CN113033974A - Digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on improved LSTM network - Google Patents
Digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on improved LSTM network Download PDFInfo
- Publication number
- CN113033974A CN113033974A CN202110252202.5A CN202110252202A CN113033974A CN 113033974 A CN113033974 A CN 113033974A CN 202110252202 A CN202110252202 A CN 202110252202A CN 113033974 A CN113033974 A CN 113033974A
- Authority
- CN
- China
- Prior art keywords
- data
- steel coil
- model
- lstm network
- input data
- 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.)
- Granted
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
- G06Q10/06—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
- 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
- G06N3/044—Recurrent networks, e.g. Hopfield 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
-
- 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"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Software Systems (AREA)
- Development Economics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Entrepreneurship & Innovation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Probability & Statistics with Applications (AREA)
- Control Of Metal Rolling (AREA)
- General Factory Administration (AREA)
- Feedback Control In General (AREA)
Abstract
The invention discloses a digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on an improved LSTM network, belonging to the technical field of deep learning and comprising the steps of determining a steel coil manufacturing process and collecting data in the steel coil manufacturing process; dividing data into input data and output data according to a causal relationship; according to a deep learning principle, a deep learning model and a mechanism model are connected in series to construct a comprehensive model; training the comprehensive model according to the input data and the output data; the trained comprehensive model can predict the quality of the produced steel coil in real time; the steel coil quality predicted by the comprehensive model in real time has defects, and technological parameters are adjusted, so that the steel coil is prevented from having defects. The method can pre-judge the defects in advance and make up the defects in time, reduce the quality of the finally formed steel coil to the maximum and the fastest degree, and reduce the quality defects to the minimum.
Description
Technical Field
The invention relates to the technical field of deep learning, in particular to a digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on an improved LSTM network.
Background
At present, the quality of hot rolled coil can be predicted, but the dynamic adjustment of process parameters cannot be implemented, for example, the quality prediction is carried out by the following schemes:
CN 102033523A is a method for predicting the quality of band steel, early warning the furnace condition and diagnosing the fault based on partial least square. Firstly, judging the modulation degree of the currently produced strip steel, and calling model parameters of corresponding models; then, the collected data are subjected to standardization processing, the head and tail of the strip steel are respectively obtained and taken into an off-line model, and the hardness of the head and tail of the strip steel is predicted, so that the quality of the strip steel is predicted. The technology has the disadvantages that data obtained each time must be acquired on site, meanwhile, the model is an off-line model and does not have a learning function, and after the data change, the model is also a model which is essentially established by a least square algorithm and can only provide regression prediction, and the accuracy cannot necessarily reach the optimum.
CN110264079A the method mainly uses CNN algorithm in artificial neural network and Lasso regression model to predict the quality of hot-rolled products, the method cleans and models the data based on historical data, and substitutes the key input variable of training data into the feature vector model to obtain the input variable for substituting into Lasso regression model; and determining the optimal regularization factor of the Lasso regression model, and training the Lasso regression model by using the input variables in the S3 to obtain an unmodified hybrid prediction model for prediction. The method has the disadvantages that the CNN model does not have a time sequence prediction function, the CNN model is trained on the basis of historical data, and in addition, each set of data needs to be cleaned and normalized, so that the processing workload of the data is greatly improved, and the CNN model is not beneficial to quickly obtaining a set of proper models. Meanwhile, the model does not have a dynamic adjusting function and belongs to an off-line simulation model state.
Disclosure of Invention
Aiming at the defects existing in the problems, the invention provides a digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on an improved LSTM network.
In order to achieve the above object, the present invention provides a method for predicting the characteristic quality of a digital steel coil and dynamically adjusting process parameters based on an improved LSTM network, comprising:
determining a steel coil production manufacturing process, and collecting data in the steel coil production manufacturing process;
dividing the data into input data and output data according to a causal relationship;
according to a deep learning principle, a deep learning model and a mechanism model are connected in series to construct a comprehensive model;
training the comprehensive model according to the input data and the output data;
the trained comprehensive model can predict the quality of the produced steel coil in real time;
and the comprehensive model predicts the quality of the steel coil in real time to have defects, and adjusts and predicts process parameters to enable the steel coil to avoid the defects.
