CN112560337B - Intelligent modeling method, device, equipment and storage medium for digital twin system of complex industrial process - Google Patents
Intelligent modeling method, device, equipment and storage medium for digital twin system of complex industrial process Download PDFInfo
- Publication number
- CN112560337B CN112560337B CN202011435289.1A CN202011435289A CN112560337B CN 112560337 B CN112560337 B CN 112560337B CN 202011435289 A CN202011435289 A CN 202011435289A CN 112560337 B CN112560337 B CN 112560337B
- Authority
- CN
- China
- Prior art keywords
- model
- deep learning
- parameters
- industrial process
- intelligent
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 88
- 238000000034 method Methods 0.000 title claims abstract description 73
- 238000003860 storage Methods 0.000 title claims abstract description 7
- 238000013135 deep learning Methods 0.000 claims abstract description 67
- 230000007246 mechanism Effects 0.000 claims abstract description 51
- 238000005312 nonlinear dynamic Methods 0.000 claims abstract description 47
- 238000013136 deep learning model Methods 0.000 claims description 122
- 238000012937 correction Methods 0.000 claims description 87
- CPLXHLVBOLITMK-UHFFFAOYSA-N Magnesium oxide Chemical compound [Mg]=O CPLXHLVBOLITMK-UHFFFAOYSA-N 0.000 claims description 44
- 230000008569 process Effects 0.000 claims description 29
- 238000003723 Smelting Methods 0.000 claims description 28
- 239000000395 magnesium oxide Substances 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 14
- 238000010276 construction Methods 0.000 claims description 5
- 238000004886 process control Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 description 21
- 230000006870 function Effects 0.000 description 19
- 230000008859 change Effects 0.000 description 9
- 238000013528 artificial neural network Methods 0.000 description 6
- 238000010168 coupling process Methods 0.000 description 6
- 238000005859 coupling reaction Methods 0.000 description 6
- 238000002844 melting Methods 0.000 description 6
- 230000008018 melting Effects 0.000 description 6
- 230000008878 coupling Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 210000004027 cell Anatomy 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 4
- 239000012535 impurity Substances 0.000 description 3
- 239000002245 particle Substances 0.000 description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 2
- CLOMYZFHNHFSIQ-UHFFFAOYSA-N clonixin Chemical compound CC1=C(Cl)C=CC=C1NC1=NC=CC=C1C(O)=O CLOMYZFHNHFSIQ-UHFFFAOYSA-N 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 239000000155 melt Substances 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000008278 dynamic mechanism Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 229910052742 iron Inorganic materials 0.000 description 1
- 239000001095 magnesium carbonate Substances 0.000 description 1
- ZLNQQNXFFQJAID-UHFFFAOYSA-L magnesium carbonate Chemical compound [Mg+2].[O-]C([O-])=O ZLNQQNXFFQJAID-UHFFFAOYSA-L 0.000 description 1
- 229910000021 magnesium carbonate Inorganic materials 0.000 description 1
- 235000014380 magnesium carbonate Nutrition 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000005272 metallurgy Methods 0.000 description 1
- -1 metallurgy Substances 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 235000010755 mineral Nutrition 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- 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
-
- 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
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Geometry (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Feedback Control In General (AREA)
Abstract
The invention provides an intelligent modeling method, device, equipment and storage medium for a digital twin system in a complex industrial process. The intelligent modeling method of the digital twin system of the complex industrial process comprises the following steps: establishing a mechanism model of a complex industrial process, wherein the mechanism model comprises an identifiable model and an unmodeled dynamic model; estimating parameters of the identifiable model to obtain an identification model; adopting an identification model error and the unmodeled dynamics to form an unknown nonlinear dynamic system; establishing an intelligent model of the unknown nonlinear dynamic system; establishing an intelligent model of the complex industrial process digital twin system by adopting the identification model and an intelligent model of the unknown nonlinear dynamic system; the identification model error is a model output error caused when the parameters in the identifiable model are replaced by the identification values of the parameters. Aiming at the problem that the precision of the digital twin system of the complex industrial process is difficult to guarantee, a system identification method based on a mechanism model is combined with a deep learning method based on big data, and an end-to-side cloud cooperative mode is adopted, so that an intelligent model of the digital twin system of the complex industrial process is established, the modeling difficulty of the digital twin system of the complex industrial process is solved, and the modeling precision is improved.
Description
Technical Field
The invention belongs to the technical field of industrial artificial intelligence, and particularly relates to an intelligent modeling method, device, equipment and storage medium of a digital twin system in a complex industrial process.
Background
There are a large number of complex industrial processes in the flow industries of iron and steel, metallurgy, mineral separation, petrifaction, electric power, etc., and the mechanism models of these industrial processes have the following complexity: the model contains nonlinear terms between input and output variables, multivariable strong coupling, unknown frequently-changed interference, unknown partial model structure, unknown input and output variable orders and unknown dynamic characteristics, so that the existing system identification method based on the mechanism model cannot establish dynamic models of the industrial processes. The interaction of material flow, information flow and energy flow in the production process causes the unknown change of the dynamic characteristics of the industrial processes along with the production time, so that the input and output data of the processes are in a changed, open and uncertain information space, and therefore, the existing deep learning technology in the complete information space can not establish the dynamic model of the industrial processes. Currently, the operation condition identification of the industrial processes and the decision of the set value of the process control system still depend on the manual identification and decision of operators and engineering technicians based on experience and knowledge. Because of the difficulty of timely and accurate sensing of working condition information and processing of multi-source heterogeneous information, and the subjectivity and uncertainty of people, high-performance control and operation optimization of industrial processes are difficult to realize, and unstable product quality and high energy consumption and material consumption are caused. The establishment of a digital twin system of an industrial process is a key for realizing the integration of the optimized operation and control of the industrial process. However, due to the above characteristics of complex industrial processes, it is difficult to build a digital twin system meeting the accuracy requirement by using an existing mechanism modeling method or a deep learning method.
Disclosure of Invention
The present invention aims to solve at least one of the technical problems in the related art to some extent. The technical scheme of the invention is as follows:
an intelligent modeling method of a digital twin system of a complex industrial process comprises the following steps:
establishing a mechanism model of a complex industrial process, wherein the mechanism model comprises an identifiable model and an unmodeled dynamic model;
estimating parameters of the identifiable model to obtain an identification model;
adopting an identification model error and the unmodeled dynamics to form an unknown nonlinear dynamic system;
establishing an intelligent model of the unknown nonlinear dynamic system;
establishing an intelligent model of the complex industrial process digital twin system by adopting the identification model and an intelligent model of the unknown nonlinear dynamic system;
the identification model error is a model output error caused when the parameters in the identifiable model are replaced by the identification values of the parameters.
