CN114418177B - New product material distribution prediction method based on digital twin workshops for generating countermeasure network - Google Patents

New product material distribution prediction method based on digital twin workshops for generating countermeasure network Download PDF

Info

Publication number
CN114418177B
CN114418177B CN202111522679.7A CN202111522679A CN114418177B CN 114418177 B CN114418177 B CN 114418177B CN 202111522679 A CN202111522679 A CN 202111522679A CN 114418177 B CN114418177 B CN 114418177B
Authority
CN
China
Prior art keywords
production
data
new product
workshop
network
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
CN202111522679.7A
Other languages
Chinese (zh)
Other versions
CN114418177A (en
Inventor
曹洪新
李露
王美玲
王玉成
叶晓东
孔令成
崔云强
毛吴俊
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.)
Hefei Institutes of Physical Science of CAS
Original Assignee
Hefei Institutes of Physical Science of CAS
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 Hefei Institutes of Physical Science of CAS filed Critical Hefei Institutes of Physical Science of CAS
Priority to CN202111522679.7A priority Critical patent/CN114418177B/en
Publication of CN114418177A publication Critical patent/CN114418177A/en
Application granted granted Critical
Publication of CN114418177B publication Critical patent/CN114418177B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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"
    • 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
    • G06N3/042—Knowledge-based neural networks; Logical representations of neural 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/047—Probabilistic or stochastic 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • 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
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06313—Resource planning in a project environment
    • 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
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06315—Needs-based resource requirements planning or analysis
    • 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
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • 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
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/04—Manufacturing
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Business, Economics & Management (AREA)
  • Marketing (AREA)
  • Tourism & Hospitality (AREA)
  • Development Economics (AREA)
  • General Health & Medical Sciences (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Biodiversity & Conservation Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Manufacturing & Machinery (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a new product material distribution prediction method of a digital twin workshop based on a generation countermeasure network, which comprises the following steps: 1. acquiring uncertain factor sample data and takt sample data affecting the production process in a digital twin workshop, and preprocessing; 2. constructing and generating an countermeasure network model, and introducing similarity calculation; 3. training and optimizing a generated countermeasure network model for predicting the production beats of the new product based on similar product historical sample data and new product real-time monitoring sample data to obtain the working beats of the production units of the digital twin workshops; 4. calculating material distribution requirements in the production process of new products and controlling the production rhythm of the digital twin workshop. The invention combines uncertain influencing factors and a countermeasure generation method in the production process, thereby improving the accuracy of the production beat prediction of the new product and improving the production efficiency of the new product.