Preferably, the input data includes:
production process data of the unit: the speed of the unit, the tapping temperature of the steel plate, the temperature before rolling, the initial rolling temperature, the rolling force, the reduction stroke, the rolling pass, the cooling length and the coil diameter;
operating parameters of the plant: the stroke, the speed, the power, the vibration frequency, the rotating speed, the voltage and the power of a blooming mill, the temperature, the pressure and the current of a descaling device, the current, the voltage, the rotating speed and the power of a finishing mill of a heating furnace are measured;
the steel coil data: the length, width and height of an initial billet, the temperature of the initial billet in a heating furnace, the temperature of the initial billet out of the heating furnace, the temperature before the initial billet enters a rolling mill, the thickness, the length, the temperature of the initial billet out of the rolling mill, the thickness, the length, the speed of the initial billet in the rolling mill, the speed of the initial billet out of the rolling mill, the cooling temperature, the cooling time, the coiling temperature, the coiling time, the coiling length and;
the output data includes:
the quality defect data of the steel coil: defect type, defect detection time, defect classification, defect occurrence position, defect occurrence reason and defect tracking.
Preferably, the deep learning model is an improved LSTM network model, and the establishing step includes:
one-dimensional based LSTM network structure comprising a weight W for each of said input data1、W2……WnAnd an offset coefficient h1、h2……hn；
Superposing the one-dimensional LSTM network structure on the one-dimensional LSTM network structure, so that the LSTM network structure comprises a weight W11、W21…Wn1And an offset coefficient h11,h21…hn1；
Repeatedly superposing the one-dimensional LSTM network structure until the dimension N of the input two-dimensional data is matched, and obtaining an N-dimensional-1-dimensional LSTM network structure;
and connecting the output ends of the N-dimensional-1-dimensional LSTM network structure into a softmax function.
Preferably, according to the deep learning model, two-dimensional time series matrix data input to the deep learning model is constructed:
preferably, a data table is manufactured according to the steel coil manufacturing process, and the input data and the output data are correspondingly filled in the data table.
The step of training the integrated model according to the input data and the output data comprises:
preprocessing the input data and the output data, including removing abnormal data and data impurities;
carrying out standardization and normalization processing on the preprocessed input data;
respectively setting the weight of input data to the input data, namely the weight of process data of a unit as WD1The weight of the operating parameter of the device is WD2The weight of the steel coil data is WD3And the weight WD1、WD2、WD3∈(0,1)；
Based on the weight WD1、WD2、WD3And training the input data and the output data to obtain the comprehensive model.
Preferably, the steel coil manufacturing process sequentially comprises the following steps:
the method comprises the steps of continuously casting a plate blank, a heating furnace, discharging the plate blank, heating the edge part, removing phosphorus by high-pressure water, a side press, a roughing mill set, a heat preservation roller way, a crop flying shear, secondary phosphorus removal, a finishing mill set, laminar cooling, a coiling machine, bundling and weighing, and robot spray printing.
The invention also provides a digital steel coil characteristic quality prediction and process parameter dynamic adjustment system based on the improved LSTM network, which comprises the following steps:
the system comprises an acquisition module, a data processing module and a data processing module, wherein the acquisition module is used for determining a steel coil production manufacturing process, acquiring data in the steel coil production manufacturing process, and dividing the data into input data and output data according to a causal relationship;
the comprehensive model is used for establishing the comprehensive model by connecting the deep learning model and the mechanism model in series according to the deep learning principle; training the comprehensive model according to the input data and the output data; the trained comprehensive model can predict the quality of the produced steel coil in real time;
and the adjusting module is used for adjusting and predicting process parameters according to the steel coil quality defect predicted by the comprehensive model in real time, so that the steel coil is prevented from having the defect.
Compared with the prior art, the invention has the beneficial effects that:
according to the invention, the quality problem of the final product can be predicted in the rolling process through the comprehensive model, namely the product quality problem can be predicted in advance before the finished product is finished, namely the defect can be predicted in advance and compensated in time, the quality of the finally formed steel coil is reduced to the minimum in a maximum and most rapid manner.
