Aluminum product rolling process real-time monitoring and optimizing system based on digital twinning

Abstract

The invention provides an aluminum product rolling process real-time monitoring and optimizing system based on digital twinning, and relates to the technical field of power battery aluminum foil rolling. The monitoring and optimization system comprises a data acquisition module, a digital twin modeling module, a real-time monitoring module, a dynamic optimization module, a feedback control module, a green manufacturing optimization module and a cross-alloy adaptive module. According to the aluminum product rolling process real-time monitoring and optimizing system based on digital twinning, the data acquisition module, the digital twinning modeling module, the real-time monitoring module, the dynamic optimizing module, the feedback control module, the green manufacturing optimizing module and cross-alloy self-adaption work cooperatively; the defects in the aspects of data acquisition and integration, process parameter optimization and anomaly detection, energy consumption optimization and green manufacturing in the prior art are overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power battery aluminum foil rolling, specifically to an aluminum product rolling process real-time monitoring and optimization system based on digital twinning. BACKGROUND

[0002] In the field of aluminum product rolling, especially power battery aluminum foil production, the technology development has undergone a profound change from traditional rolling to modern intelligent production. In the early days, aluminum foil production relied on mechanical rolling mills, and operators manually adjusted the rolling force and speed to control the thickness of aluminum foil, monitor the surface quality and mechanical properties. Data collection mainly relies on manual recording and simple instruments, and rolling parameters such as rolling force, speed, oil temperature are set by experience and stored in paper logs, making it difficult to realize real-time analysis and optimization. This method is inefficient and easily affected by operator experience, and problems such as thickness deviation, pinhole density and surface defects often occur, making it difficult to meet the needs of power batteries for high elongation and high strength aluminum foil. With the introduction of automation technology, numerical control rolling mills and sensor systems have gradually become popular, high-precision thickness meters, force sensors and temperature sensors are deployed at key nodes such as rolling mill entrances, exits and rolling gaps, collecting real-time process data such as thickness 8 microns, rolling force 265 tons, speed 800 meters per minute, oil temperature 47 degrees Celsius. Industrial control systems adjust parameters through servo motors and hydraulic systems, and data is recorded through local storage devices, initially realizing the automation of the rolling process. In recent years, the application of industrial Ethernet technology has significantly improved data transmission speed and reduced communication delay to less than 1 millisecond, supporting JSON format data packet transmission, including thickness, rolling force, tension and other process data. These technological advances have improved the precision and efficiency of aluminum foil production, and have initially realized the process control of power battery aluminum foil, food packaging aluminum foil and building aluminum plate, meeting the material needs of new energy vehicles, food packaging and building curtain walls.

[0003] Although the existing technology has achieved certain results in aluminum foil production, there are still the following problems in actual application:

[0004] Problem one, the existing technology has limitations in data collection and integration. The current sensor system, industrial control system and data storage system of the rolling mill equipment operate independently, and the generated data formats are not unified. For example, the thickness gauge generates continuous time series data, the force sensor generates discrete force value data, and the optical surface detector generates image data, lacking a unified standardized processing mechanism. Data integration usually relies on manual analysis or simple script processing, which is slow, can only process hundreds of data per second, and is prone to data loss or inconsistency. For example, the thickness data and surface pinhole data of 1060H18 aluminum foil may be associated incorrectly due to different timestamps, affecting the quality control of the power battery aluminum foil. In addition, the existing system lacks real-time evaluation of data integrity and timeliness, making it difficult to dynamically detect thickness deviation or pinhole density anomalies, resulting in the inability to adjust the rolling parameters in a timely manner, affecting the performance stability of 8 to 15 micron aluminum foil.

[0005] Problem two, the existing technology is insufficient in process parameter optimization and anomaly detection. Traditional rolling systems rely on fixed process parameters, such as rolling force 265 tons, speed 800 meters per minute, and oil temperature 47 degrees Celsius, lacking real-time dynamic adjustment capability. When producing 1060H18 power battery aluminum foil, it is difficult to quickly detect anomalies such as thickness deviation, pinhole density, or surface micro-cracks, with a response time usually exceeding 100 milliseconds, affecting production efficiency and product quality. The existing system lacks prediction models based on microstructure and mechanical behavior, making it difficult to analyze grain size, dislocation density, or stress-strain distribution in real time, resulting in insufficient thickness control and fracture toughness optimization. For example, 8 micron ultra-thin aluminum foil is prone to edge cracks during high-speed rolling, and existing technology is difficult to effectively solve this problem through parameter adjustment. In addition, the existing system has limited multi-alloy adaptation capability, and when switching to produce 3003 food packaging aluminum foil or 5052 building aluminum plate, manual parameter adjustment is required, which is time-consuming and prone to errors.

[0006] Problem three, the existing technology has shortcomings in energy consumption optimization and green manufacturing. Traditional rolling processes have high energy consumption, and the operation of servo motors, hydraulic systems and lubrication systems lacks optimization, making it difficult to meet the requirements of green manufacturing. The existing system does not fully analyze the coupling relationship between heat conduction, mechanical behavior and energy consumption, making it difficult to generate energy-saving adjustment instructions. For example, when adjusting the rolling speed and oil temperature, there is a lack of real-time monitoring of energy consumption and emissions, resulting in high rolling oil mist and heat emissions, making it difficult to meet green manufacturing standards. In addition, the existing technology lacks dynamic adaptation capability for aluminum alloy characteristics, such as the microstructure differences between 1060H18 high-purity aluminum, 3003 manganese-containing alloy and 5052 magnesium-containing alloy, resulting in insufficient energy consumption-quality balance optimization.

[0007] Therefore, a digital twin-based real-time monitoring and optimization system for aluminum product rolling process is needed to solve the above problems. SUMMARY

[0008] Technical problems to be solved

[0009] In view of the deficiencies in the prior art, the present application provides an aluminum product rolling process real-time monitoring and optimization system based on digital twinning, which solves the problems mentioned in the background.

[0010] Technical scheme

[0011] To achieve the above purpose, the present application realizes the following technical scheme: an aluminum product rolling process real-time monitoring and optimization system based on digital twinning, comprising a monitoring and optimization system, the monitoring and optimization system comprising a data acquisition module, a digital twinning modeling module, a real-time monitoring module, a dynamic optimization module, a feedback control module, a green manufacturing optimization module, and a cross-alloy adaptive module, the working modes of the modules in the monitoring and optimization system being as follows:

[0012] The data acquisition module collects thickness, rolling force, tension, speed, oil temperature, and oil pressure data in the rolling process through a customized high-precision sensor array deployed at key nodes of the rolling mill, generates real-time process data, and encapsulates the data as a transmission to the digital twinning modeling module through an industrial Ethernet for building a digital twinning model;

[0013] The digital twinning modeling module receives the data transmitted by the data acquisition module, parses it as process data, builds a dynamic multi-scale digital twinning model of the microstructure and mechanical behavior of the power battery aluminum foil, generates model data, and encapsulates the model data after transmission to the real-time monitoring module and the green manufacturing optimization module through an industrial Ethernet for anomaly detection and energy consumption optimization;

[0014] The real-time monitoring module receives the model data transmitted by the digital twinning modeling module, parses it as microstructure and mechanical behavior prediction data, generates early warning signals containing thickness deviation, pinhole density, and surface microcracks, and encapsulates the early warning signals after transmission to the dynamic optimization module through an industrial Ethernet for process parameter adjustment;

[0015] The dynamic optimization module receives the early warning signals transmitted by the real-time monitoring module, parses them as abnormal type and parameter deviation data, generates adjustment instructions containing rolling force, tension, speed, oil temperature, and oil pressure, and encapsulates the adjustment instructions after transmission to the feedback control module through an industrial Ethernet for closed-loop adjustment;

[0016] The feedback control module receives the adjustment instructions transmitted by the dynamic optimization module, parses them as rolling parameter adjustment values, issues them to the rolling mill controller through an industrial control system, coordinates the closed-loop adjustment of the rolling parameters, and receives execution data for updating the adjustment instructions;

[0017] The green manufacturing optimization module receives data transmitted by the digital twin modeling module and the dynamic optimization module, parses the data into heat conduction, mechanical behavior and energy consumption data, generates energy consumption optimization instructions containing rolling speed and oil temperature, and transmits the optimization instructions to the feedback control module through industrial Ethernet for energy-saving adjustment after encapsulation.

[0018] The cross-alloy adaptive module receives model data transmitted by the digital twin modeling module, parses the data into microstructure characteristics, generates rolling process parameters adapted to various aluminum alloys, and transmits the process parameters to the dynamic optimization module through industrial Ethernet for process parameter optimization.

[0019] Preferably, the control flow of the monitoring and optimization system is designed as follows: the data acquisition module acquires rolling parameters and transmits them to the digital twin modeling module, the digital twin modeling module generates a dynamic multi-scale digital twin model and transmits it to the real-time monitoring module and the green manufacturing optimization module, the real-time monitoring module generates a warning signal and transmits it to the dynamic optimization module, the dynamic optimization module generates adjustment instructions and transmits them to the feedback control module, the feedback control module issues instructions to the rolling mill controller to drive the rolling equipment and receives execution data to update the adjustment instructions, the green manufacturing optimization module generates energy consumption optimization instructions and transmits them to the feedback control module, and the cross-alloy adaptive module generates process parameters and transmits them to the dynamic optimization module.

[0020] The data acquisition module includes a sensor array unit, a signal conversion unit and a data transmission unit, and data is collected by the following steps:

[0021] A1, through the sensor array unit, high-precision thickness gauges, force sensors, temperature sensors and optical surface detectors are fixed at the inlet and outlet of the rolling mill and key nodes to collect thickness, rolling force, tension, speed, oil temperature and oil pressure data, generate real-time process data, and the collection frequency is 300Hz, the key nodes include the area where the rolling mill directly contacts the aluminum foil, the traction rollers and guide rollers between the inlet and outlet of the rolling mill, and the positions where the rolling oil is sprayed and circulated in the rolling mill;

[0022] A2, through the signal conversion unit, the analog signals collected by the sensor array unit are converted into digital signals by an analog-to-digital converter to generate transmissible process data;

[0023] A3, through the data transmission unit, the digital signals are encapsulated and transmitted to the digital twin modeling module through the industrial Ethernet module, and 1000 data packets are processed per second.

[0024] Preferably, the digital twin modeling module includes a data analysis unit, a microstructure prediction unit, a macro-mechanical prediction unit, a coupling exchange unit and a model transmission unit; and a dynamic multi-scale digital twin model is built by the following steps:

[0025] B1, the data analysis unit receives the data transmitted by the data acquisition module, and analyzes the thickness, rolling force, tension, speed, oil temperature and oil pressure data for micro and macro behavior prediction;

[0026] B2, the microstructure prediction unit calculates the grain size and dislocation density evolution based on the aluminum foil characteristics through the crystal plasticity analysis algorithm, generates a microstructure feature vector for describing the material microstructure;

[0027] B3, the macro mechanics prediction unit calculates the stress-strain distribution based on the rolling force 265T and the tension 4.0kg / mm² through the macro mechanics analysis algorithm, generates a mechanical behavior feature vector for describing the material mechanical properties;

[0028] B4, the coupling interaction unit synchronously interacts the microstructure feature vector and the mechanical behavior feature vector in real time through the time step iteration algorithm, generates a dynamic multi-scale digital twin model with an update frequency of 300Hz, for simulating the real-time state of the rolling process;

[0029] B5, the model transmission unit encapsulates the digital twin model through industrial Ethernet, transmits it to the real-time monitoring module and the green manufacturing optimization module, processes 500 model data per second, for anomaly detection and energy consumption optimization.

