Intelligent automation control system for complete metallurgical equipment

Abstract

The present application relates to a kind of metallurgical complete equipment intelligent automation control system, specifically relates to the field of metallurgical complete equipment automation control, through multi-module closed loop linkage, the quality stability and resource utilization efficiency of metallurgical production are significantly improved;Causal analysis module accurately links equipment exception and quality index, significantly improves the fault positioning accuracy;Decision optimization module uses advanced algorithm, realizes multi-objective optimization, reduces energy consumption and reduces zinc layer thickness fluctuation;Knowledge evolution module greatly shortens the standard cycle of new production line cold start by fusing the experience of multiple plants;Control module dynamically adjusts process parameters, solves the problem of equipment and process fragmentation in traditional system, effectively reduces the failure quality loss and reduces material waste, the system improves production efficiency, reduces energy consumption and waste, with significant economic benefits and innovation value.

Description

Technical Field

[0001] This invention relates to the field of automated control of complete sets of metallurgical equipment, and more specifically, to an intelligent automated control system for complete sets of metallurgical equipment. Background Technology

[0002] In modern hot-dip galvanizing production, precise process control is crucial for ensuring product quality. In a hot-dip galvanizing production line, annealing furnaces, air knife systems, and other equipment work together to ensure the uniformity and quality of the galvanized layer. However, fluctuations in equipment operating conditions, such as abnormal fluctuations in air knife pressure, often affect the production process, leading to uneven zinc layer thickness, sometimes exceeding tolerance limits, thus impacting product quality and production efficiency. Fluctuations in air knife pressure are often related to factors such as aging solenoid valves and excessively high moisture content in compressed air. These equipment anomalies do not occur in isolation and are often not promptly reported or effectively adjusted during production, causing interruptions in production continuity and fluctuations in quality. Although some existing systems provide initial alerts through equipment alarms, the lack of a linkage mechanism between the equipment management system and the process control system prevents synchronized anomaly handling and process optimization, resulting in resource waste and economic losses during production.

[0003] Currently, existing hot-dip galvanizing production line systems generally suffer from a disconnect between equipment management systems and process control systems. Specifically, the equipment management system and the advanced process control system use independent databases, and there is no direct mapping between fault codes and process parameters. This prevents equipment anomaly information from being effectively transmitted to the process control module for coordinated adjustments. For example, when air knife pressure fluctuates, although equipment alarms are triggered, the process control system fails to adjust the annealing furnace temperature setpoint or other relevant process parameters in a timely manner to compensate for the impact of the anomaly. Furthermore, existing systems typically rely on predefined compensation rules to respond to common faults, but their ability to handle unknown anomalies is weak, making them unable to cope with complex and ever-changing production environments. Therefore, how to construct a multi-objective optimization framework that can detect equipment anomalies in real time and automatically adjust process parameters has become a core technological challenge. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing an intelligent automated control system for complete sets of metallurgical equipment. Through multi-module closed-loop linkage, it significantly improves the quality stability and resource utilization efficiency of metallurgical production, thereby solving the problems mentioned in the background art.

[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: it specifically includes: a causal analysis module, a decision optimization module, a knowledge evolution module, and a control module;

[0006] The causal analysis module, when receiving an abnormal equipment signal, performs manifold space mapping on real-time process parameters and historical fault data by embedding a tensor kernel function with metallurgical thermodynamic constraints, and generates a causal correlation matrix C between equipment faults and quality indicators.

[0007] The decision optimization module, when at least one element value c in the causal correlation matrix C pq >Preset threshold Θ th At that time, the quantum-inspired Hamiltonian optimizer performs multi-objective optimization of process parameters and outputs a set of quality compensation strategies P that satisfy safety constraints. * ;

[0008] The knowledge evolution module, in the set of quality compensation strategies P * After execution, a federated learning mechanism is used to aggregate the decision gradients of multiple plants, and the process compensation effect is encoded into a global knowledge base K in the form of tensor continued fractions. global ;