Description

New product material distribution prediction method based on digital twin workshops for generating countermeasure network
Technical Field
The invention belongs to the field of intelligent manufacturing, and particularly relates to a new product material distribution prediction method of a digital twin workshop based on an antagonistic network.
Background
Along with the development of integration of intelligent manufacturing and industrial Internet, the data volume generated in the production process of a manufacturing workshop is increased explosively, complex and changeable production states in the manufacturing workshop are difficult to analyze and process by means of manual mode, a mathematical twin workshop is interacted with a virtual workshop through bidirectional real-time mapping of a physical workshop, production activity planning, production element management, production process control and the like of the manufacturing workshop are realized under the driving of twin data, analysis and processing can be carried out on various saved data after production, and learning optimization of an actual production control process is realized.
At present, in the new product manufacturing production process, factors such as the influence of production beats and the constraint of inventory are mainly considered on the basis of the traditional material demand plan, the production beats of the new product are calculated through methods such as comparison with similar products, visual observation and the like, and then the material demands in the production process are quantitatively calculated according to the production beats. In the actual production process, a plurality of uncertainty factors exist, which cause fluctuation of production beats and influence balance of the whole production process. The traditional production material demand plan is mainly calculated according to the material required by the product structure, and although some influencing factors in the actual production process are considered, the organic combination of the product structure demand and the dynamic demand of the actual production process is difficult to realize, the dynamic calculation capability of the material demand in the actual production process is lacking, the change of the material required by the actual production process cannot be accurately predicted, the idle waiting or backlog blocking of a production station is caused, the production cost is increased, the production period is prolonged, and the production efficiency is reduced.
Disclosure of Invention
Aiming at the problems or the shortcomings, the invention provides a new product material distribution prediction method based on a digital twin workshop for generating an countermeasure network, so that the production beats in the new product production process can be generated with high accuracy, the material dynamic requirements in the new product production process can be met with high adaptability, the accurate prediction capability of the material requirements in the actual production process is improved, and an important data basis is provided for material prediction of new product manufacture and reasonable arrangement of subsequent optimization and production plans.
In order to achieve the aim of the invention, the invention adopts the following technical scheme:
the invention discloses a new product material distribution prediction method based on a digital twin workshop for generating an countermeasure network, which is characterized by comprising the following steps:
step 1, acquiring original data of a new product and similar products thereof in an actual production process by utilizing a digital twin system of a production workshop and preprocessing the original data;
step 1.1, normalizing the product state index data produced in the workshop in the original data to obtain normalized product state index data, and marking the normalized product state index data as P= [ P ] 1 p 2 p 3 ]Wherein p is 1 ，p 2 ，p 3 Respectively representing the input amount of raw materials, the output amount of products and the qualification rate of the products;
step 1.2, normalizing the workshop equipment state index data of the production unit i in the original data to obtain normalized workshop equipment state index data, and marking the normalized workshop equipment state index data as V i ＝[v i1 v i2 v i3 v i4 ]Wherein v is i1 ，v i2 ，v i3 ，v i4 Respectively representing the equipment running time, the equipment processing time, the equipment failure rate and the unit product energy consumption of the production unit i;
step 1.3, normalizing workshop personnel state index data of the production unit i in the original data to obtain normalized workshop personnel state index data, and marking the normalized workshop personnel state index data as H i ＝[h i1 h i2 h i3 h i4 ]Wherein h is i1 ，h i2 ，h i3 ，h i4 Respectively representing the working time, the number of personnel, the working efficiency and the absenteeism rate of the production unit i;
step 1.4, normalizing the workshop production environment state index data in the original data to obtain normalized workshop production environment state index data, which is recorded as E= [ E ] 1 e 2 e 3 ]Wherein e is 1 ，e 2 ，e 3 Respectively representing a thermal environment, an acoustic environment, and a light environment;
