A system and method for cross-line collaborative optimization of secondary metal smelting

Abstract

This invention discloses a cross-production line collaborative optimization system and method for recycled metal smelting, relating to the fields of intelligent manufacturing and collaborative control technology. The system includes a physical execution layer, a digital twin layer, and an intelligent decision-making layer. The digital twin layer contains twin models corresponding to aluminum and copper production lines respectively, as well as a cross-production line coupling analysis module for calculating global coupling indices; the intelligent decision-making layer contains a multi-objective optimization engine. The method involves real-time data acquisition, driving the twin models, and calculating a real-time comprehensive cost index C=α(t)·P. Al +β(t)·P Cu +γ·P Com The core coupling index (where α(t) and β(t) are dynamic weights related to real-time electricity prices) is used to generate a collaborative optimization instruction set with the goal of minimizing C, which is then issued to the execution layer to form a closed-loop control. This invention achieves global economic collaborative optimization of heterogeneous production lines, effectively reducing overall costs while ensuring quality and environmental protection.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing and collaborative control technology in non-ferrous metal smelting, and more specifically, to a collaborative optimization system and method for two heterogeneous smelting production lines for recycled aluminum and recycled copper. Background Technology

[0002] In the recycled metal resource utilization industry, recycled aluminum smelting and casting production lines (which typically include smelting, refining, and casting processes) and recycled copper continuous casting production lines (which typically include melting, holding, crystallization, and upward casting processes) often coexist in the same plant area. Currently, these two production lines, which differ significantly in production processes, equipment, and control objectives, generally adopt an independent operation and management model.

[0003] In this model, the two production lines and their supporting facilities (such as centralized dust removal systems) typically operate independently, making it difficult to optimize from a plant-wide perspective. Regarding energy costs: the two production lines and their supporting facilities operate independently, hindering plant-wide optimization, especially during periods of power shortages and time-of-use pricing, making it impossible to intelligently adjust the total load to achieve the lowest overall electricity cost. Secondly, regarding production quality and efficiency, passive responses to the composition of incoming waste materials lead to significant fluctuations in the metal smelting process, affecting final product quality and metal recovery rates. Furthermore, localized optimization of a single production line may come at the expense of the efficiency of the other production line or the overall plant benefits. Finally, in terms of environmental operation, the supporting environmental protection facilities typically operate at maximum capacity or in a fixed mode, resulting in high energy consumption and making precise, coordinated energy-saving control based on the total pollution load generated by the two production lines in real time impossible.

[0004] While existing technologies offer energy-saving control solutions for individual smelting equipment (such as melting furnaces) or automation upgrades for single production lines, and general digital twin modeling and artificial intelligence optimization technologies also exist, the challenge remains to be solved in the industry. This challenge lies in treating aluminum and copper production lines—which have different process principles, varying key parameters, and potentially conflicting optimization objectives—as an organic whole. Constructing a collaborative control system capable of achieving a dynamic balance between multiple objectives—comprehensive plant-wide economic benefits, stable product quality, and environmental compliance—remains a critical technical hurdle. Simply combining two independent advanced control systems is insufficient to achieve a global collaborative optimization effect across production lines. Summary of the Invention

[0005] 1. Technical problem to be solved: The present invention aims to overcome the shortcomings of the prior art and provide a cross-production line collaborative optimization system and method for recycled metal smelting, so as to solve the problems of low global resource utilization efficiency, high comprehensive production cost and weak collaborative control capability caused by the independent operation of two production lines.

[0006] 2. Technical Solution: To solve the above problems, the present invention adopts the following technical solution.

[0007] This invention provides a cross-production line collaborative optimization system for recycled metal smelting, which includes: a physical execution layer, a digital twin layer, and an intelligent decision-making layer.

[0008] The physical execution layer includes a first production line assembly for recycled aluminum melting and casting and a second production line assembly for recycled copper continuous casting.

