Glass bottle lightweight forming process optimization system

Abstract

The invention provides a glass bottle lightweight forming process optimization system, belongs to the technical field of glass product production, and solves the technical problems that the existing lightweight glass bottle is low in forming efficiency, process parameters cannot be dynamically adapted and the like. Comprising a hardware execution module, a global data sensing module, a digital twinning optimization module, an enterprise-level cloud management module and a man-machine collaborative operation and maintenance module. The global data sensing module comprises an equipment three-dimensional information scanning and collecting unit, an equipment working state sensor network unit, an energy consumption and resource monitoring unit, a data transmission unit, a data storage unit and a carbon footprint accounting unit; the digital twinning optimization module comprises a virtual simulation unit, a software simulation optimization unit and a dynamic feedback optimization unit. According to the invention, an integrated process optimization system integrating an intelligent algorithm, digital twinning, green carbon reduction, flexible production, full-dimensional detection, predictive maintenance and man-machine collaboration is adopted, and efficient, accurate, green and flexible production of lightweight glass bottles is realized.

Description

Technical Field

[0001] This invention belongs to the field of glass product manufacturing technology, and relates to a glass product forming process optimization system, particularly a lightweight glass bottle forming process optimization system. Background Technology

[0002] The lightweighting process of glass bottles faces numerous technical bottlenecks, including the contradiction between strength and weight, insufficient molding precision, and reliance on experience for quality control. Under traditional processes, the internal pressure resistance decreases significantly after the bottle wall is thinned, resulting in a high breakage rate in drop tests and large errors in wall thickness uniformity, leading to a low pass rate.

[0003] Existing technologies also have limitations: Offline and Non-Real-Time: Optimization of process parameters relies heavily on pre-production simulation and post-production testing, making real-time, proactive adjustments during production impossible. Information Silos: There is a lack of deep integration and collaborative analysis between equipment status, process parameters, and product quality data. Feedback control is mostly limited to localized fine-tuning, making global optimization difficult. Traditional Methods Rely on Physical Trial and Error: For new bottle types or equipment layouts, companies need to spend significant time and materials on physical debugging, resulting in high costs and long cycles.

[0004] Therefore, there is a need for a system and method that can closely integrate software simulation with hardware production to achieve end-to-end, self-learning, and dynamic optimization. Summary of the Invention

[0005] The purpose of this invention is to address the aforementioned problems in existing technologies by proposing a lightweight glass bottle molding process optimization system. The technical problem this invention aims to solve is: how to achieve lightweight production of high-end packaging glass bottles of various categories, and realize dynamic collaborative optimization of the entire process, including process, equipment, quality, energy consumption, and carbon footprint.

[0006] The objective of this invention can be achieved through the following technical solutions: A lightweight glass bottle molding process optimization system includes a hardware execution module, a global data perception module, a digital twin optimization module, an enterprise-level cloud management module, and a human-machine collaborative operation and maintenance module. The hardware execution module performs the glass bottle production, five-dimensional quality inspection and precise control process sequentially according to the simulation optimization parameters, and utilizes waste heat to save energy, while also ensuring production continuity. It includes a molding equipment unit, an inspection equipment unit, a control unit, a waste heat recovery unit and a redundancy design unit. The full-domain data perception module collects equipment status, material quality, and energy consumption data in real time and comprehensively, and uses 5G / industrial Ethernet for transmission and blockchain storage. It also integrates carbon footprint accounting to provide a full-element data foundation for the construction of digital twins, including a three-dimensional information scanning and acquisition unit for equipment, a sensor network unit for equipment working status, an energy consumption and resource monitoring unit, a data transmission unit, a data storage unit, and a carbon footprint accounting unit. The digital twin optimization module simulates the entire production process by constructing a high-fidelity virtual model, uses simulation and AI algorithms to pre-simulate and optimize process parameters, and dynamically adjusts parameters and predicts maintenance based on real-time data comparison in actual production, realizing a continuous closed-loop iteration of simulation-optimization-feedback, including a virtual simulation unit, a software simulation optimization unit, and a dynamic feedback optimization unit. The enterprise-level cloud management module integrates order, production and logistics data by connecting to systems other than ERP, enabling closed-loop management from virtual trial production scheduling and full lifecycle quality traceability to carbon footprint and cost accounting. It provides unified data support for enterprise decision-making and supply chain collaboration, including a supply chain collaboration interface unit, a full lifecycle quality traceability unit and a carbon footprint and cost accounting unit. The human-machine collaborative operation and maintenance module, through technologies including but not limited to AR, remote monitoring, and one-click optimization, enables visualization of equipment status, remote intervention, and intelligent parameter adjustment. Combined with hierarchical permissions and data security protection, it improves operation and maintenance efficiency and system security. The module includes an AR operation and maintenance unit, a remote monitoring unit, a one-click optimization unit, a hierarchical access permission unit, and a data security protection unit.

[0007] The working principle of this invention is as follows: Through the deep collaboration of five major modules, a closed-loop intelligent control system is constructed from physical production to digital space. Its core working principle is as follows: the full-domain data perception module acts as the system's "sensory nerves," collecting real-time production data from all elements; the digital twin optimization module acts as the "decision brain," using high-fidelity simulation and AI algorithms to pre-simulate and iteratively optimize process parameters; the optimized instructions are then issued to the hardware execution module for precise production and quality inspection; the enterprise-level cloud management module achieves vertical integration with upper-level information systems, supporting full lifecycle management and supply chain collaboration; finally, the human-machine collaborative operation and maintenance layer provides intuitive, secure, and efficient interaction and operation and maintenance support for the entire system.

[0008] The five modules form a complete closed loop of "perception-decision-execution-management-operation and maintenance", realizing continuous optimization of the production process, precise control of energy and carbon, and comprehensive improvement of operation and maintenance efficiency.

[0009] The molding equipment unit is configured according to the layout scheme and size output by the software simulation optimization unit, including: raw material preparation equipment: including but not limited to crushing equipment and stirring equipment; Melting pool furnace: It adopts zoned heating control to adapt to the melting needs of glass molten material of various materials. It is equipped with an intelligent power adjustment device to reduce the heat preservation power during off-peak production periods and reduce energy consumption. Feeder: Equipped with servo-controlled shearing speed to ensure uniform material drop weight; Molding machine: It adopts standardized interface design and quick locking mechanism, is equipped with multiple mold libraries, and is matched with an automatic mold changing robot. The mold changing time is ≤5 minutes. Annealing furnace: Equipped with an infrared temperature sensor to achieve precise control of the cooling rate.

[0010] With the above structure, the forming equipment unit is the core physical actuator in glass bottle production, its function being to precisely transform raw materials into lightweight glass bottle products that meet the requirements. Through a complete process chain of raw material preparation, melting, feeding, forming, and annealing, the simulated and optimized process parameters are specifically achieved: the melting furnace ensures the quality of the molten glass, the feeder provides uniformly weighted droplets, the forming machine completes the rapid shaping and diversified production of bottles, and the annealing furnace effectively eliminates internal stress, thus jointly ensuring high-quality, high-efficiency, and low-energy production of the products.

[0011] The testing equipment unit forms a five-dimensional testing system covering appearance, size, internal structure, strength, and material properties, conducting quality inspections on finished glass bottles and key production processes. This system includes a laser thickness gauge, drop test device, internal pressure resistance test device, machine vision inspection device, ultrasonic flaw detector, and rapid LIBS material composition testing device.

[0012] By adopting the above structure, a five-dimensional, end-to-end quality inspection system is constructed, covering "appearance, dimensions, internal structure, strength, and material." Through a series of high-precision testing devices, glass bottles undergo comprehensive, thorough inspection to ensure that products meet high standards in physical properties, structural integrity, and material composition, thereby achieving precise quality control and rapid traceability of problems. Laser thickness gauges and machine vision inspection devices control dimensional accuracy and appearance defects, respectively. Ultrasonic flaw detection devices focus on detecting internal microscopic defects imperceptible to the human eye. Drop and internal pressure resistance tests verify the product's structural strength and durability. LIBS material composition testing ensures the purity of raw materials from the source, preventing batch quality issues. This system collectively constitutes a solid line of defense for product quality.

[0013] The control unit ensures that the equipment operates according to optimized parameters, including a servo motor controller, a temperature controller, and a pressure controller.

[0014] By adopting the above structure, the process parameters issued by the digital twin optimization module are accurately received and executed. Through closed-loop control of key actuators such as servo motors, temperature, and pressure, the core process links such as mold opening and closing, melting and annealing temperature, and blowing pressure are ensured to operate strictly according to the optimized settings, thereby seamlessly transforming the optimization decisions in the virtual world into high-quality products in the physical world.

[0015] The waste heat recovery unit collects waste heat from the flue gas of the pool kiln and the heat dissipation of the annealing furnace through a heat exchanger, and heats the raw material pretreatment water or workshop heating through the heat exchanger, with a waste heat utilization rate of ≥30%.