[0030] Preferably, the real-time monitoring module comprises a data receiving unit, an anomaly detection unit and a signal transmission unit, and generates an early warning signal through the following steps:

[0031] C1, the data receiving unit receives the digital twin model data transmitted by the digital twin modeling module, and analyzes the microstructure and mechanical behavior prediction data for anomaly analysis;

[0032] C2, the anomaly detection unit compares the prediction data with the real-time acquisition data through the spatiotemporal correlation analysis algorithm, generates an early warning signal containing thickness deviation, pinhole density and surface microcrack, the signal specifies the type, location and degree of anomaly, generates time 3ms, for triggering process parameter adjustment;

[0033] C3, the signal transmission unit encapsulates the early warning signal through industrial Ethernet, transmits it to the dynamic optimization module, processes 1000 early warning signals per second, for the dynamic optimization module to generate adjustment instructions.

[0034] Preferably, the dynamic optimization module comprises a signal analysis unit, an instruction generation unit and an instruction transmission unit, and generates adjustment instructions through the following steps:

[0035] D1, the signal analysis unit receives the early warning signal transmitted by the real-time monitoring module, analyzes the anomaly type and parameter deviation data, for determining the process parameter adjustment requirement;

[0036] D2, the instruction generation unit generates adjustment instructions containing rolling force 265T, tension 4.0 kg / mm², speed 800 m / min, oil temperature 47°C and oil pressure 4.0 Bar based on the abnormal type and parameter deviation data by analyzing the optimization algorithm, for optimizing the rolling process parameters;

[0037] D3, the instruction transmission unit encapsulates the adjustment instructions through industrial Ethernet and transmits them to the feedback control module, which processes 500 instructions per second, for driving the rolling mill controller.

[0038] Preferably, the feedback control module includes an instruction receiving unit, a control execution unit, a feedback acquisition unit and an instruction updating unit, and realizes closed-loop regulation through the following steps:

[0039] E1, the instruction receiving unit receives the adjustment instructions transmitted by the dynamic optimization module and analyzes them into rolling parameter adjustment values, for driving the rolling mill equipment;

[0040] E2, the control execution unit issues the adjustment instructions to the rolling mill controller through the industrial control system, drives the servo motor and hydraulic system to adjust the rolling force, tension and oil pressure, for realizing closed-loop regulation of the process parameters;

[0041] E3, the feedback acquisition unit receives the execution data fed back by the rolling mill controller, containing the actual thickness, tensile strength and elongation, processes 1000 feedback data per second, for verifying the adjustment effect;

[0042] E4, the instruction updating unit updates the adjustment instructions based on the execution data, with an update period of 5 ms, for continuously optimizing the rolling parameters.

[0043] Preferably, the green manufacturing optimization module includes a data receiving unit, an energy consumption analysis unit and an instruction transmission unit, and generates energy consumption optimization instructions through the following steps:

[0044] F1, the data receiving unit receives the data transmitted by the digital twin modeling module and the dynamic optimization module and analyzes them into heat conduction, mechanical behavior and energy consumption data, for energy consumption and emission analysis;

[0045] F2, the energy consumption analysis unit calculates the energy loss and emission during the rolling process through the heat-force-energy consumption coupling analysis model, generates energy consumption optimization instructions containing rolling speed 800 m / min and oil temperature 47°C, for energy-saving regulation;

[0046] F3, the instruction transmission unit encapsulates the energy consumption optimization instructions through industrial Ethernet and transmits them to the feedback control module, which processes 500 instructions per second, for driving the rolling mill controller.

[0047] Preferably, the green manufacturing optimization module further includes a thermal-mechanical linkage unit, an optimization generation unit, and a data feedback unit, and further optimizes energy consumption through the following steps:

[0048] G1, the thermo-mechanical linkage unit receives thermo-mechanical simulation data transmitted by the digital twin modeling module, and analyzes it into temperature field and stress field distribution for the purpose of analyzing the impact on energy consumption;

[0049] G2, the optimization generation unit uses an energy consumption-quality balance analysis algorithm to generate energy-saving adjustment instructions, including a rolling speed of 800m / min and an oil temperature of 47℃, based on the solid solution characteristics of aluminum foil, to optimize energy consumption and quality.

[0050] G3, the data feedback unit transmits energy-saving adjustment commands to the feedback control module via industrial Ethernet, processing 100 commands per minute to drive energy-saving adjustments.

[0051] Preferably, the cross-alloy adaptive module includes a data parsing unit, a parameter generation unit, and a parameter transmission unit, and generates adaptation parameters through the following steps:

[0052] H1. The data parsing unit receives the model data transmitted by the digital twin modeling module and parses it into the microstructure characteristics of the aluminum alloy for process parameter adaptation.

[0053] H2. The parameter generation unit uses a transfer learning algorithm to establish a mapping rule from grain distribution to tension parameters based on the solid solution differences of aluminum alloys, and generates process parameters suitable for thicknesses of 8-20μm to support multi-alloy rolling.

[0054] H3. The parameter transmission unit encapsulates the adaptation parameters via industrial Ethernet and transmits them to the dynamic optimization module, processing 100 parameter data points per second for optimizing process parameters.

[0055] Preferably, the cross-alloy adaptive module further includes a parameter database unit, a cloud-edge collaboration unit, and an instruction issuing unit, and optimizes process parameters through the following steps:

[0056] I1, the parameter database unit stores the microstructure characteristics of aluminum alloys, containing 1000 alloy configurations, with 100 configurations updated every 12 hours to support process parameter adaptation;

[0057] I2. The cloud-edge collaborative unit receives feedback data from the real-time monitoring module and the dynamic optimization module via industrial Ethernet, and adjusts the adaptation parameters to optimize multi-alloy processes.

[0058] I3, the instruction issuing unit, transmits the adjusted adaptation parameters to the rolling mill controller via industrial Ethernet, processing 100 instructions per minute to drive the rolling mill equipment.

[0059] It should be noted that although the names of data parsing unit, data receiving unit, and instruction transmission unit are repeated in different modules, the function, input data, output results, and application scenarios of each unit are clearly different. The units in the digital twin modeling module process raw process data, the units in the real-time monitoring module focus on quality anomaly detection, the units in the green manufacturing optimization module focus on energy consumption optimization, and the units in the cross-alloy adaptive module adapt parameters to alloy characteristics. These have been clearly demonstrated in the step descriptions above. Furthermore, data transmission between modules can be encapsulated into JSON format data packets before transmission, and the next-layer module can parse and receive the required data using a compatible JSON parser; of course, the data format is not limited here.

[0060] Beneficial effects

[0061] This invention provides a real-time monitoring and optimization system for the aluminum product rolling process based on digital twins. It has the following beneficial effects:

[0062] This system, a real-time monitoring and optimization system for the aluminum product rolling process based on digital twins, addresses the shortcomings of existing technologies in data acquisition and integration, process parameter optimization and anomaly detection, energy consumption optimization, and green manufacturing through data acquisition, digital twin modeling, real-time monitoring, dynamic optimization, feedback control, green manufacturing optimization, and cross-alloy adaptive collaborative operation. It provides an efficient, accurate, and green solution for the rolling production of aluminum alloys, including 1060H18, 3003, and 5052, with a wide range of applications. The data acquisition module utilizes sensor array units deployed at key nodes of the rolling mill, such as the inlet, outlet, roll gap, tension control points, lubrication system interface, and cooling system interface. High-precision thickness gauges, force sensors, temperature sensors, and optical surface inspectors are fixed with M8 stainless steel bolts. Data is acquired at a frequency of 300 Hz, including thickness, rolling force (265 tons), speed (800 meters per minute), oil temperature (47 degrees Celsius), and oil pressure. The system generates unified JSON data packets, processing 1000 data entries per second, and transmits them to the digital twin modeling module via industrial Ethernet with a communication latency of less than 1 millisecond. The digital twin modeling module parses data packets using a JSON parser and combines alloy configurations from a cloud database, such as 1060H18 aluminum content exceeding 99.6%, to generate microstructure and mechanical behavior prediction data. It also synchronizes thickness deviation and pinhole density in real time, solving problems such as inconsistent data formats, slow processing speed, and missing data. This provides reliable input for the precise production of 1060H18 power battery aluminum foil and the surface quality optimization of 3003 and 5052 aluminum foils.

[0063] The real-time monitoring module in this system uses a spatiotemporal correlation analysis algorithm to compare the predicted data from the digital twin model with the real-time acquired data every 3 milliseconds, generating early warning signals that include thickness deviation, pinhole density, and surface microcracks. For example, it can detect a thickness deviation of 0.3 micrometers in 8-micrometer 1060H18 aluminum foil or a surface scratch length of 0.5 millimeters in 30-micrometer 3003 aluminum foil, significantly outperforming the 100-millisecond response time of traditional systems. The dynamic optimization module uses an analytical optimization algorithm based on alloy configurations in the cloud database, such as 1.0 to 1.5% manganese content in 3003 and 2.2 to 2.8% magnesium content in 5052, to generate adjustment commands such as a rolling force of 268 tons, a speed of 780 meters per minute, and a tension of 4.2 kilograms per square millimeter. This drives the servo motor and hydraulic system to adjust the roll gap and traction roll speed, ensuring thickness uniformity and a pinhole density of less than 0.1 per square meter, thus solving the edge crack problem of 1060H18 aluminum foil. The cross-alloy adaptive module uses a transfer learning algorithm to update 100 alloy configurations from the cloud database every 12 hours, generating adaptation parameters to support rapid switching between 1060H18, 3003, and 5052 alloys, with a switching time of less than 100 milliseconds, thus solving the problem of inefficiency in multi-alloy adaptation. The feedback control module updates instructions every 5 milliseconds, driving servo motors and hydraulic systems based on execution data from the rolling mill controller, such as actual thickness and tensile strength. This meets the requirements of an elongation greater than 5% for 1060H18 aluminum foil and the fracture toughness requirements for 5052 aluminum plates, addressing the issues of insufficient process parameter optimization and anomaly detection.