[0009] The digital twin layer includes: a first digital twin model communicatively connected to the first production line device set, used for high-fidelity dynamic simulation of the smelting, refining and casting processes of recycled aluminum; a second digital twin model communicatively connected to the second production line device set, used for high-fidelity dynamic simulation of the melting, crystallization and upward drawing processes of recycled copper; and a cross-production line coupling analysis module, which is data-connected to the first and second digital twin models, used to receive and process the simulation output data of both, and calculate at least one global coupling index characterizing the overall operating status of the two production lines.

[0010] The intelligent decision-making layer includes: a multi-objective optimization engine connected to the cross-production line coupling analysis module, used to generate a set of collaborative optimization decision instructions by optimizing the global coupling index as the core objective and satisfying the independent process constraints of the two production lines as boundary conditions; and an instruction distribution module used to accurately distribute the set of collaborative optimization decision instructions to the corresponding controlled devices in the physical execution layer.

[0011] Furthermore, the global coupling index calculated by the cross-production line coupling analysis module is the real-time comprehensive cost index, whose mathematical expression is defined as: C=α(t)·P Al +β(t)·P Cu +γ·P Com Where C represents the real-time comprehensive cost index, P Al P Cu P Com These represent the predicted or actual power of the first production line (aluminum wire), the second production line (copper wire), and public facilities in the next scheduling cycle, respectively, with γ being a fixed weighting coefficient. α(t) and β(t) are dynamic weighting coefficients, whose values ​​are correlated with the externally input real-time electricity price signal and dynamically adjusted over time t. This formula quantifies the external market signal (real-time electricity price) into dynamic weighting coefficients for the internal optimization objective, transforming the optimization objective from simply minimizing physical energy consumption to optimizing overall economic cost.

[0012] Furthermore, the collaborative optimization decision instruction set specifically includes three types of control instructions: The first type of instruction is sent to the controller of the smelting furnace or regenerative combustion system in the first production line device set to adjust its heating power or combustion reversal frequency. The second type of instruction is sent to the controller of the holding furnace or crystallizer cooling system in the second production line unit set to adjust its temperature setpoint or the flow rate of the cooling medium (such as water); The third type of instruction is sent to the frequency converter of the common facility (such as the main dust removal fan) that serves both production lines, in order to coordinate and adjust its operating frequency.

[0013] This invention provides a method for cross-production line collaborative optimization of recycled metal smelting based on the above system.

[0014] S1: Real-time collection of production operation data from the first production line equipment set and the second production line equipment set; S2: Based on the production operation data, drive the first digital twin model and the second digital twin model to perform state simulation and generate at least one global coupling index; S3: With the optimization of the global coupling index as the main objective, and under the condition of satisfying the independent process constraints of each production line, the multi-objective optimization engine is used to solve the problem and generate a collaborative optimization decision instruction set. S4: Send the collaborative optimization decision instruction set to the physical execution layer for execution, and return to step S1 to form closed-loop control.

[0015] The present invention provides an electronic device and a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the method described thereon.

[0016] 3. Beneficial effects: Compared with the prior art, the technical solution provided by this invention has the following advantages: (1) Optimizing overall economic benefits: By constructing a real-time comprehensive cost index C model linked to real-time electricity prices, the system can generate cross-production line load adjustment instructions during peak electricity price periods, guiding energy consumption to shift to lower-cost periods, thereby reducing the overall production cost of the entire plant.

[0017] (2) Improve the stability of production quality and resource recovery rate: By using the first and second digital twin models to make advance predictions of key process states (such as the final composition of aluminum liquid and the solidification structure of copper rod), and combined with global optimization decisions, it is possible to more proactively and collaboratively compensate for the interference caused by raw material fluctuations, and effectively improve the stability of total metal yield and product quality.

[0018] (3) Achieving precise coordination between environmental compliance and operational energy consumption: The system can dynamically and collaboratively control the operating power of public environmental protection facilities (such as dust removal fans) based on the real-time operating conditions and total pollution load prediction of the two production lines, ensuring that emissions meet standards while minimizing their energy consumption.

