequence matrix Y; extracting the geometric characteristics of each arranged part to obtain a geometric characteristic matrix X; arranging to obtain a stock layout sequencing data set PRD-T; establishing a Packing-Sort-Model of a deep learning Model; inputting a PRD-T data set, and training to obtain a Packing-Sort-Model capable of outputting the layout sequence of the parts to be laid; extracting geometric characteristics of the part to be subjected to stock layout to obtain a geometric characteristic matrix A of the part to be subjected to stock layout; inputting the geometric characteristic matrix of the part to be subjected to stock layout into the trained deep learning model; calculating to obtain a stock layout sequence matrix B of the parts to be stock layout; and (4) performing one-by-one approach connection layout on the parts to be subjected to layout according to the layout sequence to finish layout. The invention can realize sequencing of the parts to be arranged in the sample arranging process, and has good sample arranging effect and high efficiency.
Description
Technical Field
The invention relates to a contour stock ordering method, in particular to a two-dimensional contour stock ordering method based on deep learning.
Background
The stock layout problem generally exists in the field of manufacturing industry, and the solving of the problem is greatly helpful for improving the material utilization rate and reducing the enterprise cost, so that the problems related to the stock layout problem have extremely important research value.
The most widely studied in the present lay-out field is the two-dimensional contour lay-out problem, which in turn includes the positioning problem and the sequencing problem, and the lay-out order is a main factor for determining the quality of the final lay-out for a plurality of parts. The common two-dimensional profile layout sequencing method mainly comprises the following steps: a stock layout sequencing method based on a heuristic algorithm and a stock layout sequencing method based on an intelligent optimization algorithm.
The heuristic algorithm usually determines the sequence number of the next stock layout part according to the current stock layout condition and the prior knowledge, for example, a part with the lowest stock layout height after stock layout is selected as the next stock layout part, the algorithm speed of the method is high, but the final stock layout utilization rate is low because the whole stock layout distribution condition is not comprehensively considered.
The method for solving the problem of stock layout sequencing based on the intelligent optimization algorithm mainly solves the problem of the combination of stock layout sequences by adopting the intelligent algorithms such as genetic algorithm GA, simulated annealing algorithm SA, hill climbing algorithm and the like to obtain the optimal solution. The method is high in time complexity and easy to fall into a local optimal solution.
Disclosure of Invention
In order to overcome the defects of the conventional two-dimensional contour layout sequencing method and improve the final quality of layout, the invention provides a two-dimensional contour layout sequencing method based on deep learning. The method has no requirement on the shape of the stock layout parent metal, can be used for solving the stock layout problem of irregular outlines on rectangular, round and irregular parent metals, and has the characteristics of fast stock layout sequencing and good stock layout effect.
In order to solve the technical problems of the method, the invention is implemented by the following technical scheme, and the method for sequencing the two-dimensional contour layout based on the deep learning specifically comprises the following steps: s1: acquiring historical big data existing in the stock layout field; s2: preprocessing the stock layout history big data, calibrating the stock layout sequence of m stock layout parts, and obtaining a stock layout sequence matrix Y of the m stock layout parts, wherein Y is a matrix of 1 x m; s3: extracting the geometric characteristics of each layout part in the historical data, and forming a geometric characteristic matrix X of the layout parts in all the historical data, wherein the matrix X is { X ═ X }
i1,2 … m, geometric characteristic X of the ith laid part
i={g1
i,g2
i…gj
i…gn
i}
T,gj
iThe method is characterized in that the method is a jth geometric feature in ith stock layout part geometric features, n is the number n of the extracted geometric features which is not less than 1, namely X is a matrix of n X m; s4: packing and sorting the layout sequence matrix Y obtained in the step S2 and the geometric feature matrix X obtained in the step S3 to obtain a training data set PRD-T related to the two-dimensional contour layout sequencing problem; s5: building a deep learning Model Packing-Sort-Model, and training the deep learning Model Packing-Sort-Model by utilizing the training data set PRD-T obtained in S4, so that the Model can predict the stock layout sequence of the sample to be stock-laid; s6: traversing each part to be laid, and extracting the geometric characteristics of each part to be laid to formA geometric characteristic matrix A of the part to be arranged; s7: inputting the geometric characteristic matrix A of the part to be subjected to stock layout obtained in the step S6 into a Packing-Sort-Model of the deep learning Model obtained in the step S5, and predicting to obtain a stock layout sequence matrix B of the part to be subjected to stock layout; s8: and formulating the layout sequence of the pieces to be laid according to the layout sequence matrix B of the pieces to be laid S7, and performing one-by-one abutting layout on the pieces to be laid according to the layout sequence.
Preferably, the geometric features of the layout parts described in S3 include all of the geometric features of the parts that can be described in mathematical language.