[0064] The green manufacturing optimization module in this system analyzes the temperature field, stress field, and motor power consumption through a thermo-mechanical-energy consumption coupled analytical model and an energy consumption-mass balance analytical algorithm. It generates energy-saving commands such as adjusting the speed to 780 meters per minute and the oil temperature to 46.5 degrees Celsius, processing 500 JSON-formatted commands per second. This drives the lubrication and cooling systems to optimize oil temperature and coolant flow, reducing rolling oil mist and heat emissions. Unit energy consumption is reduced to 120 kWh per ton, meeting green manufacturing standards. The emergency shutdown logic stops the servo motor and hydraulic system within 5 milliseconds when a surge in pinhole density or microcrack depth exceeds a safety threshold, protecting the rolling mill equipment and reducing production risks. The green manufacturing optimization module, combined with alloy configuration data from a cloud database, optimizes the thermal conductivity of 1060H18 high-purity aluminum and the strength characteristics of 5052 magnesium-containing alloys, ensuring a balance between energy consumption and mass, and improving the cycle life of power battery aluminum foil and the performance of building aluminum panels. In summary, through the coordinated operation of its various modules, the system achieves efficient monitoring, precise optimization, and green production of the aluminum foil rolling process, providing a reliable and flexible solution for the fields of power batteries, food packaging, and building curtain walls. Attached Figure Description

[0065] Figure 1 This is a system framework diagram of the present invention;

[0066] Figure 2 This is the core flowchart of the system of the present invention;

[0067] Figure 3 This is a system data interaction diagram of the present invention. Detailed Implementation

[0068] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Specific Implementation Example 1:

[0070] like Figures 1 to 3 As shown, the specific application method of this solution is as follows:

[0071] This system is designed for aluminum product rolling processes and is applicable to various aluminum alloys, including 1060H18, 3003, and 5052, with thicknesses ranging from 8 to 50 micrometers. It covers applications such as power batteries, food packaging, and building curtain walls. Deployed on an industrial rolling production line, the system utilizes digital twin technology to construct a dynamic multi-scale model of microstructure and mechanical behavior through data acquisition, digital twin modeling, real-time monitoring, dynamic optimization, feedback control, green manufacturing optimization, and cross-alloy adaptive modules. This allows for real-time monitoring and optimization of the rolling process, addressing issues such as surface quality fluctuations, thickness control, and fracture toughness. The system transmits JSON data packets via industrial Ethernet with a communication latency of less than 1 millisecond, enabling precise adjustment of process parameters, energy consumption optimization, and multi-product adaptation. The following details the system's application flow in various product scenarios, covering the adoption of digital twin technology, real-time monitoring and optimization mechanisms, specific machine control execution methods, the usage logic of alloy configuration in the cloud database, the operating logic of each module, data processing, and control flow.

[0072] The system is deployed on a four-roll irreversible cold rolling mill production line, equipped with servo motors, hydraulic systems, lubrication systems, and cooling systems. The production targets include 1060H18 aluminum foil for power batteries, with a thickness of 8 to 15 micrometers, a width of 800 to 1200 millimeters, and a roll weight of no more than 5 tons; 3003 aluminum foil for food packaging, with a thickness of 15 to 30 micrometers and a width of 600 to 1000 millimeters; and 5052 aluminum plates for building curtain walls, with a thickness of 30 to 50 micrometers and a width of 1000 to 1500 millimeters. This system is primarily applicable to aluminum materials used in batteries, but it is also suitable for aluminum foil boxes used in food packaging and building materials, demonstrating its wide range of applications. The data acquisition module uses a sensor array unit to collect real-time process data at key nodes of the rolling mill. These key nodes include the mill inlet (the position of the aluminum material before entering the rolls), the mill outlet (the position of the aluminum material after leaving the rolls), the roll gap area (the area where the rolls directly contact the aluminum material), the tension control point (located at the traction and guide rolls between the inlet and outlet), the lubrication system interface (the location of rolling oil injection and circulation), and the cooling system interface (the location of coolant injection). The sensor array unit uses M8 stainless steel bolts to fix a high-precision thickness gauge, force sensor, temperature sensor, pressure sensor, and optical surface inspector. The high-precision thickness gauge measures the aluminum material thickness, the force sensor detects rolling force and tension, the temperature sensor monitors oil and coolant temperatures, the pressure sensor detects oil pressure, and the optical surface inspector scans for surface defects including pinholes, microcracks, and scratches. The sensors collect data at a frequency of 300 Hz, generating real-time process data, including thickness, rolling force, tension, speed, oil temperature, and oil pressure, adapting to different product requirements. For example, for power battery aluminum foil, the focus is on pinholes and thickness uniformity; for food packaging aluminum foil, the focus is on surface smoothness; and for architectural aluminum panels, the focus is on tensile strength. The signal conversion unit converts analog signals into digital signals through an analog-to-digital converter, generating transmittable process data. The data transmission unit encapsulates the digital signals into JSON format data packets via an industrial Ethernet module, processing 1000 data packets per second, adding timestamps and node identifiers, such as time July 15, 2025, 17:50:00:123 milliseconds, at the node output, transmitting the data to the digital twin modeling module, providing input for building multi-product adapted digital twin models.

[0073] The digital twin modeling module employs digital twin technology. By receiving JSON format data packets transmitted from the data acquisition module, it constructs a dynamic, multi-scale digital twin model of the microstructure and mechanical behavior of aluminum materials, supporting various alloys such as 1060H18, 3003, and 5052. The data parsing unit uses a JSON parser to parse the data packets into process data including thickness, rolling force (reference value 265 tons), tension (reference value 4.0 kg / mm²), speed (reference value 800 m / min), oil temperature (reference value 47 degrees Celsius), and oil pressure (reference value 4.0 bar), which are then provided to the microstructure prediction unit and the macroscopic mechanical prediction unit. The microstructure prediction unit extracts alloy configurations from a cloud database using a crystal plasticity analysis algorithm. Examples include high-purity aluminum (aluminum content above 99.6%) for 1060H18, manganese content (1.0-1.5%) for 3003, and magnesium content (2.2-2.8%) for 5052. It calculates grain size and dislocation density evolution, generating microstructure feature vectors to describe changes in the material's microstructure. For instance, fine grains in 1060H18 facilitate thickness control, larger grains in 3003 affect surface quality, and grain boundary strengthening in 5052 enhances tensile strength. The macromechanical prediction unit uses a macromechanical analysis algorithm, based on a rolling force of 265 tons and a tension of 4.0 kg / mm², combined with alloy mechanical properties from the cloud database, to calculate stress-strain distribution and generate mechanical behavior feature vectors. These vectors describe the deformation resistance and fracture risk of different products; for example, aluminum foil for power batteries requires high elongation, while aluminum panels for construction require high strength. The coupling interaction unit uses a time-stepping iterative algorithm with a time step of 3 milliseconds to synchronize microstructure feature vectors (grain size and dislocation density) and mechanical behavior feature vectors (stress-strain field) in real time, generating a dynamic multi-scale digital twin model with an update frequency of 300 Hz. This simulates the real-time state of different product rolling processes. For example, the power battery aluminum foil model optimizes thickness uniformity, the food packaging aluminum foil model reduces surface defects, and the architectural aluminum panel model improves fracture toughness. The digital twin model runs on an edge computing server, retrieving alloy configuration updates from a cloud database every 12 hours, including grain size, alloy element ratios, and mechanical properties, ensuring the model is adapted to the latest alloy characteristics. The model transmission unit encapsulates the digital twin model into JSON format data packets via industrial Ethernet, processing 500 model data entries per second, adding timestamps and node identifiers, and transmitting them to the real-time monitoring module and green manufacturing optimization module for real-time monitoring and energy consumption optimization.

[0074] The real-time monitoring module utilizes digital twin technology to monitor the rolling process in real time. It receives JSON-formatted model data from the digital twin modeling module and parses it using a JSON parser into predicted data for microstructure (grain size and dislocation density) and mechanical behavior (stress-strain field), adapting to different product characteristics. The data receiving unit provides the parsed data to the anomaly detection unit. The anomaly detection unit uses a spatiotemporal correlation analysis algorithm to compare the predicted data with the real-time data acquired by the data acquisition module, identifying thickness deviations (8-50 micrometers off target thickness), pinhole density (micropores on the surface), surface microcracks (abnormal length or depth), and scratches (surface abrasions). The anomaly type is adjusted according to product requirements; for example, for power battery aluminum foil, pinholes and thickness deviations are prioritized; for food packaging aluminum foil, scratches are monitored; and for architectural aluminum panels, microcracks are prioritized. When an anomaly is detected, a warning signal is generated. The signal includes the anomaly type (e.g., thickness deviation, pinhole, microcrack, or scratch), location (e.g., inlet, outlet, or roll gap), and degree (e.g., deviation amount or defect density), generated within 3 milliseconds. For example, when producing 8-micron 1060H18 aluminum foil, if the exit thickness is 0.3 microns higher than expected, a warning signal will specify the anomaly type as thickness deviation, the location as exit, and the extent as 0.3 microns higher. When producing 30-micron 3003 aluminum foil, if surface scratches are detected, a warning signal will specify the anomaly type as scratch, the location as exit, and the extent as length as 0.5 millimeters. When producing 50-micron 5052 aluminum plate, if microcracks are detected, a warning signal will specify the anomaly type as microcracks, the location as roll gap, and the extent as depth as 0.2 millimeters. The signal transmission unit encapsulates the warning signals into JSON format data packets via industrial Ethernet, processes 1000 warning signals per second, and transmits them to the dynamic optimization module to trigger process parameter optimization. If a serious anomaly is detected, such as a surge in pinhole density or a microcrack depth exceeding the safety threshold, the system generates an emergency stop command, which is sent to the mill controller through the industrial control system to stop the servo motors and hydraulic system, protecting the mill equipment.

[0075] The dynamic optimization module optimizes process parameters using digital twin technology. It receives JSON-formatted early warning signals from the real-time monitoring module and parses them into anomaly types such as thickness deviation, pinholes, microcracks, or scratches, and parameter deviation data such as deviation amount or defect density. The signal parsing unit provides the parsed data to the instruction generation unit. The instruction generation unit, using the parsing optimization algorithm, generates adjustment instructions adapted to different products based on the anomaly type, deviation data, and alloy configuration in the cloud database. These instructions include a rolling force of 265 tons, tension of 4.0 kg / mm², speed of 800 m / min, oil temperature of 47 degrees Celsius, and oil pressure of 4.0 bar. For example, when producing 1060H18 aluminum foil, if the thickness deviation is 0.3 micrometers too high, the algorithm, taking into account the high-purity aluminum characteristics of 1060H18, increases the rolling force to 270 tons or reduces the speed to 780 meters per minute; when producing 3003 aluminum foil, if surface scratches occur, the algorithm, taking into account the manganese content of 3003, reduces the speed to 770 meters per minute; when producing 5052 aluminum plate, if the risk of microcracks increases, the algorithm, taking into account the magnesium content of 5052, reduces the tension to 3.8 kg / mm². The instruction generation unit sets priorities based on product characteristics, for example, prioritizing thickness control for power battery aluminum foil, surface quality for food packaging aluminum foil, and fracture toughness for building aluminum plates, while adhering to process constraints such as a rolling force not exceeding 280 tons, ensuring that adjustment instructions are stable and executable. The instruction transmission unit encapsulates the adjustment instructions into JSON format data packets via industrial Ethernet, processes 500 instructions per second, adds timestamps and node identifiers, and transmits them to the feedback control module to drive the mill controller to perform optimization.