[0016] By adopting the above structure, waste heat from the production process is efficiently recovered and utilized. The high-temperature flue gas (≥800℃) from the pool kiln and the waste heat emitted from the annealing furnace are converted into directly usable thermal energy through a heat exchanger. This energy is used to preheat raw materials or provide heating for the workshop, thereby reducing the system's demand for external energy and achieving the goals of energy conservation and green production. Its designed waste heat utilization rate is no less than 30%, improving the energy efficiency of the entire production system.

[0017] The redundant design unit employs dual-path backup for key sensors, automatically switching to the backup sensor in case of primary sensor failure. The equipment's electrical system has overvoltage and overcurrent protection functions to ensure production continuity.

[0018] By adopting the above structure, a fault emergency and safety protection mechanism is built through dual-path backup of key sensors and overvoltage and overcurrent protection of the electrical system. When the main sensor fails, it can automatically and seamlessly switch to the backup sensor, and cut off the protection in time when there is an electrical abnormality, thereby minimizing unplanned downtime and ensuring the continuity and stability of the production process.

[0019] The equipment 3D information scanning and acquisition unit uses a 3D laser scanner to accurately acquire the 3D geometric dimensions, spatial positions, and physical properties of each device and mold during system initialization, and to construct an initial digital model.

[0020] The above structure provides a high-precision initial physical world data foundation for the digital twin system. By using a 3D laser scanner, the geometric dimensions, spatial layout, and physical properties of the equipment and molds are accurately captured during system initialization, thereby constructing an initial digital model that is completely consistent with the physical production environment, ensuring the accuracy and reliability of virtual simulation and optimization analysis.

[0021] The device operating status sensor network unit includes: Material flow sensors include an online raw material composition analyzer, a glass melt viscosity and temperature sensor, a droplet weight monitor, and a billet infrared thermal imager; Equipment status sensors include mold temperature and clamping force sensors, servo motor current and vibration sensors, air pressure and flow sensors, and thermocouples for each temperature zone of the annealing furnace. Quality inspection sensors include high-precision laser thickness gauges, high-speed industrial vision cameras, internal pressure testing machines, and multi-angle drop testing machines; Temperature sensors monitor the temperature of the tank furnace and annealing furnace; Pressure sensor to collect initial / final blowing pressure.

[0022] This architecture provides the digital twin system with real-time, high-precision production data covering the entire process of materials, equipment, and quality. Through various sensors deployed at key nodes of the production line, it monitors information across all dimensions, from raw material characteristics and molten glass state to equipment operating parameters and product quality, providing a reliable data foundation for subsequent simulation optimization, closed-loop control, and decision analysis.

[0023] The energy consumption and resource monitoring unit is equipped with smart meters and gas flow meters at key nodes such as the furnace, air compressor, and annealing furnace to monitor energy consumption data in real time.

[0024] By adopting the above structure, real-time and accurate energy data collection is carried out on key energy-consuming nodes such as pool kilns, air compressors, and annealing furnaces. This provides a precise data foundation for energy efficiency optimization, cost accounting, and carbon footprint management of the entire system, which is a key link in realizing green manufacturing and refined energy management.

[0025] The data transmission unit uses 5G / industrial Ethernet communication to upload data to the dynamic feedback optimization unit in real time.

[0026] The above structure serves as an "information highway" connecting physical devices and digital systems. By utilizing high-bandwidth, low-latency communication technologies such as 5G / Industrial Ethernet, real-time collected production data is stably and efficiently uploaded to the dynamic feedback optimization unit, providing reliable data flow assurance for the system's real-time analysis, decision-making, and closed-loop control.

[0027] The data storage unit uses blockchain technology to record equipment operation logs, process parameter adjustment records, and material quality traceability information, with a traceability period of no less than one year.

[0028] By adopting the above structure and leveraging the immutability and traceability of blockchain technology, key production data such as equipment operation, process adjustment, and quality traceability can be securely and reliably recorded for a long period of time, providing a reliable data foundation for data traceability, process analysis, quality improvement, and compliance auditing for a period of no less than one year.

[0029] The carbon footprint accounting unit connects to the EU EPDR and the domestic dual-carbon regulatory platform to meet compliance requirements. Based on the raw material carbon emission database, equipment energy consumption carbon emission coefficient and transportation carbon emission model, it calculates the carbon footprint of a single bottle in real time and generates a carbon footprint report.

[0030] By adopting the above structure and integrating internal and external carbon emission factor databases, the carbon emission data of a single glass bottle product can be quantified and tracked in real time and accurately throughout its entire life cycle. At the same time, it automatically generates carbon footprint reports that comply with domestic and international regulations, providing key data support for optimizing carbon emissions in the production process and achieving green compliance for enterprises.

[0031] The virtual simulation unit includes: Equipment modeling sub-unit: Based on the entered 3D equipment information, construct a high-fidelity virtual production line; Process modeling sub-unit: Based on the entered equipment working information, simulate the entire glass bottle processing process in a virtual environment; Physics-field coupling analysis sub-unit: Integrating computational fluid dynamics and finite element analysis, it performs high-precision simulation of the flow, temperature distribution, forming process and stress evolution of molten glass during the simulation process, and predicts wall thickness uniformity, structural strength and potential defects; Material property library sub-unit: Contains glass material models with different ratios, accurately reflecting their thermophysical and rheological properties.

[0032] Using the above structure, a high-fidelity digital twin is constructed and driven. Through comprehensive modeling of integrated equipment geometry, process logic, multiphysics, and material properties, the entire glass bottle production process is accurately reproduced and simulated in a virtual environment. This enables forward-looking prediction and evaluation of product performance, potential defects, and production feasibility, providing an accurate simulation basis and scientific evidence for subsequent process optimization.

[0033] The software simulation optimization unit is used to pre-optimize the equipment layout, dimensions, and molding process parameters, including: Information input subunit: Receives equipment 3D model information, equipment operating parameter information, multi-material physical performance parameters, and environmental parameters. The equipment 3D model information covers the structural dimensions, assembly relationships, and material parameters of the furnace, feeder, forming machine, annealing furnace, and testing equipment. The equipment operating parameter information includes glass melting temperature, blowing pressure range, mold opening and closing speed, and annealing cooling rate. The multi-material physical performance parameters cover the melting point, viscosity-temperature curve, and strength index of ordinary glass, glass-ceramic composite materials, nano-SiO2 reinforced glass, and lead-free environmentally friendly glass. Simulation Analysis Subunit: Based on the entered information, a virtual production environment is constructed, a 1:1 digital twin of the physical production line, simulating the entire processing of glass bottles from raw material melting, initial molding, shaping and strengthening, annealing to inspection. ANSYS finite element analysis software is used to perform stress analysis and process bottleneck identification on the initial blank and finished bottle, focusing on optimizing the stress concentration area at the bottom of the bottle. An LSTM long short-term memory network model and reinforcement learning algorithm are embedded to train a quality prediction model, enabling advanced prediction and evaluation of key quality indicators, with predictions possible 5-10 seconds in advance. Simultaneously, the collaborative working status of various electrical components is analyzed, process connection bottlenecks are identified, and the interval time between each process step is optimized to minimize process connection time. Coupled analysis of process parameters is performed, simulating the interaction between glass melt temperature, blowing pressure, and cooling time through iterative simulation. Output subunit: Through multiple rounds of simulation iteration, it outputs the optimal processing equipment size, layout scheme, and molding process parameters. The process parameters include raw material ratio, melting temperature, molding temperature, initial blowing pressure curve, final blowing pressure, mold opening and closing speed, annealing temperature curve, cooling rate, and time nodes of each process, ensuring that the process interval is minimized and the coupling interference of process parameters is minimized; it outputs the Pareto optimal solution and the carbon footprint pre-control target. Transfer learning sub-unit: When switching bottle type or material, the system can reuse the training model of similar products, and only needs to fine-tune the model with a small amount of new data, shortening the debugging cycle to 2-3 days; establish a standardized mold library and parameter library linkage mechanism.

[0034] By adopting the above structure, a digital twin that is completely consistent with the physical production line is constructed. By integrating multiphysics simulation and AI algorithms, multiple rounds of iterative optimization and forward-looking prediction of equipment layout, process parameters and production process are carried out in a virtual environment. Finally, the optimal production plan that takes into account quality, efficiency, energy consumption and carbon footprint is output, and transfer learning support is provided for rapid production changeover. Thus, the system performance can be accurately designed and continuously optimized before it is put into actual production.