Further, preferably, the intelligent model of the unknown nonlinear dynamic system comprises an offline deep learning model, an online deep learning model, a deep learning correction model and a self-correction mechanism; establishing the offline deep learning model by adopting LSTM; establishing the online deep learning model by adopting the same structure as the offline deep learning model; establishing the deep learning correction model by adopting the same structure as the offline deep learning model; when the error between the output of the intelligent model of the digital twin system of the complex industrial process and the actual output of the complex industrial process is larger than a set threshold value, adopting a self-correction mechanism to correct the connection weight parameters and the bias parameters of the online deep learning model by using the connection weight parameters and the bias parameters of the deep learning correction model; wherein the deep learning correction model uses more historical data than the online deep learning model.
Further, preferably, the online deep learning model and the deep learning correction model each include an input layer, a hidden layer, a fully connected layer, and an output layer; fixing the connection weight parameters and the bias parameters of the hidden layer in the online deep learning model, and correcting the connection weight parameters and the bias parameters of the full-connection layer in the online deep learning model online; training the connection weight parameters and the bias parameters of the hidden layer and the full-connection layer in the deep learning correction model on line; when the error between the output of the intelligent model of the complex industrial process digital twin system and the actual output of the complex industrial process is larger than a set threshold, adopting a self-correction mechanism to replace the connection weight parameters and the bias parameters of the hidden layer in the online deep learning model by the connection weight parameters and the bias parameters of the hidden layer of the deep learning correction model, and replacing the connection weight parameters and the bias parameters of the full connection layer in the online deep learning model by the connection weight parameters and the bias parameters of the full connection layer of the deep learning correction model.
Further preferably, the complex industrial process is an electric fused magnesia smelting process.
An intelligent modeling device for a digital twin system of a complex industrial process, comprising:
the mechanism model modeling module is used for establishing a mechanism model of the complex industrial process, and the mechanism model comprises an identifiable model and an unmodeled dynamic model;
the parameter identification module is used for estimating parameters of the identifiable model to obtain an identification model;
the unknown nonlinear dynamic system construction module is used for constructing an unknown nonlinear dynamic system by adopting an identification model error and the unmodeled dynamics;
the intelligent model modeling module of the unknown nonlinear dynamic system is used for building an intelligent model of the unknown nonlinear dynamic system;
the digital twin system intelligent model modeling module is used for establishing an intelligent model of the complex industrial process digital twin system by adopting the identification model and the intelligent model of the unknown nonlinear dynamic system;
the identification model error is a model output error caused when the parameters in the identifiable model are replaced by the identification values of the parameters.
Further, as a preference, the unknown nonlinear dynamic system intelligent model modeling module comprises an offline deep learning model modeling module, an online deep learning model modeling module, a deep learning correction model modeling module and a self-correction module; the offline deep learning model modeling module establishes the offline deep learning model by adopting LSTM; the online deep learning model modeling module establishes the online deep learning model by adopting the same structure as the offline deep learning model; the deep learning correction model modeling module establishes the deep learning correction model by adopting the same structure as the offline deep learning model; the self-correction module corrects the connection weight parameters and the bias parameters of the online deep learning model by adopting a self-correction mechanism when the error between the output of the intelligent model of the digital twin system of the complex industrial process and the actual output of the complex industrial process is larger than a set threshold value; wherein the deep learning correction model uses more historical data than the online deep learning model.
Further, preferably, the online deep learning model and the deep learning correction model each include an input layer, a hidden layer, a fully connected layer, and an output layer; the online deep learning model modeling module fixes the connection weight parameters and the bias parameters of the hidden layer in the online deep learning model, and corrects the connection weight parameters and the bias parameters of the full-connection layer in the online deep learning model online; the deep learning correction model modeling module trains the connection weight parameters and the bias parameters of the hidden layer and the full-connection layer in the deep learning correction model on line; the self-correction module is used for replacing the connection right parameters and the bias parameters of the hidden layer in the online deep learning model by adopting a self-correction mechanism when the error between the output of the intelligent model of the digital twin system of the complex industrial process and the actual output of the complex industrial process is larger than a set threshold value, and replacing the connection right parameters and the bias parameters of the full-connection layer in the online deep learning model by using the connection right parameters and the bias parameters of the hidden layer of the deep learning correction model.
Further preferably, the complex industrial process is an electric fused magnesia smelting process.
An apparatus for implementing an intelligent modeling method for a digital twin system of a complex industrial process, comprising: an end side sub-device, an edge side sub-device, and a cloud side sub-device;
the terminal equipment is used for obtaining input data of the online deep learning model and the deep learning correction model;
the edge side sub-equipment is used for running the identification model and the online deep learning model;
the cloud side sub-equipment is used for running the deep learning correction model;
the end side sub-equipment is a process control system of the complex industrial process, the edge side sub-equipment is edge computing equipment, and the cloud side sub-equipment is an industrial cloud.
A computer readable storage medium storing a computer program which when executed by a processor implements the above-described intelligent modeling method of a digital twin system of a complex industrial process.
According to the invention, a system identification method based on a mechanism model is combined with a deep learning method based on big data, and an end-to-side cloud cooperative mode is adopted to establish an intelligent model of the digital twin system of the complex industrial process, so that the modeling difficulty of the digital twin system of the complex industrial process is solved, and the modeling precision is improved.
Drawings
FIG. 1 is a flow chart of an intelligent modeling method for a digital twin system of a complex industrial process according to an embodiment of the present invention;
FIG. 2 is a flow chart of an implementation of a digital twin system intelligent modeling method for an electric smelting magnesia smelting process according to an embodiment of the invention;
FIG. 3 is a diagram of the LSTM network architecture of one embodiment of the invention;
FIG. 4 is a diagram of LSTM cell nodes according to one embodiment of the invention;
FIG. 5 is a graph showing error versus neuron number for one embodiment of the present invention;
FIG. 6 is a graph showing error versus number of cell nodes according to one embodiment of the present invention;
FIG. 7 is a graph showing error versus network layer number according to one embodiment of the present invention;
FIG. 8 is a graph showing error versus time series window length for one embodiment of the present invention;
FIG. 9 is a block diagram of a digital twin system for an fused magnesia smelting process according to one embodiment of the invention;
FIG. 10 is a schematic diagram of the architecture of an intelligent modeling apparatus for a digital twin system for a complex industrial process according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
FIG. 1 is a flowchart of an intelligent modeling method for a digital twin system for a complex industrial process, which comprises the following steps:
s1: a mechanism model of the complex industrial process is established, wherein the mechanism model comprises an identifiable model and an unmodeled dynamic model.