Drawings
FIG. 1 is a schematic flow chart of a digital steel coil characteristic quality prediction and process parameter dynamic adjustment method based on an improved LSTM network in the invention;
FIG. 2 is a frame diagram of a digital steel coil characteristic quality prediction and process parameter dynamic adjustment system based on an improved LSTM network according to the present invention;
FIG. 3 is a schematic flow chart of a steel coil manufacturing process according to the present invention;
FIG. 4 is a block diagram of a deep learning model according to the present invention;
FIG. 5 is an architecture diagram of the digital steel coil characteristic quality prediction and process parameter dynamic adjustment system based on the improved LSTM network according to the present invention;
fig. 6 is a flow chart of the digital steel coil characteristic quality prediction and process parameter dynamic adjustment system based on the improved LSTM network in the present invention.
Reference numerals:
1. an acquisition module; 2. synthesizing the model; 3. and an adjusting module.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In the description of the present invention, it should also be noted that, unless otherwise explicitly specified or limited, the terms "mounted," "connected," and "connected" are to be construed broadly, e.g., as meaning either a fixed connection, a removable connection, or an integral connection; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
The invention provides a digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on an improved LSTM network.
Referring to fig. 1, the invention provides a digital steel coil characteristic quality prediction and process parameter dynamic adjustment method based on an improved LSTM network, comprising:
determining a steel coil production manufacturing process, and collecting data in the steel coil production manufacturing process;
specifically, as shown in fig. 3, the steps of the steel coil manufacturing process sequentially include:
continuous casting plate blank, heating furnace, plate blank discharging, edge heating, high-pressure water dephosphorization, side press, roughing mill group, heat preservation roller way, crop flying shear, secondary dephosphorization, finishing mill group, laminar cooling, coiling machine, bundling weighing and robot spray printing
Dividing data into input data and output data according to a causal relationship;
specifically, the input data includes:
the process data of the machine set are as follows: the speed of the unit, the tapping temperature of the steel plate, the temperature before rolling, the initial rolling temperature, the rolling force, the reduction stroke, the rolling pass, the cooling length and the coil diameter;
operating parameters of the plant: the stroke, the speed, the power, the vibration frequency, the rotating speed, the voltage and the power of a blooming mill, the temperature, the pressure and the current of a descaling device, the current, the voltage, the rotating speed and the power of a finishing mill of a heating furnace are measured;
the steel coil data: the length, width and height of an initial billet, the temperature of the initial billet in a heating furnace, the temperature of the initial billet out of the heating furnace, the temperature before the initial billet enters a rolling mill, the thickness, the length, the temperature of the initial billet out of the rolling mill, the thickness, the length, the speed of the initial billet in the rolling mill, the speed of the initial billet out of the rolling mill, the cooling temperature, the cooling time, the coiling temperature, the coiling time, the coiling length and;
the output data includes:
the quality defect data of the steel coil: defect type, defect detection time, defect classification, defect occurrence position, defect occurrence reason and defect tracking.
According to a deep learning principle, a deep learning model and a mechanism model are connected in series to construct a comprehensive model;
specifically, as shown in fig. 4, the deep learning model is an improved LSTM network model, and the establishing step includes:
one-dimensional LSTM-based network structure including a weight W for each input data1、W2……WnAnd an offset coefficient h1、h2……hn；
Superimposing a one-dimensional LSTM network structure on a one-dimensional LSTM network structure such that the LSTM network structure includes a weight W11、W21…Wn1And an offset coefficient h11,h21…hn1；
Repeatedly superposing the one-dimensional LSTM network structure until the dimension N of the input two-dimensional data is matched, and obtaining an N-dimensional-1-dimensional LSTM network structure;
and (4) connecting the output ends of the N-dimensional-1-dimensional LSTM network structure into the softmax function.
The input data in the deep learning model can be two-dimensional time sequence matrix data, so that the training speed of the network is greatly accelerated; the two-dimensional time sequence matrix data establishment comprises the following steps:
and according to the steel coil manufacturing process, manufacturing a data table, and correspondingly filling the input data and the output data into the data table.
Training the comprehensive model according to the input data and the output data;
the trained comprehensive model can predict the quality of the produced steel coil in real time;
the steel coil quality predicted by the comprehensive model in real time has defects, and the predicted process parameters are adjusted, so that the steel coil is prevented from having defects.