[0035] The dynamic feedback optimization unit is communicatively connected to the software simulation optimization unit, the hardware execution module, and the global data perception module. It uses a PLC and a central controller to construct the control logic. Its reinforcement learning algorithm uses the highest pass rate, lowest energy consumption, lowest cost, and lowest carbon footprint as multi-objective reward functions to automatically iterate and optimize the combination of process parameters. It integrates AI algorithms to achieve adaptive parameter adjustment, defect tracing, and predictive equipment maintenance. The dynamic feedback optimization unit includes: Data distribution subunit: Distributes the optimal set of process parameters obtained by the optimization engine to the central controller of the hardware execution module to guide actual production, and receives equipment status and product quality information uploaded by the equipment working status sensor network unit in real time during the production process; Data comparison and analysis subunit: compares the real-time collected equipment status data and material quality data with the preset parameters output by the software simulation and optimization unit to identify deviations; Parameter Adaptive Adjustment Subunit: When a deviation is detected, parameters are automatically fine-tuned based on reinforcement learning algorithms: If the wall thickness uniformity is unqualified, the pressure of the corresponding side blowing head or the mold closing gap is adjusted; if the glass melt viscosity change leads to poor initial blowing effect, the coefficients k and b in the initial blowing pressure formula P1=kln(μ)+b are updated, where k and b are adjustable coefficients determined by automatic iterative optimization based on reinforcement learning algorithms, historical production data, and real-time detected glass melt viscosity changes. The initial values ​​of k and b are determined by the software simulation optimization unit through iterative pre-optimization using digital twin simulation. The updated values ​​in actual production are automatically calculated and generated by the reinforcement learning model based on real-time detected glass melt viscosity and product quality feedback, and the optimized parameters are fed back to the simulation model parameter library to achieve closed-loop self-learning optimization. μ is the glass melt viscosity; if the residual stress after annealing exceeds the standard, the cooling rate of the annealing furnace or equipment parameters are adjusted; if the carbon footprint exceeds the standard, the crushed glass ratio is increased or the tank furnace temperature is reduced; if the energy consumption is too high, the annealing cooling rate or tank furnace insulation power is optimized. Predictive maintenance subunit: Based on LSTM algorithm, analyze equipment vibration, temperature and wear data to predict the remaining life of the equipment and generate maintenance suggestions; establish equipment maintenance knowledge base to record fault causes and solutions, forming a prediction-maintenance-feedback closed loop; Model and parameter library update sub-unit: Feeds back the adjusted optimal parameters to the software simulation optimization unit, updates the parameter library of the simulation model, and provides accurate basic data for pre-optimization in subsequent production. Early warning and traceability sub-unit: Sets multi-level parameter threshold early warning, including quality threshold and carbon footprint threshold. When equipment parameters exceed the safe range or the material non-conforming rate exceeds 5% for three consecutive times, it triggers an audible and visual alarm and suspends production. Operation resumes after the parameters are adjusted to meet the requirements. In case of sudden failure, it automatically executes emergency adjustment logic to ensure that the non-conforming rate does not exceed 1%. The data server records equipment operation logs, process parameter adjustment records, and material quality information, with a traceability period of not less than 1 year, which facilitates fault diagnosis and process iteration.

[0036] By adopting the above structure, production data and simulation preset values ​​are compared in real time. AI algorithms such as reinforcement learning are used to automatically diagnose deviations, trace defects, and dynamically adjust process parameters. At the same time, predictive maintenance of equipment and system early warning are carried out. Finally, the optimal parameters verified in practice are fed back to update the simulation model, forming an adaptive optimization closed loop of "perception-decision-execution-learning" that runs through the virtual and physical worlds, continuously improving the quality, efficiency and greenness of production.

[0037] The supply chain collaboration interface unit is not limited to connecting to enterprise ERP systems to obtain order information. For new bottle type orders, the platform can conduct virtual trial production and production changeover simulations in advance, providing a basis for accurate production scheduling.

[0038] The above structure serves as a bridge connecting the production system and enterprise information management. By connecting with upper-level systems such as ERP to obtain order data, and using digital twin technology to conduct virtual trial production and changeover simulation for new orders, business needs are transformed into accurate and feasible production plans, thereby achieving efficient and scientific collaboration from order to production scheduling.

[0039] The full lifecycle quality traceability unit assigns a unique QR code to each product batch, linking it to the entire chain of data from raw material formula, process parameters, test results to outgoing logistics, forming an immutable quality blockchain.

[0040] By adopting the above structure and utilizing QR code and blockchain technology, a unique and tamper-proof "digital ID card" is established for each product batch. This allows for the complete and reliable recording and association of data across the entire chain, from raw materials and production to logistics, thereby enabling precise traceability of quality, rapid identification of responsibility for problems, and transparent management of the production process.

[0041] The carbon footprint and cost accounting unit automatically calculates the carbon emissions and production costs per unit of product based on real-time energy consumption and material data, providing data support for green manufacturing and refined management.

[0042] By adopting the above structure, and through the automatic collection and correlation of real-time production data, the resource consumption, carbon emissions and cost composition of each unit product can be accurately quantified, providing enterprises with quantifiable decision-making basis for implementing green manufacturing transformation and refined cost control, and achieving a balance between economic and environmental benefits.

[0043] The AR operation and maintenance unit is equipped with AR glasses that operators can wear, enabling them to view the internal status of the equipment, parameter deviations, and maintenance instructions in real time, thus lowering the barrier to operation and maintenance.

[0044] By adopting the above structure and using augmented reality technology, digital information such as the internal status of the equipment, real-time parameters, and maintenance instructions can be intuitively overlaid on the physical view, realizing the visualization of key operation and maintenance information. This reduces the reliance on personnel experience, improves the accuracy and efficiency of maintenance operations, and effectively lowers the operation and maintenance threshold.

[0045] The remote monitoring unit allows managers to view production line operation data in real time via PC or mobile device, and managers can remotely issue adjustment instructions.

[0046] By adopting the above structure, the limitations of physical space are broken, and the core indicators of the production line (output, pass rate, energy consumption, carbon footprint) are displayed in real time through multiple terminals. In case of anomalies, early warnings are pushed out proactively, enabling managers to grasp the production status anytime and anywhere and make remote interventions (such as parameter fine-tuning and emergency shutdown), thereby greatly improving the management response speed and decision-making efficiency.

[0047] When production efficiency or pass rate declines, the operator can trigger the one-click optimization unit. The system will automatically call the digital twin and AI algorithm to output the optimal parameters and perform adjustments.

[0048] By adopting the above structure, the complex system optimization process is simplified into a "one-click" operation. When production efficiency or pass rate declines, the system can automatically call the digital twin and AI algorithm for rapid diagnosis and global optimization, and automatically perform parameter adjustments. This significantly reduces the technical threshold and time cost of optimization operations, and quickly restores and ensures the optimal operating state of the production system.

[0049] The hierarchical access control unit sets three levels of permissions: operators, administrators, and R&D personnel. Operators can only view operation-related parameters, administrators can view all data, and R&D personnel can only access the simulation model to prevent the leakage of core technologies.

[0050] By adopting the above structure and dividing permissions into three levels—operators, managers, and R&D personnel—layered control of data and functions can be achieved. This ensures system information security, effectively isolates the core simulation model from the production site, prevents the leakage of key technologies, and ensures that each type of user can only access data and functions within their scope of responsibility.

[0051] The data security protection unit performs regular data backups and security audits, in compliance with relevant data security regulations.

[0052] By adopting the above structure and implementing two key measures—regular data backup and security audit—the integrity, availability, and compliance of system data are ensured, data loss and security risks are effectively prevented, and the requirements of laws and regulations such as the "Data Security Law" and the "Administrative Measures for Industrial Data Security" are met, providing reliable data security guarantees for the stable operation of the entire platform.

[0053] Compared with existing technologies, this optimized lightweight glass bottle molding process system has the following advantages: This system uses digital twin technology for virtual simulation and AI pre-optimization to accurately set the optimal process parameters. Combined with a five-dimensional full-process quality inspection system, it enables switching between multiple bottle types and materials to ensure product wall thickness uniformity, high strength, and low defect rate.

[0054] This system significantly reduces production energy consumption through intelligent power regulation, waste heat recovery, and energy consumption monitoring. The AI ​​algorithm-driven parameter optimization and cost accounting modules enable refined management of raw materials and energy, further reducing costs.

[0055] This system constructs a complete closed loop of "perception-decision-execution-feedback". The dynamic feedback optimization unit can automatically adjust process parameters and predict equipment maintenance based on real-time production data using reinforcement learning algorithms, and feed back the verified optimal parameters to the simulation model, enabling the system to have continuous evolution capabilities and ensuring that production is always in the optimal state.

[0056] This system integrates carbon footprint accounting, accurately tracks the carbon footprint of a single bottle, and generates compliance reports, effectively supporting enterprises in achieving their "dual carbon" goals.

[0057] This system lowers the barrier to operation and maintenance and improves response speed and operational security through AR operation and maintenance, remote monitoring and one-click optimization. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the optimized system process of the present invention.

[0059] Figure 2 This is a schematic diagram of the system flow of the hardware execution module in this invention.

[0060] Figure 3 This is a schematic diagram of the system flow of the global data perception module in this invention.

[0061] Figure 4 This is a schematic diagram of the system flow of the digital twin optimization module in this invention.

