CN111915054A - Offline neural network-based rewinding elongation rate self-adaptive optimization method - Google Patents
Offline neural network-based rewinding elongation rate self-adaptive optimization method Download PDFInfo
- Publication number
- CN111915054A CN111915054A CN202010512698.0A CN202010512698A CN111915054A CN 111915054 A CN111915054 A CN 111915054A CN 202010512698 A CN202010512698 A CN 202010512698A CN 111915054 A CN111915054 A CN 111915054A
- Authority
- CN
- China
- Prior art keywords
- steel
- neural network
- steel coil
- coil
- elongation
- 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.)
- Pending
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 75
- 238000005457 optimization Methods 0.000 title claims abstract description 19
- 238000000034 method Methods 0.000 title claims abstract description 17
- 229910000831 Steel Inorganic materials 0.000 claims abstract description 153
- 239000010959 steel Substances 0.000 claims abstract description 153
- 238000004519 manufacturing process Methods 0.000 claims abstract description 29
- 230000003044 adaptive effect Effects 0.000 claims abstract description 7
- 230000006870 function Effects 0.000 claims description 25
- 239000002436 steel type Substances 0.000 claims description 24
- 238000012546 transfer Methods 0.000 claims description 15
- 238000012549 training Methods 0.000 claims description 8
- 238000003062 neural network model Methods 0.000 claims description 5
- 238000004801 process automation Methods 0.000 claims description 5
- 238000005097 cold rolling Methods 0.000 abstract description 13
- 238000004364 calculation method Methods 0.000 abstract description 11
- 238000005272 metallurgy Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 241000277275 Oncorhynchus mykiss Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 238000003466 welding Methods 0.000 description 1
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"
-
- 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
- G06N3/048—Activation functions
-
- 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
-
- 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
- G06Q10/06395—Quality analysis or management
-
- 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
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Operations Research (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Educational Administration (AREA)
- Life Sciences & Earth Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
- Manufacturing & Machinery (AREA)
- Winding, Rewinding, Material Storage Devices (AREA)
- Control Of Metal Rolling (AREA)
Abstract
The invention relates to a rewinding elongation self-adaptive optimization method based on an offline neural network, and belongs to the technical field of metallurgy automation. The technical scheme is as follows: the method comprises the steps of selecting a steel group number, the thickness of a steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel which have influences on the elongation of a tension leveler of a recoiling unit as input layer nodes of a neural network, selecting the elongation of the tension leveler of the recoiling unit as output layer nodes of the neural network, and performing off-line optimization on a tension leveler elongation calculation model of the recoiling unit to realize off-line adaptive control on the elongation of the tension leveler of the steel coil tension leveler of a cold-rolling recoiling production line. The method ensures the calculation accuracy of the elongation set value of the steel coil straightening and withdrawal machine of the cold-rolling heavy coil production line, thereby ensuring the quality of the finished steel coil of the cold-rolling heavy coil production line.
Description
Technical Field
The invention relates to a rewinding elongation self-adaptive optimization method based on an offline neural network, and belongs to the technical field of metallurgy automation.
Background
The heavy coil production line is an important component in the modern strip steel production, and is unfolded by an uncoiler and sent to a head cutting shear to cut off the part of the strip steel head and tail which does not meet the requirements. The strip head is sent to a welding machine to be welded with the last strip tail, the plate shape is improved through a straightening machine and a tension straightening machine, and the edge is cut through a circle shear. And then the strip steel arrives at a strip steel inspection station for surface inspection. And then the strip steel is sent to an electrostatic oiling machine for oiling treatment, and then is sent to a coiling machine for coiling. And when the coil weight reaches a specified value, dividing the coil by slitting. The calculation accuracy of the elongation rate set value of the withdrawal and straightening unit of the heavy coil unit directly influences the quality of the finished steel coil of the heavy coil production line. If the elongation set value of the withdrawal and straightening machine is too large, the internal structure of the finished steel coil and the performance of the finished steel coil are easily influenced. On the contrary, if the elongation set value of the withdrawal and straightening machine is too small, the wave-shaped defect of the steel coil cannot be eliminated. The elongation of the withdrawal and straightening machine in the prior art is a range value obtained through manual calculation, has no self-adaptive optimization capability, has higher time cost and labor cost, and directly influences the qualification rate of finished steel coils.