Preferably, the S6 specifically includes the following steps:
(1) extracting geometric characteristics of various stock layout parts including area, number of edges, rectangle degree and concave-convex property;
(2) the geometric characteristic matrix A of the part to be arranged is { a ═ a
i1,2 … m, geometric characteristic a of the ith part to be laid
i={g1
i,g2
i,g3
i…gn
i}
TAnd A is a matrix of n x m.
The invention has the following beneficial effects: the method is not limited by the shape of the stock layout base material and the outline shape of the part to be stock-laid, and can be applied to stock layout of rectangular, circular and two-dimensional irregular patterns; the stock layout effect is good, efficient.
Drawings
FIG. 1 is a flow chart of a stock layout sequencing method according to the present invention;
FIG. 2 is an exemplary diagram of a history stock template used in an embodiment of the invention;
FIG. 3 is a Packing-Sort-Model of the present invention; and
FIG. 4 is a schematic diagram of the present invention for performing simple sequencing runs.
Detailed Description
The technical contents, structural features, attained objects and effects of the present invention are explained in detail below with reference to the accompanying drawings.
The problem of stock layout sequencing is solved based on an intelligent optimization algorithm, and the material utilization rate can be improved while the enterprise cost is reduced. The invention designs a two-dimensional contour layout sequencing method based on deep learning, which specifically comprises the following steps as shown in figure 1:
s1: acquiring historical big data of m stock layout parts in the stock layout field;
s2: preprocessing the stock layout history big data, and calibrating the stock layout sequence of all stock layout parts to obtain a stock layout sequence matrix Y of the stock layout parts, wherein Y is a row vector of 1 × m;
s3: extracting geometric features of various stock layout parts in historical data, wherein the geometric features of the various parts, such as area, number of edges, rectangle degree and concavity and convexity, which can be described by mathematical languages, can be properly selected according to specific stock layout problems; forming a geometric feature matrix X, X ═ X of the arranged parts in all historical data
i1,2 … m, wherein the geometric feature of the ith laid part is X
i,X
i={g1
i,g2
i…gj
i…gn
i}
TGji is the jth geometric feature in the ith geometric feature of the layout part, n is the number n of the extracted geometric features is more than or equal to 1, namely X is a matrix of n × m;
s4: packing and sorting the layout sequence matrix Y obtained in S2 and the geometric feature matrix X obtained in S3 to obtain a training data set PRD-T related to the two-dimensional contour layout sequencing problem, as shown in FIG. 2;
s5: as shown in fig. 3, a deep learning Model Packing-Sort-Model is set up, and the deep learning Model Packing-Sort-Model is trained by using a training data set PRD-T obtained in S4, so that the Model can predict the stock layout sequence of the pieces to be stock-laid;
s6: traversing each part to be arranged, extracting the geometric characteristics of each part to be arranged, and forming a geometric characteristic matrix A of the part to be arranged, wherein A is { a ═ a }
iI-1, 2 … m, wherein the geometric feature a of the ith part to be laid out
i={g1
i,g2
i,g3
i…gn
i}
TI.e. a is a matrix of n x m.
S7: as shown in fig. 4, inputting the geometric feature matrix a of the part to be subjected to stock layout obtained in S6 into the deep learning Model Packing-Sort-Model obtained in S5, and predicting to obtain a stock layout sequence matrix B of the part to be subjected to stock layout;
s8: and formulating the layout sequence of the pieces to be laid according to the S7 layout sequence matrix B, and performing one-by-one abutting layout on the pieces to be laid according to the layout sequence.