[0062] Figure 5 This is a schematic diagram of the system flow of the enterprise-level cloud management module in this invention.

[0063] Figure 6 This is a schematic diagram of the system flow of the human-machine collaborative operation and maintenance module in this invention. Detailed Implementation

[0064] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0065] like Figures 1-6 As shown, this lightweight glass bottle molding process optimization system includes a hardware execution module, a global data perception module, a digital twin optimization module, an enterprise-level cloud management module, and a human-machine collaborative operation and maintenance module. The hardware execution module performs glass bottle production, five-dimensional quality inspection and precise control processes sequentially based on simulation optimization parameters. It also utilizes waste heat for energy saving and ensures production continuity. It includes molding equipment unit, inspection equipment unit, control unit, waste heat recovery unit and redundancy design unit. The full-domain data perception module collects equipment status, material quality, and energy consumption data in real time and comprehensively, and uses 5G / industrial Ethernet for transmission and blockchain storage. It also integrates carbon footprint accounting to provide a full-element data foundation for the construction of digital twins, including equipment 3D information scanning and acquisition unit, equipment working status sensor network unit, energy consumption and resource monitoring unit, data transmission unit, data storage unit, and carbon footprint accounting unit. The digital twin optimization module simulates the entire production process by constructing a high-fidelity virtual model, uses simulation and AI algorithms to pre-analyze and optimize process parameters, and dynamically adjusts parameters and predicts maintenance based on real-time data comparison in actual production, realizing a continuous closed-loop iteration of simulation-optimization-feedback, including a virtual simulation unit, a software simulation optimization unit, and a dynamic feedback optimization unit; The enterprise-level cloud management module integrates order, production and logistics data by connecting to systems other than ERP, enabling closed-loop management from virtual trial production scheduling and full lifecycle quality traceability to carbon footprint and cost accounting. It provides unified data support for enterprise decision-making and supply chain collaboration, including a supply chain collaboration interface unit, a full lifecycle quality traceability unit and a carbon footprint and cost accounting unit. The human-machine collaborative operation and maintenance module uses technologies such as AR, remote monitoring, and one-click optimization to achieve visualization of equipment status, remote intervention, and intelligent parameter adjustment. Combined with hierarchical permissions and data security protection, it improves operation and maintenance efficiency and system security. It includes an AR operation and maintenance unit, a remote monitoring unit, a one-click optimization unit, a hierarchical access permission unit, and a data security protection unit.

[0066] Through the deep collaboration of five major modules, a closed-loop intelligent control system is constructed, encompassing physical production and the digital space. Its core working principle is as follows: the full-domain data perception module acts as the system's "sensory nerves," collecting real-time production data from all elements; the digital twin optimization module acts as the "decision brain," using high-fidelity simulation and AI algorithms to pre-simulate and iteratively optimize process parameters; the optimized instructions are then issued to the hardware execution module for precise production and quality inspection; the enterprise-level cloud management module achieves vertical integration with upper-level information systems, supporting full lifecycle management and supply chain collaboration; and finally, the human-machine collaborative operation and maintenance layer provides intuitive, secure, and efficient interaction and operation and maintenance support for the entire system.

[0067] The five modules form a complete closed loop of "perception-decision-execution-management-operation and maintenance", realizing continuous optimization of the production process, precise control of energy and carbon, and comprehensive improvement of operation and maintenance efficiency.

[0068] The molding equipment unit is configured according to the layout scheme and size based on the software simulation optimization unit output, including: Raw material preparation equipment: crushing equipment, mixing equipment, etc.; Melting pool furnace (used for raw material melting): adopts zoned heating control (accuracy ±3℃), adapts to the melting needs of glass liquid of various materials, and is equipped with intelligent power adjustment device to reduce heat preservation power during off-peak production periods and reduce energy consumption; Feeder (used for shearing molten glass into droplets): Equipped with servo-controlled shearing speed to ensure uniform droplet weight; Molding machine (for finished product shaping and transfer): It adopts a standardized interface design and a quick locking mechanism, and is equipped with multiple mold libraries (suitable for 50-1000ml capacity, round / square / irregular bottle shapes), and is equipped with an automatic mold changing robot with a mold changing time of ≤5 minutes; Annealing furnace (for low-temperature annealing): Equipped with an infrared temperature sensor to achieve precise control of the cooling rate (reducing the rate by 15%-25% compared to traditional processes). The forming equipment unit is the core physical actuator in glass bottle production, its function being to precisely transform raw materials into lightweight finished glass bottles that meet the requirements. Through a complete process chain of raw material preparation, melting, feeding, forming, and annealing, the simulated and optimized process parameters are specifically implemented: the melting furnace ensures the quality of the molten glass, the feeder provides uniformly weighted droplets, the forming machine completes the rapid shaping and diversified production of bottles, and the annealing furnace effectively eliminates internal stress, thus jointly ensuring high-quality, high-efficiency, and low-energy production of the products.

[0069] The testing equipment unit forms a five-dimensional testing system covering appearance, dimensions, internal structure, strength, and material properties, conducting quality inspections on finished glass bottles and key production processes, including: Laser thickness gauge (measurement accuracy ±0.02mm, used for full bottle wall thickness scanning). Drop test device (test height 1.2m-1.5m); Internal pressure resistance testing device (test pressure ≥ 60 kPa); Machine vision inspection device (shooting frame rate ≥100fps, defect recognition accuracy ≥0.1mm, recognizing defects such as bottle scratches, bubbles, cracks, and bottle mouth burrs); Ultrasonic flaw detection device (detects internal microbubbles with a diameter ≥0.3mm, enabling fully online detection of internal micro-defects); The LIBS (Laser-Induced Breakdown Spectrometer) material composition rapid detection device verifies the purity of raw materials in real time, avoiding quality problems caused by mixed batches.

[0070] We have established a five-dimensional, end-to-end quality inspection system covering "appearance, size, internal structure, strength, and material." Through a series of high-precision testing devices, we conduct comprehensive and thorough inspections of glass bottles to ensure that the products leaving the factory meet high standards in terms of physical properties, structural integrity, and material composition. This enables precise quality control and rapid traceability of problems.

[0071] Laser thickness gauges and machine vision inspection devices control dimensional accuracy and appearance defects, respectively.

[0072] Ultrasonic flaw detection devices are designed to detect internal microscopic defects that are invisible to the human eye.

[0073] Drop and internal pressure testing devices verify the structural strength and durability of the product.

[0074] LIBS material composition testing ensures the purity of raw materials from the source, preventing batch quality issues.

[0075] This system together forms a solid defense for product quality.

[0076] The control unit ensures that the equipment operates according to optimized parameters, including: Servo motor controller (controls the opening and closing speed of the mold); Temperature controller (controls the temperature of the tank furnace and annealing furnace); Pressure controller (controls initial / final blowing pressure).

[0077] It accurately receives and executes the process parameters issued by the digital twin optimization module. Through closed-loop control of key actuators such as servo motors, temperature, and pressure, it ensures that core process links such as mold opening and closing, melting and annealing temperature, and blowing pressure operate strictly according to the optimized settings, thereby seamlessly transforming the optimization decisions in the virtual world into high-quality products in the physical world.

[0078] The waste heat recovery unit collects waste heat (≥800℃) from the flue gas of the pool kiln and the heat dissipation from the annealing furnace through a heat exchanger. The heat exchanger is used to heat the raw material pretreatment water or workshop heating, with a waste heat utilization rate of ≥30%.

[0079] This system efficiently recovers and utilizes waste heat from the production process, converting the high-temperature flue gas (≥800℃) from the pool kiln and the waste heat emitted from the annealing furnace into directly usable thermal energy through a heat exchanger. This energy is used to preheat raw materials or provide heating for the workshop, thereby reducing the system's demand for external energy and achieving the goals of energy conservation and green production. Its designed waste heat utilization rate is no less than 30%, improving the energy efficiency of the production system.

[0080] The redundant design unit employs dual-path backup for critical sensors (pressure, temperature). In the event of a primary sensor failure, it automatically switches to the backup sensor. The equipment's electrical system has overvoltage and overcurrent protection functions to ensure production continuity.

[0081] By implementing dual-path backup for key sensors and overvoltage and overcurrent protection for the electrical system, a fault emergency and safety protection mechanism is constructed. When the main sensor fails, it can automatically and seamlessly switch to the backup sensor, and promptly cut off protection in case of electrical abnormalities, thereby minimizing unplanned downtime and ensuring the continuity and stability of the production process.

[0082] The equipment 3D information scanning and acquisition unit uses a 3D laser scanner to accurately acquire the 3D geometric dimensions, spatial position, and physical properties of each piece of equipment and mold during system initialization, and to build an initial digital model.

[0083] It provides a high-precision initial physical world data foundation for digital twin systems. By using a 3D laser scanner, the geometric dimensions, spatial layout, and physical properties of equipment and molds are accurately captured during system initialization, thereby constructing an initial digital model that is completely consistent with the physical production environment, ensuring the accuracy and reliability of virtual simulation and optimization analysis.