Specifically, the unknown constant F is adopted to represent a nonlinear function F (-) with model parameters of unknown change, and a new variable is adoptedRepresenting the input u in the mechanism model i (k) And output y i (k) The nonlinear term of the complex industrial process is represented by unknown nonlinear function V (,) to represent the unknown part of the model structure and the unknown order of the input and output variables, unknown interference, unknown coupling and unknown variation dynamic characteristics, so that the mechanism model of the complex industrial process is represented by an identifiable model and an unmodeled dynamic.
S2: and estimating parameters of the identifiable model to obtain the identification model.
Specifically, input/output data u of a process is utilized i (k)、y i (k) Solving for new variable data representing nonlinear term between input and outputBy input/output data u i (k)、y i (k)And new variable data->And estimating parameters of the identifiable model by adopting an identification algorithm, thereby obtaining the identification model.
S3: and adopting an identification model error and the unmodeled dynamics to form an unknown nonlinear dynamic system.
Specifically, the identification model error is a model output error caused when the parameters in the identifiable model are replaced by the identification values thereof. An unknown nonlinear dynamic system expressed by the following formula is formed by adopting an identification model error and unmodeled dynamics:
wherein,
f is a nonlinear function of unknown variation, y i (k) The ith phase output of the mechanism model at k time, u i (k) An i-th phase input for the k-time mechanism model,to express u i (k) And y is i (k) A new variable of the nonlinear term in between, for the output of the recognition model->Is k-1 output of the intelligent model of the unknown nonlinear dynamic system at moment, n being the order of the unknown nonlinear dynamic system.
S4: and establishing an intelligent model of the unknown nonlinear dynamic system.
Specifically, the intelligent model of the unknown nonlinear dynamic system comprises an offline deep learning model, an online deep learning model, a deep learning correction model and a self-correction mechanism.
Selecting an input variable in the formula (1) as the input of a single neuron, the order n as the number of the neurons, v (k) as tag data, adopting the input data and the output data of the formula (1) to form a large data sample, adopting a training algorithm, and enabling the tag data v (k) to be output with an offline deep learning modelv(k) Error delta of (2)v(k) And determining the number n of neurons of the offline deep learning model, the number h of unit nodes of the LSTM, the number L of the neural network layers, the connection weight parameters and the bias parameters of each layer as small as possible, thereby establishing the offline deep learning model.
And establishing an online deep learning model by adopting the same structure as the offline deep learning model, namely establishing the online deep learning model by adopting an LSTM, wherein the number of neurons, the number of unit nodes and the number of neural network layers of the online deep learning model are the same as those of the offline deep learning model. Initial values of connection weight parameters and bias parameters of all layers of the online deep learning model adopt connection weight parameter values and bias parameters of corresponding layers of the offline deep learning model, connection weight parameters and bias parameters of hidden layers in the online deep learning model are fixed, real-time updated input data and tag data with a time sequence window length of N are adopted to online correct connection weight parameters and bias parameters of all connection layers of the online deep learning model, wherein the time sequence window length of N is obtained by training an algorithm, and tag data v (k) and output of the online deep learning modelThe error Δv (k) of (a) is as small as possible.
And establishing a deep learning correction model by adopting the same structure as the offline deep learning model. Namely, the LSTM is adopted to build a deep learning correction model, and the number of neurons, the number of unit nodes and the number of layers of the neural network of the deep learning correction model are the same as those of the offline deep learning model. And training all connection right parameters and bias parameters of each layer of the deep learning correction model by adopting the input data and the output data of the formula (1) at the current moment and all the previous moments as the input data and the label data of the deep learning correction model.
The self-correction mechanism sets the upper boundary of the interval of the error to be delta. When the error |Deltar (k) | > delta between the output obtained by overlapping the output of the online deep learning model and the output of the identification model at the moment k and the real output value of the real industrial process at the moment k, the connection weight parameters and the bias parameters of all layers of the deep learning correction model are adopted to replace the connection weight parameters and the bias parameters of the corresponding layers of the online deep learning model.
The input data of the online deep learning model and the deep learning correction model are obtained by an end-process control system, the online deep learning model operates on an edge-to-edge computing device, and the deep learning correction model operates on a cloud-industrial cloud.
S5: and establishing an intelligent model of the digital twin system of the complex industrial process by adopting the identification model and the intelligent model of the unknown nonlinear dynamic system.
Further, in one embodiment, the intelligent modeling method of the digital twin system of the complex industrial process can be used for the fused magnesia smelting process.
The fused magnesia smelting process is a strong nonlinear and strong coupling industrial process taking the rotation direction and frequency of a three-phase motor as input and the current of a three-phase electrode as output. The smelting process adopts a submerged arc mode, raw ore is fed while being melted, the height of a molten pool changes along with the continuous feeding and melting of the raw ore, the length of raw ore particles and the change of impurity components, and the resistivity of the molten pool changes along with the change of the temperature of the molten pool, the length of the raw ore particles and the change of the impurity components. Therefore, the model order of the dynamic model of the melting current is unknown and the dynamic characteristic change is unknown, and a mathematical model cannot be established by adopting a system identification method based on a mechanism model. The input and output data of the process are in a variable, open and uncertain information space, so that the existing deep learning technology of the complete information space cannot be adopted to build a dynamic model of the process.
FIG. 2 is a flowchart of an implementation of a digital twin system intelligent modeling method for an electric smelting magnesia smelting process according to an embodiment of the invention, which includes the steps of:
s1': and establishing an electrode current mechanism model in the fused magnesia smelting process, and representing the mechanism model by using an identifiable model and an unknown nonlinear function.