[0076] The feedback control module implements closed-loop execution of optimization commands. It receives JSON-formatted adjustment commands from the dynamic optimization module and parses them into rolling parameter adjustment values, including rolling force, tension, speed, oil temperature, and oil pressure, adapting to different product processes. The command receiving unit provides the parsed data to the control execution unit. The control execution unit sends the adjustment commands to the mill controller through the industrial control system, driving the equipment to execute specific control actions. The servo motor receives rolling force adjustment commands and controls the roll gap by adjusting the motor speed. For example, when increasing the rolling force to 270 tons, the servo motor speed increases by 10 revolutions per minute, decreasing the roll gap by 0.01 mm. It also receives tension adjustment commands and controls the tension by adjusting the traction roll motor speed. For example, when reducing the tension to 3.8 kg / mm², the traction roll motor speed decreases by 10 revolutions per minute. The hydraulic system receives oil pressure adjustment commands and controls the oil pressure by adjusting the hydraulic pump valve opening. For example, when adjusting the oil pressure to 4.0 bar, the valve opening increases by 5 degrees. The lubrication system receives oil temperature adjustment commands and controls the oil temperature by adjusting the heater power and oil pump flow rate. For example, when adjusting the oil temperature to 47 degrees Celsius, the heater power increases by 100 watts, and the oil pump flow rate increases by 0.2 liters per minute. The cooling system receives coolant temperature adjustment commands and controls the temperature by adjusting the cooling pump valve and coolant circulation flow rate. For example, when adjusting the coolant temperature to 40 degrees Celsius, the cooling pump valve opening increases by 3 degrees, and the flow rate increases by 0.1 liters per minute. The feedback acquisition unit receives execution data from the rolling mill controller, including actual thickness, tensile strength, and elongation. It processes 1000 feedback data points per second to verify the adjustment effect, such as confirming that the thickness of 1060H18 aluminum foil reaches 8 micrometers, or that the tensile strength of 5052 aluminum plate meets construction requirements. The instruction update unit updates and adjusts instructions based on the executed data. For example, if the thickness of 1060H18 aluminum foil is still 0.2 micrometers too high, the rolling force is adjusted to 272 tons, and the servo motor further reduces the roll gap by 0.005 millimeters. If the scratches on the surface of 3003 aluminum foil are not eliminated, the speed is reduced to 770 meters per minute, and the traction roll motor decelerates by 5 revolutions per minute. The update cycle is 5 milliseconds, ensuring that closed-loop regulation continuously optimizes process parameters.

[0077] The green manufacturing optimization module optimizes energy consumption and emissions using digital twin technology. It receives JSON data from the digital twin modeling module and the dynamic optimization module, and uses a JSON parser to parse it into thermal conduction (temperature field distribution), mechanical behavior (stress-strain field), and energy consumption data (power consumption of servo motors, hydraulic systems, and cooling systems), adapting it to the energy consumption requirements of different products. The data receiving unit provides the parsed data to the energy consumption analysis unit and the thermo-mechanical linkage unit. The energy consumption analysis unit uses a coupled thermo-mechanical-energy consumption analytical model, combined with alloy configurations from a cloud database, to calculate energy losses during the rolling process, such as motor and oil pump power consumption, and emissions, such as rolling oil mist and heat emissions. It then generates energy consumption optimization instructions including a rolling speed of 800 meters per minute and an oil temperature of 47 degrees Celsius. For example, when producing 1060H18 aluminum foil, the speed is reduced to 780 meters per minute, and the traction roller motor is reduced by 8 revolutions per minute to reduce motor power consumption; the oil temperature is adjusted to 46.5 degrees Celsius, and the heater power is reduced by 50 watts to reduce oil mist emissions; when producing 5052 aluminum plates, the coolant temperature is optimized to 40 degrees Celsius, and the cooling pump valve opening is increased by 2 degrees to reduce heat emissions. The thermo-mechanical linkage unit analyzes the thermo-mechanical simulation data from the digital twin modeling module, extracts the temperature and stress field distributions, and analyzes the impact of rolling speed, oil temperature, and coolant temperature on energy consumption. The optimization generation unit uses an energy consumption-mass balance analysis algorithm, based on alloy configurations in the cloud database such as the thermal conductivity of 1060H18 high-purity aluminum, the thermal stability of 3003 containing manganese, and the strength of 5052 containing magnesium, to generate energy-saving adjustment instructions including rolling speed and oil temperature, ensuring a balance between energy consumption and quality. For example, low energy consumption is prioritized for power battery aluminum foil, while strength is prioritized for architectural aluminum plates. The instruction transmission unit encapsulates energy-saving instructions into JSON format data packets via industrial Ethernet, processing 500 instructions per second and transmitting them to the feedback control module. This drives the mill controller to perform energy-saving adjustments, such as adjusting the servo motor speed or hydraulic pump valves. The data feedback unit processes 100 instructions per minute, feeding back the energy-saving effects to the feedback control module to optimize subsequent adjustments.

[0078] The cross-alloy adaptive module achieves multi-alloy adaptation through digital twin technology. It receives JSON-formatted model data transmitted from the digital twin modeling module and parses it into microstructure characteristics of aluminum alloys such as 1060H18, 3003, and 5052 using a JSON parser. This includes grain distribution and alloy element ratios, such as manganese and magnesium content. The data parsing unit provides the parsed data to the parameter generation unit. The parameter generation unit extracts alloy configurations from a cloud database using a transfer learning algorithm. For example, 1060H18 has an aluminum content higher than 99.6%, 3003 has a manganese content of 1.0 to 1.5%, and 5052 has a magnesium content of 2.2 to 2.8%. It establishes a mapping rule from grain distribution to tension parameters and generates process parameters suitable for thicknesses of 8 to 50 micrometers. For example, 1060H18 aluminum foil has a tension of 4.0 kg / mm², 3003 aluminum foil has a tension of 4.5 kg / mm² to prevent breakage, and 5052 aluminum plate has a tension of 3.8 kg / mm² to enhance strength. The parameter database unit stores 1000 alloy configurations, including 1060H18, 3003, and 5052. Every 12 hours, it updates 100 configurations via a cloud database, including the latest grain size, alloy element ratios, and mechanical properties. After the update, the parameter generation unit recalculates the process parameters based on the latest configuration. For example, if the manganese content of the 3003 alloy is updated to 1.2%, the tension is adjusted to 4.4 kg / mm². The cloud-edge collaboration unit receives early warning signals from the real-time monitoring module and adjustment instructions from the dynamic optimization module via industrial Ethernet. It then adjusts the adaptation parameters based on the cloud database configuration. For example, based on feedback regarding the pinhole density of the 3003 aluminum foil, the tension is reduced to 4.3 kg / mm², and the traction roller motor speed is reduced by 6 revolutions per minute. The parameter transmission unit encapsulates the adaptation parameters into JSON format data packets via industrial Ethernet, processes 100 parameter data entries per second, and transmits them to the dynamic optimization module for the instruction generation unit to generate adjustment instructions. The instruction issuing unit transmits the adjusted parameters to the rolling mill controller via industrial Ethernet, processing 100 instructions per minute to drive servo motors and traction rollers to perform parameter adjustments, ensuring stable production of multiple alloys.

[0079] The system operates within an industrial control network, equipped with an edge computing server to process the digital twin model, a cloud database to store alloy configurations, and an industrial Ethernet network to achieve communication latency of less than 1 millisecond. The core of the digital twin technology lies in generating microstructure feature vectors through crystal plasticity analysis algorithms, generating mechanical behavior feature vectors through macroscopic mechanical analysis algorithms, and synchronizing microscopic and macroscopic properties through time-stepping iterative algorithms. Combined with the alloy configurations in the cloud end database, it simulates the rolling process in real time. After the mill starts, the data acquisition module collects real-time process data at a frequency of 300 Hz and transmits it to the digital twin modeling module. The digital twin modeling module updates the multi-scale model every 3 milliseconds and transmits it to the real-time monitoring module and the green manufacturing optimization module. The real-time monitoring module generates an early warning signal every 3 milliseconds and transmits it to the dynamic optimization module. The dynamic optimization module generates adjustment instructions every 5 milliseconds and transmits them to the feedback control module. The feedback control module drives the mill controller and receives execution data every 5 milliseconds, updating the instructions. The green manufacturing optimization module generates energy-saving instructions every second and transmits them to the feedback control module. The cross-alloy adaptive module generates adaptation parameters every second and transmits them to the dynamic optimization module. The entire process forms a closed loop: data collection, digital twin model construction, anomaly detection, parameter optimization, adjustment execution, and feedback updates. Simultaneously, the green manufacturing optimization module and the cross-alloy adaptive module optimize energy consumption and multi-alloy compatibility in parallel. In a real-world scenario, when producing 8-micron 1060H18 aluminum foil, the system detects an exit thickness deviation of 0.3 microns, generating a warning signal: anomaly type thickness deviation, location exit, degree increased by 0.3 microns. The dynamic optimization module generates an instruction to increase the rolling force to 268 tons. The feedback control module drives the servo motor to increase its speed by 10 revolutions per minute and decrease the roll gap by 0.01 millimeters. The green manufacturing optimization module adjusts the speed to 780 meters per minute and reduces the traction roller motor speed by 8 revolutions per minute to reduce energy consumption. The cross-alloy adaptive module optimizes the tension to 4.2 kilograms per square millimeter based on the 1060H18 configuration and reduces the traction roller motor speed by 5 revolutions per minute. When producing 30-micron 3003 aluminum foil, a surface scratch was detected, generating an abnormal scratch warning signal. The scratch was located at the exit point, with a depth of 0.5 mm. The dynamic optimization module reduced the speed to 770 m / min, and the traction roller motor slowed down by 5 rpm. The cross-alloy adaptive module adjusted the tension to 4.3 kg / mm² based on the 3003 configuration, and the traction roller motor slowed down by 6 rpm. When producing 50-micron 5052 aluminum plate, a microcrack risk was detected, generating an abnormal microcrack warning signal. The microcrack was located at the roll gap, with a depth of 0.2 mm. The dynamic optimization module reduced the tension to 3.8 kg / mm², and the traction roller motor slowed down by 10 rpm. The feedback control module drove the traction roller to adjust the tension, and the green manufacturing optimization module adjusted the coolant temperature to 40 degrees Celsius. The cooling pump valve opening was increased by 2 degrees to reduce heat dissipation. The system continued to run until the product met requirements, with no pinholes, microcracks, or scratches.If a serious anomaly is detected, such as a surge in pinhole density, the system generates an emergency stop command to halt the servo motors and hydraulic system, protecting the rolling mill equipment. Specific Implementation Example 2:

[0081] like Figures 1 to 3 As shown, the following details the algorithm logic of each module and its specific application in the system:

[0082] The data acquisition module utilizes a core signal processing algorithm, specifically designed to convert raw signals from sensors into process data usable in the digital twin model, providing precise input for subsequent modeling. Input consists of analog signals captured by the sensor array, including voltage signals from the thickness gauge, strain signals from the force sensor, thermocouple signals from the temperature sensor, pressure signals from the pressure sensor, and image signals from the optical surface inspection instrument, covering key parameters of the rolling process. Output is digital process parameter data, including thickness, rolling force, tension, speed, oil temperature, and oil pressure, with accuracy reaching 0.01 micrometers, 0.1 tons, 0.1 kilograms per square millimeter, 0.1 meters per minute, 0.1 degrees Celsius, and 0.1 bar. The algorithm filters out signal noise and performs linear mapping to convert analog signals into digital values, processing at a frequency of 300 Hz to ensure real-time capture of dynamic process changes. Within the system, the algorithm's output data is encapsulated as JSON data packets and transmitted via industrial Ethernet to the digital twin modeling module, providing foundational data for constructing multi-scale models. The algorithm is adapted to different product requirements. For example, it provides high-precision thickness data for power battery aluminum foil to support uniformity optimization, provides image data related to surface defects for food packaging aluminum foil to improve smoothness, and provides mechanical data related to tensile strength for building aluminum panels to enhance mechanical performance, thus laying the foundation for modeling and monitoring of subsequent modules.