Disclosure of Invention
The invention aims to provide an off-line neural network-based recoiling elongation rate self-adaptive optimization method, which adopts an off-line neural network control mode, selects a steel group number, the thickness of a steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel which have influence on the elongation rate of a withdrawal straightening machine of a recoiling unit as input layer nodes of a neural network, selects the elongation rate of the withdrawal straightening machine of the recoiling unit as output layer nodes of the neural network, adopts the off-line neural network control mode, selects qualified steel coil data produced in a period, the calculation model of the elongation of the tension leveler of the recoiling unit is optimized off-line, the off-line adaptive control of the elongation of the steel coil tension leveler of the cold-rolling heavy coil production line is realized, the calculation accuracy of the elongation set value of the steel coil tension leveler of the cold-rolling heavy coil production line is ensured, thereby ensuring the quality of the finished steel coil of the cold rolling heavy coil production line and effectively solving the problems in the background technology.
The technical scheme of the invention is as follows: a rewinding elongation rate self-adaptive optimization method based on an offline neural network comprises the following steps:
step a, determining a network structure and a network layer number of a neural network;
b, selecting qualified steel coil data of the heavy coil production line, grouping the steel coils according to steel types according to actual conditions, dividing the steel types with similar performance into a group, numbering, and selecting the data of the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil, the speed of strip steel, the elongation of a tension leveler of a heavy coil unit and the like of the steel coils grouped in each group;
step c, defining a network input layer node and a network output layer node of the neural network, wherein the input layer node of the neural network selects a steel type group number, the thickness of a steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel, and the output layer node of the neural network selects the elongation of a tension leveler of a recoiling unit;
d, determining a transfer function of the neural network, selecting a Sigmoid function as the transfer function, setting the learning rate to be 0.01, and selecting smaller random numbers as initial weights of an input layer and an output layer;
and e, adopting an off-line self-adaptive training mode, setting a learning period to be 7 days, selecting 100 groups of steel coil data in the step b every time, taking the steel type group number, the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel as input layer nodes, taking the elongation of a tension leveler of the recoiling unit as output layer nodes, carrying out self-adaptive learning training on the neural network, storing the optimized neural network model into a recoiling process automation system, and carrying out production application in the next production.
In the step a, a three-layer neural network structure is adopted.
In the step b, according to the actual situation of the heavy coil production line, steel grades are grouped according to performance, the steel grades with similar performance are divided into one group, the steel grades are sequentially numbered, and 20 groups of steel grades are grouped in total. And then selecting the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil, the speed of the strip steel and the elongation of the withdrawal straightening machine of the recoiling unit of each group of steel type grouping.
In the step d, the hidden layer transfer function of the neural network selects a symmetric Sigmoid function with positive and negative symmetry:
the transfer function of the output layer of the neural network selects a non-negative Sigmoid function:
the learning rate of the neural network is set to 0.01, and a smaller value is randomly selected as an initial weight of an input layer and an output layer of the neural network, such as 0.1256 or 0.3436.
The invention has the beneficial effects that: the method comprises the steps of selecting a steel group number, the thickness of a steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel which have influences on the elongation of a tension leveler of a recoiling unit as input layer nodes of a neural network, selecting the elongation of the tension leveler of the recoiling unit as output layer nodes of the neural network, selecting qualified steel coil data according to periods by adopting an off-line neural network control mode, carrying out off-line optimization on a tension leveler elongation calculation model of the recoiling unit, realizing off-line adaptive control on the elongation of the tension leveler of the steel coil of a cold-rolling recoiling production line, and ensuring the calculation accuracy of the elongation set value of the steel coil tension leveler of the cold-rolling recoiling production line so as to ensure the quality of a finished cold-rolling recoiling.
Drawings
FIG. 1 is a diagram of an off-line neural network model architecture according to an embodiment of the present invention;
FIG. 2 is a control flow diagram of an embodiment of the present invention;
FIG. 3 is a schematic diagram illustrating the network error optimization effect according to an embodiment of the present invention;
in fig. 1: the steel coil is divided into a steel group number 1, a steel coil width 2, a steel coil thickness 3, a steel coil inner diameter 4, a steel coil density 5, a strip steel speed 6, a hidden layer 7 and an elongation rate 8.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions of the embodiments of the present invention with reference to the drawings of the embodiments, and it is obvious that the described embodiments are a small part of the embodiments of the present invention, rather than all embodiments, and all other embodiments obtained by a person of ordinary skill in the art without creative work based on the embodiments of the present invention belong to the protection scope of the present invention.