Specifically, during the smelting process of the fused magnesia, the fused bath is continuously increased along with the continuous addition and the smelting of the raw ore, and the height h (DEG) of the fused bath and the vibration frequency of an electric vibration feederThe relation between the two is:
wherein y is i (t) is the electrode current of the ith phase at time t, y is the optimal melting current; k (k) h (y,y i (t)) is a conversion coefficient between the cumulative charge amount and the bath height, which conversion coefficient is dependent on the electrode current y i (t) varies around fluctuations in the optimal melting current y; m is m 0 Is the initial charge; b (B) 1 Is the length of raw ore particles; b (B) 2 Is an impurity component of raw ore; k (k) m (B 1 ,B 2 ) Is a conversion coefficient between the electric vibration frequency and the charging speed, and the conversion coefficient is as follows B 1 And B 2 Is changed by a change in (a);the vibration frequency of the electric vibration feeder at the time t; sigma (t) is a start-stop sign of the electric vibration feeder, sigma (t) =1 represents start, and sigma (t) =0 represents stop; />And sigma (t) according to real-time smeltingAnd (5) determining a state.
When h (-) is changed in a formula (4) with feeding and melting, the current control system enables the electrode current to stably track the optimal melting current set value to melt magnesite by dragging the motor to adjust the distance between the electrode and the molten pool. The electric smelting process of the fused magnesia is characterized in that the rotation direction and the frequency u of a motor are adopted i (t) as input, electrode current y i The electrode current dynamic mechanism model with the output of (t) is as follows:
wherein i=1, 2,3 respectively represent A, B, C three phases, L i Is the self-inductance of the ith phase circuit, F i (. Cndot.) is the unknown nonlinear function of the ith phase circuit, s is the motor slip, r d The radius of an equivalent gear of the lifting mechanism is p is the pole pair number of the motor, g 0 Is the arc conductivity constant, r a Is the arc column radius, T 0 Is the ionization temperature constant of the gas, T 1 For arc gap temperature, u i (t) is the motor rotation direction and frequency of the ith phase circuit, U i For the phase voltage of the i-th phase circuit,the effect on the i-th phase current for a three-phase circuit coupling.
F i (. Cndot.) is represented by formula (6):
wherein Δh ie (. Cndot.) is the variation of the height of the electrode when the electrode is subjected to the electromagnetic force generated by the feeding interference and the varying mutual inductance; ρ (·) is the resistivity of the molten pool, B, as a function of the melt temperature 1 And B 2 Is changed by a change in (a); f (f) 0 Is the proportional coefficient of the resistance of the molten pool; s is the cross-sectional area of the molten pool.
As shown in formula (7):
wherein M is 12 (. Cndot.) is the mutual inductance between A, B phase circuits, M 13 (. Cndot.) is the mutual inductance between A, C phase circuits, M 23 (. Cndot.) is the mutual inductance between B, C phase circuits, M 12 (·)、M 13 (. Cndot.) and M 23 (. Cndot.) as the raw ore is added and melted, the temperature of the melt changes.
Unknown nonlinear function F in equation (5) i (. Cndot.) with unknown constant F i Instead, the resulting model error, the dynamics of the unknown changes, the unknown nonlinear function for unknown change couplingThe expression, i.e., (5), can be expressed as:
wherein,the method comprises the following steps:
discretizing the formula (8) by adopting an Euler method to obtain the product:
wherein,is an unknown nonlinear function shown in the formula (9)>Is a discrete form of (c).
To eliminate u in formula (10) i (k 0 ) Is divided by y for both sides of (10) i (k) And the subtraction of the expression at time k-1 can be obtained:
due to the moment y of k i (k+1) is unknown, so let k in formula (11) equal k-1, the discretized current mechanism model is:
using new variablesAnd->Representing a nonlinear term between input and output in equation (12), namely:
using an unknown nonlinear function V i (k) Representing the unknown part of the model structure and the unknown order of the input and output variables, unknown interference, unknown variation coupling and unknown variation dynamic characteristics, and enablingWill Q i Substituting the formula (13) into the formula (12) so as to model the electrode current mechanism of the fused magnesia smelting process by using an identifiable model and an unknown nonlinear function V in the formula (12) i (k) Expressed as:
wherein V is i (k) The method comprises the following steps:
s2': and estimating the parameters of the identifiable model to obtain the identification model.
Specifically, the data u is input and output by using the actual process i (k)、y i (k) Obtaining new variable data representing a nonlinear term between input and output in the expression (14) from the expression (13)And->Estimating a parameter Q in a recognizable model (14) using a least squares recognition algorithm i Obtaining the estimated value +.>The method comprises the following steps:
thus, the output of the identification model can be obtainedThe method comprises the following steps:
s3': using an identification model error and an unknown nonlinear function V i (k) Construction of unknown nonlinear dynamic System v i (k) The mechanism model is expressed as the sum of the recognition model output and the unknown nonlinear dynamic system output.
In particular, the error of the identification model is adoptedAnd an unknown nonlinear function V i (k) Unknown nonlinear dynamic system v constituting the following expression i (k)：
Thus v i (k) Nonlinear function f that can be varied by unknown i The expression is that:
wherein the input data vectorThe method comprises the following steps:
from (14), (17) and (18), an unknown nonlinear dynamic system output v is obtained i (k) The method comprises the following steps:
the current mechanism model of the smelting process of the fused magnesia obtained by the method (21) can be expressed as an identification model outputOutput v of unknown nonlinear dynamic system i (k) The sum is that:
S4': establishing an unknown nonlinear dynamic system v i (k) (i=1, 2, 3).
Specifically, the intelligent model consists of an offline deep learning model, an online deep learning model, a deep learning correction model and a self-correction mechanism.
S41': v is established by adopting a multilayer long and short period memory network LSTM architecture as shown in figure 3 i (k) Is specific to the offline deep learning model:
selecting (19) an input data vectorAs the number of neurons per layer; vector dataN neurons as input data to the 1 st layer, respectively; tag data is v (k) = [ v ] when training the network 1 (k),v 2 (k),v 3 (k)] T . The node number of LSTM single neuron is recorded as h, the network layer number is recorded as L, the current time is k time, and the nth l Layer n i The output of the individual neurons is +.>(n i ＝1,…,n，n l =1, …, L). Fig. 4 shows a cell node diagram of LSTM.
For layer 1 (n l =1), nth i The inputs to the individual neurons are:
layer 1 nth i Output of individual neuronsThe method comprises the following steps:
wherein, as indicated by the letter "; tanh is a hyperbolic tangent function, i.e., tanh (x) = (e) x -e -x )/(e x +e -x )；For outputting gate output, +.>Is in LSTM unit state, as shown in the following formula:
where σ is a sigmoid function, i.e., σ (x) = (1+e) -x ) -1 ；Output for forgetting gate, ++>For input gate output, +.>Is a state candidate, as shown in the following formula:
(25) In the formulae (1) and (26),and->For h× (h+15) dimensional weight matrix, +.>Andthe column vector is biased for h x 1 dimensions.