[0083] The digital twin modeling module comprises three core algorithms that collectively construct a dynamic, multi-scale digital twin model, driving real-time simulation and optimization of the rolling process. The first algorithm is the crystal plasticity analytical algorithm, a term defined in this invention, referring to a computational method for simulating the microstructural changes of metallic materials (such as aluminum foil) during rolling. Based on the crystal plasticity theory of materials science and combined with rapid analytical techniques, it predicts changes in grain size and dislocation density within the aluminum foil. This algorithm improves upon traditional crystal plasticity simulation methods (such as the crystal plasticity finite element method) by simplifying calculations and enhancing real-time performance, making it suitable for rapid response in production environments. The algorithm analyzes the microscopic deformation behavior of the aluminum foil under rolling force, speed, and other process conditions, simulating how grains shrink or how dislocations (crystal defects) increase. It utilizes the material properties of aluminum foil (such as alloy composition) and rolling parameters to quickly generate characteristic data describing the material's microstructure, similar to drawing a "map of changes in the material's internal structure." Inputs include process parameters from the data acquisition module, such as rolling force of 265 tons and speed of 800 meters per minute, and alloy configurations from a cloud database, such as 1060H18 with aluminum content higher than 99.6%, 3003 with manganese content of 1.0 to 1.5%, and 5052 with magnesium content of 2.2 to 2.8%. Outputs are microstructure feature vectors, including grain size and dislocation density. The algorithm generates feature vectors describing the material's microstructure by analyzing the influence of alloying elements on grain growth and dislocation movement. In the system, the algorithm's output provides microscopic prediction data to the real-time monitoring module for detecting anomalies related to microstructure. For example, pinholes in power battery aluminum foil are related to grain size, surface smoothness in food packaging aluminum foil is related to dislocation density, and fracture toughness in architectural aluminum panels is related to grain boundary characteristics, thus supporting the accuracy of anomaly detection.

[0084] Processing steps:

[0085] The initial properties of the aluminum foil, such as its crystal structure and alloy composition, are read. Based on parameters such as rolling force and speed, the deformation of the crystals is analyzed. The changes in grain size and the trend of dislocation increase are calculated. The results are then organized into an eigenvector for use in subsequent modules.

[0086] In the microstructure prediction unit, the algorithm simulates the microstructural changes of aluminum foil (such as 1060H18) in real time at a frequency of 300 times per second, helping to build a digital twin model. For example, when the rolling force increases, the algorithm predicts that the grains will become smaller or the number of dislocations will increase, which may lead to material hardening and can be used to detect anomalies such as pinholes. This improves the accuracy of aluminum foil production for power batteries, ensuring thickness uniformity and surface quality.

[0087] The second algorithm is a macroscopic mechanical analytical algorithm, a term defined in this invention, referring to a calculation method for assessing the overall mechanical properties (such as stress distribution) of aluminum foil during the rolling process. Based on continuum mechanics and combined with rapid analytical techniques, it reduces the complex calculations of traditional finite element analysis and is suitable for real-time monitoring of aluminum foil deformation behavior. This algorithm analyzes the influence of forces (such as rolling force and tension) on the aluminum foil during the rolling process, calculating how the material is stretched or deformed under external forces, generating characteristic data describing the overall stress and deformation. This is analogous to predicting the deformation of a bridge by measuring the forces acting on it, ensuring the aluminum foil remains stable during rolling. Inputs include: process parameters (such as rolling force 265 tons, tension 4.0 kg / mm²) and the mechanical properties of the aluminum foil (such as strength and elasticity, obtained from a cloud database). Output: a mechanical behavior feature vector (a numerical set describing stress and deformation).

[0088] Processing steps: Collect process parameters such as rolling force and tension, as well as material properties. Analyze the stress distribution of the material under these forces (e.g., where the stress is greatest). Calculate the deformation of the material (e.g., degree of tension or compression). Organize the results into a feature vector for use by subsequent modules.

[0089] In the macroscopic mechanical prediction unit, this algorithm works in conjunction with the microscopic algorithm to generate the macroscopic part of the digital twin model. For example, for an 8-micrometer-thick aluminum foil for power batteries, it predicts stress concentration areas, indicates potential microcracks, and triggers the dynamic optimization module to adjust the tension (e.g., 4.0 kg / mm²) to prevent fracture. This algorithm ensures that the system simulates the rolling process at a frequency of 300 times per second, improving the stability of the material's mechanical properties.

[0090] The third algorithm is the time-stepping iterative algorithm, a term defined in this invention, referring to a computational method for real-time synchronization of microscopic and macroscopic data. Based on time discretization technology, it ensures the digital twin model reflects the rolling process in real time by progressively updating the model in chronological order. This algorithm decomposes the rolling process into small time intervals (e.g., every 3 milliseconds), progressively updating microscopic structure (e.g., grain size) and macroscopic mechanical (e.g., stress distribution) data. Through repeated checks and adjustments, it ensures consistency between the two sets of data, much like updating frames in an animation, generating a dynamic digital twin model. Inputs include microscopic structure feature vectors (grain size, dislocation density) and mechanical behavior feature vectors (stress, deformation). Output: a dynamic multi-scale digital twin model (a synchronized set of microscopic and macroscopic data).

[0091] Processing steps:

[0092] Set a time interval (e.g., 3 milliseconds). Update microscopic data (e.g., changes in grain size). Synchronize macroscopic data (e.g., changes in stress distribution). Check for consistency between microscopic and macroscopic data; adjust if inconsistencies exist. Generate a complete dynamic model for use by subsequent modules.

[0093] Within the coupled interactive unit, the algorithm synchronizes microscopic and macroscopic data at a frequency of 300 times per second, supporting real-time simulation of the rolling process. For example, in the production of 3003 alloy aluminum foil, it predicts the impact of speed variations on grain size and stress, enabling the green manufacturing optimization module to adjust parameters (such as a speed of 780 m / min). This algorithm ensures real-time model updates, addressing the latency issues inherent in traditional simulations.

[0094] The real-time monitoring module uses a spatiotemporal correlation analysis algorithm, a term defined in this invention, referring to a calculation method for detecting anomalies during the rolling process. Based on statistical analysis techniques, it identifies anomalies such as thickness deviation and pinholes by comparing the predicted data from the digital twin model with the actual acquired data. The algorithm analyzes the temporal and spatial differences in the data to determine whether the prediction matches the actual result. For example, it checks whether the thickness of the aluminum foil at the mill inlet and outlet deviates from the target value, or whether there are abnormal pinholes on the surface, similar to finding defects by comparing photos and the actual product. Its specific application lies in comparing the predicted data from the digital twin model with real-time process data to generate early warning signals to trigger optimization. The input includes the predicted data from the digital twin modeling module, including microstructure and mechanical behavior feature vectors, and the real-time process data from the data acquisition module. The output is an early warning signal, containing anomaly types such as thickness deviation, pinholes, microcracks or scratches, locations such as inlet, outlet, or roll gap, and degrees such as deviation amount or defect density. The algorithm identifies anomalies by analyzing the spatiotemporal differences between the predicted data and the real-time data, generating the signal in 3 milliseconds. The algorithm adjusts its detection priorities based on product characteristics. For example, for power battery aluminum foil, it prioritizes detecting thickness deviations and pinholes to ensure battery performance; for food packaging aluminum foil, it prioritizes detecting scratches to ensure packaging smoothness; and for building aluminum panels, it prioritizes detecting microcracks to improve structural safety. In the system, the algorithm's warning signals are transmitted to the dynamic optimization module via industrial Ethernet, triggering adjustments to process parameters. If a serious anomaly is detected, the algorithm generates an emergency stop command, which is issued through the industrial control system to protect the equipment. The algorithm ensures the real-time and targeted nature of anomaly detection, supporting rapid response across multiple product scenarios.

[0095] The algorithm of the dynamic optimization module is an analytical optimization algorithm, a term defined in this invention, referring to an optimization method for generating process parameter adjustment instructions. Based on optimization theory, it generates optimal rolling parameter adjustment schemes by rapidly analyzing abnormal data, similar to automatically adjusting machine settings based on problems. This algorithm calculates optimal adjustment values ​​for parameters such as rolling force and speed based on the anomaly type (e.g., thickness deviation) and aluminum foil characteristics, ensuring rapid recovery to normal production. It finds the most suitable adjustment scheme by analyzing the impact of anomalies while adhering to equipment limitations (e.g., rolling force not exceeding 280 tons). Its specific application lies in generating optimization instructions based on anomaly type and alloy characteristics to adjust rolling parameters to improve product quality. Inputs include early warning signals from the real-time monitoring module, including anomaly type and deviation data, as well as alloy configurations from the cloud database. Outputs are adjustment instructions, including rolling force, tension, speed, oil temperature, and oil pressure. The algorithm generates optimization instructions by analyzing the impact of anomalies on process parameters and combining them with alloy characteristics. For example, it adjusts the rolling force to control the thickness of power battery aluminum foil, adjusts the speed to reduce scratches for food packaging aluminum foil, and adjusts the tension to reduce the risk of breakage for building aluminum panels. The algorithm adheres to process constraints, such as rolling force not exceeding 280 tons, ensuring the execution of instructions. Within the system, adjustment instructions from the algorithm are transmitted to the feedback control module via industrial Ethernet, driving the equipment to perform optimizations and ensuring that different products meet target quality requirements.

[0096] The feedback control module comprises two algorithms that drive closed-loop regulation for continuous optimization of process parameters. The first algorithm is the equipment control algorithm, which takes as input the adjustment commands from the dynamic optimization module and outputs equipment control signals. The algorithm maps rolling parameters to equipment action signals; for example, it maps rolling force adjustment to servo motor signals, tension adjustment to traction roll signals, and oil pressure, oil temperature, and coolant temperature adjustments to hydraulic system, lubrication system, and cooling system signals. In the system, the algorithm's output drives the equipment to execute parameter adjustments, ensuring that process parameters reach target values. For example, it adjusts the thickness of aluminum foil for power batteries and optimizes the surface quality of aluminum foil for food packaging. The second algorithm is the feedback update algorithm, which takes as input the execution data fed back by the equipment, including actual thickness, tensile strength, and elongation, as well as target parameters, and outputs updated adjustment commands. The algorithm generates new commands by comparing actual data with target values, optimizing subsequent adjustments. For example, it generates commands to further adjust the rolling force for aluminum foil for power batteries and commands to reduce the speed for aluminum foil for food packaging. In the system, the algorithm implements closed-loop regulation with an update cycle of 5 milliseconds, ensuring continuous optimization of process parameters and adapting to various product requirements.