A rewinding elongation rate self-adaptive optimization method based on an offline neural network comprises the following steps:
step a, determining a network structure and a network layer number of a neural network;
b, selecting qualified steel coil data of the heavy coil production line, grouping the steel coils according to steel types according to actual conditions, dividing the steel types with similar performance into a group, numbering, and selecting the data of the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil, the speed of strip steel, the elongation of a tension leveler of a heavy coil unit and the like of the steel coils grouped in each group;
step c, defining a network input layer node and a network output layer node of the neural network, wherein the input layer node of the neural network selects a steel type group number, the thickness of a steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel, and the output layer node of the neural network selects the elongation of a tension leveler of a recoiling unit;
d, determining a transfer function of the neural network, selecting a Sigmoid function as the transfer function, setting the learning rate to be 0.01, and selecting smaller random numbers as initial weights of an input layer and an output layer;
and e, adopting an off-line self-adaptive training mode, setting a learning period to be 7 days, selecting 100 groups of steel coil data in the step b every time, taking the steel type group number, the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel as input layer nodes, taking the elongation of a tension leveler of the recoiling unit as output layer nodes, carrying out self-adaptive learning training on the neural network, storing the optimized neural network model into a recoiling process automation system, and carrying out production application in the next production.
In the step a, a three-layer neural network structure is adopted.
In the step b, according to the actual situation of the heavy coil production line, steel grades are grouped according to performance, the steel grades with similar performance are divided into one group, the steel grades are sequentially numbered, and 20 groups of steel grades are grouped in total. And then selecting the data of the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil, the speed of the strip steel, the elongation of a tension leveler of the recoiling unit and the like of each group of steel type grouping.
In the step d, the hidden layer transfer function of the neural network selects a symmetric Sigmoid function with positive and negative symmetry:
the transfer function of the output layer of the neural network selects a non-negative Sigmoid function:
the learning rate of the neural network is set to 0.01, and a smaller value is randomly selected as an initial weight of an input layer and an output layer of the neural network, such as 0.1256 or 0.3436.
Examples
An off-line neural network is adopted to establish a tension leveler elongation rate control model, as shown in fig. 1-3, and C # programming is applied to development and implementation in a rewinding process automation system, and the method comprises the following steps:
step a, setting a network structure and a network layer number of a neural network, wherein the invention adopts a three-layer neural network structure; the main codes are as follows:
Int BPNN
IN=6;H=4;Out=1;
In1=nn1;In2=nn2;In3=nn3; In4=nn4; In5=nn5;In6=nn6;
Int U=[x];
In1=0; In2=0; In3=0; In4=0; In5=0; In6=0;
Int OUT =[0];
OUT=0;
end;
b, selecting qualified steel coil data of the heavy coil production line, grouping the steel coils according to steel types, dividing the steel types with similar performance into a group for numbering, and dividing the group into 20 groups in total, and selecting the data of the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil, the elongation of a tension leveler of the heavy coil unit and the like of each group of steel types; the main codes are as follows:
Int U=[i];
U[i]=[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20];
string str = "select * from PDO";
DataTable dt1 = new DataTable();
DB.FillDataTable(str,dt1);
end;
step c, defining a network input layer node and a network output layer node of the neural network, wherein the input layer node of the neural network selects a steel type group number, the thickness of a steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel, and the output layer node of the neural network selects the elongation of a tension leveler of a recoiling unit; the main codes are as follows:
In1=U[i];