The invention is further described with reference to the following examples:
the invention provides a two-dimensional contour layout sequencing method based on deep learning, a flow chart of the method is shown as the attached figure 1, and the method comprises the following concrete implementation steps:
(1) historical big data existing in the stock layout field is obtained, and the data used in the embodiment is provided for a laboratory where the stock layout field is located. As shown in FIG. 1, we adopt the template which has been arranged before as the training data source, this embodiment needs a large amount of the arranged template data shown in FIG. 1, and because the amount of data required for training is large, we only provide one of the templates shown in FIG. 1 as an introduction example;
(2) preprocessing the historical big data acquired in the step (1), calibrating the stock layout sequence of the stock layout in the historical data, and sorting to obtain a stock layout sequence matrix Y. As shown in fig. 1, extracting a sequential matrix of the arranged samples of the template, taking m as 25, and obtaining Y as [1,2,3,4,5.. 25], where Y is a matrix of 1 × 25;
(3) and (2) extracting the geometric features of each sample arranged piece in the historical data obtained in the step (1), wherein the area S, the number r of edges and the perimeter l of the case are selected as main geometric features, namely n is 3, so as to form a geometric feature matrix X. We extract the geometric feature matrix of the arranged parts shown in the attached figure 1 to obtain the geometric feature matrix
Wherein S is
iDenotes the area of the ith sample
iIndicates the number of sides of the ith sample-arranged piece, l
iRepresents the perimeter of the ith sample-arranged piece, X being a 3 × 25 matrix in this embodiment;
(4) and (4) leading the matrix Y and the matrix X obtained in the step (2) and the step (3) into matlab for sorting to obtain a training data set PRD-T related to the two-dimensional contour layout sequencing problem. In order to facilitate the extraction in the training process, the PRD-T is made into a table data type;
(5) and building a deep learning Model Packing-Sort-Model as shown in the attached figure 3. The packaging-Sort-Model is a deep neural network with 4 layers, wherein Layer1 is an input Layer, Layer2 and Layer3 are hidden layers, Layer4 is an output Layer, and the number of nodes of the input Layer is the geometric characteristic number n which is 3;
(6) training the Packing-Sort-Model built in the step (5) by taking the training data set PRD-T obtained in the step (4) as an input, so that the Packing-Sort-Model can predict the layout sequence of the output layout parts;
(7) 3 parts are selected as parts to be arranged in the case, and the shapes of the parts are shown in the attached figure 4. Comprises a circular sample piece, a concave hexagonal sample piece and a regular octagonal sample piece. Respectively extracting the geometric features of the 3 samples to be arranged according to the geometric features in the step (3) to obtain a geometric feature matrix of the samples to be arranged
(8) And (4) inputting the geometric characteristic matrix A of the to-be-arranged samples obtained in the step (7) into the Model Packing-Sort-Model obtained in the step (6), and calculating to obtain a arranging sequence matrix B [3, 2, 1] of the 3 to-be-arranged samples selected in the embodiment. According to the result obtained by the sequence matrix B, the sample arrangement sequence of the three sample pieces to be arranged is that the regular octagon sample pieces, the concave hexagon sample pieces and the circular sample pieces are arranged from front to back in sequence;
(9) and (4) performing successive sample arrangement on the sample to be arranged one by one according to the sequence matrix B [3, 2, 1] obtained in the step (8), selecting a rectangle as a template for sample arrangement of the embodiment, wherein the flow is shown in figure 4, and the sample arrangement results of the three parts are shown in figure 4, so that the sample arrangement is completed.
The foregoing is a preferred embodiment of the present application and is not intended to limit the scope of the invention, it should be understood that various modifications and adaptations can be made by those skilled in the art without departing from the principles of the present application and should be considered as within the scope of the present application.
Claims (3)
1. A two-dimensional contour layout sequencing method based on deep learning is characterized by specifically comprising the following steps of:
s1: acquiring historical big data existing in the stock layout field;
s2: preprocessing the stock layout history big data, calibrating the stock layout sequence of m stock layout parts, and obtaining a stock layout sequence matrix Y of the m stock layout parts, wherein Y is a matrix of 1 x m;
s3: extracting the geometric characteristics of each layout part in the historical data, and forming a geometric characteristic matrix X of the layout parts in all the historical data, wherein the matrix X is { X ═ X }
i1,2 … m, geometric characteristic X of the ith laid part
i={g1
i,g2
i…gj
i…gn
i}
T,gj
iThe method is characterized in that the method is a jth geometric feature in ith stock layout part geometric features, n is the number n of the extracted geometric features which is not less than 1, namely X is a matrix of n X m;
s4: packing and sorting the layout sequence matrix Y obtained in the step S2 and the geometric feature matrix X obtained in the step S3 to obtain a training data set PRD-T related to the two-dimensional contour layout sequencing problem;
s5: building a deep learning Model Packing-Sort-Model, and training the deep learning Model Packing-Sort-Model by utilizing the training data set PRD-T obtained in S4, so that the Model can predict the stock layout sequence of the parts to be stock-laid;
s6: traversing each part to be subjected to stock layout, and extracting the geometric characteristics of each part to be subjected to stock layout to form a geometric characteristic matrix A of the part to be subjected to stock layout;
s7: inputting the geometric characteristic matrix A of the part to be subjected to stock layout obtained in the step S6 into a Packing-Sort-Model of the deep learning Model obtained in the step S5, and predicting to obtain a stock layout sequence matrix B of the part to be subjected to stock layout; and
s8: and formulating the layout sequence of the parts to be laid according to the layout sequence matrix B of the parts to be laid according to S7, and performing one-by-one approach layout on the parts to be laid according to the layout sequence.
2. The two-dimensional profile layout sequencing method based on deep learning of claim 1, wherein the geometric features of each layout part in S3 comprise all geometric features of each part which can be described by a mathematical language.