The step of training the integrated model based on the input data and the output data comprises:
preprocessing input data and output data, including removing abnormal data and data impurities;
carrying out standardization and normalization processing on the preprocessed input data;
separately set inputThe weight of the data on the input data, i.e. the process data of the plant, is WD1The operating parameters of the device are weighted by WD2The weight of the steel coil data is WD3And weight WD1、WD2、WD3∈(0,1)；
Based on weight WD1、WD2、WD3And training the input data and the output data to obtain a comprehensive model.
By the method, the final quality prediction result can be estimated through data of the steel coil in a certain section of the production line, and if certain quality defect behaviors, namely data changes, are found in a certain section of the production process, the finally generated quality problems caused by the defect behaviors can be known in advance according to the comprehensive model rules.
Because the model can be established in the rolling process to predict the quality problem of the final product, namely the quality problem of the product can be predicted in advance before the finished product is produced, when the parameter change of the quality defect is found, for example, the rolling force of a first pass rolling mill of a finishing mill cannot reach the expected target, and the rolling force of 200 kilograms is lacked, after the defect of the first rolling action is found, the rolling force of 200 kilograms is immediately and timely increased by using a finishing mill of a second rack for compensation, the quality of the final rolling is optimized, and the final finished steel coil is not allowed to generate an ultra-thick result. Therefore, the rolling defects can be pre-judged in advance and compensated in time according to the arrangement of the rolling mill, the quality of the finally formed steel coil is reduced to the maximum and the most rapid degree, and the quality defects are reduced to the minimum. Meanwhile, in the rolling process, point-to-point linkage data can be established between the data collected by the system and the final finished product quality, the linkage data is used as a basis for judging the quality in advance, a rolling experience model system based on a knowledge graph can be generated, the system can establish a rolling experience model for each type of steel coil through rolling conditions, rolling factors and the final quality, the model learns the data in each stage, and the formed experience data is used as a sediment of a knowledge base to guide the specific rolling behavior of the rolling mill. For example, for rolling under certain conditions, under the condition that the rolling experience model cannot be passed, rolling is stopped so as to effectively monitor and supervise the rolling behavior, and thus, the rolling behavior can be ensured to be always in the supervision and training learning state of the model so as to ensure that the quality of the final steel coil reaches the standard.
Referring to fig. 2, the present invention further provides a digital steel coil characteristic quality prediction and process parameter dynamic adjustment system based on the improved LSTM network, including:
the acquisition module 1 is used for determining a steel coil manufacturing process, acquiring data in the steel coil manufacturing process, and dividing the data into input data and output data according to a causal relationship;
the comprehensive model 2 is used for establishing a comprehensive model by connecting the deep learning model and the mechanism model in series according to a deep learning principle; training the comprehensive model according to the input data and the output data; the trained comprehensive model can predict the quality of the produced steel coil in real time;
and the adjusting module 3 is used for adjusting the input data according to the steel coil quality defect predicted by the comprehensive model in real time, so that the steel coil is prevented from having the defect.
Specifically, the system may construct a three-layer infrastructure, as shown in fig. 5, including an application layer, a model layer, and a data layer, where the adjustment module 3 is disposed on the application layer, the comprehensive model 2 is disposed on the model layer, and the acquisition module 1 is disposed on the data layer. For example, a data layer is provided with an acquisition device and a PLC, the PLC is connected with the acquisition device for acquiring each data through an OPC-UA protocol, and the PLC transmits the acquired data into a database through the OPC-UA protocol; the comprehensive model 2 is arranged in the model layer, the comprehensive model 2 extracts data in the database, and the steel coil quality is predicted according to the data; and a quality prediction system and a dynamic process parameter adjustment system are arranged on the application layer, the quality of the steel coil is predicted according to the quality prediction system, whether input data needs to be adjusted or not is judged, if the data needs to be adjusted, the dynamic process parameter adjustment system adjusts the process parameters, and the steps are repeated after adjustment until the adjustment is not needed, wherein the specific flow is shown in fig. 6.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes will occur to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (7)
1. A digital steel coil characteristic quality prediction and process parameter dynamic adjustment method based on an improved LSTM network is characterized by comprising the following steps:
determining a steel coil production manufacturing process, and collecting data in the steel coil production manufacturing process;
dividing the data into input data and output data according to a causal relationship;
according to a deep learning principle, a deep learning model and a mechanism model are connected in series to construct a comprehensive model;
training the comprehensive model according to the input data and the output data;
the trained comprehensive model can predict the quality of the produced steel coil in real time;
and the comprehensive model predicts the quality of the steel coil in real time to have defects, and adjusts and predicts process parameters to enable the steel coil to avoid the defects.