[0019] (4) Achieve closed-loop collaborative optimization: The system can continuously optimize based on real-time data through a closed-loop process of “data perception - twin prediction - coupling analysis - global optimization - instruction execution - feedback update”, reducing the reliance on fixed manual operation experience.

[0020] It should be noted that the structures not described in this invention are not related to the design points and improvement directions of this invention, and are the same as or can be implemented using existing technologies, so they will not be elaborated here. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall architecture of the cross-production line collaborative optimization system for recycled metal smelting provided in an embodiment of the present invention.

[0022] Figure 2 This is a schematic diagram of the process flow and main monitoring and control points of the recycled aluminum melting and casting production line in an embodiment of the present invention.

[0023] Figure 3 This is a schematic diagram of the process flow and main monitoring and control points of the recycled copper continuous casting production line in an embodiment of the present invention.

[0024] Figure 4 This is a complete flowchart of the cross-production line collaborative optimization method in an embodiment of the present invention.

[0025] Figure 5 This is a schematic diagram illustrating the data flow and decision-making logic between various modules of the system in a specific collaborative optimization scenario, as shown in an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The specific parameters, communication protocols, and implementation methods mentioned in the following description are merely examples and should not be construed as the sole limitation of this invention.

[0027] To make the description of the technical solution of this invention clearer, the key parameters and variables appearing in the specification are defined uniformly below: 1. Production status parameters: T current，Al : Indicates the current real-time temperature of the aluminum melt in the smelting furnace of the first production line (recycled aluminum casting line), in degrees Celsius (°C).

[0028] [Si] current: Indicates the current real-time mass percentage content of silicon (Si) in the aluminum melt.

[0029] T mold，array : This indicates the temperature field distribution data in the crystallizer area of ​​the second production line (recycled copper upper drawing continuous casting line).

[0030] F pull : Represents the real-time traction force during the copper upward continuous casting process, measured in Newtons (N).

[0031] 2. Energy consumption and cost parameters: P Al P Cu P Com : These represent the real-time or predicted power of the first production line, the second production line, and public auxiliary facilities (such as dust removal systems), respectively, in kilowatts (kW).

[0032] C: Represents the real-time comprehensive cost index defined in this invention, which is the objective function for global optimization. It is calculated as C = α(t)·P Al +β(t)·P Cu +γ·P Com .

[0033] Price(t): Represents the real-time electricity price signal obtained from the external energy system, in yuan / kWh, and is a key economic variable affecting the cost index C.

[0034] α(t), β(t): represent weighting coefficients that are dynamically associated with Price(t), used to convert physical energy consumption into economic cost.

[0035] 3. Model prediction and decision parameters: T pred，Al [Si] pred These are the predicted future temperature of the aluminum melt and the predicted trend of silicon content change, respectively, output by the first digital twin model.

[0036] Front pred : The predicted location of the solidification interface of the copper rod output by the second digital twin model.

[0037] Min C: This represents the core objective of the multi-objective optimization engine in this invention, namely, to seek decisions that minimize the cost index C.

[0038] ΔP Al ΔT Cu ΔF: These represent the power adjustment command, temperature setting adjustment command, and fan frequency adjustment command generated by the optimization engine and to be sent to the execution device, respectively.

[0039] Example 1: Detailed System Configuration and Data Interaction Implementation like Figure 1 As shown, this system can be physically deployed on a central server in the factory and connected to the field device layer through an industrial IoT gateway and network.

[0040] 1. Specific composition of the physical execution layer: First production line assembly (recycled aluminum line): Main controlled equipment (actuators): intelligent smelting furnace (including supporting regenerative combustion system), refining powder spraying machine, intelligent aluminum alloy ingot casting machine.