[0084] The device operating status sensor network unit includes: Material flow sensors include an online raw material composition analyzer, a glass melt viscosity and temperature sensor, a droplet weight monitor, and a billet infrared thermal imager; Equipment status sensors include mold temperature and clamping force sensors, servo motor current and vibration sensors, air pressure and flow sensors, and thermocouples for each temperature zone of the annealing furnace. Quality inspection sensors include high-precision laser thickness gauges, high-speed industrial vision cameras (for detecting surface defects), internal pressure testing machines, and multi-angle drop testing machines; Temperature sensors (accuracy ±1℃) monitor the temperature of the tank furnace and annealing furnace; The pressure sensor (accuracy ±0.001MPa) collects the initial / final blowing pressure.

[0085] It provides digital twin systems with real-time, high-precision production data covering the entire process of materials, equipment, and quality. Through various sensors deployed at key nodes of the production line, it monitors information in real time from raw material characteristics, molten glass state, equipment operating parameters to product quality, providing a reliable data foundation for subsequent simulation optimization, closed-loop control, and decision analysis.

[0086] The energy consumption and resource monitoring unit installs smart meters and gas flow meters at key nodes such as the pool furnace, air compressor, and annealing furnace to monitor energy consumption data in real time.

[0087] By collecting real-time and accurate energy data from key energy-consuming nodes such as furnaces, air compressors, and annealing furnaces, a precise data foundation is provided for the energy efficiency optimization, cost accounting, and carbon footprint management of the entire system. This is a crucial link in achieving green manufacturing and refined energy management.

[0088] The data transmission unit uses 5G / Industrial Ethernet communication to upload data to the dynamic feedback optimization unit in real time. As an "information superhighway" connecting physical devices and digital systems, it utilizes high-bandwidth, low-latency communication technologies such as 5G / Industrial Ethernet to stably and efficiently upload real-time collected production data to the dynamic feedback optimization unit, providing reliable data flow assurance for the system's real-time analysis, decision-making, and closed-loop control.

[0089] The data storage unit uses blockchain technology to record equipment operation logs, process parameter adjustment records, and material quality traceability information, with a traceability period of no less than one year.

[0090] Leveraging the immutable and traceable characteristics of blockchain technology, key production data such as equipment operation, process adjustment, and quality traceability can be securely and reliably recorded for a long period of time, providing a reliable data foundation for data traceability, process analysis, quality improvement, and compliance auditing for a period of no less than one year.

[0091] The carbon footprint accounting unit connects to the EU EPDR and the domestic dual-carbon regulatory platform to meet compliance requirements. Based on the raw material carbon emission database, equipment energy consumption carbon emission coefficient, and transportation carbon emission model, it calculates the carbon footprint of a single bottle in real time. It connects to the raw material carbon emission database (such as 0.15kg / kg of quartz sand and 0.8kg / kg of soda ash), the equipment energy consumption carbon emission coefficient (such as 0.6kg / kWh of electricity), and the transportation carbon emission model (calculated by distance and load) to generate a carbon footprint report.

[0092] By integrating internal and external carbon emission factor databases, the system can quantify and track the carbon emission data of individual glass bottle products throughout their entire life cycle in real time and with high accuracy. It can also automatically generate carbon footprint reports that comply with domestic and international regulations, providing key data support for optimizing carbon emissions in the production process and achieving green compliance for enterprises.

[0093] The virtual simulation unit includes: Equipment modeling sub-unit: Based on the entered 3D equipment information, construct a high-fidelity virtual production line; Process modeling sub-unit: Based on the input equipment working information (such as the blowing pressure curve P1=kln(μ)+b, mold opening and closing speed, temperature setting), the entire glass bottle processing process is simulated in a virtual environment; Physics-field coupling analysis sub-unit: Integrating computational fluid dynamics and finite element analysis, it performs high-precision simulation of the flow, temperature distribution, forming process and stress evolution of molten glass during the simulation process, and predicts wall thickness uniformity, structural strength and potential defects (such as uneven wall thickness and stress concentration). Material property library sub-unit: It contains glass material models with different proportions (such as 25%-35% quartz glass, 1-3% ZrO2, 0.5-1.5% Li2O), which accurately reflect their thermophysical and rheological properties.

[0094] By constructing and driving a high-fidelity digital twin, and through comprehensive modeling of equipment geometry, process logic, multiphysics, and material properties, the entire glass bottle production process can be accurately reproduced and simulated in a virtual environment. This enables forward-looking prediction and evaluation of product performance, potential defects, and production feasibility, providing a precise simulation basis and scientific evidence for subsequent process optimization.

[0095] The software simulation optimization unit is used to pre-optimize equipment layout, dimensions, and molding process parameters, including: Information input subunit: Receives equipment 3D model information, equipment operating parameter information, physical performance parameters of multiple materials, and environmental parameters. The equipment 3D model information covers the structural dimensions, assembly relationships, and material parameters of the furnace, feeder, forming machine, annealing furnace, and testing equipment. The equipment operating parameter information includes the glass melting temperature, blowing pressure range, mold opening and closing speed, and annealing cooling rate. The physical performance parameters of multiple materials cover the melting point, viscosity-temperature curve, and strength index of ordinary glass, glass-ceramic composite materials, nano-SiO2 reinforced glass, and lead-free environmentally friendly glass. The simulation analysis subunit constructs a virtual production environment based on a 1:1 digital twin of the physical production line using entered information. It simulates the entire glass bottle manufacturing process, from raw material melting, initial molding, shaping and strengthening, annealing to inspection. ANSYS finite element analysis software is used to perform stress analysis and process bottleneck identification on the initial preform and finished bottle, focusing on optimizing the stress concentration area at the bottle bottom. An LSTM long short-term memory network model and reinforcement learning algorithm are embedded to train a quality prediction model, enabling advanced prediction and evaluation of key quality indicators, with predictions possible 5-10 seconds in advance. Simultaneously, the unit analyzes the collaborative working status of various electrical components (servo motors, temperature controllers, pressure controllers), identifies process connection bottlenecks, optimizes the interval time between process steps, and minimizes process connection time. Coupled analysis of process parameters is performed, simulating the iterative effects of glass melt temperature, blowing pressure, and cooling time. Output subunit: Through multiple rounds of simulation iteration, it outputs the optimal processing equipment size, layout scheme (such as mold spacing, material transfer path) and molding process parameters. The process parameters include raw material ratio, melting temperature (1500-1550℃), molding temperature (1150-1250℃), initial blowing pressure curve (P1=kln(μ)+b, k and b are simulation optimization coefficients, P1=0.28-0.32MPa), final blowing pressure (0.48-0.52MPa), mold opening and closing speed (accuracy ±1mm / s), annealing temperature curve, cooling rate (15%-25% lower than traditional process), and time nodes of each process, ensuring that the process interval is minimized and the coupling interference of process parameters is minimized; it outputs the Pareto optimal solution of "energy consumption-quality-carbon footprint" and the carbon footprint pre-control target (such as single bottle carbon footprint ≤0.5kgCO2e). In the transfer learning sub-unit, when switching bottle types or materials, the system can reuse the training model of similar products, requiring only a small amount of new data for model fine-tuning, shortening the debugging cycle to 2-3 days; a standardized mold library and parameter library linkage mechanism is established to adapt to 50-1000ml capacity, round / square / irregular shapes and more than 10 kinds of bottle types, as well as the production of multiple materials such as ordinary glass and glass-ceramic composite materials.

[0096] By constructing a digital twin that is completely consistent with the physical production line, and integrating multiphysics simulation and AI algorithms, the system performs multiple rounds of iterative optimization and forward-looking prediction of equipment layout, process parameters and production flow in a virtual environment. Ultimately, it outputs the optimal production solution that takes into account quality, efficiency, energy consumption and carbon footprint, and provides transfer learning support for rapid production changeover. This allows for precise design and continuous optimization of system performance before it is put into actual production.