[0009] The control module, when a new production line is started or equipment is replaced, retrieves information from the global knowledge base K. global Compensation rules based on fractal topology constraints are applied to dynamically adjust the zinc bath temperature T. zn The air knife pressure parameter P is used to achieve a thickness error ΔG < δ. G ;

[0010] In a preferred embodiment, the specific operation of manifold space mapping in the causal analysis module is as follows:

[0011] A1. Reduce the dimensionality of the equipment manual text dataset, sensor time series stream, and maintenance record table to an isothermal compressed embedded manifold M using the Ricci flow algorithm;

[0012] A2. Using the causal tensor kernel function constrained by the Navier-Stokes equations, calculate the partial derivative matrix between the air knife pressure P and the zinc layer thickness G.

[0013] A3. Generate a causal relationship matrix C = [c pq ] P×Q .

[0014] In a preferred embodiment, the calculation process of the causal tensor kernel function satisfies:

[0015] A1. The kernel function output value has a negative exponential relationship with the distance of the device parameters in the manifold space M;

[0016] A2. The second-order partial derivative determinant of the zinc liquid flow potential function Φ is used as a physical constraint term;

[0017] The physical constraint relationship of the zinc liquid flow potential function Φ satisfies the viscous fluid dynamics equation.

[0018] In a preferred embodiment, the specific operation of multi-objective optimization in the decision optimization module is as follows:

[0019] A1. Construct the Hamiltonian H(x) containing the Lévy flight mass approximation term based on the failure probability distribution P(Y|F);

[0020] A2. By simulating quantum tunneling evolution using FPGA voltage pulses, screen steel samples with energy consumption per ton of steel ≤ 100 kJ and zinc layer fluctuation ≤ δ. G Pareto solution set P * .

[0021] In a preferred embodiment, the construction rule for the Hamiltonian is:

[0022] A1, Correlation Strength of Process Parameters J ij Obtained by normalizing the elements of the causal relationship matrix;

[0023] A2. The Lévy norm is used to calculate the deviation between the measured value and the set value in the quality objective function.

[0024] In a preferred embodiment, the federated learning operation in the knowledge evolution module includes:

[0025] A1. Gradient of the effectiveness of compensation strategies in each branch plant ▽J k Privacy mask matrix M k After encryption, a tensor block A is formed. k ;

[0026] A2. Generating a global knowledge base K by fusing tensor blocks with a progressively continued fraction structure. global .

[0027] In a preferred embodiment, the asymptotic continued fraction structure satisfies:

[0028] A1. The denominator identity matrix I is used to maintain the numerical stability of the knowledge representation;

[0029] A2. The depth of the continuous fractional level is dynamically trimmed based on the factory reputation weight.

[0030] In a preferred embodiment, the fractal topological constraint implementation of the control module includes:

[0031] A1. Equipment group layout information is converted into fractal dimension D. f ;

[0032] A2. The effective radius of the compensation rule decreases with a power-law relationship with the topological distance d between devices.

[0033] In a preferred embodiment, the calculation rule for the power-law decay is as follows:

[0034] A1. The attenuation operator ψ(d) is implemented through the negative power integral of the distance;

[0035] A2. The attenuation index α is set to 2.3 ± 0.5 to adapt to the layout characteristics of metallurgical workshops.

[0036] In a preferred embodiment, the causal correlation matrix C in the causal analysis module is updated periodically by t. c The strip thickness h is dynamically adjusted based on real-time data, where h is obtained from the thickness gauge at the mill exit, and the rolling cycle t is the rolling rhythm period. c ∈[5,60]s;

[0037] The quantum-heuristic Hamiltonian optimizer of the decision optimization module injects the latest fault probability distribution P(Y|F) before each iteration;

[0038] The global knowledge base K of the knowledge evolution module global Lie group gradient aggregation is performed every 24 hours.