2. The method for digital steel coil characteristic quality prediction and dynamic process parameter adjustment based on the improved LSTM network as claimed in claim 1, wherein the input data includes:
production process data of the unit: the speed of the unit, the tapping temperature of the steel plate, the temperature before rolling, the initial rolling temperature, the rolling force, the reduction stroke, the rolling pass, the cooling length and the coil diameter;
operating parameters of the plant: the stroke, the speed, the power, the vibration frequency, the rotating speed, the voltage and the power of a blooming mill, the temperature, the pressure and the current of a descaling device, the current, the voltage, the rotating speed and the power of a finishing mill of a heating furnace are measured;
the steel coil data: the length, width and height of an initial billet, the temperature of the initial billet in a heating furnace, the temperature of the initial billet out of the heating furnace, the temperature before the initial billet enters a rolling mill, the thickness, the length, the temperature of the initial billet out of the rolling mill, the thickness, the length, the speed of the initial billet in the rolling mill, the speed of the initial billet out of the rolling mill, the cooling temperature, the cooling time, the coiling temperature, the coiling time, the coiling length and;
the output data includes:
the quality defect data of the steel coil: defect type, defect detection time, defect classification, defect occurrence position, defect occurrence reason and defect tracking.
3. The method for predicting the characteristic quality of the digital steel coil and dynamically adjusting the process parameters based on the improved LSTM network as claimed in claim 2, wherein the deep learning model is an improved LSTM network model, and the establishing step comprises:
one-dimensional based LSTM network structure comprising a weight W for each of said input data1、W2……WnAnd an offset coefficient h1、h2……hn；
Superposing the one-dimensional LSTM network structure on the one-dimensional LSTM network structure, so that the LSTM network structure comprises a weight W11、W21…Wn1And an offset coefficient h11,h21…hn1；
Repeatedly superposing the one-dimensional LSTM network structure until the dimension N of the input two-dimensional data is matched, and obtaining an N-dimensional-1-dimensional LSTM network structure;
and connecting the output ends of the N-dimensional-1-dimensional LSTM network structure into a softmax function.
4. The digital steel coil characteristic quality prediction and process parameter dynamic adjustment method based on the improved LSTM network as claimed in claim 3, wherein according to the deep learning model, two-dimensional time sequence matrix data input into the deep learning model is constructed:
and manufacturing a data table according to the steel coil manufacturing process, and correspondingly filling the input data and the output data into the data table.
5. The method for digital steel coil characteristic quality prediction and dynamic process parameter adjustment based on the improved LSTM network as claimed in claim 4, wherein the step of training the comprehensive model according to the input data and the output data comprises:
preprocessing the input data and the output data, including removing abnormal data and data impurities;
carrying out standardization and normalization processing on the preprocessed input data;
respectively setting the weight of input data to the input data, namely the weight of process data of a unit as WD1The weight of the operating parameter of the device is WD2The weight of the steel coil data is WD3And the weight WD1、WD2、WD3∈(0,1)；
Based on the weight WD1、WD2、WD3And training the input data and the output data to obtain the comprehensive model.
6. The digital steel coil characteristic quality prediction and process parameter dynamic adjustment method based on the improved LSTM network as claimed in claim 1, wherein the steps of the steel coil production process sequentially include:
the method comprises the steps of continuously casting a plate blank, a heating furnace, discharging the plate blank, heating the edge part, removing phosphorus by high-pressure water, a side press, a roughing mill set, a heat preservation roller way, a crop flying shear, secondary phosphorus removal, a finishing mill set, laminar cooling, a coiling machine, bundling and weighing, and robot spray printing.