step 1.5, calculating the structural material demands of n parts of the new product in workshop production and normalizing according to the design parameters of the new product, thereby obtaining a normalized structural material demand matrix R= [ R ] of the new product 1 ··· r i ··· r n ]Wherein r is i The structure material requirement of the ith part of the new product when the production unit i is used for production is represented, and n is the total number of parts to be processed and produced in the new product;
step 1.6, dividing the production efficiency comprehensive evaluation index of the workshop into two grades, wherein the method comprises the following steps: the index of high production efficiency is 1 grade, and the index of low production efficiency is 0 grade;
step 1.7, normalizing the production takt data of n production units in the workshop in the original data, extracting data corresponding to an index with high workshop production efficiency, and obtaining normalized real production takt data, which is recorded as T= [ T ] 1 ··· t i ··· t n ]Wherein t is i Tact data representing the production unit i;
step 2, calculating a weight vector A= [ a ] of uncertainty influencing factors contained in the 4-type state index data according to the expert knowledge of the system 1 ··· a m ]Wherein a is m A weight indicating the mth state index, m=6+4×n×2 being the number of state indexes; thereby constructing the production state index data set B= [ P V ] 1 H 1 ··· V i H i ··· V n H n E]And weightingProcessing to obtain a weighted production state index data set z=b..a;
step 3, constructing and generating an countermeasure network model, which comprises the following steps: generating a network G and a discrimination network D, wherein the generation network G and the discrimination network D are based on a multi-layer BP neural network;
step 4, counting part parameters of similar products, including: the part design and the process parameters of similar products are used for training a part similarity prediction network model based on a multi-layer BP neural network;
step 5, counting the part parameters of the new product, and calculating the similarity of the parts in the new product by using the part similarity prediction network model to obtain a similarity matrix F= [ F ] of the parts of the new product 1 ··· f i ··· f n ]Wherein f i Similarity of the ith part of the new product;
step 6, calculating predicted tact data y according to formula (1):
y＝G(z|R).*F (1)
in formula (1), G () represents generating a network, Z is a set of samples in the data set Z;
step 7, constructing a loss function V (D, G) for generating the countermeasure network model by using the formula (2):
in the formula (2), x is a group of takt data, x-P, among the real takt data T r Representing true tact data obeying probability distribution P r ，Expects true tact data; y-P g Representing predicted tact data compliance probability distribution P g ，/>Expected for predicting tact data;
step 8, a network G is fixedly generated, and a matrix x=t.r is used as an input of the discrimination network D, so that an output label of the discrimination network D is 1; meanwhile, taking predicted production takt data y as input of the discrimination network D, enabling an output label of the discrimination network D to be 0, and training and optimizing the discrimination network D so as to maximize a loss function V (D, G);
step 9, fixing a discrimination network D, taking predicted production beat data y as input of the discrimination network D, enabling an output label of the discrimination network D to be 1, and training and optimizing the generation network G so as to minimize a loss function V (D, G);
step 10, training the generating network G and the judging network D alternately and iteratively according to the process of the steps 6-9 until the probability distribution P of the production beat data is predicted g Probability distribution P with real tact data r Equality, thus obtaining a trained generated countermeasure network model;
step 11, counting production state related data in the production process of new products, which comprises the following steps: product state index data P', plant state index data V of production unit i i Workshop personnel status index data H 'of production unit i' i The workshop production environment state index data E ', the new product structure material demand matrix R' and the new product structure material demand matrix R 'are input into a trained generation network for generating an countermeasure network model, so that predicted production takt data y' of the new product is obtained;
and step 12, calculating the material supply quantity and the material supply time of n production units in the digital twin workshop, and transmitting the predicted production takt data y' to corresponding equipment in the new product manufacturing production line so as to control the working rhythm of the corresponding equipment.
Compared with the prior art, the invention has the beneficial effects that:
1. according to the invention, comprehensive analysis is carried out on the production process in the digital twin workshop, uncertain influence factors in the actual production process are considered, relevant factor data influencing the production state in the digital twin workshop is collected and processed, the influence degree of relevant factors on the production beats is statistically analyzed according to expert knowledge, the influence degree is used as input data of a new product production beat prediction model, the requirements of new product structural materials are used as condition input for generating an antagonism network model, similarity calculation is introduced on the basis of generating the antagonism network model, the complex change condition in the actual production process is accurately described, and the accuracy of new product production beat generation and material supply prediction is improved.
2. The invention predicts the production beats of the new product by using the generated countermeasure network model, trains the generated countermeasure network model by using the historical data of the similar product, and absorbs the historical experience of the similar product; meanwhile, data are collected in real time in the production process of the new product, and the antagonism network model is generated by dynamic updating and optimizing, so that the calculation complexity in the production beat generation process of the new product is reduced, and the prediction efficiency of material distribution of the new product is improved.
Drawings
FIG. 1 is a schematic diagram of a new product production application scenario in a digital twin shop;
FIG. 2 is a schematic diagram of an index system affecting tact;
FIG. 3 is a schematic diagram of the generation of an impedance network model;
FIG. 4 is a schematic diagram of a generated network structure in a generated reactance network model;
FIG. 5 is a schematic diagram of a discrimination network structure in a generation reactance network model;
FIG. 6 is a schematic diagram of a part similarity prediction network model;
fig. 7 is a flow chart of a new product material distribution prediction method.
Detailed Description
In this embodiment, a method for predicting new product material distribution in a digital twin workshop based on generating an countermeasure network is provided, firstly, selecting influencing factors of production beats in the digital twin workshop as variables, acquiring corresponding variable values from production history data of similar products, and using the variable values to train a model for generating the countermeasure network for predicting new product material distribution, and meanwhile, collecting sample data of uncertain factors influencing the production process of the new product, so as to optimize the model for generating the countermeasure network for predicting new product material distribution, and the method for predicting new product material distribution specifically comprises the following steps:
step 1, acquiring production state indexes of similar products in an actual production process in a production workshop shown in fig. 1 through a digital twin system, and carrying out statistics and preprocessing on original data of the production state indexes of the similar products as shown in fig. 2; removing abnormal data combination, normalizing the original data of each state index to [0,1] according to formula (1) relative to the maximum planning number of each state index in the production plan,
in the formula (1), Y is a numerical value normalized by each index of the production state, X is a numerical value of the corresponding index of the acquired production state, and X max The maximum value of the corresponding index in the production plan; x is X min Is the minimum value of the corresponding index in the production plan.
Step 1.1, extracting product state index data produced by workshops, and recording the normalized product state index data as P= [ P ] 1 p 2 p 3 ]，p 1 ，p 2 ，p 3 Respectively representing the input amount of raw materials, the output amount of products and the qualification rate of the products.
Step 1.2, extracting workshop equipment state index data, and marking the equipment state index data in the normalized production unit i as V i ＝[v i1 v i2 v i3 v i4 ]，v i1 ，v i2 ，v i3 ，v i4 Respectively representing the running time of the equipment, the processing time of the equipment, the failure rate of the equipment and the energy consumption of unit products.
Step 1.3, extracting workshop personnel state index data, and marking the personnel state index data in the normalized production unit i as H i ＝[h i1 h i2 h i3 h i4 ]，h i1 ，h i2 ，h i3 ，h i4 Respectively representing working time, personnel number, working efficiency and duty failure rate.
Step 1.4, extracting and normalizing workshop production environment state index data, and recording the normalized environment state index data asE＝[e 1 e 2 e 3 ]，e 1 ，e 2 ，e 3 Respectively representing a thermal environment, an acoustic environment, and a light environment.
Step 2, calculating the structural material requirement of new product parts in the production of a workshop production unit i according to new product design parameters, constructing a new product structural material requirement matrix and normalizing R= [ R ] 1 ··· r i ··· r n ]Wherein r is i The structure material requirement of the ith part of the new product when the production unit i is used for production is represented, and n is the total number of parts to be processed and produced in the new product.