[0039] The beneficial effects of this invention are as follows: through multi-module closed-loop linkage, the quality stability and resource utilization efficiency of metallurgical production are significantly improved; the causal analysis module accurately correlates equipment anomalies with quality indicators, significantly improving the accuracy of fault location; the decision optimization module adopts advanced algorithms to achieve multi-objective optimization, reducing energy consumption and zinc layer thickness fluctuations; the knowledge evolution module, by integrating the experience of multiple branch plants, greatly shortens the cold start-up period for new production lines; the control module dynamically adjusts process parameters, solving the problem of equipment and process separation in traditional systems, effectively reducing fault quality losses and material waste. This system improves production efficiency, reduces energy consumption and waste, and has significant economic benefits and innovative value. Attached Figure Description

[0040] Figure 1 This is a flowchart of the method of the present invention;

[0041] Figure 2 This is a block diagram of the system structure of the present invention. Detailed Implementation

[0042] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0044] In the description of this application, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed in this application.

[0045] Example 1

[0046] This embodiment provides, for example Figure 1-2 The intelligent automated control system for a complete set of metallurgical equipment shown includes: a causal analysis module, a decision optimization module, a knowledge evolution module, and a control module.

[0047] The causal analysis module, when receiving abnormal equipment signals, such as air knife pressure sensor alarms, uses a tensor kernel function with embedded metallurgical thermodynamic constraints to perform manifold space mapping on real-time process parameters and historical fault data, generating a causal correlation matrix C between equipment faults and quality indicators.

[0048] The decision optimization module, when at least one element value c in the causal correlation matrix C... pq >Preset threshold Θ th At that time, the quantum-inspired Hamiltonian optimizer performs multi-objective optimization of process parameters and outputs a set of quality compensation strategies P that satisfy safety constraints. * The physical implementation of its quantum-inspired Hamiltonian optimizer is as follows: an industrial server is configured with an FPGA acceleration card, and the evolution of the Hamiltonian is simulated by voltage pulses.

[0049] The knowledge evolution module, in the quality compensation strategy set P * After execution, a federated learning mechanism is used to aggregate the decision gradients of multiple plants, and the process compensation effect is encoded into a global knowledge base K in the form of tensor continued fractions. global ;

[0050] The control module, when a new production line is started or equipment is replaced, retrieves data from the global knowledge base K. global Compensation rules based on fractal topology constraints are applied to dynamically adjust the zinc bath temperature T. zn The air knife pressure parameter P is used to achieve a thickness error ΔG < δ. G , where δ G For thickness tolerance.

[0051] In this embodiment, the causal analysis module needs to be specifically explained. The specific operation of manifold space mapping is as follows:

[0052] A1. Combine the equipment manual text dataset O and the sensor time sequence stream (S = {s}) t The maintenance record table B (|t=1,2,...,T}) is reduced to an isothermal compressed embedded manifold M using the Ricci flow algorithm, while satisfying curvature constraints. The number of iterations was set to 100-200, the word vector dimension was 300, the word vector mapping adopted the Word2Vec model, and the training corpus contained a list of fault description keywords from equipment manuals. The expression of its manifold M is as follows:

[0053] Where M represents the isothermal compressed embedded manifold, A represents the candidate set of manifolds to be optimized, used for iterative optimization, and its value ranges within Q, where Q represents the smooth manifold space, ▽ R Represents the Ricci curvature tensor operator, used to constrain the rate of change of manifold curvature, ||·|| F denoted by Frobenius norm, used to quantify the strength of curvature constraints; λ represents the smoothing factor, ranging from [0.2, 0.8], used to adjust the weights of physical constraints and data distribution; D... Wasser The Wasserstein distance is used to measure the difference between the original data distribution and the manifold structure. The function of this formula is to embed the device data (O,S,B) into the low-dimensional manifold M through a geometric transformation with physical constraints, while preserving the metallurgical thermodynamic topological properties.