7. The system for improving the digital steel coil characteristic quality prediction and process parameter dynamic adjustment method of the LSTM network according to any one of claims 1 to 6, comprising:
the system comprises an acquisition module, a data processing module and a data processing module, wherein the acquisition module is used for determining a steel coil production manufacturing process, acquiring data in the steel coil production manufacturing process, and dividing the data into input data and output data according to a causal relationship;
the comprehensive model is used for establishing the comprehensive model by connecting the deep learning model and the mechanism model in series according to the deep learning principle; training the comprehensive model according to the input data and the output data; the trained comprehensive model can predict the quality of the produced steel coil in real time;
and the adjusting module is used for adjusting and predicting process parameters according to the steel coil quality defect predicted by the comprehensive model in real time, so that the steel coil is prevented from having the defect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110252202.5A CN113033974B (en) | 2021-03-08 | 2021-03-08 | Digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on improved LSTM network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110252202.5A CN113033974B (en) | 2021-03-08 | 2021-03-08 | Digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on improved LSTM network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113033974A true CN113033974A (en) | 2021-06-25 |
| CN113033974B CN113033974B (en) | 2022-02-18 |
Family
ID=76466921
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110252202.5A Active CN113033974B (en) | 2021-03-08 | 2021-03-08 | Digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on improved LSTM network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113033974B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113753343A (en) * | 2021-07-23 | 2021-12-07 | 阳春新钢铁有限责任公司 | Finished product coil anti-extrusion deformation control system and method |
| CN114943415A (en) * | 2022-04-15 | 2022-08-26 | 上海波士内智能科技有限公司 | Metal welding defect root cause analysis method based on knowledge graph |
| CN115452842A (en) * | 2022-10-20 | 2022-12-09 | 颖态智能技术(上海)有限公司 | Fold detection method for valve bag packaging machine |
| CN116466657A (en) * | 2023-03-17 | 2023-07-21 | 浙江立群汽车配件制造有限公司 | Automatic production and processing control system for universal joint assembly |
| CN116579650A (en) * | 2023-04-28 | 2023-08-11 | 华院计算技术(上海)股份有限公司 | Continuous casting quality judging method and device, computer readable storage medium and terminal |
| CN118070983A (en) * | 2024-04-19 | 2024-05-24 | 苏州城市学院 | Industrial machinery production optimization method and system based on deep learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110174409A (en) * | 2019-06-14 | 2019-08-27 | 北京科技大学 | A kind of cut deal periodicity defect control method based on real-time detection result |
| CN111079906A (en) * | 2019-12-30 | 2020-04-28 | 燕山大学 | Cement product specific surface area prediction method and system based on long-time and short-time memory network |
| US20200166909A1 (en) * | 2018-11-20 | 2020-05-28 | Relativity Space, Inc. | Real-time adaptive control of manufacturing processes using machine learning |
| CN111235709A (en) * | 2020-03-18 | 2020-06-05 | 东华大学 | Online detection system for spun yarn evenness of ring spinning based on machine vision |
| CN111676365A (en) * | 2020-06-09 | 2020-09-18 | 邯郸钢铁集团有限责任公司 | LSTM-based method for predicting annealing temperature set value of transition steel coil |
-
2021
- 2021-03-08 CN CN202110252202.5A patent/CN113033974B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200166909A1 (en) * | 2018-11-20 | 2020-05-28 | Relativity Space, Inc. | Real-time adaptive control of manufacturing processes using machine learning |
| CN110174409A (en) * | 2019-06-14 | 2019-08-27 | 北京科技大学 | A kind of cut deal periodicity defect control method based on real-time detection result |
| CN111079906A (en) * | 2019-12-30 | 2020-04-28 | 燕山大学 | Cement product specific surface area prediction method and system based on long-time and short-time memory network |
| CN111235709A (en) * | 2020-03-18 | 2020-06-05 | 东华大学 | Online detection system for spun yarn evenness of ring spinning based on machine vision |
| CN111676365A (en) * | 2020-06-09 | 2020-09-18 | 邯郸钢铁集团有限责任公司 | LSTM-based method for predicting annealing temperature set value of transition steel coil |