[0041] Key monitoring equipment (sensors): Multiple thermocouples embedded in the smelting furnace (used to measure the molten pool temperature T). current，Al An online direct-reading spectrometer installed in the flow channel (used for real-time analysis of the aluminum melt composition, such as silicon content [Si]). current ), a continuous online flue gas monitoring system (used to measure NOx, SO2 concentrations and total flue gas volume).

[0042] Communication Interface: The programmable logic controllers (PLCs) of the aforementioned devices generally support the Modbus TCP / IP protocol. For example, the heating power setting of the smelting furnace corresponds to a specific holding register address within its PLC.

[0043] Second production line assembly (recycled copper wire): Main controlled equipment: power frequency induction melting furnace, copper liquid holding furnace, and upward continuous casting unit (the core of which is the crystallizer, traction machine and take-up device).

[0044] Key monitoring equipment: A multi-point temperature sensor array arranged around the crystallizer (used to monitor the temperature field distribution T in the crystallizer). mold，array The real-time traction speed V, fed back by the thermocouple of the insulation furnace and the traction machine servo system. pull With tension F pull .

[0045] Communication Interface: The main control system of the upward continuous casting unit mostly adopts industrial real-time Ethernet protocols such as Profinet. The opening setpoint of the crystallizer cooling water regulating valve and the traction speed setpoint are both accessible as process variables on the network.

[0046] Public facilities: Main equipment: a centralized pulse bag filter serving the entire workshop (its main fan is driven by a high-voltage frequency converter).

[0047] Key interface: Dust collector fan frequency converters typically support protocols such as Modbus RTU, and their output frequency settings can be read and written through the corresponding register address.

[0048] 2. Construction and operation of the digital twin layer model: First digital twin model (aluminum wire): Model input: Real-time collected molten pool temperature T current，Al Spectral composition [Si] current Gas flow rate, etc.

[0049] Model output: Predicted final temperature T of molten aluminum within a certain time window in the future. pred，Al And predicted values ​​of key element content [Si] pred This model is built upon the principles of thermodynamics, fluid dynamics, and metallurgical reaction kinetics, and can be implemented using a machine learning model trained on historical data or a simplified mechanistic model.

[0050] Second digital twin model (copper wire): Model input: Crystallizer temperature field data T mold，array traction speed V pull Cooling water parameters, etc.

[0051] Model output: Predicted solidification interface location of the copper rod (Front) pred and predicted grain size grade pred This model is primarily based on the heat transfer and solidification mechanisms within the crystallizer.

[0052] The core algorithm of the cross-production line coupling analysis module: This module receives the prediction outputs of the two twin models in real time and connects to the external real-time electricity price signal Price(t).

[0053] It is based on the formula C=α(t)·P Al +β(t)·P Cu +γ·P Com Calculate the global coupling index. Where P... Al P Cu P is the energy consumption for the next cycle calculated by the twin model based on the current operating conditions. Com This refers to the predicted energy consumption of public facilities based on the total flue gas volume.

[0054] The strategy for determining the dynamic weighting coefficients α(t) and β(t) can be designed according to actual needs. As an example strategy, α(t) = β(t) = K * Price(t), where K is a proportionality coefficient. This means that when the electricity price Price(t) increases, the weights increase accordingly, and the system will pay more attention to the energy consumption at that moment during optimization, tending to make more proactive energy-saving scheduling. More complex strategies can also incorporate factors such as the urgency of production line plans.

[0055] 3. Intelligent decision-making layer workflow: The multi-objective optimization engine can employ deep reinforcement learning algorithms. Its state includes: various predicted values ​​from the twin model, current energy consumption, real-time electricity price, etc. Its actions are the adjustments required by various control commands (such as power adjustment percentage, temperature adjustment value, frequency adjustment value). The core design principle of its reward function is to reward a reduction in the overall cost index C, while also rewarding the stability of product quality indicators and compliance with environmental protection standards.

[0056] The instruction distribution module is responsible for translating the abstract "action" vector output by the optimization engine into specific control instructions that conform to the communication protocol of the field devices, and ensuring that they are verified by security logic before being issued.