[0054] Furthermore, the specific formula for the Ricci flow algorithm is as follows:

[0055] P proj =R(D fused |Μ);

[0056] Where R(·) denotes the Ricci flow dimensionality reduction operator, M denotes the isothermal compressible embedded manifold with curvature K = 0.5, and P proj =(ξ k ,η l ,ζ m ) Μ Represents the coordinates of the projected manifold;

[0057] A2. Using the causal tensor kernel function constrained by the Navier-Stokes equations, calculate the partial derivative matrix of the air knife pressure P and the zinc layer thickness G. The formula for calculating the causal tensor kernel function is:

[0058]

[0059] Among them, K causel The kernel function representing the causal relationship strength, with values ​​ranging from [0,1], is used to quantify the strength of causal interactions between device parameters. (X) i ,X j Let represent the device parameter vector on the manifold, and γ represent the manifold curvature adjustment factor, with a value range of 0.3 ≤ γ ≤ 1.8, used to control the rate at which the association strength decays with manifold distance. M Let represent the distance on the manifold M, reflecting the similarity of parameters in the metallurgical feature space. Φ represents the zinc liquid flow potential function, used to describe the dynamic behavior of the zinc liquid in the galvanizing process. The boundary condition for the zinc liquid flow potential function Φ is Φ| t＝0 =βP0, where P0 represents the rated pressure of the air knife. The coordinates represent the direction of zinc liquid flow and are used to define the direction of the air knife pressure field. det(·) represents the determinant operator and is used to inject Navier-Stokes equation constraints.

[0060] A3. Generate a causal relationship matrix C = [c pq ] P×Q Element update period t c The element values ​​are adaptively adjusted according to the strip thickness h. Where Z represents the normalization factor, and Z is calculated according to Z... t =0.85Z t-1 +(0.15)·max(K causal Dynamically updated, where Z t Z represents the current value. t-1 Represents historical values, max(K) causal The formula represents the current maximum causal strength. Its function is to control the weight of historical data through a forgetting factor α (ranging from [0.8, 0.95]) when metallurgical equipment ages or its operating conditions change abruptly, so that the normalization factor Z quickly responds to the latest causal correlation strength max(K). causal To avoid system misjudgment; δ represents the Kronecker function, P represents the total number of fault types, and Q represents the number of quality indicators;

[0061] The calculation process of the causal tensor kernel function satisfies:

[0062] A1. The kernel function output value has a negative exponential relationship with the distance of the device parameters in the manifold space M;

[0063] A2. The second-order partial derivative determinant of the zinc liquid flow potential function Φ is used as a physical constraint term;

[0064] The physical constraint relationship of the zinc liquid flow potential function Φ satisfies the viscous fluid dynamics equation, and its expression is:

[0065]

[0066] Where t represents the time variable, u represents the flow velocity field, ν represents the viscosity coefficient, and β = 0.05 represents the pressure coupling coefficient (based on the zinc liquid density ρ = 6.67 g / cm³). 3 And the viscosity coefficient ν = 0.038 cm 2 / s experimental calibration), P represents the air knife pressure, the industrial setting range is 100-500 kPa, and the boundary conditions of the viscous fluid dynamics equation for the zinc liquid flow potential function Φ are:

[0067] air knife exit

[0068] strip interface

[0069] Where, ρ=6.67g / cm 3 The density of the zinc liquid is ν = 0.038 cm³. 2 / s represents the viscosity coefficient.

[0070] In this embodiment, the decision optimization module needs to be specifically explained. The specific operation of multi-objective optimization is as follows:

[0071] A1. Based on the failure probability distribution P(Y|F), a Hamiltonian H(x) containing a Lévy flight mass approximation term is constructed. This rule is used to optimize and control multiple relevant variables in complex production processes. The failure probability distribution P(Y|F) is generated using a fractal dissipative Bayesian inversion algorithm, and the specific formula is as follows:

[0072]

[0073] Among them, D f W represents the fractal dimension. k Represents the convolution weight matrix, * t Let σ represent the time parameter, α represent the activation function, α represent the decay exponent, and Y and F represent the fault and quality vectors. Represents the time-varying gradient of the causal matrix. This indicates that the impact of a failure in device k decreases according to a power law with respect to distance d. This indicates that the temporal gradient of the causality matrix C is convolved and filtered to capture the dynamics of fault propagation.