3. The deep learning-based two-dimensional profile layout sequencing method according to claim 1, wherein the S6 specifically comprises the following steps:
(1) extracting geometric characteristics of each part to be arranged, including area, number of edges, rectangle degree and concave-convex property;
(2) the geometric characteristic matrix A of the part to be arranged is { a ═ a
i1,2 … m, geometric characteristic a of the ith part to be laid
i={g1
i,g2
i,g3
i…gn
i}
TAnd A is a matrix of n x m.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811384644.XA CN109543899B (en) | 2018-11-20 | 2018-11-20 | Two-dimensional contour layout sequencing method based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811384644.XA CN109543899B (en) | 2018-11-20 | 2018-11-20 | Two-dimensional contour layout sequencing method based on deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109543899A CN109543899A (en) | 2019-03-29 |
CN109543899B true CN109543899B (en) | 2020-02-11 |
Family
ID=65848674
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811384644.XA Active CN109543899B (en) | 2018-11-20 | 2018-11-20 | Two-dimensional contour layout sequencing method based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109543899B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110827260B (en) * | 2019-11-04 | 2023-04-21 | 燕山大学 | Cloth defect classification method based on LBP characteristics and convolutional neural network |
CN111242363A (en) * | 2020-01-08 | 2020-06-05 | 广东工业大学 | PCB order splicing and typesetting prediction method and system based on machine learning |
CN111428776B (en) * | 2020-03-19 | 2021-03-16 | 燕山大学 | Two-dimensional irregular contour layout method based on skeleton line matching |
CN113704841B (en) * | 2021-08-31 | 2023-11-14 | 浙江大学 | Bias-based optimized layout method |
CN114898012B (en) * | 2022-04-29 | 2023-03-24 | 广东工业大学 | In-circle loading rectangle optimization method based on simulated annealing and Skyline |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009116596A (en) * | 2007-11-06 | 2009-05-28 | Canon Inc | Print processing system |
CN103500255A (en) * | 2013-10-16 | 2014-01-08 | 南通大学 | Intelligent layout method used for rectangular part |
CN106971242A (en) * | 2017-03-21 | 2017-07-21 | 上海大学 | A kind of clothes Automatic Optimal discharging method |
CN106926212A (en) * | 2017-04-28 | 2017-07-07 | 燕山大学 | Two-dimensional silhouette Nesting based on profile framework characteristic |
CN107864405B (en) * | 2017-11-14 | 2020-12-01 | 国家新闻出版广电总局广播科学研究院 | Viewing behavior type prediction method, device and computer readable medium |
CN108427666A (en) * | 2018-02-27 | 2018-08-21 | 广州多普网络科技有限公司 | A kind of print publishing system and method based on deep learning |
-
2018
- 2018-11-20 CN CN201811384644.XA patent/CN109543899B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN109543899A (en) | 2019-03-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109543899B (en) | Two-dimensional contour layout sequencing method based on deep learning | |
CN107862411B (en) | Large-scale flexible job shop scheduling optimization method | |
CN106203395A (en) | Face character recognition methods based on the study of the multitask degree of depth | |
WO2016169286A1 (en) | Workshop layout method for discrete manufacturing system | |
CN103838836B (en) | Based on discriminant multi-modal degree of depth confidence net multi-modal data fusion method and system | |
CN101377854B (en) | Method for simulating Chinese characters hand-written handwriting by a computer | |
CN108197294A (en) | A kind of text automatic generation method based on deep learning | |
CN101604409B (en) | Dispatching method applied to multi-enterable complex manufacturing system | |
CN110619419B (en) | Passenger flow prediction method for urban rail transit | |
CN104778458B (en) | A kind of textile pattern search method based on textural characteristics | |
CN107958298A (en) | A kind of choosing method of the logistics node based on clustering algorithm | |
CN103678681B (en) | The Multiple Kernel Learning sorting technique of the auto-adaptive parameter based on large-scale data | |
CN111079208B (en) | Particle swarm algorithm-based CAD model surface corresponding relation identification method | |
CN114758180A (en) | Knowledge distillation-based light flower recognition method | |
Su et al. | Cabin placement layout optimisation based on systematic layout planning and genetic algorithm | |
CN111062511B (en) | Aquaculture disease prediction method and system based on decision tree and neural network | |
CN106096327A (en) | Gene character recognition methods based on the study of the Torch supervised degree of depth | |
CN117807873A (en) | Layout method and system of two-dimensional irregular parts based on contour matching | |
CN110705650B (en) | Sheet metal layout method based on deep learning | |
CN109388875A (en) | A kind of design implementation method of elastic element module | |
CN110688722B (en) | Automatic generation method of part attribute matrix based on deep learning | |
CN103143123A (en) | System and method for beam direction multi-target optimization based on ant colony algorithm | |
CN106650916B (en) | A kind of mesh segmentation method based on ant group optimization | |
CN111222529A (en) | GoogLeNet-SVM-based sewage aeration tank foam identification method | |
CN110826604A (en) | Material sorting method based on deep learning |
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 |