Dt1= Steel Grade Grouping;
Dt2= Coil thickness;
Dt3= Coil width;
Dt4= Coil diameter;
Dt5= Coil density;
Dt6= Strip speed;
Yout= Elongation ratio;
and d, determining a transfer function of the neural network, selecting a Sigmoid function by the transfer function, setting the learning rate to be 0.01, selecting a smaller random number from the initial weights of the input layer and the output layer, wherein the smaller random number is a randomly selected smaller value, such as 0.3215. (ii) a The main codes are as follows:
Int Learning rate=0.01;
Wi=[0.2352 0.4335 0.6732 -0.2368 -0.2474
-0.4377 0.3486 0.4638 0.4755 0.9753
0.2356 0.2362 0.4367 0.2356 0.2362
0.5456 0.4372 0.7442 0.2356 0.2367
0.2352 0.2362 0.5485 0.6453 0.7452
0.3462 0.4326 0.5324 0.5474 0.4632];
w1=wi1;w2=wi2;w3=wi3; w4=wi4; w5=wi5; w6=wi6;
Wo=[-0.2462 -0.5473 -0.2362 -0.8457]
W7=w01;
int error=0;
x(1)=error(x)-error_1;
x(2)=error(x);
x(3)=error(x)-2*error_1+error_2;
error(x)=[x(1);x(2);x(3)];
F(x)=wi*wo;
R(t)=(exp(Y(t))-exp(-Y(t)))/(exp(h(t))+exp(-Y(t)));
U(x)=1.2*(1-0.8*exp(-0.01*t));
for Y=1:1:h(t)
H(t)=exp(h(t))/(exp(h(t))+exp(-h(t)));
h(t)1=x(1); h(t)2=x(2); h(t)3=x(3);
h(t)4=x(4); h(t)5=x(5); h(t)6=x(6);
Y=[ x(1), x(2), x(3), x(4), x(5), x(6)];
Yout(t)=sign((yout(x)-y_1)/(h(t)-dt*0.01));
end;
step e, adopting an off-line self-adaptive training mode, setting a learning period to be 7 days, selecting 100 groups of steel coil data in the step b each time, taking the steel type group number, the steel coil thickness, the steel coil width, the steel coil inner diameter, the steel coil density and the strip steel speed as input layer nodes, taking the elongation of a tension leveler of a recoiling unit as output layer nodes, carrying out self-adaptive learning training on a neural network, storing an optimized neural network model into a recoiling process automation system, and carrying out production application in the next production; the main codes are as follows:
Int Period=7;
Int PDO Coil data [x]=100;
string str = "select * from PDO";
DataTable dt1 = new DataTable();
error2=error_1;
error_1=error(x);
U(x)=2/(exp(h(t))+exp(-h(t)))^3;
for H=1:1:In(x);
yout(x)=error(x)*efs(x)*rate(t)*U(x);
for X=1:1:Out;
for k=1:1:In;
dwi=refs* exp 3(x)*F(x)+ exp *(wi1*wi2*wi3);
wo=rate *(wo1*wo2*wo3* wo4*wo5*wo6);
for x=1:1:out;
dh(x)=4/(exp(H(x))+exp(-H(x)))^2;
refs= rate2*wi;
h(x)=d(x)*rate(x);
wi=xite* rate*wi;
wi=rate*(wi1*wi2*wi3*wi1*wi2*wi3);
yout(t)=y(x);
w1=wi1;w2=wi2;w3=wi3; w4=wi4; w5=wi5; w6=wi6;
W7=wo1;
end;
the invention is characterized in that: 1. and selecting qualified steel coil data according to periods by adopting an off-line neural network control mode, and carrying out off-line optimization on the tension leveler elongation calculation model of the recoiling unit. 2. The steel coil performance data is added into the neural network control model, so that the elongation optimization result is more accurate. 3. The transportability is strong, and other cold rolling heavy coil production lines can be universally transplanted.
The method comprises the steps of network structure composition of an offline neural network control model, network input layer node selection of the neural network control model, output layer node selection of the neural network control model, transfer function setting of the neural network control model, learning rate setting of the neural network control model, performance index selection of the neural network control model, input layer data transmission of the neural network control model, output layer data transmission of the neural network control model and self-adaptive optimization of the neural network control model. The method comprises the steps of selecting a steel group number, the thickness of a steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel which have influences on the elongation of a tension leveler of a recoiling unit as input layer nodes of a neural network, selecting the elongation of the tension leveler of the recoiling unit as output layer nodes of the neural network, selecting qualified steel coil data according to periods by adopting an off-line neural network control mode, carrying out off-line optimization on a tension leveler elongation calculation model of the recoiling unit, realizing off-line adaptive control on the elongation of the tension leveler of the steel coil of a cold-rolling recoiling production line, and ensuring the calculation accuracy of the elongation set value of the steel coil tension leveler of the cold-rolling recoiling production line so as to ensure the quality of a finished cold-rolling recoiling.