[0074] A2. By simulating quantum tunneling evolution using FPGA voltage pulses, screen steel samples with energy consumption per ton of steel ≤ 100 kJ and zinc layer fluctuation ≤ δ. G Pareto solution set P * In this application δ G =0.7μm, its δ G The specific value will be determined based on the actual product standard.

[0075] The FPGA pulse parameter mapping process is as follows:

[0076] The Hamiltonian evolution step size Δt = 0.01 ms corresponds to a voltage pulse width of 5 ns and an amplitude V. amp =k·|J ij (k = 0.2V is the calibration coefficient);

[0077] The construction rules for Hamiltonians are as follows:

[0078] A1, Correlation Strength of Process Parameters J ij Obtained by normalizing the elements of the causal correlation matrix, where the correlation strength J of the process parameters is... ij It reflects the mutual influence between process parameters. After normalization, it can more accurately describe the degree of correlation between various process parameters.

[0079] A2. The quality objective function uses the Lévy norm to calculate the deviation between the measured value and the set value, thereby quantifying the achievement of the quality objective. This helps to minimize the deviation and improve the quality control level of the production process during optimization.

[0080] The formula for calculating the Hamiltonian H(x) is:

[0081] H(x)=-∑ (i,j) J ij x i x j +μ||G opt -f(x)||Lévy;

[0082] Among them, J ij This represents the correlation strength matrix of normalized process parameters. (C pq (from the causal relationship matrix), x i x j This represents the process parameters to be optimized, such as zinc melt temperature and mill pressure. μ represents the mass weighting coefficient, with a value ranging from 0.1 to 0.9. Specifically, it is determined through energy-mass Pareto front analysis. Where ΔG is the real-time thickness error, δ G=0.7μm is the tolerance threshold, ||·||Lévy represents the norm calculation based on the Lévy distribution, and the Lévy norm is defined as: ||x||Lévy=∑|x i -μ| 1.5 G opt This represents a vector of target quality indicators, such as the ideal zinc coating thickness, G. opt =[G1,G2,...] T ;

[0083] The formula for calculating the quality objective function is:

[0084]

[0085] Among them, X * The optimal process parameters are connected, where <θ|...|X> represents the quantum tunneling probability amplitude, where θ is the ground state, X is the solution state, and T represents the optimization time window, and T≤0.1s;

[0086] Its quantum tunneling probability amplitude is decomposed using the Suzuki-Trotter method:

[0087]

[0088] In practical applications, compared with traditional genetic algorithms, energy consumption per ton of steel is reduced by 18.2% (see Table 1);

[0089] Table 1. Performance Comparison of Optimization Algorithms:

[0090] Genetic Algorithm 102.5 0.75 320 Quantum heuristics 83.7 0.68 95 .

[0091] In this embodiment, the knowledge evolution module needs to be specifically described. The operations of federated learning include:

[0092] A1. Gradient of the effectiveness of compensation strategies in each branch plant ▽J k Privacy mask matrix M k After encryption, a tensor block A is formed. k ;

[0093] A2. Generating a global knowledge base K by fusing tensor blocks with a progressively continued fraction structure. global ;

[0094] The expression for the asymptotic continued fraction structure is:

[0095]

[0096] Where I represents the identity matrix, and the hierarchy depth L = [∑ω i The factory's reputation weight is dynamically determined.

[0097] The global knowledge base update process executes the Lie group gradient aggregation algorithm, and its knowledge evolution process satisfies

[0098] Where g represents the Lie algebra characterizing the motion constraints of the metallurgical equipment, with an update cycle of 24 hours, and ω i This represents the factory reputation weight, the value of which is dynamically calculated based on the data quality of each branch factory (ω). i =1 - data missing rate), ad g Represent the accompaniment representation of Lie algebras;

[0099] Privacy mask matrix M k The confusion factor ρ ≥ 0.85, and its test method is as follows:

[0100] 1) Generate 1000 sets of stochastic gradients ▽J k

[0101] 2) Calculate the KL divergence between the original data and the encrypted data: KL(P) orig ||P enc >1.2;

[0102] 3) Attacker's reconstruction error > 15%;

[0103] The asymptotic continued fraction structure satisfies:

[0104] A1. The denominator identity matrix I is used to maintain the numerical stability of the knowledge representation;

[0105] A2. The depth of the continuous fractional level is dynamically trimmed based on the factory reputation weight.