When n is l When not less than 2, the nth l Layer n i The output of each neuron is shown as (24) to (26) in the 1 st layer and the n th layer i The computation of the individual neuron outputs is identical, except for the weight matrixAnd->Is h× (h+h), nth i The input of the neuron is n l -layer 1 nth i The output of the individual neurons, namely:
wherein, in calculating the input and output of the 1 st neuron of each layer,and->The value of (1) is determined by random initialization when k=0, and when k is equal to or greater than 1, the value is equal to the output and state of the last neuron of the layer at the previous moment, namely
Output of the nth neuron of the L th layerInput to the full connection layer, and the output of the full connection layer is the output of the offline deep learning modelv(k) The following formula is shown:
wherein W is d ∈R 3×h And b d ∈R 3 Weights and offsets for the full connection layer.
The actual process data of two heats (10 hours for each heat and 36000 groups of data) are respectively used as a training set and a test set, M (M=36000) input and output data of (19) are adopted to form a big data sample, and the following training algorithm is adopted to enable the tag data v (k) and an offline deep learning model to be outputv(k) Model error delta of (2)v(k) Training the network as little as possible, determining n, h and L, the specific operations are as follows:
let the network layer number l=1, the objective function of the training algorithm is:
wherein delta isv(k)＝v(k)-v(k) V (k) andv(k) As shown in the formulas (30) and (31):
wherein,the result is obtained by the expression (24). Based on error back propagation, a gradient descent algorithm is adopted to train network weightsW d And bias->b d . All weights and bias training algorithms are the same, and gates are output belowWeight->Describing training of (a) as an example, the objective function J is about +.>The partial derivatives of (2) are:
updating the output gate weights according to
Wherein η is the learning rate.
Other weights and biases are determined using the same training algorithm as equations (32) through (33).
Let the number h of cell nodes of LSTM equal to the input time sequenceThe number of layers of the neural network L is equal to 1. Increasing n from 1, and calculating model errors delta respectivelyv(k) Mean Absolute Error (MAE) of (a). As can be seen from fig. 5, Δ when n=3v(k) The order n of the dynamic system is 3, i.e. the number of neurons per layer is 3.
Fixing n=3, l=1, increasing h from 15, and calculating Δ respectivelyv(k) Is a MAE of (c). As can be seen from fig. 6, Δ when h=1770v(k) The MAE of (2) is the smallest, so the number of unit nodes h takes 1770.
Fixing n=3, h=1770, increasing L from 1, and calculating Δ respectivelyv(k) Is a MAE of (c). As can be seen from fig. 7, Δ when l=7v(k) The MAE of (2) is the smallest, so the number of layers of the neural network L is 7.
S42': an online deep learning model is built by adopting the same structure as the offline deep learning model, and the method is characterized in that:
an LSTM is adopted to build an online deep learning model, the number of neurons, the number of unit nodes and the number of layers of a neural network of the online deep learning model are the same as those of the offline deep learning model, initial values of connection right parameters and bias parameters of all layers of the online deep learning model adopt connection right parameter values and bias parameters of corresponding layers of the offline deep learning model, connection right parameters and bias parameters of 1 st to 7 th layers in the online deep learning model are fixed, and connection right parameters W of all connection layers of the online deep learning model are corrected online by adopting real-time updated input data and tag data with a time sequence window length of N d (k) And bias parameter b d (k) Obtaining the output of the online deep learning model of v (k)The method comprises the following steps:
wherein W is d (k)∈R 3×h And b d (k)∈R 3 For the weights and offsets of the full connection layer,is the output of the layer 7, 3 neuron.
At time k+1, the updated dataset is
The method comprises the following steps: />
Online lineCorrection W d (k+1) and b d The objective function and correction algorithm of (k+1) are shown in the equations (36) and (37), respectively:
wherein,representing tag data v (k) and online deep learning model output +.>Model error of +.>And->The following formula is shown:
n=3, h=1770, and l=7 are fixed, and N is incremented from 1 by a training algorithm represented by formulas (36) to (38), and Root Mean Square Errors (RMSE) of Δv (k) are calculated, respectively. As can be seen from fig. 8, RMSE of Δv (k) is minimum when n=1330, and thus the time-series window length N is 1330.
S43': the deep learning correction model shown in the following formula is built using exactly the same structure as the offline deep learning model, i.e., n=3, h=1770, l=7.
Wherein,the output of the correction model is deep learned for time k+1. The k+1 time takes all the data of all the past times, and the updated data set is +.>All connection right parameters and bias parameters of each layer of the deep learning correction model are trained on line.
Setting the upper boundary of the interval of the error as delta=100deg.A, and at the time k+1, if the output of the online deep learning model is overlapped with the output of the identification model, obtaining the output of the electrode current intelligent modelTrue value y of electrode current in fused magnesia smelting process i Error between (k+1)>And correcting the connection right parameters and the bias parameters of the corresponding layers of the online deep learning model by adopting all the connection right parameters and the bias parameters of each layer of the deep learning correction model.
The input data of the online deep learning model and the deep learning correction model are obtained by a PLC control system in the end-fused magnesia smelting process, the identification model and the online deep learning model operate in an edge-edge computing device, and the deep learning correction model operates in a cloud-industrial cloud.
S5': and an electrode current intelligent model of the digital twin system in the fused magnesia smelting process is established by adopting an identification model and an intelligent model of an unknown nonlinear dynamic system.
Specifically, an electrode current intelligent model of the digital twin system in the fused magnesia smelting process is as follows:
/>
wherein, the identification modelOutput ofThe intelligent model output of the unknown nonlinear dynamic system is obtained by the formula (17)>The result is obtained by the expression (34). The digital twin system structure of the fused magnesia smelting process is shown in figure 9.
Selecting 300 groups of test set data, and outputting by adopting an identification modelAnd outputting the identification model in the embodiment of the invention>Output of intelligent model of unknown nonlinear dynamic system>The two current models established by addition are compared, and the model outputs are respectively +.>And-> And->The result is obtained by the expression (17) and the expression (34), respectively. The model accuracy was evaluated using the performance evaluation index Root Mean Square Error (RMSE) and the Mean Absolute Percentage Error (MAPE) shown in the formulas (41) and (42), and the results are shown in table 1.