[0097] The dynamic feedback optimization unit communicates with the software simulation optimization unit, hardware execution module, and global data perception module. It uses a PLC and central controller to construct the control logic. Its reinforcement learning algorithm uses a multi-objective reward function of "highest pass rate + lowest energy consumption + lowest cost + lowest carbon footprint" to automatically iteratively optimize process parameter combinations. It integrates AI algorithms to achieve adaptive parameter adjustment, defect tracing, and predictive equipment maintenance. The dynamic feedback optimization unit includes: Data distribution subunit: Distributes the optimal set of process parameters obtained by the optimization engine to the central controller of the hardware execution module to guide actual production, and receives "equipment status" and "product quality" information uploaded by the equipment working status sensor network unit in real time during the production process; Data comparison and analysis subunit: Compares the real-time collected equipment status data and material quality data with the preset parameters output by the software simulation and optimization unit to identify deviations (such as wall thickness exceeding ±0.02mm, blowing pressure deviating from the preset range, and equipment temperature fluctuation exceeding ±5℃). Parameter adaptive adjustment subunit: When a deviation is detected, parameters are automatically fine-tuned based on a reinforcement learning algorithm: If the wall thickness uniformity is unqualified, the pressure of the corresponding side blowing head or the mold closing gap is adjusted; if the glass melt viscosity change leads to poor initial blowing effect, the coefficients k and b in the initial blowing pressure formula P1=kln(μ)+b are updated, where k and b are adjustable coefficients determined by automatic iterative optimization based on reinforcement learning algorithm according to historical production data and real-time detected glass melt viscosity changes. The initial values ​​of k and b are determined by the software simulation optimization unit through iterative pre-optimization using digital twin simulation. The updated values ​​are automatically generated by the reinforcement learning model based on the real-time detection of the glass melt viscosity and product quality feedback. The optimized parameters are then fed back to the simulation model parameter library to achieve closed-loop self-learning optimization. μ is the glass melt viscosity. If the residual stress exceeds the standard after annealing, adjust the annealing furnace cooling rate or equipment parameters (mold closing gap, servo motor control accuracy). If the carbon footprint exceeds the standard, the cullet glass content can be increased to the upper limit allowed by the process (e.g., 35%) or the tank furnace temperature can be reduced (10-20℃). If the energy consumption is too high, optimize the annealing cooling rate or the tank furnace insulation power. Predictive maintenance subunit: Based on LSTM algorithm, analyze equipment vibration, temperature and wear data to predict the remaining life of the equipment (e.g., warning when the remaining life of the mold wear is ≥5000 cycles), generate maintenance suggestions (e.g., mold grinding, motor oil change); establish equipment maintenance knowledge base, record fault causes and solutions, and form a closed loop of "prediction-maintenance-feedback"; Model and parameter library update sub-unit: Feeds back the adjusted optimal parameters to the software simulation optimization unit, updates the parameter library of the simulation model, provides accurate basic data for the pre-optimization of subsequent production, and realizes closed-loop adaptive optimization of process and equipment; Early warning and traceability sub-unit: Sets multi-level parameter threshold early warning, quality threshold (e.g., wall thickness ±0.02mm), carbon footprint threshold (e.g., ≤0.5kgCO2e / bottle). When equipment parameters exceed the safe range or the material non-conforming rate exceeds 5% for three consecutive times, an audible and visual alarm is triggered and production is suspended. Operation resumes after the parameters are adjusted to be within acceptable limits. In case of sudden failure, emergency adjustment logic is automatically executed (e.g., reducing production speed, stabilizing blowing pressure) to ensure that the non-conforming rate does not exceed 1%. The data server records equipment operation logs, process parameter adjustment records, and material quality information, with a traceability period of no less than one year, facilitating fault diagnosis and process iteration.

[0098] By comparing production data with simulation presets in real time, AI algorithms such as reinforcement learning are used to automatically diagnose deviations, trace defects, and dynamically adjust process parameters. At the same time, predictive maintenance of equipment and system early warning are carried out. Finally, the optimal parameters verified in practice are fed back to update the simulation model, forming an adaptive optimization closed loop of "perception-decision-execution-learning" that runs through the virtual and physical worlds, continuously improving the quality, efficiency and greenness of production.

[0099] The supply chain collaboration interface unit can connect to various systems beyond just enterprise ERP systems to obtain order information. For orders of new bottle types, the platform can conduct virtual trial production and production changeover simulations in advance, providing a basis for accurate production scheduling.

[0100] As a bridge connecting the production system and enterprise information management, it obtains order data by connecting with upper-level systems such as ERP, and uses digital twin technology to conduct virtual trial production and production changeover simulation for new orders, transforming business needs into accurate and feasible production plans, thereby achieving efficient and scientific collaboration from order to production scheduling.

[0101] The full lifecycle quality traceability unit assigns a unique QR code to each product batch, linking it to the entire chain of data from raw material formula, process parameters, test results to outgoing logistics, forming an immutable quality blockchain.

[0102] By using QR codes and blockchain technology, a unique and tamper-proof "digital ID card" is created for each product batch, which completely and reliably records and links its data from raw materials and production to logistics, thereby achieving precise traceability of quality, rapid determination of responsibility for problems, and transparent management of the production process.

[0103] The carbon footprint and cost accounting unit automatically calculates the carbon emissions and production costs per unit of product based on real-time energy consumption and material data, providing data support for green manufacturing and refined management.

[0104] By automatically collecting and linking real-time production data, the resource consumption, carbon emissions, and cost composition of each unit product can be accurately quantified, providing enterprises with quantifiable decision-making basis for implementing green manufacturing transformation and refined cost control, thereby achieving a balance between economic and environmental benefits.

[0105] The AR maintenance unit is equipped with AR glasses that operators can wear to view the internal status of the equipment (such as mold wear), parameter deviations, and maintenance instructions (such as mold disassembly steps) in real time, thus lowering the barrier to maintenance.

[0106] Augmented reality technology is used to intuitively overlay digital information such as the internal status of equipment, real-time parameters, and maintenance instructions onto the physical field of view, thereby visualizing key operation and maintenance information, reducing reliance on personnel experience, improving the accuracy and efficiency of maintenance operations, and effectively lowering the threshold for operation and maintenance.

[0107] The remote monitoring unit allows managers to view production line operation data (output, pass rate, energy consumption, carbon footprint) in real time via PC and mobile devices. It pushes early warning information (SMS, APP notification) when anomalies occur, and managers can remotely issue adjustment instructions (emergency shutdown, parameter fine-tuning).

[0108] Breaking the limitations of physical space, the system displays core production line indicators (output, pass rate, energy consumption, carbon footprint) in real time through multiple terminals and proactively pushes early warnings when anomalies occur. This enables managers to grasp the production status anytime, anywhere and make remote interventions (such as parameter fine-tuning and emergency shutdown), thereby greatly improving management response speed and decision-making efficiency.

[0109] When production efficiency or yield rate declines, operators can trigger one-click optimization. The system automatically calls up the digital twin and AI algorithm to output the optimal parameters and perform adjustments.

[0110] The complex system optimization process is simplified into a "one-click" operation. When production efficiency or pass rate declines, the system can automatically call upon digital twins and AI algorithms for rapid diagnosis and global optimization, and automatically adjust parameters. This significantly reduces the technical threshold and time cost of optimization operations, and quickly restores and ensures the optimal operating state of the production system.

[0111] The hierarchical access control unit sets up three levels of permissions: operators, administrators, and R&D personnel. Operators can only view operation-related parameters, administrators can view all data, and R&D personnel can only access the simulation model, preventing the leakage of core technologies. By dividing permissions into three levels, layered control of data and functions is achieved. This ensures system information security while effectively isolating the core simulation model from the production site, preventing the leakage of key technologies, and ensuring that each type of user can only access data and functions within their scope of responsibility.

[0112] The data security protection unit regularly performs data backups and security audits, complying with relevant data security regulations. By regularly implementing these two key measures, the integrity, availability, and compliance of system data are ensured, effectively preventing data loss and security risks. This meets the requirements of regulations such as the "Data Security Law" and the "Industrial Data Security Management Measures," providing reliable data security assurance for the stable operation of the entire platform.

[0113] The workflow steps of this system are as follows: S1: System initialization, inputting equipment 3D information, initial process parameters and material formula library through the global data perception module.

[0114] S2: The digital twin optimization module performs virtual production simulation. Through a multi-objective optimization engine, it obtains the optimal equipment layout, mold structure, process parameter set, and predicted production cycle time.

[0115] S3: Based on the optimization results of S2, configure the hardware execution module and start production.

[0116] S4: During the production process, the global data sensing unit collects all data in real time and uploads it to the digital twin optimization module.

[0117] S5: The digital twin optimization module performs dynamic feedback control. Short-term feedback: Based on real-time quality data, process parameters are fine-tuned using a machine learning proxy model.

[0118] Mid-term feedback: Based on equipment health prediction, maintenance windows are planned, and based on raw material market price fluctuations, the optimal cost-effective formula fine-tuning scheme is simulated and recommended.

[0119] Long-term learning: Using actual production data as new samples, the simulation model and surrogate model are continuously trained and updated, enabling the system to have self-learning and evolution capabilities.

[0120] S6: The enterprise-level cloud management module integrates all information to achieve transparent management, quality traceability, and carbon accounting throughout the entire process from order to delivery.