Claims (4)
1. A rewinding elongation rate self-adaptive optimization method based on an offline neural network is characterized by comprising the following steps:
step a, determining a network structure and a network layer number of a neural network;
b, selecting qualified steel coil data of the heavy coil production line, grouping the steel coils according to steel types according to actual conditions, dividing the steel types with similar performance into a group, numbering, and selecting the data of the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil, the speed of strip steel, the elongation of a tension leveler of a heavy coil unit and the like of the steel coils grouped in each group;
step c, defining a network input layer node and a network output layer node of the neural network, wherein the input layer node of the neural network selects a steel type group number, the thickness of a steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel, and the output layer node of the neural network selects the elongation of a tension leveler of a recoiling unit;
d, determining a transfer function of the neural network, selecting a Sigmoid function as the transfer function, setting the learning rate to be 0.01, and selecting smaller random numbers as initial weights of an input layer and an output layer;
and e, adopting an off-line self-adaptive training mode, setting a learning period to be 7 days, selecting 100 groups of steel coil data in the step b every time, taking the steel type group number, the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil and the speed of strip steel as input layer nodes, taking the elongation of a tension leveler of the recoiling unit as output layer nodes, carrying out self-adaptive learning training on the neural network, storing the optimized neural network model into a recoiling process automation system, and carrying out production application in the next production.
2. The offline neural network-based rewinding elongation rate adaptive optimization method according to claim 1, wherein: in the step a, a three-layer neural network structure is adopted.
3. The offline neural network-based rewinding elongation rate adaptive optimization method according to claim 1, wherein: in the step b, according to the actual situation of the heavy coil production line, steel types are grouped according to performances, the steel types with similar performances are divided into a group, the steel types are sequentially numbered, 20 groups of steel types are grouped in total, and then the thickness of the steel coil, the width of the steel coil, the inner diameter of the steel coil, the density of the steel coil, the speed of the strip steel and the elongation of the tension leveler of the heavy coil unit of each group of steel types are selected.
4. The offline neural network-based rewinding elongation rate adaptive optimization method according to claim 1, wherein: in the step d, the hidden layer transfer function of the neural network selects a symmetric Sigmoid function with positive and negative symmetry:
the transfer function of the output layer of the neural network selects a non-negative Sigmoid function:
the learning rate of the neural network is set to 0.01, and a smaller value is randomly selected as an initial weight of an input layer and an output layer of the neural network, such as 0.1256 or 0.3436.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010512698.0A CN111915054A (en) | 2020-06-08 | 2020-06-08 | Offline neural network-based rewinding elongation rate self-adaptive optimization method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010512698.0A CN111915054A (en) | 2020-06-08 | 2020-06-08 | Offline neural network-based rewinding elongation rate self-adaptive optimization method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111915054A true CN111915054A (en) | 2020-11-10 |
Family
ID=73237641
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010512698.0A Pending CN111915054A (en) | 2020-06-08 | 2020-06-08 | Offline neural network-based rewinding elongation rate self-adaptive optimization method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111915054A (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101618402A (en) * | 2008-06-30 | 2010-01-06 | 宝山钢铁股份有限公司 | Method for controlling planeness of cold-rolling strip steel |
| CN104942019A (en) * | 2014-03-31 | 2015-09-30 | 宝山钢铁股份有限公司 | Automatic control method for width of steel strips during cold rolling |
| CN106353196A (en) * | 2016-10-31 | 2017-01-25 | 中冶赛迪工程技术股份有限公司 | Metal strip stretching and straightening deformation experimental device and experimental method |
| CN106424170A (en) * | 2016-09-28 | 2017-02-22 | 邯钢集团邯宝钢铁有限公司 | Pickling line tandem cold mill withdrawal and straightening machine control method |