Step 3, calculating the weight vector A= [ a ] of the factors contained in each state according to the expert knowledge of the system 1 ··· a m ]Wherein a is m A weight indicating the mth state index, m=6+4×n×2 being the number of state indexes; construction of production status index data setsk is the total number of samples in B; weighting the production state index data set B to obtain weighted data sets of each group of samples +.>
And 4, dividing the comprehensive evaluation index of the production efficiency of the workshop into two grades, wherein the grade index with high production efficiency is 1, and the grade index with low production efficiency is 0.
Step 5, extracting data corresponding to high workshop production efficiency from the production beat data of n production units in the workshop in the original data, and performing normalization processing to obtain normalized real production beat data, wherein the normalized real production beat data is recorded as T= [ T ] 1 ··· t i ··· t n ]，t i The tact data representing the production unit i.
And 6, constructing a generated countermeasure network model, as shown in fig. 3, including a generated network G and a discrimination network D, wherein the generated network G and the discrimination network D are constructed based on a multi-layer BP neural network, as shown in fig. 4 and 5.
Step 7, counting the part design and process parameters of similar products, and training a part similarity prediction network model based on a multi-layer BP neural network, as shown in figure 6; counting the part parameters of a new product, and calculating the similarity of the parts in the new product by using a part similarity prediction network model to obtain a similarity matrix F= [ F ] of the parts of the new product 1 ··· f i ··· f n ]，f i Is the similarity of the ith part of the new product.
Step 8, calculating and generating the hidden layer node quantity of the network G, the discrimination network D and the part similarity prediction network model by using the formula (2):
in the formula (2), M is the number of hidden layer nodes, M 1 For inputting the number of layer nodes, M 2 The number of the nodes is the number of the output layer.
Step 9, calculating a predicted takt y according to formula (3) using the generation network G and the similarity calculation:
y＝G(z|R).*F (3)
in formula (3), G () represents the generation network, Z is a set of samples in the data set Z, z=b j .*A。
Step 10, constructing and generating a loss function V (D, G) of the countermeasure network model by using the formula (4):
in the formula (4), x is a group of takt data, x-P, among the real takt data T r Representing true tact data obeying probability distribution P r ，Expects true tact data; y-P g Representing predicted tact data compliance probability distribution P g ，/>To predict the tact data expectations.
Step 11, a network G is fixedly generated, and a matrix x=t.r is used as an input of a discrimination network D, so that an output label of the discrimination network D is 1; meanwhile, taking the predicted production takt data y as the input of the discrimination network D, enabling the output label of the discrimination network D to be 0, thereby training and optimizing the discrimination network D and maximizing the loss function V (D, G);
and step 12, fixing the discrimination network D, taking the predicted production beat data y as the input of the discrimination network D, and enabling the output label of the discrimination network D to be 1, so as to train and optimally generate a network G, and minimizing a loss function V (D, G).
Step 13, alternately and iteratively training the generation network G and the discrimination network D according to the process of the steps 9-12 until the probability distribution P of the production takt data is predicted g Probability distribution P with real tact data r And the same is achieved, so that a trained generation countermeasure network model is obtained.
Step 14, counting production state related data in the production process of new products, including: product state index data P', plant state index data V of production unit i i Workshop personnel status index data H 'of production unit i' i And workshop production environment state index data E ', a new product structure material demand matrix R', and inputting the new product structure material demand matrix R 'into a trained generation network for generating an countermeasure network model, thereby obtaining predicted production takt data y' of the new product.
And 15, calculating the material supply quantity and the material supply time of n production units in the digital twin workshop, and transmitting the predicted production takt data y' to corresponding equipment in the new product manufacturing production line so as to control the working rhythm of the corresponding equipment.
Step 16, in the new product production process, obtaining the high-efficiency production state related data of the new product through a digital twin system of a production workshop, training and generating a discrimination network D and a generation network G in an countermeasure network model by using the steps 1 to 15, and optimizing and predicting the production takt y' to obtain better material supply quantity and supply time, as shown in fig. 7.