[0121] Example 1 This embodiment takes the mass production of a 500ml lightweight liquor bottle (ordinary glass material) and a 300ml irregularly shaped cosmetic bottle (glass-ceramic composite material) as examples, and adopts the glass bottle lightweight molding process optimization system of the present invention. The specific steps are as follows: 1. Software simulation intelligent pre-optimization Input multi-dimensional information: Equipment 3D information (tank volume 5m³) 3 Modular prototyping / forming mold components, annealing furnace length 10m), working parameters (raw material ratio: quartz sand 60%, soda ash 15%, limestone 10%, crushed glass 30%, ZrO2 2%, Li2O 1%; melting temperature 1520℃, forming temperature 1200℃), multi-material parameters (melting point of ordinary glass 1500℃, melting point of glass-ceramic composite material 1550℃), basic carbon footprint data (carbon emissions of quartz sand 0.15kg / kg, carbon emissions of electricity 0.6kg / kWh); Digital Twin and AI Simulation: A 1:1 digital twin was constructed to simulate the entire production process of a 500ml liquor bottle. After training with historical data, the LSTM model predicted that the wall thickness deviation of the initial blank would be ≤±0.01mm. The reinforcement learning algorithm iteratively optimized the formula for initial blowing pressure, P1=0.05ln(μ)+0.25, final blowing pressure, and annealing cooling rate of 5℃ / min (20% lower than the traditional method). The energy consumption simulation output melting temperature of 1520℃ was the optimal solution for "energy consumption-mass". Flexible parameter output: linear layout of equipment for 500ml liquor bottles (spacing 1.5m-2.0m), mold parameters and process adjustment scheme for 300ml irregular-shaped bottles (initial blowing pressure 0.3MPa, final blowing pressure 0.52MPa), mold changing adaptation parameters (automatic mold changing robot path planning); carbon footprint pre-control target is ≤0.45kg / bottle.

[0122] 2. Modular hardware deployment and production execution Molding equipment deployment: Tank kiln, modular mold components, automatic mold changing robot, and waste heat recovery heat exchanger are installed according to an optimized layout; the molds adopt H13 steel with diamond-like coating (liquor bottles) and H13 steel with ceramic coating (irregular-shaped bottles). Inspection equipment deployment: Five-dimensional inspection system (laser thickness gauge, high-speed vision camera, ultrasonic flaw detector, 1.5m drop test device, LIBS material detector); Production switchover: After the 500ml liquor bottle is produced, an automatic mold change command is triggered. The robot changes the mold of the irregular bottle according to the digital twin planning path. The mold change time is 4.5 minutes. The software automatically calls up the process parameters of the irregular bottle and adjusts the furnace temperature to 1550℃ to adapt to the glass-ceramic composite material.

[0123] 3. Real-time data acquisition and dynamic optimization Data acquisition: Real-time acquisition of glass melt viscosity μ=1000Pa·s (liquor bottle) / 1200Pa·s (irregularly shaped bottle), initial blowing pressure 0.3MPa / 0.32MPa, wall thickness uniformity ±0.015mm / ±0.018mm, single bottle carbon footprint 0.43kg / 0.48kg, equipment vibration value 0.2mm / s, and mold wear 0.01mm; Dynamic adjustment: When the carbon footprint of the irregularly shaped bottle is close to the threshold of 0.48 kg, the system automatically increases the proportion of crushed glass from 30% to 35%, and the carbon footprint is reduced to 0.46 kg after adjustment; when a slight burr is detected at the mouth of the irregularly shaped bottle, the closing gap of the shaping mold is associated with it and reduced by 0.01 mm, the defect is eliminated. Equipment maintenance warning: When the cumulative wear of the mold reaches 0.05mm, the system predicts that the remaining lifespan is 8000 cycles and pushes a grinding and maintenance suggestion.

[0124] 4. Human-machine collaboration and full lifecycle traceability Operation and maintenance: Operators use AR glasses to check the wear of the molds and complete the polishing according to the instructions; managers use mobile devices to monitor the pass rate (96.5% / 95.8%), energy consumption and carbon footprint data of the two bottle production lines in real time. Quality traceability: Each bottle of product is affixed with a unique QR code, which is linked to the raw material batch, equipment number, process parameters, test results, and carbon footprint information, enabling full traceability from raw materials to finished products to consumers.

[0125] 5. Production Results This system is used to produce 500ml liquor bottles (380g / bottle) and 300ml irregularly shaped cosmetic bottles (280g / bottle). Product specifications: wall thickness uniformity error ≤ ±0.02mm, internal pressure resistance ≥65kPa, 1.5m drop breakage rate ≤3%, and appearance defect missed rate 0.08%. Production efficiency: 600 liquor bottles / hour, 450 irregularly shaped bottles / hour. Cost per bottle: 0.55 yuan reduction for liquor bottles, 0.4 yuan reduction for irregularly shaped bottles. Carbon footprint per bottle: 0.43kg for liquor bottles, 0.46kg for irregularly shaped bottles. Annual cost savings: 6.5 million yuan, and annual carbon dioxide emission reduction of 2200 tons.

[0126] This embodiment demonstrates that the glass bottle lightweight molding process optimization system of the present invention, through technologies such as intelligent algorithms, digital twins, and modular design, achieves efficient, precise, and green production of multiple bottle types and materials, improves production efficiency, product quality, and environmental protection levels, and reduces costs and maintenance difficulties, thus possessing extremely strong industrial application value.

[0127] In summary, this system utilizes digital twin technology for virtual simulation and AI pre-optimization to accurately set optimal process parameters. Combined with a five-dimensional full-process quality inspection system, it improves product qualification rate and production efficiency. It enables rapid switching between multiple bottle types and materials, reducing mold changeover time to less than 5 minutes, and ensuring product wall thickness uniformity, high strength, and low defect rate.

[0128] This system significantly reduces production energy consumption through intelligent power regulation, waste heat recovery (utilization rate ≥30%), and energy consumption monitoring. Simultaneously, the AI ​​algorithm-driven parameter optimization and cost accounting modules enable refined management of raw materials and energy, reducing per-bottle costs and bringing considerable economic benefits to the enterprise.

[0129] This system constructs a complete closed loop of "perception-decision-execution-feedback". The dynamic feedback optimization unit can automatically adjust process parameters and predict equipment maintenance based on real-time production data using reinforcement learning algorithms, and feed back the verified optimal parameters to the simulation model, enabling the system to have continuous evolution capabilities and ensuring that production is always in an optimal state.

[0130] This system integrates carbon footprint accounting, accurately tracks the carbon footprint of each bottle, and generates compliance reports, effectively supporting enterprises in achieving their "dual carbon" goals. Supply chain collaboration and transfer learning units enable rapid virtual trial production of new orders, shortening the new product debugging cycle to 2-3 days and greatly enhancing market adaptability and flexible production capabilities.

[0131] This system lowers the barrier to entry for operations and maintenance (O&M) and improves response speed and operational security through features such as AR O&M, remote monitoring, and one-click optimization. Combined with hierarchical access control and blockchain data storage, it enhances O&M efficiency while ensuring the security of core technical data and the traceability of the entire production process.

[0132] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. A lightweight glass bottle molding process optimization system, characterized in that, It includes a hardware execution module, a global data perception module, a digital twin optimization module, an enterprise-level cloud management module, and a human-machine collaborative operation and maintenance module; The hardware execution module performs the glass bottle production, five-dimensional quality inspection and precise control process sequentially according to the simulation optimization parameters, and uses waste heat to save energy to ensure production continuity. It includes a molding equipment unit, an inspection equipment unit, a control unit, a waste heat recovery unit and a redundancy design unit. The full-domain data perception module collects equipment status, material quality, and energy consumption data in real time and comprehensively, and uses 5G / industrial Ethernet for transmission and blockchain storage. It also integrates carbon footprint accounting to provide a full-element data foundation for the construction of digital twins, including a three-dimensional information scanning and acquisition unit for equipment, a sensor network unit for equipment working status, an energy consumption and resource monitoring unit, a data transmission unit, a data storage unit, and a carbon footprint accounting unit. The digital twin optimization module simulates the entire production process by constructing a high-fidelity virtual model, uses simulation and AI algorithms to pre-simulate and optimize process parameters, and dynamically adjusts parameters and predicts maintenance based on real-time data comparison in actual production, realizing a continuous closed-loop iteration of simulation-optimization-feedback, including a virtual simulation unit, a software simulation optimization unit, and a dynamic feedback optimization unit. The enterprise-level cloud management module integrates order, production and logistics data by connecting to systems other than ERP, enabling closed-loop management from virtual trial production and scheduling, full lifecycle quality traceability to carbon footprint and cost accounting. It provides unified data support for enterprise decision-making and supply chain collaboration, including a supply chain collaboration interface unit, a full lifecycle quality traceability unit and a carbon footprint and cost accounting unit. The human-machine collaborative operation and maintenance module, through technologies including but not limited to AR, remote monitoring, and one-click optimization, enables visualization of equipment status, remote intervention, and intelligent parameter adjustment. It also incorporates hierarchical permissions and data security protection, including an AR operation and maintenance unit, a remote monitoring unit, a one-click optimization unit, a hierarchical access permission unit, and a data security protection unit.