[0097] The green manufacturing optimization module includes two algorithms to optimize energy consumption and emissions for green manufacturing. The first algorithm is the thermo-mechanical-energy coupling analytical algorithm, a term defined in this invention, referring to a computational method for analyzing the interaction between heat conduction, mechanical behavior, and energy consumption during the rolling process. Based on existing thermo-mechanical coupling analysis techniques (such as heat conduction equations and finite element analysis) and energy consumption calculation models, it emphasizes "coupling analysis" to highlight its ability to quickly analyze the interplay of temperature, mechanics, and energy consumption, generating energy-saving and low-emission process parameter adjustment schemes. This algorithm is applicable to the green manufacturing optimization module, supporting real-time optimization of energy consumption and emissions in aluminum foil production. It comprehensively analyzes temperature changes (e.g., roll oil temperature 47°C), mechanical parameters (e.g., rolling force 265T), and energy consumption data (e.g., motor power consumption) during the rolling process, calculating how they interact to generate optimized process parameters (e.g., speed 780m / min). It is similar to analyzing the heat generation, stress, and power consumption of a machine to find an energy-saving and efficient operating mode. The algorithm reduces the computational load of traditional numerical simulations through a rapid analytical method. Inputs include heat conduction data (e.g., roll oil temperature 47℃, aluminum foil surface temperature, acquired from the data acquisition module) and mechanical behavior data (e.g., rolling force 265T, tension 4.0kg / mm², speed 800m / min, acquired from a cloud database or real-time monitoring module) from the digital twin modeling module, energy consumption data from the dynamic optimization module, and alloy configurations from the cloud database. Outputs are energy consumption optimization commands, including rolling speed, oil temperature, and coolant temperature. The algorithm analyzes the influence of temperature and stress fields on energy consumption, combined with alloy characteristics, to generate energy-saving commands. For example, it reduces the speed of aluminum foil used in power batteries to reduce motor power consumption, and optimizes the coolant temperature of aluminum panels used in construction to reduce heat emissions. In the system, the algorithm's commands are transmitted to the feedback control module via industrial Ethernet to drive the equipment to perform energy-saving adjustments. The second algorithm is the Energy Consumption-Quality Balance Analytical Algorithm, a term defined in this invention, referring to a calculation method for optimizing energy consumption and product quality. Based on multi-objective optimization technology, this algorithm generates energy-saving instructions by balancing energy consumption (such as motor power) and quality (such as thickness uniformity). It analyzes temperature, mechanical, and energy consumption data during the rolling process to find the optimal solution for reducing energy consumption while ensuring product quality (such as the absence of pinholes). Similar to reducing fuel consumption while maintaining car performance, it ensures the achievement of green manufacturing goals. The inputs are thermal-mechanical distribution data and alloy configurations from a cloud database, and the output is energy-saving adjustment instructions. The algorithm generates instructions by balancing energy consumption and quality requirements; for example, prioritizing low energy consumption for power battery aluminum foil to reduce production costs, and prioritizing strength for building aluminum panels to meet structural requirements. In the system, the algorithm optimizes energy consumption and emissions to ensure green manufacturing goals. In the green manufacturing optimization module, the energy consumption-quality balance analysis algorithm generates energy-saving instructions.For example, when producing 1060H18 aluminum foil, it adjusts the speed to 780m / min and the oil temperature to 46.5℃, reducing energy consumption to 120kWh / t, while ensuring thickness uniformity and meeting green manufacturing standards.

[0098] The cross-alloy adaptive module employs a transfer learning algorithm, specifically designed to generate process parameters suitable for multiple alloys, ensuring consistent optimization across different alloys. Inputs include the microstructure characteristics of the digital twin modeling module, alloy configurations from the cloud database, and early warning signals from the real-time monitoring module. Outputs are process parameters, including tension and rolling force. The algorithm analyzes differences in alloy elements to establish a mapping rule between grain distribution and process parameters, generating parameters suitable for thicknesses ranging from 8 to 50 micrometers. For example, different tension values ​​are generated for 1060H18, 3003, and 5052 to optimize thickness, surface quality, or strength. The algorithm, combined with the cloud database configuration, dynamically adjusts parameters to adapt to alloy characteristics. For instance, tension is optimized based on the high-purity aluminum characteristics of 1060H18 to control thickness, while tension is adjusted based on the magnesium content of 5052 to enhance strength. Within the system, the algorithm's output is transmitted via industrial Ethernet to the dynamic optimization module, where optimization and adjustment instructions ensure stable multi-alloy production. Specific Implementation Example 3:

[0100] like Figures 1 to 3 As shown, the following are specific use cases of the content described in the above embodiments:

[0101] Case 1:

[0102] Scene description:

[0103] An aluminum processing plant deployed a real-time monitoring and optimization system for the aluminum product rolling process based on digital twins on a four-roll irreversible cold rolling mill production line. The system is equipped with a sensor array, edge computing server, and cloud database, and is connected to an industrial Ethernet network with a communication latency of less than 1 millisecond.

[0104] Application process:

[0105] After system startup, the data acquisition module collects real-time process data from key nodes of the rolling mill using a core signal processing algorithm. These key nodes include the mill inlet, outlet, roll gap area, tension control point, lubrication system interface, and cooling system interface. The sensor array uses M8 stainless steel bolts to secure high-precision thickness gauges, force sensors, temperature sensors, pressure sensors, and optical surface inspectors. The algorithm input consists of analog signals captured by the sensors, such as thickness gauge voltage signals, force sensor strain signals, and temperature sensor thermocouple signals. The output is digital process data, including a thickness of 8 micrometers, rolling force of 265 tons, speed of 800 meters per minute, and oil temperature of 47 degrees Celsius. The algorithm processes the data at a frequency of 300 Hz, generating JSON data packets at a rate of 1000 packets per second, with timestamps such as 2025-06-16 09:00:00.123. The data is then transmitted to the digital twin modeling module at the node output, providing thickness and pinhole detection data for the construction of the 1060H18 aluminum foil model.

[0106] The digital twin modeling module uses a crystal plasticity analysis algorithm. Inputting process data and a cloud database of 1060H18 alloy configuration (aluminum content higher than 99.6%), it outputs a microstructure feature vector including grain size and dislocation density, generating fine grain data. This data is transmitted to the real-time monitoring module for pinhole detection. The macroscopic mechanical analysis algorithm takes process data and the mechanical properties of 1060H18 as input and outputs a mechanical behavior feature vector including stress-strain distribution, generating high elongation data, which is transmitted to the real-time monitoring module for thickness deviation detection. The time-stepping iterative algorithm takes microscopic and mechanical feature vectors as input and outputs a dynamic multi-scale digital twin model. With a time step of 3 milliseconds, it generates JSON data packets, processing 500 packets per second, which are transmitted to the real-time monitoring module and the green manufacturing optimization module to provide thickness optimization predictions for 1060H18 aluminum foil.

[0107] The real-time monitoring module uses a spatiotemporal correlation analysis algorithm, taking digital twin model prediction data and real-time process data as input, and outputting early warning signals. These signals include anomaly types such as thickness deviation, location (e.g., exit point), and severity (e.g., deviation amount of 0.3 micrometers). If the algorithm detects a 0.3-micrometer deviation in exit thickness, it generates an early warning signal, which is transmitted via industrial Ethernet to the dynamic optimization module in 3 milliseconds, processing 1000 signals per second and triggering thickness adjustments. If a surge in pinhole density is detected, an emergency stop command is generated and transmitted to the mill controller to protect the equipment.

[0108] The dynamic optimization module analyzes the optimization algorithm, taking the early warning signal and the 1060H18 alloy configuration in the cloud database as inputs, and outputting adjustment instructions, including a rolling force of 268 tons and a speed of 780 meters per minute. The algorithm analyzes the impact of thickness deviation and generates instructions to increase the rolling force, while adhering to the constraint that the rolling force does not exceed 280 tons. These instructions are transmitted to the feedback control module via industrial Ethernet, processing 500 instructions per second to drive thickness optimization.

[0109] The feedback control module uses equipment control algorithms, taking adjustment commands as input and equipment control signals as output. It maps rolling force adjustment to servo motor signals and speed adjustment to traction roller signals. The feedback update algorithm takes the actual thickness data and target parameters fed back by the equipment as input and outputs updated adjustment commands, such as further adjusting the rolling force to 270 tons. The update cycle is 5 milliseconds and is transmitted via industrial Ethernet to ensure that the thickness of 1060H18 aluminum foil reaches 8 micrometers.

[0110] The green manufacturing optimization module employs a thermo-mechanical-energy consumption coupled analytical algorithm. Inputting data includes thermal conduction and mechanical properties from a digital twin model and the configuration of the 1060H18 alloy. The output is energy consumption optimization commands, including parameters such as a speed of 780 meters per minute and an oil temperature of 46.5 degrees Celsius. The algorithm analyzes the impact of the temperature field on energy consumption and transmits this data to the feedback control module via industrial Ethernet, processing 500 commands per second to reduce motor power consumption. The energy consumption-mass balance analytical algorithm takes thermal-mechanical distribution data as input and outputs energy-saving commands, balancing the energy consumption and thickness uniformity of the 1060H18 aluminum foil to optimize production costs.

[0111] The cross-alloy adaptive module uses a transfer learning algorithm. The input includes the microstructure characteristics of 1060H18, alloy configuration from the cloud database, and early warning signals. The output includes process parameters such as tension of 4.2 kg / mm². The algorithm establishes a mapping rule from grain distribution to tension, which is transmitted to the dynamic optimization module via industrial Ethernet. It processes 100 parameter data per second, supporting the optimization of surface finish for 3003 aluminum foil and the improvement of fracture toughness for 5052 aluminum plates.

[0112] Application results:

[0113] The system enables real-time monitoring and optimization of 1060H18 aluminum foil, achieving precise thickness control, timely pinhole detection, and significantly improved production efficiency. The digital twin model updates every 3 milliseconds, with anomaly detection latency below 3 milliseconds, parameter adjustment cycle of 5 milliseconds, data processing rate of 1000 data entries per second, and communication latency below 1 millisecond. Energy consumption optimization reduces production costs, and the green manufacturing module reduces motor power consumption and heat emissions. The cross-alloy adaptive module supports the production of 3003 aluminum foil and 5052 aluminum plates, adapting the system to multiple product scenarios and greatly improving production stability. Specific Implementation Example 4:

[0115] likeFigures 1 to 3 As shown, the following are the core hardware components and supplementary hardware descriptions:

[0116] The edge computing server, the core of the system's data processing, is deployed in a rack next to the production line. Equipped with a multi-core high-performance processor, graphics processing unit, and 32GB of memory, it supports multi-scale modeling of digital twins and real-time data processing. The server runs data acquisition, digital twin modeling, real-time monitoring, dynamic optimization, and cross-alloy adaptive modules. It processes JSON data packets, executes crystal plasticity analytical algorithms to generate microstructure feature vectors, macroscopic mechanical analytical algorithms to generate mechanical behavior feature vectors, and time-stepping iterative algorithms to construct dynamic multi-scale models. The server updates the model every 3 milliseconds, processing 500 data packets per second. It supports thickness control of 1060H18 aluminum foil, surface finish optimization of 3003 aluminum foil, and fracture toughness improvement of 5052 aluminum plates. The server interacts with a cloud database via industrial Ethernet to obtain alloy configurations such as 1060H18 aluminum content higher than 99.6%, 3003 manganese content 1.0 to 1.5%, and 5052 magnesium content 2.2 to 2.8%, supporting transfer learning algorithms to generate adaptation parameters. The server is equipped with redundant power supplies and cooling systems to ensure continuous operation and meet the computing needs for real-time monitoring and optimization.

[0117] The industrial Ethernet module utilizes gigabit Ethernet switches and fiber optic communication equipment, deployed in the production line control room. It connects sensor arrays, edge computing servers, rolling mill controllers, and cloud database servers, ensuring communication latency of less than 1 millisecond. The module supports JSON format data packet transmission, including process data, model data, warning signals, and adjustment instructions, processing 1000 data packets per second. High-bandwidth communication is achieved via fiber optic links, enabling the data acquisition module to transmit real-time data to the digital twin modeling module, the real-time monitoring module to transmit warning signals to the dynamic optimization module, and the feedback control module to issue instructions to the rolling mill controller, ensuring smooth data flow across multiple product scenarios. The module features network redundancy to prevent communication interruptions and supports rapid transmission of emergency stop commands.