Claims (1)

1. A new product material distribution prediction method based on a digital twin shop generating an countermeasure network is characterized by comprising the following steps:
step 1, acquiring original data of a new product and similar products thereof in an actual production process by utilizing a digital twin system of a production workshop and preprocessing the original data;
step 1.1, normalizing the product state index data produced in the workshop in the original data to obtain normalized product state index data, and marking the normalized product state index data as P= [ P ] 1 p 2 p 3 ]Wherein p is 1 ，p 2 ，p 3 Respectively representing the input amount of raw materials, the output amount of products and the qualification rate of the products;
step 1.2, normalizing the workshop equipment state index data of the production unit i in the original data to obtain normalized workshop equipment state index data, and marking the normalized workshop equipment state index data as V i ＝[v i1 v i2 v i3 v i4 ]Wherein v is i1 ，v i2 ，v i3 ，v i4 Respectively representing the equipment running time, the equipment processing time, the equipment failure rate and the unit product energy consumption of the production unit i;
step 1.3, normalizing workshop personnel state index data of the production unit i in the original data to obtain normalized workshop personnel state index data, and marking the normalized workshop personnel state index data as H i ＝[h i1 h i2 h i3 h i4 ]Wherein h is i1 ，h i2 ，h i3 ，h i4 Respectively representing the working time, the number of personnel, the working efficiency and the absenteeism rate of the production unit i;
step 1.4, normalizing the workshop production environment state index data in the original data to obtain normalized workshop production environment state index data, which is recorded as E= [ E ] 1 e 2 e 3 ]Wherein e is 1 ，e 2 ，e 3 Respectively representing a thermal environment, an acoustic environment, and a light environment;
step 1.5, calculating the structural material demands of n parts of the new product in workshop production and normalizing according to the design parameters of the new product, thereby obtaining a normalized structural material demand matrix R= [ R ] of the new product 1 …r i …r n ]Wherein r is i The structure material requirement of the ith part of the new product when the production unit i is used for production is represented, and n is the total number of parts to be processed and produced in the new product;
step 1.6, dividing the production efficiency comprehensive evaluation index of the workshop into two grades, wherein the method comprises the following steps: the index of high production efficiency is 1 grade, and the index of low production efficiency is 0 grade;
step 1.7, normalizing the production takt data of n production units in the workshop in the original data, extracting data corresponding to an index with high workshop production efficiency, and obtaining normalized real production takt data, which is recorded as T= [ T ] 1 …t i …t n ]Wherein t is i Tact data representing the production unit i;
step 2, calculating a weight vector A= [ a ] of uncertainty influencing factors contained in the 4-type state index data according to the expert knowledge of the system 1 …a m ]Wherein a is m A weight indicating the mth state index, m=6+4×n×2 being the number of state indexes; thereby constructing the production state index data set B= [ P V ] 1 H 1 …V i H i …V n H n E]Weighting to obtain a weighted production state index data set Z=B.A;
step 3, constructing and generating an countermeasure network model, which comprises the following steps: generating a network G and a discrimination network D, wherein the generation network G and the discrimination network D are based on a multi-layer BP neural network;
step 4, counting part parameters of similar products, including: the part design and the process parameters of similar products are used for training a part similarity prediction network model based on a multi-layer BP neural network;
step 5, counting the part parameters of the new product, and utilizing the part similarity prediction network model to enter parts in the new productCalculating the line similarity to obtain a similarity matrix F= [ F ] of the new product part 1 …f i …f n ]Wherein f i Similarity of the ith part of the new product;
step 6, calculating predicted tact data y according to formula (1):
y＝G(z|R).*F (1)
in formula (1), G () represents generating a network, Z is a set of samples in the data set Z;
step 7, constructing a loss function V (D, G) for generating the countermeasure network model by using the formula (2):
in the formula (2), x is a group of takt data, x-P, among the real takt data T r Representing true tact data obeying probability distribution P r ，Expects true tact data; y-P g Representing predicted tact data compliance probability distribution P g ，/>Expected for predicting tact data;
step 8, a network G is fixedly generated, and a matrix x=t.r is used as an input of the discrimination network D, so that an output label of the discrimination network D is 1; meanwhile, taking predicted production takt data y as input of the discrimination network D, enabling an output label of the discrimination network D to be 0, and training and optimizing the discrimination network D so as to maximize a loss function V (D, G);
step 9, fixing a discrimination network D, taking predicted production beat data y as input of the discrimination network D, enabling an output label of the discrimination network D to be 1, and training and optimizing the generation network G so as to minimize a loss function V (D, G);
step (a)10. Training the generating network G and the judging network D alternately and iteratively according to the process of the step 6-the step 9 until the probability distribution P of the production beat data is predicted g Probability distribution P with real tact data r Equality, thus obtaining a trained generated countermeasure network model;
step 11, counting production state related data in the production process of new products, which comprises the following steps: product state index data P', plant state index data V of production unit i i ' workshop personnel status index data of production unit iThe workshop production environment state index data E ', a new product structure material demand matrix R ' and a trained generation network for generating an countermeasure network model are input, so that predicted production takt data y ' of a new product is obtained;
and step 12, calculating the material supply quantity and the material supply time of n production units in the digital twin workshop, and transmitting the predicted production takt data y' to corresponding equipment in the new product manufacturing production line so as to control the working rhythm of the corresponding equipment.
CN202111522679.7A 2021-12-13 2021-12-13 New product material distribution prediction method based on digital twin workshops for generating countermeasure network Active CN114418177B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111522679.7A CN114418177B (en) 2021-12-13 2021-12-13 New product material distribution prediction method based on digital twin workshops for generating countermeasure network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111522679.7A CN114418177B (en) 2021-12-13 2021-12-13 New product material distribution prediction method based on digital twin workshops for generating countermeasure network