| CN108637020A (en) * | 2018-05-09 | 2018-10-12 | 北京科技大学 | A kind of TSP question PSO-BP neural networks strip profile prediction technique |
| CN109034390A (en) * | 2018-08-07 | 2018-12-18 | 河北工业大学 | Phase angular amplitude PID adaptive approach based on BP neural network Three-Dimensional Magnetic feature measurement |
| CN110266228A (en) * | 2019-07-05 | 2019-09-20 | 长安大学 | Surface permanent magnetic Synchronous Machine Models forecast Control Algorithm based on BP neural network |
| CN110814050A (en) * | 2019-10-24 | 2020-02-21 | 唐山钢铁集团有限责任公司 | Rolling mill model control method based on BP neural network |
-
2020
- 2020-06-08 CN CN202010512698.0A patent/CN111915054A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101618402A (en) * | 2008-06-30 | 2010-01-06 | 宝山钢铁股份有限公司 | Method for controlling planeness of cold-rolling strip steel |
| CN104942019A (en) * | 2014-03-31 | 2015-09-30 | 宝山钢铁股份有限公司 | Automatic control method for width of steel strips during cold rolling |
| CN106424170A (en) * | 2016-09-28 | 2017-02-22 | 邯钢集团邯宝钢铁有限公司 | Pickling line tandem cold mill withdrawal and straightening machine control method |
| CN106353196A (en) * | 2016-10-31 | 2017-01-25 | 中冶赛迪工程技术股份有限公司 | Metal strip stretching and straightening deformation experimental device and experimental method |
| CN108637020A (en) * | 2018-05-09 | 2018-10-12 | 北京科技大学 | A kind of TSP question PSO-BP neural networks strip profile prediction technique |
| CN109034390A (en) * | 2018-08-07 | 2018-12-18 | 河北工业大学 | Phase angular amplitude PID adaptive approach based on BP neural network Three-Dimensional Magnetic feature measurement |
| CN110266228A (en) * | 2019-07-05 | 2019-09-20 | 长安大学 | Surface permanent magnetic Synchronous Machine Models forecast Control Algorithm based on BP neural network |
| CN110814050A (en) * | 2019-10-24 | 2020-02-21 | 唐山钢铁集团有限责任公司 | Rolling mill model control method based on BP neural network |
Non-Patent Citations (1)
| Title |
|---|
| 马伟然等: "《基于神经网络的拉矫机控制模型建立》", 《重工机械》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108637020B (en) | Self-adaptive variation PSO-BP neural network strip steel convexity prediction method | |
| CN112884239A (en) | Aerospace detonator production scheduling method based on deep reinforcement learning | |
| CN107694337A (en) | Coal unit SCR denitrating flue gas control methods based on network response surface | |
| CN108108855B (en) | Conveying line path planning method | |
| CN111160755B (en) | Real-time scheduling method for aircraft overhaul workshop based on DQN | |
| CN111915054A (en) | Offline neural network-based rewinding elongation rate self-adaptive optimization method | |
| CN112836974B (en) | Dynamic scheduling method for multiple field bridges between boxes based on DQN and MCTS | |
| CN112446130A (en) | Strip steel deviation simulation system of continuous hot galvanizing unit annealing furnace and control method | |
| CN111400870A (en) | Heavy coil production line tension group model calculation method based on BP neural network | |
| CN111401611B (en) | Route optimization method for routing inspection point of chemical plant equipment | |
| CN110064653A (en) | A kind of method and apparatus applied to the rolling of cold continuous rolling changeover portion | |
| CN106865418B (en) | A kind of control method of coil of strip reservoir area loop wheel machine equipment | |
| CN105128893B (en) | The generation method and system of a kind of whole time of running information of train | |
| CN111459014A (en) | Neural network PID-based crown block swing angle model control method | |
| CN112941298A (en) | Method for optimizing and calculating production time of steel coil in cold rolling process and application thereof | |
| CN110880049A (en) | Regional multi-wind-field operation and maintenance scheduling method and system | |
| Ledet et al. | The manufacturing game | |
| Naples | Business failures and the expenditure multiplier, or how recessions become depressions | |
| CN114936778B (en) | Component job shop scheduling method and device | |
| CN104014597B (en) | For the section cooling method of hot continuous rolling | |
| Pearson | POLICY, HISTORY, THEORY: Labour and the environment: An historical perspective | |
| Pozniak | Plug the Gap: Retrain for Net Zero [climate change-careers] | |
| CN107341325A (en) | A kind of discrete event system suboptimum monitoring controller generation method | |
| Verzier | Nothing Is Automatic: Producing More‐Than‐Human Relations in the Pearl River Delta | |
| Jin et al. | Chaotic salp swarm algorithm: Application to parameter identification for MIMO Hammerstein model under heavy tail noise |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201110 |