Wherein N is 0 =300 is the number of time steps, y i (k) To be a true value of the value,and outputting a value for the model.
Table 1 three-phase current model accuracy performance evaluation table
As can be seen from Table 1, the modeling accuracy of the embodiment of the invention is significantly better than that of the method using the identification model output, three-phase current y 1 (k)、y 2 (k)、y 3 (k) The RMSE was reduced by 93.14%, 91.82%, 93.72%, and the MAPE was reduced by 96.22%, 95.23%, 96.14%, respectively. The three-phase current model precision established by the embodiment of the invention meets the requirement of establishing a digital twin system in the fused magnesia smelting process, and creates conditions for realizing intelligent optimization control of the fused magnesia smelting process.
In one embodiment, as shown in FIG. 10, there is provided an intelligent modeling apparatus for a digital twin system of a complex industrial process, comprising: the system comprises a mechanism model modeling module, a parameter identification module, an unknown nonlinear dynamic system construction module, an unknown nonlinear dynamic system intelligent model modeling module and a digital twin system intelligent model modeling module, wherein:
the mechanism model modeling module is used for establishing a mechanism model of the complex industrial process, and the mechanism model comprises an identifiable model and an unmodeled dynamic model;
the parameter identification module is used for estimating parameters of the identifiable model to obtain an identification model;
the unknown nonlinear dynamic system construction module is used for constructing an unknown nonlinear dynamic system by adopting an identification model error and the unmodeled dynamics;
the intelligent model modeling module of the unknown nonlinear dynamic system is used for establishing an intelligent model of the unknown nonlinear dynamic system;
the digital twin system intelligent model modeling module is used for establishing an intelligent model of the complex industrial process digital twin system by adopting the identification model and the intelligent model of the unknown nonlinear dynamic system;
the identification model error is a model output error caused when the parameters in the identifiable model are replaced by the identification values of the parameters.
In one embodiment, the unknown nonlinear dynamic system intelligent model modeling module comprises an offline deep learning model modeling module, an online deep learning model modeling module, a deep learning correction model modeling module and a self-correction module; the offline deep learning model modeling module establishes the online deep learning model by adopting LSTM; the online deep learning model modeling module establishes the online deep learning model by adopting the same structure as the offline deep learning model; the deep learning correction model modeling module establishes the deep learning correction model by adopting the same structure as the offline deep learning model; the self-correction module corrects the connection weight parameters and the bias parameters of the online deep learning model by adopting a self-correction mechanism when the error between the output of the intelligent model of the digital twin system of the complex industrial process and the actual output of the complex industrial process is larger than a set threshold value; wherein the deep learning correction model uses more historical data than the online deep learning model.
In one embodiment, the online deep learning model and the deep learning correction model each include an input layer, a hidden layer, a fully connected layer, and an output layer; the online deep learning model modeling module fixes the connection weight parameters and the bias parameters of the hidden layer in the online deep learning model, and corrects the connection weight parameters and the bias parameters of the full-connection layer in the online deep learning model online; the deep learning correction model modeling module trains the connection weight parameters and the bias parameters of the hidden layer and the full-connection layer in the deep learning correction model on line; the self-correction module is used for replacing the connection right parameters and the bias parameters of the hidden layer in the online deep learning model by adopting a self-correction mechanism when the error between the output of the intelligent model of the digital twin system of the complex industrial process and the actual output of the complex industrial process is larger than a set threshold value, and replacing the connection right parameters and the bias parameters of the full-connection layer in the online deep learning model by using the connection right parameters and the bias parameters of the hidden layer of the deep learning correction model.
In one embodiment, the complex industrial process is an fused magnesia smelting process.
For specific limitations on the intelligent modeling apparatus of the digital twin system of the complex industrial process, reference may be made to the above limitation on the intelligent modeling method of the digital twin system of the complex industrial process, and no further description is given here. The above-described individual modules in the complex industrial process digital twin system intelligent modeling apparatus may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory of the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, there is provided an apparatus for implementing the complex industrial process digital twin system intelligent modeling method in each of the above embodiments, comprising: an end side sub-device, an edge side sub-device, and a cloud side sub-device; the terminal equipment is used for obtaining input data of the online deep learning model and the deep learning correction model; the edge side sub-equipment is used for running the identification model and the online deep learning model; the cloud side sub-equipment is used for running the deep learning correction model; the end side sub-equipment is a process control system of the complex industrial process, the edge side sub-equipment is edge computing equipment, and the cloud side sub-equipment is an industrial cloud.
In one embodiment, a computer readable storage medium is provided, which stores a computer program which when executed by a processor implements the complex industrial process digital twin system intelligent modeling method in the above embodiments.
Those skilled in the art can combine and combine the features of the different embodiments described in this specification and of the different embodiments without contradiction.
In summary, the intelligent modeling method, the device and the equipment for the digital twin system of the complex industrial process provided by the embodiment of the invention solve the problem that the precision of the digital twin system of the complex industrial process is difficult to guarantee, combine the system identification method based on the mechanism model with the deep learning method based on big data, establish the intelligent model of the digital twin system of the complex industrial process by adopting a terminal cloud cooperative mode, solve the modeling problem of the digital twin system of the complex industrial process and improve the modeling precision.
The foregoing is merely illustrative of the present invention, and the present invention is not limited thereto, and any person skilled in the art will readily recognize that various changes and substitutions are possible within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (6)
1. An intelligent modeling method for a digital twin system of a complex industrial process, which is characterized by comprising the following steps:
establishing a mechanism model of a complex industrial process, wherein the mechanism model comprises an identifiable model and an unmodeled dynamic model;
estimating parameters of the identifiable model to obtain an identification model;
an unknown nonlinear dynamic system is formed by adopting an identification model error and the unmodeled dynamic, wherein the identification model error is a model output error caused when parameters in an identifiable model are replaced by identification values of the parameters;
establishing an intelligent model of the unknown nonlinear dynamic system;
establishing an intelligent model of the complex industrial process digital twin system by adopting the identification model and an intelligent model of the unknown nonlinear dynamic system;
the intelligent model of the unknown nonlinear dynamic system comprises an offline deep learning model, an online deep learning model, a deep learning correction model and a self-correction mechanism;
establishing the offline deep learning model by adopting LSTM;
establishing the online deep learning model by adopting the same structure as the offline deep learning model;
establishing the deep learning correction model by adopting the same structure as the offline deep learning model;
when the error between the output of the intelligent model of the digital twin system of the complex industrial process and the actual output of the complex industrial process is larger than a set threshold value, adopting a self-correction mechanism to correct the connection weight parameters and the bias parameters of the online deep learning model by using the connection weight parameters and the bias parameters of the deep learning correction model;
the history data used by the deep learning correction model is more than the history data used by the online deep learning model, and the complex industrial process is an electric smelting magnesia smelting process.