Non-Patent Citations (2)
| Title |
|---|
| 王东城等: ""数据驱动的热轧带钢边部线状缺陷智能预报模型"", 《钢铁》 * |
| 路杨: ""基于时间序列的连退多目标动态鲁棒操作优化方法"", 《中国优秀硕士论文全文数据库 (基础科学辑)》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113753343A (en) * | 2021-07-23 | 2021-12-07 | 阳春新钢铁有限责任公司 | Finished product coil anti-extrusion deformation control system and method |
| CN114943415A (en) * | 2022-04-15 | 2022-08-26 | 上海波士内智能科技有限公司 | Metal welding defect root cause analysis method based on knowledge graph |
| CN115452842A (en) * | 2022-10-20 | 2022-12-09 | 颖态智能技术(上海)有限公司 | Fold detection method for valve bag packaging machine |
| CN116466657A (en) * | 2023-03-17 | 2023-07-21 | 浙江立群汽车配件制造有限公司 | Automatic production and processing control system for universal joint assembly |
| CN116466657B (en) * | 2023-03-17 | 2023-09-19 | 浙江立群汽车配件制造有限公司 | Automatic production and processing control system for universal joint assembly |
| CN116579650A (en) * | 2023-04-28 | 2023-08-11 | 华院计算技术(上海)股份有限公司 | Continuous casting quality judging method and device, computer readable storage medium and terminal |
| CN116579650B (en) * | 2023-04-28 | 2024-04-26 | 华院计算技术(上海)股份有限公司 | Continuous casting quality judging method and device, computer readable storage medium and terminal |
| CN118070983A (en) * | 2024-04-19 | 2024-05-24 | 苏州城市学院 | Industrial machinery production optimization method and system based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113033974B (en) | 2022-02-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113033974B (en) | Digital steel coil characteristic quality prediction and process parameter dynamic adjustment method and system based on improved LSTM network | |
| CN107179749A (en) | Hot dip zinc product whole process method of quality control | |
| US7050875B2 (en) | System and method for detecting an anomalous condition | |
| CN112418538A (en) | Continuous casting billet inclusion prediction method based on random forest classification | |
| CN109675935B (en) | Online fault diagnosis method for IPCA rolling process with variable control limit | |
| CN102581244B (en) | Online control system and control method for surface quality of continuous casting billet | |
| US6546310B1 (en) | Process and device for controlling a metallurgical plant | |
| CN112100745A (en) | Automobile girder steel mechanical property prediction method based on LDA theory | |
| US20050004684A1 (en) | System and method for adjusting a control model | |
| CN106649755A (en) | Threshold self-adaption setting abnormity detection method for multi-dimensional real-time power transformation device data | |
| CN115358281B (en) | Machine learning-based cold and hot all-in-one machine monitoring control method and system | |
| CN106845826B (en) | PCA-Cpk-based cold continuous rolling production line service quality state evaluation method | |
| JP2023086616A (en) | Crack determination method of high-temperature steel, rolling method of high-temperature steel, and generation method of crack determination model of high-temperature steel | |
| CN115169453A (en) | Hot continuous rolling width prediction method based on density clustering and depth residual error network | |
| CN116346870B (en) | Industrial Internet of things system based on real-time monitoring of processing quality and control method | |
| US6980874B2 (en) | System and method for detecting an anomalous condition in a multi-step process | |
| CN117548911A (en) | Connector welding intelligent supervision system based on artificial intelligence | |
| CN113435794B (en) | Intelligent monitoring method for post-processing of nozzle casting based on image processing | |
| CN115106384B (en) | Thick plate rolling roll gap correction method based on random forest | |
| CN110174409B (en) | Medium plate periodic defect control method based on real-time detection result | |
| CN117171936A (en) | Slab quality prediction method for extracting real-time characteristic value of crystallizer based on defect mechanism | |
| CN202639268U (en) | Online control system for surface quality of continuous casting billet | |
| CN115446276B (en) | Continuous casting steel leakage early warning method based on convolutional neural network recognition crystallizer copper plate V-shaped bonding characteristics | |
| CN117472001A (en) | Double-rack leveling multi-parameter forecasting method and device | |
| CN117974069B (en) | Method, equipment and medium for formulating production data acquisition scheme of injection bottle |
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 |