[0057] Example 2: A complete collaborative optimization operation scenario This embodiment combines Figure 4 and Figure 5 Taking a typical weekday afternoon operation as an example, the workflow of the system of the present invention will be specifically explained.

[0058] Scenario Initialization: Time: 14:15 PM. External Conditions: Power grid is in peak power hours, real-time electricity price Price(t) is 1.2 yuan / kWh. Production Status: The aluminum wire monitoring system detects the current silicon content [Si] of the furnace charge. current There is a fluctuating upward trend; copper wire production is stable and the heat capacity of the insulation furnace is sufficient.

[0059] Step 1: Comprehensive data perception and twin prediction.

[0060] The system periodically collects data from the entire process. The first digital twin model, based on the current data, predicts the final silicon content [Si] of the molten aluminum if the current process is maintained. pred It may exceed the upper limit. The model also simulates that if the refining power is increased by 3%, it can be controlled within the acceptable range. This will reduce P... Al The predicted value is rising. The second digital twin model predicts: the copper wire insulation furnace at the current temperature T... Cu，current Due to its significant thermal inertia, lowering the set temperature by 25°C within the next 30 minutes will not affect the front of the copper rod. pred and Grain pred And it can make P Cu The predicted value decreased significantly. The public facilities model predicts P based on the total flue gas volume. Com .

[0061] Step 2: Cross-production line coupling analysis and problem construction.

[0062] The coupling analysis module receives the predicted data. Because it's during peak power hours, it uses a higher dynamic weighting coefficient (e.g., α(t) = β(t) = 0.96). Calculations show that although the aluminum wire requires a slight increase in power consumption (ΔP... Al (positive), but the energy-saving potential achievable by copper wire (ΔP) Cu (The negative value is even greater.) After weighted calculation, a scheme that "allows a slight increase in aluminum wire consumption to maintain quality, while instructing deep energy saving in copper wire" is likely to reduce the global cost index C. The module constructs this scenario as a clear multi-objective optimization problem and submits it to the decision engine.

[0063] Step 3: Intelligent collaborative decision generation.

[0064] The multi-objective optimization engine aims to minimize the cost index C over the next 30 minutes, and rapidly solves the problem under multiple constraints, including aluminum liquid composition, copper rod mass, and emission concentration. The engine ultimately outputs a set of collaborative optimization decision instructions, such as: 1. Send an instruction to the aluminum wire melting furnace: Increase the power of the refining section by 3% (Type 1 instruction, to deal with quality fluctuations).

[0065] 2. Send a command to the copper wire insulation furnace: lower the temperature setting from 1150°C to 1125°C (second type of command, using thermal inertia to save energy).

[0066] 3. Send a command to the main dust collector fan inverter: reduce the operating frequency from 48Hz to 46.5Hz (third type of command, coordinated energy saving).

[0067] Step 4: Instruction distribution, execution, and closure.

[0068] The command distribution module securely and accurately sends three commands to three different controlled devices using the corresponding industrial network protocols (Modbus TCP, Profinet, Modbus RTU). After the devices execute the commands, the system enters the next data acquisition cycle and begins a new round of optimization based on the new state, thereby achieving continuous self-optimization and collaboration.

[0069] In this scenario, the system of this invention, based on the real-time electricity price Price(t) and the prediction results of the twin model, incorporates the quality control requirements of the aluminum production line and the energy-saving potential of the copper production line into a unified optimization objective C through a cross-production line coupling analysis module. The collaborative decision-making instruction set generated by the multi-objective optimization engine instructs the aluminum production line to slightly increase refining power to stabilize the composition, while instructing the copper production line to utilize thermal inertia to reduce the insulation temperature, and simultaneously coordinating a reduction in the operating frequency of public utilities. This decision-making process demonstrates that, while ensuring the process constraints of each production line, the system can optimize the overall cost index C by dynamically allocating loads across production lines and responding to external economic signals.