[0106] In this embodiment, the control module needs to be specifically described, and the fractal topology constraint implementation methods include:

[0107] A1. Equipment group layout information is converted into fractal dimension D. f The fractal dissipation Bayesian inversion algorithm includes a multi-scale fractal decay operator, which introduces the fractal dimension D of the device group when calculating the topological distance d between devices. f Its expression is:

[0108]

[0109] The minimum number of spheres N(0.5) with a radius of r = 0.5 required for the coverage device topology map is collected by a laser rangefinder, and the fractal dimension D is... f The measurements include:

[0110] Establish a polar coordinate system with the air knife as the center, and divide the area into annular regions with a radius of 0.5m;

[0111] The number of spheres covering all devices is N (0.5), when D fWhen ∈[1.6,2.1], the attenuation exponent α is 2.3;

[0112] A2. The effective radius of the compensation rule decreases exponentially with respect to the topological distance d between devices;

[0113] The calculation rule for power-law decay is as follows:

[0114] A1. The attenuation operator ψ(d) is realized through the negative power integral of the distance. The specific calculation formula for the attenuation operator ψ(d) is as follows:

[0115]

[0116] Among them, D f The value represents the fractal number of the equipment group, ranging from 1.6 to 2.1; α represents the attenuation index, ranging from 2.3 ± 0.5; and Z represents the normalization factor, used to ensure that the thickness error ΔG < δ. G ;

[0117] A2. The attenuation index α is set to 2.3±0.5 to adapt to the layout characteristics of metallurgical workshops. Specifically, the attenuation index α is obtained by fitting the measured data of multi-workshop layouts, and the value range of 2.3±0.5 covers 90% of metallurgical scenarios (see verification data table 2). When the equipment spacing d>5m, the upper limit of 2.8 is taken, and the lower limit of 1.8 is taken in dense areas (d<2m).

[0118] Table 2. Validation data for the attenuation index α:

[0119]

[0120] The causal correlation matrix C in the causal analysis module is updated over a period of t. c The strip thickness h is dynamically adjusted based on real-time data, where h is obtained from the thickness gauge at the mill exit, and the rolling cycle t is the rolling rhythm period. c ∈[5,60]s, specifically:

[0121] When the strip thickness h is less than or equal to 0.5 mm, the rolling cycle is 5 seconds.

[0122] When the strip thickness h is between 0.5 mm and 2.0 mm, the rolling cycle is 30 seconds.

[0123] When the strip thickness h is greater than 2.0 mm, the rolling cycle is 60 seconds;

[0124] The quantum-heuristic Hamiltonian optimizer of the decision optimization module injects the latest failure probability distribution P(Y|F) before each iteration;

[0125] The global knowledge base K of the knowledge evolution module global Lie group gradient aggregation is performed every 24 hours.

[0126] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0127] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0128] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0129] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0130] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0131] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0132] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An intelligent automated control system for a complete set of metallurgical equipment, characterized in that, Specifically, it includes: The module includes causal analysis, decision optimization, knowledge evolution, and control. The causal analysis module, upon receiving an equipment anomaly signal, uses a tensor kernel function embedded with metallurgical thermodynamic constraints to perform manifold space mapping on real-time process parameters and historical fault data, generating a causal correlation matrix between equipment faults and quality indicators. ; The decision optimization module, when the causal correlation matrix at least one of them The quantum-inspired Hamiltonian optimizer performs multi-objective optimization of process parameters and outputs a set of quality compensation strategies that satisfy safety constraints. The specific operation is as follows: A1. Based on fault probability distribution Constructing Hamiltonian of the flight mass approximation term Failure probability distribution It is generated using the fractal dissipative Bayesian inversion algorithm, and the specific formula is as follows: in, Denotes the fractal dimension. Represents the convolution weight matrix. Indicates time parameter, This represents the activation function. Indicates the decay index, Represents the fault and quality vector. Represents the time-varying gradient of the causal matrix. Indicates equipment The impact of the fault varies with distance Power-law decay Represents the causal matrix The temporal gradient is used for convolutional filtering to capture fault propagation dynamics; A2. Simulate quantum tunneling evolution using FPGA voltage pulses to screen energy consumption per ton of steel. And zinc layer fluctuation Pareto solution set ; The knowledge evolution module, in the set of quality compensation strategies After execution, a federated learning mechanism is used to aggregate the decision gradients of multiple plants, and the process compensation effect is encoded into a global knowledge base in the form of tensor continued fractions. ; The control module, when a new production line is started or equipment is replaced, retrieves information from the global knowledge base. Compensation rules based on fractal topology constraints are applied to dynamically adjust the temperature of the zinc bath. The air knife pressure parameter P is used to achieve thickness error. .