[0118] The cloud-based database server is deployed in a remote data center, equipped with a high-performance storage array and multi-core processors, storing 1000 alloy configurations, including grain size, alloy element ratios, and mechanical properties of 1060H18, 3003, and 5052. The server interacts with the edge computing server via industrial Ethernet, transmitting JSON-formatted data packets. It provides alloy configuration support for a digital twin modeling module to generate microscopic and macroscopic feature vectors, and a cross-alloy adaptive module to generate adaptation parameters such as the 1060H18 tensile strength of 4.2 kg / mm². The server processes 100 data packets per second with a communication latency of less than 1 millisecond and is equipped with a data encryption module to ensure secure transmission of alloy configurations. The server supports an energy-mass balance analysis algorithm, optimizing the energy consumption of 1060H18 aluminum foil and the strength of 5052 aluminum plates by providing thermal conductivity and strength data.

[0119] The emergency shutdown device is integrated next to the mill controller, including a hardware circuit breaker and relays, and receives emergency shutdown commands from the real-time monitoring module. When a serious anomaly is detected, such as a surge in pinhole density or abnormal microcrack depth, the device cuts off the power to the servo motors and hydraulic system within 5 milliseconds, stopping the mill operation and protecting the equipment and aluminum material. The device is equipped with a manual reset switch to ensure safe resumption of production after a reset, and is compatible with various product scenarios such as 1060H18, 3003, and 5052.

[0120] The data storage unit is deployed within an edge computing server, equipped with a combination of high-speed solid-state drives (SSDs) and hard disk drives (HDDs) to store real-time process data, digital twin model data, and early warning signals. The SSDs store frequently accessed data, such as the thickness data of 1060H18 aluminum foil, with a read / write speed of 500MB / s. The HDDs store historical data, such as surface defect records of 3003 aluminum foil, with a capacity of 10TB. The unit supports the storage requirements of the data acquisition module (1000 data packets per second) and the recording of 1000 early warning signals per second by the real-time monitoring module, ensuring data traceability and assisting in the verification and optimization of the digital twin model.

[0121] The human-machine interface terminal is deployed in the production line control room, equipped with a touchscreen display and industrial keyboard, showing real-time process data, digital twin model status, and early warning signals. The terminal receives JSON-formatted data from the edge computing server, displaying the thickness and pinhole detection results of 1060H18 aluminum foil, the surface finish of 3003 aluminum foil, and the mechanical properties of 5052 aluminum plate. The terminal supports operator input of adjustment commands, which are transmitted to the feedback control module with a response time of 500 milliseconds. The terminal is equipped with an alarm that emits an audible and visual alarm when an emergency stop command is triggered, prompting the operator to handle the anomaly. Specific Implementation Example 5:

[0123] like Figures 1 to 3 As shown, the following are the specific experimental data:

[0124] Table 1: System Performance Display Table

[0125] Aluminum alloy type Target thickness (microns) Actual thickness (microns) Pinhole density (number / meter2) Surface scratch length (mm / meter2) Tensile strength (MPa) Rolling force (tons) Speed (m / min) Oil temperature (degrees Celsius) Energy consumption (kWh / ton) Data acquisition time Early warning signal generation time Adjustment instruction execution time 1060H18 8 8.02 0.5 0 70 268 780 46.5 120 2025-6-16 2025-6-16 2025-6-16 3003 20 20.15 1.0 0.2 150 270 770 47.0 130 2025-6-16 2025-6-16 2025-6-16 5052 40 40.30 0.8 0 220 265 790 47.5 140 2025-6-16 2025-6-16 2025-6-16

[0126] The experimental data reflects the system's performance in producing 1060H18 aluminum foil, 3003 aluminum foil, and 5052 aluminum plate, and verifies the supporting role of hardware and algorithms in thickness control, pinhole detection, surface finish, fracture toughness, and energy consumption optimization, thus serving as a demonstration.

[0127] The experimental data comes from actual measurements taken on the production line on June 16, 2025. The system is deployed on a four-roll irreversible cold rolling mill, equipped with a sensor array (high-precision thickness gauge, force sensor, temperature sensor, pressure sensor, optical surface inspector, and M8 bolts fixed to key nodes), an edge computing server, an industrial Ethernet module, a mill controller, and a cloud database server. The sensor array collects data at a frequency of 300 Hz, generating JSON format data packets. The edge computing server processes the digital twin model and early warning signals. The cloud database provides configurations for 1060H18, 3003, and 5052 alloys. The industrial Ethernet ensures communication latency of less than 1 millisecond. Data acquisition time, early warning signal generation time, and adjustment command execution time are recorded using timestamps. Tensile strength is measured by the force sensor, and energy consumption is calculated by the green manufacturing optimization module.

[0128] The following are additional details regarding this plan:

[0129] The sensor array employs a dual-redundancy design, deploying two sets of high-precision thickness gauges, force sensors, temperature sensors, pressure sensors, and optical surface inspectors at each critical node. In the event of a primary sensor failure, automatic switching to the backup sensor occurs within 10 milliseconds, ensuring continuous data acquisition. The sensors are equipped with electromagnetic interference shielding to prevent electromagnetic noise during mill operation from affecting signal quality and ensuring the stability of the 300 Hz acquisition frequency. The mill controller integrates dual-backup microprocessors. When the primary processor fails, the backup processor takes over control tasks with a 5-millisecond switching time, supporting rapid execution of emergency stop commands. It is compatible with 1060H18 aluminum foil thickness control, 3003 aluminum foil surface finish optimization, and 5052 aluminum plate fracture toughness improvement. The lubrication and cooling systems are equipped with backup pumps and valves, automatically switching in case of failure with a 100-millisecond response time, ensuring stable oil temperature at 47 degrees Celsius and oil pressure at 4.0 bar, reducing energy consumption and emissions.

[0130] The data storage unit adopts a distributed storage architecture. Real-time process data and digital twin model data are stored on the solid-state drives of the edge computing server, while simultaneously being synchronized hourly to a backup storage array on the cloud database server. This array has a capacity of 50TB and a synchronization speed of 200MB / s. Backup data is appended with a checksum to ensure data integrity, with a verification time of 10 milliseconds. When an abnormal event (such as a surge in pinhole density) is triggered, the data storage unit generates a snapshot, saving warning signals and adjustment instructions, with a storage period of 5 minutes, supporting post-event analysis. The data recovery module retrieves backup data from the cloud database server, with a recovery time of less than 1 second, ensuring the continuity of production data for 1060H18, 3003, and 5052. The storage unit supports JSON format data packet storage, processing 1000 data entries per second, adapting to the warning signal recording requirements of the real-time monitoring module.

[0131] The emergency shutdown logic is triggered by the spatiotemporal correlation analysis algorithm of the real-time monitoring module. Specific conditions include a pinhole density exceeding 5 per square meter, a microcrack depth exceeding 0.5 mm, and a thickness deviation exceeding 1 micrometer. The algorithm generates an emergency shutdown command, which is transmitted to the mill controller via industrial Ethernet. The command includes the anomaly type, location, and severity, in JSON data packet format, with a transmission time of less than 1 millisecond. The emergency shutdown device cuts off the power to the servo motor and hydraulic system via a hardware circuit breaker, stopping the mill operation and protecting 1060H18 aluminum foil, 3003 aluminum foil, and 5052 aluminum plate. The recovery process requires the operator to confirm the anomaly has been eliminated via the reset switch on the human-machine interface terminal. The terminal displays the anomaly log and process data. After the engineer verifies that the thickness, pinhole, and microcrack indicators are normal, they input the recovery command, and the mill controller restarts the equipment. The recovery time is 30 seconds, ensuring production continuity.

[0132] The industrial Ethernet module integrates a load balancer to dynamically allocate data traffic, prioritizing warning signals (1000 messages per second), adjustment commands (500 messages per second), model data (500 messages per second), and process data (1000 messages per second). The balancer allocates bandwidth based on packet type and node identifiers (e.g., ingress and egress) to prevent data congestion, maintaining communication latency below 1 millisecond. The edge computing server is equipped with a traffic monitoring module that analyzes data processing load in real time. If the load exceeds 80%, it automatically reduces the transmission frequency of non-critical data (such as historical logs), prioritizing data flows for 1060H18 thickness control, 3003 surface finish optimization, and 5052 fracture toughness improvement. The load balancer supports an energy-quality balance analysis algorithm to ensure timely transmission of energy-saving commands from the green manufacturing optimization module.

[0133] Comparative example:

[0134] Traditional aluminum foil rolling monitoring and optimization system:

[0135] In an aluminum processing plant, a traditional aluminum foil rolling system is deployed on a four-roll irreversible cold rolling mill production line to produce 1060H18 aluminum foil with a target thickness of 8μm and a width of 800mm, used in power batteries; 3003 aluminum foil with a target thickness of 20μm, used in food packaging; and 5052 aluminum plates with a target thickness of 40μm, used in building curtain walls. This system employs traditional automation technology, including independently operating sensors, a CNC rolling mill, and local storage devices. It relies on manual analysis and fixed process parameter adjustments, lacking digital twin technology, real-time multi-scale modeling, and cross-alloy adaptive capabilities.

[0136] The system uses a thickness gauge, force sensor, and temperature sensor, fixed with M8 bolts, to collect data at a frequency of 100Hz. It collects data such as thickness (8.4μm), rolling force (265T), speed (800m / min), and oil temperature (47℃), storing it in CSV format at a rate of 200 records per second. It uses RS-485 communication with a 10ms delay. However, the lack of a unified data format makes it prone to data loss and timestamp discrepancies. Monitoring relies on a PLC and a human-machine interface terminal. Operators manually judge thickness deviations of 0.5μm, pinholes of 1.2 per m², or microcracks, with a response time of 100ms, lacking prediction of microstructure and mechanical behavior. Process optimization is based on a fixed parameter table, requiring manual adjustment of the rolling force (270T) or speed, taking 5-10 seconds with a 50ms execution delay. The lack of closed-loop control results in incomplete elimination of thickness deviations of 0.4μm and surface scratches of 0.5mm / m². Energy management is lacking, with unit energy consumption of 180-200kWh / t, high rolling oil mist and heat emissions, failing to meet green manufacturing standards. Switching between multiple alloys, such as from 1060H18 to 3003, requires manual parameter adjustment, which takes 5 minutes and relies on experience. The fracture toughness of 5052 aluminum plate is less than 200MPa.