2. The method of claim 1, wherein the online deep learning model and the deep learning correction model each comprise an input layer, a hidden layer, a fully connected layer, and an output layer;
fixing the connection weight parameters and the bias parameters of the hidden layer in the online deep learning model, and correcting the connection weight parameters and the bias parameters of the full-connection layer in the online deep learning model online;
training the connection weight parameters and the bias parameters of the hidden layer and the full-connection layer in the deep learning correction model on line;
when the error between the output of the intelligent model of the complex industrial process digital twin system and the actual output of the complex industrial process is larger than a set threshold, adopting a self-correction mechanism to replace the connection weight parameters and the bias parameters of the hidden layer in the online deep learning model by the connection weight parameters and the bias parameters of the hidden layer of the deep learning correction model, and replacing the connection weight parameters and the bias parameters of the full connection layer in the online deep learning model by the connection weight parameters and the bias parameters of the full connection layer of the deep learning correction model.
3. An intelligent modeling device for a digital twin system of a complex industrial process, comprising:
the mechanism model modeling module is used for establishing a mechanism model of the complex industrial process, and the mechanism model comprises an identifiable model and an unmodeled dynamic model;
the parameter identification module is used for estimating parameters of the identifiable model to obtain an identification model;
the unknown nonlinear dynamic system construction module is used for constructing an unknown nonlinear dynamic system by adopting an identification model error and the unmodeled dynamics, wherein the identification model error is a model output error caused when parameters in an identifiable model are replaced by identification values of the parameters;
the intelligent model modeling module of the unknown nonlinear dynamic system is used for building an intelligent model of the unknown nonlinear dynamic system;
the digital twin system intelligent model modeling module is used for establishing an intelligent model of the complex industrial process digital twin system by adopting the identification model and the intelligent model of the unknown nonlinear dynamic system;
the intelligent model modeling module of the unknown nonlinear dynamic system comprises an offline deep learning model modeling module, an online deep learning model modeling module, a deep learning correction model modeling module and a self-correction module;
the offline deep learning model modeling module establishes the offline deep learning model by adopting LSTM;
the online deep learning model modeling module establishes the online deep learning model by adopting the same structure as the offline deep learning model;
the deep learning correction model modeling module establishes the deep learning correction model by adopting the same structure as the offline deep learning model;
the self-correction module corrects the connection weight parameters and the bias parameters of the online deep learning model by adopting a self-correction mechanism when the error between the output of the intelligent model of the digital twin system of the complex industrial process and the actual output of the complex industrial process is larger than a set threshold value;
the history data used by the deep learning correction model is more than the history data used by the online deep learning model, and the complex industrial process is an electric smelting magnesia smelting process.
4. The apparatus of claim 3, wherein the online deep learning model and the deep learning correction model each comprise an input layer, a hidden layer, a fully connected layer, and an output layer;
the online deep learning model modeling module fixes the connection weight parameters and the bias parameters of the hidden layer in the online deep learning model, and corrects the connection weight parameters and the bias parameters of the full-connection layer in the online deep learning model online;
the deep learning correction model modeling module trains the connection weight parameters and the bias parameters of the hidden layer and the full-connection layer in the deep learning correction model on line;
the self-correction module is used for replacing the connection right parameters and the bias parameters of the hidden layer in the online deep learning model by adopting a self-correction mechanism when the error between the output of the intelligent model of the digital twin system of the complex industrial process and the actual output of the complex industrial process is larger than a set threshold value, and replacing the connection right parameters and the bias parameters of the full-connection layer in the online deep learning model by using the connection right parameters and the bias parameters of the hidden layer of the deep learning correction model.
5. An apparatus for implementing the method of claim 1 or 2, the apparatus comprising: an end side sub-device, an edge side sub-device, and a cloud side sub-device;
the terminal equipment is used for obtaining input data of the online deep learning model and the deep learning correction model;
the edge side sub-equipment is used for running the identification model and the online deep learning model;
the cloud side sub-equipment is used for running the deep learning correction model;
the end side sub-equipment is a process control system of the complex industrial process, the edge side sub-equipment is edge computing equipment, and the cloud side sub-equipment is an industrial cloud.