Publications (2)

Publication Number Publication Date
CN114418177A CN114418177A (en) 2022-04-29
CN114418177B true CN114418177B (en) 2024-03-29

Family

ID=81264662

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111522679.7A Active CN114418177B (en) 2021-12-13 2021-12-13 New product material distribution prediction method based on digital twin workshops for generating countermeasure network

Country Status (1)

Country Link
CN (1) CN114418177B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115062945A (en) * 2022-06-08 2022-09-16 启明信息技术股份有限公司 Shutdown budget method based on discrete non-rigid production line station beat
CN116090702B (en) * 2023-01-18 2024-05-14 江苏盛泉环保科技发展有限公司 ERP data intelligent supervision system and method based on Internet of things
CN116560319A (en) * 2023-05-11 2023-08-08 山东山大华天软件有限公司 Intelligent production line visual prediction method and system based on production takt control

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112131782A (en) * 2020-08-27 2020-12-25 浙江大学 Multi-loop intelligent factory edge side digital twin scene coupling device
CN112183872A (en) * 2020-10-10 2021-01-05 东北大学 Blast furnace gas generation amount prediction method combining generation of countermeasure network and neural network
CN113505655A (en) * 2021-06-17 2021-10-15 电子科技大学 Bearing fault intelligent diagnosis method for digital twin system

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20210197335A1 (en) * 2019-12-26 2021-07-01 Dalian University Of Technology Data Augmentation Method Based On Generative Adversarial Networks In Tool Condition Monitoring

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112131782A (en) * 2020-08-27 2020-12-25 浙江大学 Multi-loop intelligent factory edge side digital twin scene coupling device
CN112183872A (en) * 2020-10-10 2021-01-05 东北大学 Blast furnace gas generation amount prediction method combining generation of countermeasure network and neural network
CN113505655A (en) * 2021-06-17 2021-10-15 电子科技大学 Bearing fault intelligent diagnosis method for digital twin system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于数字孪生的智能车间管控;刘义;刘晓冬;焦曼;叶春晖;周乐;;制造业自动化;20200725(第07期);148-152 *

Also Published As

Publication number Publication date
CN114418177A (en) 2022-04-29

Similar Documents

Publication Publication Date Title
CN114418177B (en) New product material distribution prediction method based on digital twin workshops for generating countermeasure network
CN112085261B (en) Enterprise production status diagnosis method based on cloud fusion and digital twin technology
US11755976B2 (en) Method for predicting oxygen load in iron and steel enterprises based on production plan
CN112800668A (en) Digital twin-based complex product operation, maintenance and manufacturing integrated overall model
CN106527373B (en) Workshop Autonomous Scheduling system and method based on multiple agent
CN107220734A (en) CNC Lathe Turning process Energy Consumption Prediction System based on decision tree
CN109255440B (en) Method for predictive maintenance of power production equipment based on Recurrent Neural Networks (RNN)
CN105574593B (en) Track state static detection and control system and method based on cloud computing and big data
CN114118673A (en) Workshop intelligent fault diagnosis early warning method based on digital twin technology
CN117389236B (en) Propylene oxide production process optimization method and system
CN115660293B (en) Comprehensive evaluation method for full life cycle of complex electromechanical product based on digital twin
CN108268970A (en) A kind of conveyer belt failure prediction method, system and device based on HTM
CN117709617A (en) MES-based intelligent scheduling system for production workshop
CN115718466A (en) Digital twin workshop fault prediction method based on random forest and analytic hierarchy process
CN114253242B (en) VPN-based cloud equipment data acquisition system for Internet of things
CN105913142B (en) Method for improving order completion period prediction accuracy by utilizing workshop RFID data
CN117952439A (en) Intelligent prediction method and system for road construction environment influence
CN113705897A (en) Product quality prediction method and system for industrial copper foil production
CN116976862A (en) Factory equipment informatization management system and method
CN111612227A (en) Load prediction method based on K-means clustering and bat optimization neural network
CN114926075B (en) Machine part production scheduling method based on man-hour prediction
CN116360339A (en) Numerical control machine tool reliability optimization method based on digital twin model of meta-action unit
CN113408764B (en) Product online technical state management and control system based on digital twins
CN115640980A (en) Power grid engineering cost dynamic management system based on target control
CN115392663A (en) Data acquisition and processing method based on big data

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