[0137] Taking 1060H18 aluminum foil as an example, the system detected a thickness deviation of 0.5μm. Manually adjusting the rolling force to 270T took 10 seconds, reducing the thickness to 8.4μm, with a pinhole density of 1.2 pins / m², energy consumption of 180kWh / t, and surface scratches of 0.3mm / m². Switching to 3003 aluminum foil took 5 minutes, and the 0.5mm / m² surface scratches were not effectively controlled. Experimental data are as follows:

[0138] Table 2: Experimental Data Display Table

[0139] Aluminum alloy type Target thickness (microns) Actual thickness (microns) Pinhole density (number / meter2) Surface scratch length (mm / meter2) Tensile strength (MPa) Energy consumption (kWh / ton) Response time (ms) Switching time 1060H18 8 8.4 1.2 0.3 65 180 100 5 min 3003 20 20.6 1.5 0.5 140 190 100 5 min 5052 40 40.8 1.0 0.2 200 200 100 5 min 5 min

[0140] Comparison with the system of this invention:

[0141] Traditional systems have shortcomings in data integration, anomaly detection, process optimization, energy consumption control, and multi-alloy adaptation. Data integration uses CSV format, processing 200 records per second with a 10ms delay, making it prone to data loss. Anomaly detection response time is 100ms, relying on manual judgment, with a pinhole density of 1.2 pins / m² and a thickness deviation of 0.4μm. Process optimization takes 5-10 seconds and lacks closed-loop adjustment. Energy consumption is 180-200kWh / t, with no optimization module and high emissions. Alloy switching takes 5 minutes, and the fracture toughness of 5052 is insufficient. The system of this invention constructs a dynamic multi-scale model using digital twin technology, with an update frequency of 300Hz, processing 1000 JSON data packets per second with a delay of less than 1ms, achieving efficient data integration. Anomaly detection takes 3ms, reducing pinhole density to 0.5 pins / m² and thickness deviation to 0.02μm. Closed-loop adjustment takes 5ms, adjusting rolling force to 268T and speed to 780m / min, achieving precise thickness control and reducing surface scratches to 0mm / m². Energy consumption is optimized to 120-140 kWh / t, reducing emissions and meeting green manufacturing standards. The cross-alloy adaptive module supports 1060H18, 3003, and 5052 alloys, with a switching time of less than 100ms and tensile strength increased to 220 MPa (5052). Experimental data comparison shows that the system of this invention is significantly superior to traditional systems in terms of thickness control, surface quality, fracture toughness, energy consumption optimization, and production flexibility, providing efficient, precise, and green solutions for power batteries, food packaging, and building curtain walls.

[0142] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising a reference structure" does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0143] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A real-time monitoring and optimization system for aluminum product rolling process based on digital twin, comprising a monitoring and optimization system, characterized in that: The monitoring and optimization system includes a data acquisition module, a digital twin modeling module, a real-time monitoring module, a dynamic optimization module, a feedback control module, a green manufacturing optimization module, and a cross-alloy adaptive module. The working modes of each module in the monitoring and optimization system are as follows: The data acquisition module collects data on thickness, rolling force, tension, speed, oil temperature and oil pressure during the rolling process through a customized high-precision sensor array deployed at key nodes of the rolling mill, generates real-time process data, and transmits the data to the digital twin modeling module after encapsulation via industrial Ethernet. The digital twin modeling module receives data transmitted from the data acquisition module, parses it into process data, constructs a dynamic multi-scale digital twin model of the microstructure and mechanical behavior of the power battery aluminum foil, generates model data, and encapsulates the model data via industrial Ethernet before transmitting it to the real-time monitoring module and the green manufacturing optimization module. The real-time monitoring module receives model data transmitted by the digital twin modeling module, parses it into microstructure and mechanical behavior prediction data, generates early warning signals including thickness deviation, pinhole density and surface microcracks, and encapsulates the early warning signals through industrial Ethernet before transmitting them to the dynamic optimization module. The dynamic optimization module receives the early warning signal transmitted by the real-time monitoring module, parses it into anomaly type and parameter deviation data, generates adjustment instructions including rolling force, tension, speed, oil temperature and oil pressure, and encapsulates the adjustment instructions through industrial Ethernet and transmits them to the feedback control module. The feedback control module receives the adjustment instructions transmitted by the dynamic optimization module, parses them into rolling parameter adjustment values, sends them to the mill controller through the industrial control system, coordinates the closed-loop adjustment of rolling parameters, and receives execution data to update the adjustment instructions. The green manufacturing optimization module receives data transmitted from the digital twin modeling module and the dynamic optimization module, parses it into heat conduction, mechanical behavior and energy consumption data, generates energy consumption optimization instructions including rolling speed and oil temperature, and encapsulates the optimization instructions through industrial Ethernet and transmits them to the feedback control module. The cross-alloy adaptive module receives model data transmitted from the digital twin modeling module, parses it into microstructure characteristics, generates rolling process parameters adapted to various aluminum alloys, and transmits them to the dynamic optimization module via industrial Ethernet.

2. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 1, characterized in that: The control flow design of the monitoring and optimization system is as follows: the data acquisition module collects rolling parameters and transmits them to the digital twin modeling module; the digital twin modeling module generates a dynamic multi-scale digital twin model and transmits it to the real-time monitoring module and the green manufacturing optimization module; the real-time monitoring module generates an early warning signal and transmits it to the dynamic optimization module; the dynamic optimization module generates adjustment instructions and transmits them to the feedback control module; the feedback control module issues instructions to the mill controller to drive the rolling equipment and receives and executes data update and adjustment instructions; the green manufacturing optimization module generates energy consumption optimization instructions and transmits them to the feedback control module; and the cross-alloy adaptive module generates process parameters and transmits them to the dynamic optimization module. The data acquisition module includes a sensor array unit, a signal conversion unit, and a data transmission unit, and acquires data through the following steps: A1. Using a sensor array unit, high-precision thickness gauges, force sensors, temperature sensors, and optical surface detectors are fixed at the mill inlet and outlet and key nodes to collect data on thickness, rolling force, tension, speed, oil temperature, and oil pressure, generating real-time process data. The acquisition frequency is 300Hz. The key nodes include the area where the rolls directly contact the aluminum foil, the traction rolls and guide rolls located between the mill inlet and outlet, and the locations of the sprayed and circulating rolling oil in the mill. A2. Through the signal conversion unit, the analog signals collected by the sensor array unit are converted into digital signals by the analog-to-digital converter to generate transmittable process data. A3. Through the data transmission unit, the digital signal is encapsulated and transmitted to the digital twin modeling module using an industrial Ethernet module.

3. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 1, characterized in that: The digital twin modeling module includes a data parsing unit, a microstructure prediction unit, a macroscopic mechanical prediction unit, a coupling and exchange unit, and a model transmission unit; and constructs a dynamic multi-scale digital twin model through the following steps: B1. The data parsing unit receives data transmitted from the data acquisition module and parses it into thickness, rolling force, tension, speed, oil temperature and oil pressure data. B2. Using the microstructure prediction unit and the crystal plasticity analytical algorithm, based on the characteristics of aluminum foil, the evolution of grain size and dislocation density is calculated to generate microstructure feature vectors. B3. Using the macroscopic mechanical prediction unit and the macroscopic mechanical analytical algorithm, based on a rolling force of 265T and a tension of 4.0kg / mm², the stress-strain distribution is calculated, and a mechanical behavior feature vector is generated. B4. By coupling interactive units and using a time-stepping iterative algorithm, the interaction between microscopic organization feature vectors and mechanical behavior feature vectors is synchronized in real time to generate a dynamic multi-scale digital twin model. B5. Through the model transmission unit, the digital twin model is encapsulated and transmitted to the real-time monitoring module and the green manufacturing optimization module using industrial Ethernet.

4. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 1, characterized in that: The real-time monitoring module includes a data receiving unit, an anomaly detection unit, and a signal transmission unit, and generates an early warning signal through the following steps: C1. Receive digital twin model data transmitted by the digital twin modeling module through the data receiving unit, and parse it into microstructure and mechanical behavior prediction data; C2. Through the anomaly detection unit, using the spatiotemporal correlation analysis algorithm, the predicted data is compared with the real-time acquired data to generate an early warning signal containing thickness deviation, pinhole density and surface microcracks. The signal specifies the anomaly type, location and degree. C3. The warning signal is encapsulated and transmitted to the dynamic optimization module via the signal transmission unit using industrial Ethernet.

5. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 1, characterized in that: The dynamic optimization module includes a signal parsing unit, an instruction generation unit, and an instruction transmission unit, and generates adjustment instructions through the following steps: D1. Receive the early warning signal transmitted by the real-time monitoring module through the signal analysis unit, and analyze it into anomaly type and parameter deviation data; D2. Through the instruction generation unit, using the analytical optimization algorithm, based on the anomaly type and parameter deviation data, an adjustment instruction containing rolling force 265T, tension 4.0kg / mm², speed 800m / min, oil temperature 47℃ and oil pressure 4.0Bar is generated. D3. Through the instruction transmission unit, the adjustment instructions are encapsulated and transmitted to the feedback control module using industrial Ethernet.

6. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 1, characterized in that: The feedback control module includes an instruction receiving unit, a control execution unit, a feedback acquisition unit, and an instruction updating unit, and achieves closed-loop regulation through the following steps: E1. Receive adjustment instructions transmitted by the dynamic optimization module through the instruction receiving unit and parse them into rolling parameter adjustment values; E2. Through the control execution unit, the industrial control system sends adjustment commands to the rolling mill controller, driving the servo motor and hydraulic system to adjust the rolling force, tension and oil pressure; E3. Receive execution data from the mill controller via the feedback acquisition unit, including actual thickness, tensile strength, and elongation. E4. Update and adjust instructions based on execution data through the instruction update unit.

7. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 1, characterized in that: The green manufacturing optimization module includes a data receiving unit, an energy consumption analysis unit, and an instruction transmission unit, and generates energy consumption optimization instructions through the following steps: F1. Receive data transmitted from the digital twin modeling module and dynamic optimization module through the data receiving unit, and parse it into heat conduction, mechanical behavior and energy consumption data; F2. Through the energy consumption analysis unit, using the thermo-mechanical-energy consumption coupled analytical model, calculate the energy loss and emissions during the rolling process, and generate energy consumption optimization instructions including a rolling speed of 800m / min and an oil temperature of 47℃. F3. Through the command transmission unit, the energy consumption optimization command is encapsulated and transmitted to the feedback control module using industrial Ethernet.

8. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 7, characterized in that: The green manufacturing optimization module also includes a thermal-mechanical linkage unit, an optimization generation unit, and a data feedback unit, and further optimizes energy consumption through the following steps: G1. Through the thermo-mechanical linkage unit, receive the thermo-mechanical simulation data transmitted by the digital twin modeling module and analyze it into temperature field and stress field distribution; G2. By optimizing the generation unit and using the energy consumption-mass balance analysis algorithm, based on the solid solution characteristics of aluminum foil, an energy-saving adjustment command containing a rolling speed of 800m / min and an oil temperature of 47℃ is generated. G3. Through the data feedback unit, energy-saving adjustment commands are transmitted to the feedback control module via industrial Ethernet.

9. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 1, characterized in that: The cross-alloy adaptive module includes a data parsing unit, a parameter generation unit, and a parameter transmission unit, and generates adaptation parameters through the following steps: H1. Through the data parsing unit, receive the model data transmitted by the digital twin modeling module and parse it into the microstructure characteristics of the aluminum alloy. H2. Using the parameter generation unit and transfer learning algorithm, based on the solid solution differences of aluminum alloy, a mapping rule from grain distribution to tension parameters is established to generate process parameters suitable for thicknesses of 8-20μm. H3. Through the parameter transmission unit, the adaptation parameters are encapsulated and transmitted to the dynamic optimization module using industrial Ethernet.

10. The real-time monitoring and optimization system for aluminum product rolling process based on digital twin as described in claim 9, characterized in that: The cross-alloy adaptive module also includes a parameter database unit, a cloud-edge collaboration unit, and a command issuance unit, and optimizes process parameters through the following steps: I1. The microstructure characteristics of aluminum alloys are stored through a parameter database unit, containing 1,000 alloy configurations, with 100 configurations updated every 12 hours. I2. Through the cloud-edge collaboration unit, feedback data from the real-time monitoring module and the dynamic optimization module is received via industrial Ethernet, and the adaptation parameters are adjusted. I3. The adjusted adaptation parameters are transmitted to the mill controller via the command issuing unit using industrial Ethernet.

Images