2. The intelligent automated control system for complete sets of metallurgical equipment according to claim 1, characterized in that: The specific operation of manifold space mapping in the causal analysis module is as follows: A1. Reduce the dimensionality of the equipment manual text dataset, sensor time series stream, and maintenance record table to an isothermal compressed embedded manifold using the Ricci flow algorithm. ; A2. Using the causal tensor kernel function constrained by the Navier-Stokes equations, calculate the partial derivative matrix between the air knife pressure P and the zinc layer thickness G. A3. Generate a causal relationship matrix between equipment failures and quality indicators. .

3. The intelligent automated control system for complete sets of metallurgical equipment according to claim 2, characterized in that: The calculation process of the causal tensor kernel function satisfies: A1. Kernel function output values ​​and device parameters in manifold space The distances are negatively exponentially related; A2. Zinc liquid flow potential function The second-order partial derivative determinant is used as a physical constraint term; The zinc liquid flow potential function The physical constraints satisfy the equations of viscous fluid dynamics.

4. The intelligent automated control system for the complete set of metallurgical equipment according to claim 3, characterized in that: The construction rules for the Hamiltonian are as follows: A1. Correlation strength of process parameters Obtained by normalizing the elements of the causal relationship matrix; A2. The quality objective function adopts... The norm is used to calculate the deviation between the measured value and the set value.

5. The intelligent automated control system for complete sets of metallurgical equipment according to claim 4, characterized in that: The federated learning operations in the knowledge evolution module include: A1. Gradient of the effects of compensation strategies in each branch plant Through privacy mask matrix Encryption forms a tensor block ; A2. Generating a global knowledge base by fusing tensor blocks with a progressively continued fraction structure. .

6. The intelligent automated control system for complete sets of metallurgical equipment according to claim 5, characterized in that: The asymptotic continued fraction structure satisfies: A1, Denominator Identity Matrix Used to maintain the numerical stability of knowledge representation; A2. The depth of the continuous fractional level is dynamically trimmed based on the factory reputation weight.

7. The intelligent automated control system for complete sets of metallurgical equipment according to claim 6, characterized in that: The fractal topology constraint implementation methods of the control module include: A1. Converting equipment group layout information into fractal dimension ; A2. Effective radius of compensation rules and topological distance between devices It exhibits a power-law decay.

8. The intelligent automated control system for complete sets of metallurgical equipment according to claim 7, characterized in that: The calculation rule for the power-law decay is as follows: A1. Attenuation Operator This is achieved through the negative power integral of the distance; A2. Decay Index The value is set to 2.3 ± 0.5 to suit the layout characteristics of metallurgical workshops.

9. The intelligent automated control system for metallurgical complete sets of equipment according to claim 8, characterized in that: The causal relationship matrix in the causal analysis module Update cycle The strip thickness h is dynamically adjusted based on real-time data, where h is obtained from the thickness gauge at the mill exit, and the rolling rhythm cycle is also considered. ; The quantum-heuristic Hamiltonian optimizer of the decision optimization module injects the latest failure probability distribution before each iteration. ; The global knowledge base of the knowledge evolution module Lie group gradient aggregation is performed every 24 hours.

Images