[0070] The above embodiments are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural modifications made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A cross-production line collaborative optimization system for recycled metal smelting, characterized in that, include: Physical execution layer, digital twin layer, and intelligent decision-making layer; The physical execution layer includes a first production line assembly for recycled aluminum melting and casting and a second production line assembly for recycled copper upward continuous casting. The digital twin layer includes: A first digital twin model, which is communicatively connected to the first production line unit set, is used to simulate the smelting-refining-casting process of recycled aluminum; A second digital twin model, which is communicatively connected to the second production line unit, is used to simulate the melting-crystallization-upward drawing process of recycled copper; The cross-production line coupling analysis module is connected to the first digital twin model and the second digital twin model respectively, and is used to receive the status data of the two and calculate at least one global coupling index. The intelligent decision-making layer includes: A multi-objective optimization engine, connected to the cross-production line coupling analysis module, is used to generate a collaborative optimization decision instruction set with the optimization of the coupling index as the main objective and the independent process indicators of the first and second production lines as constraints. The instruction distribution module is used to distribute the collaborative optimization decision instruction set to the corresponding device in the physical execution layer.

2. The system according to claim 1, characterized in that, The first digital twin model is configured as follows: Receive real-time temperature data from the smelting furnace in the first production line unit set, and melt composition data from the online spectrometer; Based on the received data, the system outputs the predicted final temperature and key element content of the molten aluminum using a built-in multiphysics coupling model.

3. The system according to claim 1, characterized in that, The second digital twin model is configured as follows: Receive temperature field distribution data from the crystallizer in the second production line assembly and real-time traction force data from the upward drawing machine; Based on the received data, the system outputs predicted values ​​for the solidification interface location and grain size grade of the copper rod through a built-in thermal-fluid-solid coupling model.

4. The system according to claim 1, characterized in that, The coupling index calculated by the cross-production line coupling analysis module is the real-time comprehensive cost index, and its expression is: C=α(t)·P Al +β(t)·P Cu +γ·P Com Where C is the real-time comprehensive cost index, and P Al P Cu P Com These are the actual or predicted power of the first production line, the second production line, and the public facilities, respectively. α(t) and β(t) are dynamic weighting coefficients associated with the real-time electricity price signal, and γ is a fixed weighting coefficient.

5. The system according to claim 1 or 4, characterized in that, The multi-objective optimization engine is configured to execute optimization algorithms based on deep reinforcement learning; The primary objective is to minimize the coupling index within a predetermined future time period. The constraints include at least the following: the aluminum liquid temperature of the first production line is within the first process range, the copper rod grain size of the second production line reaches the qualified level, and the predicted flue gas emission concentration of the two production lines is lower than the set threshold.

6. The system according to claim 5, characterized in that, The collaborative optimization decision instruction set includes: A first type of instruction is sent to the smelting furnace or regenerative combustion system in the first production line device set to adjust the heating power or combustion reversal frequency; The second type of instruction sent to the holding furnace or crystallizer cooling system in the second production line unit set is used to adjust the temperature setpoint or the cooling medium flow rate; Additionally, a third type of instruction is sent to the frequency converter of the dust collector fan shared by the two production lines to coordinate the adjustment of the operating frequency.

7. A method for cross-production line collaborative optimization of recycled metal smelting based on the system described in any one of claims 1 to 6, characterized in that, Includes the following steps: S1: Real-time collection of production operation data from the first production line equipment set and the second production line equipment set; S2: Based on the production operation data, drive the first digital twin model and the second digital twin model to perform state simulation and generate at least one global coupling index; S3: With the optimization of the global coupling index as the main objective, and under the condition of satisfying the independent process constraints of each production line, the multi-objective optimization engine is used to solve the problem and generate a collaborative optimization decision instruction set. S4: Send the collaborative optimization decision instruction set to the physical execution layer for execution, and return to step S1 to form closed-loop control.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method as described in claim 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method as described in claim 7.

Images