6. A computer readable storage medium storing a computer program, which when executed by a processor implements the method of claim 1 or 2.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011435289.1A CN112560337B (en) | 2020-12-10 | 2020-12-10 | Intelligent modeling method, device, equipment and storage medium for digital twin system of complex industrial process |
| PCT/CN2021/136323 WO2022121923A1 (en) | 2020-12-10 | 2021-12-08 | Smart modelling method and apparatus of complex industrial process digital twin system, device, and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011435289.1A CN112560337B (en) | 2020-12-10 | 2020-12-10 | Intelligent modeling method, device, equipment and storage medium for digital twin system of complex industrial process |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112560337A CN112560337A (en) | 2021-03-26 |
| CN112560337B true CN112560337B (en) | 2023-12-01 |
Family
ID=75060616
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011435289.1A Active CN112560337B (en) | 2020-12-10 | 2020-12-10 | Intelligent modeling method, device, equipment and storage medium for digital twin system of complex industrial process |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112560337B (en) |
| WO (1) | WO2022121923A1 (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112560337B (en) * | 2020-12-10 | 2023-12-01 | 东北大学 | Intelligent modeling method, device, equipment and storage medium for digital twin system of complex industrial process |
| CN113607601B (en) * | 2021-06-18 | 2022-10-28 | 东北大学 | Intelligent detection method for ore pulp concentration based on combination of identification model and deep learning |
| CN114637262B (en) * | 2022-03-10 | 2022-11-15 | 天津科技大学 | Decision control method and system of intelligent factory digital twin information based on 5G drive |
| CN114609917B (en) * | 2022-05-11 | 2022-08-05 | 曜石机器人(上海)有限公司 | Servo driver and servo system based on digital twin technology |
| CN115310228B (en) * | 2022-08-09 | 2023-06-27 | 重庆大学 | Gear shaping design method based on digital twin |
| CN115184563B (en) * | 2022-09-08 | 2022-12-02 | 北京中环高科环境治理有限公司 | Chemical workshop field data acquisition method based on digital twinning |
| CN115857360B (en) * | 2023-02-14 | 2023-07-25 | 青岛海大新星软件咨询有限公司 | Industrial system mechanism modeling simulation platform |
| CN116341396B (en) * | 2023-05-30 | 2023-08-11 | 青岛理工大学 | Complex equipment digital twin modeling method based on multi-source data fusion |
| CN117131828B (en) * | 2023-07-12 | 2024-05-03 | 合肥工业大学 | Digital twin identification method for passive parameters of boost converter |
| CN117150832B (en) * | 2023-11-01 | 2024-02-23 | 北京科技大学 | Real-time prediction method and device for cross section shape of hot-rolled digital twin strip steel |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107526294A (en) * | 2017-07-26 | 2017-12-29 | 西安理工大学 | A kind of Nonlinear Delay dynamic system model INTELLIGENT IDENTIFICATION method |
| CN110363285A (en) * | 2019-06-20 | 2019-10-22 | 广东工业大学 | Integrated transfinite learning machine and the Complex Nonlinear System modeling method of Hammerstein-Wiener |
| CN111768000A (en) * | 2020-06-23 | 2020-10-13 | 中南大学 | Industrial process data modeling method for online adaptive fine-tuning deep learning |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180136617A1 (en) * | 2016-11-11 | 2018-05-17 | General Electric Company | Systems and methods for continuously modeling industrial asset performance |
| CN110703718A (en) * | 2019-11-13 | 2020-01-17 | 东北大学 | Industrial process control method based on signal compensation |
| CN111439681B (en) * | 2020-01-17 | 2021-11-02 | 华中科技大学 | Intelligent identification method and system for unsafe operation based on tower crane |
| CN112560337B (en) * | 2020-12-10 | 2023-12-01 | 东北大学 | Intelligent modeling method, device, equipment and storage medium for digital twin system of complex industrial process |
-
2020
- 2020-12-10 CN CN202011435289.1A patent/CN112560337B/en active Active
-
2021
- 2021-12-08 WO PCT/CN2021/136323 patent/WO2022121923A1/en active Application Filing
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107526294A (en) * | 2017-07-26 | 2017-12-29 | 西安理工大学 | A kind of Nonlinear Delay dynamic system model INTELLIGENT IDENTIFICATION method |
| CN110363285A (en) * | 2019-06-20 | 2019-10-22 | 广东工业大学 | Integrated transfinite learning machine and the Complex Nonlinear System modeling method of Hammerstein-Wiener |
| CN111768000A (en) * | 2020-06-23 | 2020-10-13 | 中南大学 | Industrial process data modeling method for online adaptive fine-tuning deep learning |
Non-Patent Citations (2)
| Title |
|---|
| 一类非线性离散时间动态系统的交替辨识算法及应用;张亚军;柴天佑;杨杰;;自动化学报(第01期);第101-112页 * |
| 张亚军 ; 柴天佑 ; 杨杰 ; .一类非线性离散时间动态系统的交替辨识算法及应用.自动化学报.2017,(第01期),第101-112页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2022121923A1 (en) | 2022-06-16 |
| CN112560337A (en) | 2021-03-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112560337B (en) | Intelligent modeling method, device, equipment and storage medium for digital twin system of complex industrial process | |
| CN109946975B (en) | Reinforced learning optimal tracking control method of unknown servo system | |
| US11543787B2 (en) | Networked control system time-delay compensation method based on predictive control | |
| CN112631215B (en) | Intelligent forecasting method, device, equipment and storage medium for industrial process operation index | |
| CN111459051B (en) | Discrete terminal sliding mode model-free control method with disturbance observer | |
| Smoczek et al. | Evolutionary algorithm-based design of a fuzzy TBF predictive model and TSK fuzzy anti-sway crane control system | |
| CN109839825B (en) | Method and system for predictive control of component content in rare earth extraction process | |
| Yi et al. | Event-triggered globalized dual heuristic programming and its application to networked control systems | |
| CN112381316B (en) | Electromechanical equipment health state prediction method based on hybrid neural network model | |
| CN111428317B (en) | Joint friction torque compensation method based on 5G and cyclic neural network | |
| Ai et al. | Data-driven-based adaptive fuzzy neural network control for the antimony flotation plant | |
| ÇANAKOĞLU et al. | Induction motor parameter estimation using metaheuristic methods | |
| CN109685283B (en) | Method for predicting submerged arc furnace working condition based on confidence rule base reasoning | |
| CN111007716A (en) | Alternating current servo motor variable discourse domain fuzzy PI control method based on prediction function | |
| Bartczuk et al. | Hybrid state variables-fuzzy logic modelling of nonlinear objects | |
| CN111525548B (en) | Method and system for evaluating static voltage stability margin of power grid | |
| Fu et al. | Multi-model adaptive switching control of a nonlinear system and its applications in a smelting process of fused magnesia | |
| Abbasi Nozari et al. | Intelligent non-linear modelling of an industrial winding process using recurrent local linear neuro-fuzzy networks | |
| de Souza et al. | Soft sensors in the primary aluminum production process based on neural networks using clustering methods | |
| CN115016286A (en) | Nonlinear industrial system data-driven reinforcement learning robust control method | |
| CN115952888A (en) | Multivariable grey model-based energy carbon emission prediction method and system | |
| Tian et al. | A new incremental learning modeling method based on multiple models for temperature prediction of molten steel in LF | |
| CN115718883A (en) | Thermal early warning method and device for battery energy storage system | |
| CN114896892B (en) | Multi-heat current digital twin method, device and equipment based on end Bian Yun cooperation | |
| Yang et al. | Intelligent Forecasting System Using Grey Model Combined with Neural Network. |
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 | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20210326 Assignee: WOLONG ELECTRIC GROUP Co.,Ltd. Assignor: Northeastern University Contract record no.: X2024210000023 Denomination of invention: Intelligent modeling methods, devices, equipment, and storage media for digital twin systems in complex industrial processes Granted publication date: 20231201 License type: Common License Record date: 20240528 |
|
| EE01 | Entry into force of recordation of patent licensing contract |