2. The lightweight glass bottle molding process optimization system according to claim 1, characterized in that, The molding equipment unit is configured according to the layout scheme and size output by the software simulation optimization unit, including: Raw material preparation equipment includes, but is not limited to, crushing and mixing equipment; Melting tank furnace: It adopts zoned heating control, adapts to the melting needs of glass molten material of multiple materials, and is equipped with intelligent power adjustment device; Feeder: Equipped with servo-controlled shearing speed; Molding machine: It adopts standardized interface design and quick locking mechanism, is equipped with multiple mold libraries, and is matched with an automatic mold changing robot. The mold changing time is ≤5 minutes. Annealing furnace: equipped with an infrared temperature sensor; The testing equipment unit forms a five-dimensional testing system from appearance, size, internal structure, strength to material, to conduct quality testing on finished glass bottles and key production processes. It includes a laser thickness gauge, drop test device, internal pressure resistance test device, machine vision inspection device, ultrasonic flaw detection device, and LIBS material composition rapid detection device. The control unit ensures that the equipment operates according to optimized parameters, including a servo motor controller, a temperature controller, and a pressure controller.

3. The lightweight glass bottle molding process optimization system according to claim 2, characterized in that, The waste heat recovery unit collects waste heat from the flue gas of the pool kiln and the heat dissipation of the annealing furnace through a heat exchanger, and heats the raw material pretreatment water or workshop heating through the heat exchanger, with a waste heat utilization rate of ≥30%. The redundant design unit adopts dual-path backup for key sensors, automatically switching to the backup sensor when the main sensor fails, and the equipment electrical system has overvoltage and overcurrent protection functions. The equipment 3D information scanning and acquisition unit uses a 3D laser scanner to accurately acquire the 3D geometric dimensions, spatial positions, and physical properties of each device and mold during system initialization, and to construct an initial digital model.

4. The glass bottle lightweight molding process optimization system according to claim 3, characterized in that, The device operating status sensor network unit includes: Material flow sensors include an online raw material composition analyzer, a glass melt viscosity and temperature sensor, a droplet weight monitor, and a billet infrared thermal imager; Equipment status sensors include mold temperature and clamping force sensors, servo motor current and vibration sensors, air pressure and flow sensors, and thermocouples for each temperature zone of the annealing furnace. Quality inspection sensors include high-precision laser thickness gauges, high-speed industrial vision cameras, internal pressure testing machines, and multi-angle drop testing machines; Temperature sensors: monitor the temperature of the tank furnace and annealing furnace; Pressure sensor: Collects initial / final blowing pressure; The energy consumption and resource monitoring unit is equipped with smart meters and gas flow meters at key nodes, including but not limited to tank furnaces, air compressors and annealing furnaces, to monitor energy consumption data in real time. The data transmission unit uses 5G / industrial Ethernet communication to upload data to the dynamic feedback optimization unit in real time.

5. The glass bottle lightweight molding process optimization system according to claim 4, characterized in that, The data storage unit uses blockchain technology to record equipment operation logs, process parameter adjustment records, and material quality traceability information, with a traceability period of no less than one year. The carbon footprint accounting unit connects to the EU EPDR and the domestic dual-carbon regulatory platform. Based on the raw material carbon emission database, equipment energy consumption carbon emission coefficient and transportation carbon emission model, it calculates the carbon footprint of a single bottle in real time and generates a carbon footprint report.

6. The lightweight glass bottle molding process optimization system according to claim 5, characterized in that, The virtual simulation unit includes: Equipment modeling sub-unit: Based on the entered 3D equipment information, construct a high-fidelity virtual production line; Process modeling sub-unit: Based on the entered equipment working information, simulate the entire glass bottle processing process in a virtual environment; Physics-field coupling analysis sub-unit: Integrating computational fluid dynamics and finite element analysis, it performs high-precision simulation of the flow, temperature distribution, forming process and stress evolution of molten glass during the simulation process, and predicts wall thickness uniformity, structural strength and potential defects; Material property library sub-unit: Contains glass material models with different ratios, accurately reflecting their thermophysical and rheological properties.

7. The lightweight glass bottle molding process optimization system according to claim 6, characterized in that, The software simulation optimization unit is used to pre-optimize the equipment layout, dimensions, and molding process parameters, including: Information input subunit: Receives equipment 3D model information, equipment operating parameter information, multi-material physical performance parameters, and environmental parameters. The equipment 3D model information covers, but is not limited to, the structural dimensions, assembly relationships, and material parameters of the tank furnace, feeder, forming machine, annealing furnace, and testing equipment. The equipment operating parameter information includes the glass melting temperature, blowing pressure range, mold opening and closing speed, and annealing cooling rate. The multi-material physical performance parameters cover the melting point, viscosity-temperature curve, and strength index of ordinary glass, glass-ceramic composite materials, nano-SiO2 reinforced glass, and lead-free environmentally friendly glass. Simulation Analysis Subunit: Based on the entered information, a virtual production environment is constructed, a 1:1 digital twin of the physical production line, simulating the entire processing of glass bottles from raw material melting, forming, annealing to inspection. ANSYS finite element analysis software is used to perform stress analysis and process bottleneck identification on the initial blank and finished bottle. An LSTM long short-term memory network model and reinforcement learning algorithm are embedded to train a quality prediction model, enabling advanced prediction and evaluation of key quality indicators. Simultaneously, the collaborative working status of various electrical components is analyzed, process connection bottlenecks are identified, the interval time of each process step is optimized, and the time consumption of process connections is minimized. Coupled analysis of process parameters is performed, simulating the interaction between glass melt temperature, blowing pressure, and cooling time through iterative simulation. Output sub-unit: Through multiple rounds of simulation iteration, output the optimal processing equipment size, layout scheme, and molding process parameters. The process parameters include raw material ratio, melting temperature, molding temperature, initial blowing pressure curve, final blowing pressure, mold opening and closing speed, annealing temperature curve, cooling rate, and time nodes of each process, ensuring that the process interval is minimized and the coupling interference of process parameters is minimized; output the optimal solution; output the carbon footprint pre-control target; Transfer learning sub-unit: When switching bottle type or material, the system can reuse the training model of similar products, and only needs to fine-tune the model with a small amount of new data, shortening the debugging cycle to 2-3 days; establish a standardized mold library and parameter library linkage mechanism.

8. The glass bottle lightweight molding process optimization system according to claim 7, characterized in that, The dynamic feedback optimization unit is communicatively connected to the software simulation optimization unit, the hardware execution module, and the global data perception module. It uses a PLC and a central controller to construct the control logic. Its reinforcement learning algorithm uses the highest pass rate, lowest energy consumption, lowest cost, and lowest carbon footprint as multi-objective reward functions to automatically iterate and optimize the combination of process parameters. It integrates AI algorithms to achieve adaptive parameter adjustment, defect tracing, and predictive equipment maintenance. The dynamic feedback optimization unit includes: Data distribution subunit: Distributes the optimal set of process parameters obtained by the optimization engine to the central controller of the hardware execution module to guide actual production, and receives equipment status and product quality information uploaded by the equipment working status sensor network unit in real time during the production process; Data comparison and analysis subunit: compares the real-time collected equipment status data and material quality data with the preset parameters output by the software simulation and optimization unit to identify deviations; Parameter adaptive adjustment subunit: When a deviation is detected, parameters are automatically fine-tuned based on reinforcement learning algorithm: If the wall thickness uniformity is not up to standard, the pressure of the corresponding side blowing head or the mold closing gap is adjusted; if the glass melt viscosity change leads to poor initial blowing effect, the coefficients k and b in the initial blowing pressure formula P1=kln(μ)+b are updated, where k and b are adjustable coefficients determined by automatic iterative optimization based on reinforcement learning algorithm according to historical production data and real-time detected glass melt viscosity changes, and μ is the glass melt viscosity; if the residual stress after annealing exceeds the standard, the cooling rate of the annealing furnace or equipment parameters are adjusted; if the carbon footprint exceeds the standard, the proportion of crushed glass is increased or the temperature of the tank furnace is reduced; if the energy consumption is too high, the annealing cooling rate or the tank furnace insulation power is optimized. Predictive maintenance subunit: Based on LSTM algorithm, analyze equipment vibration, temperature and wear data to predict the remaining life of the equipment and generate maintenance suggestions; establish equipment maintenance knowledge base to record fault causes and solutions, forming a prediction-maintenance-feedback closed loop; Model and parameter library update sub-unit: Feeds back the adjusted optimal parameters to the software simulation optimization unit to update the parameter library of the simulation model; Early warning and traceability sub-unit: Set multi-level parameter threshold early warning, quality threshold, carbon footprint threshold. When equipment parameters exceed the safe range or the material non-conforming rate exceeds 5% for 3 consecutive times, trigger an audible and visual alarm and suspend production. Operation will resume after the parameters are adjusted to meet the requirements. In case of sudden failure, the emergency adjustment logic will be automatically executed to ensure that the non-conforming rate does not exceed 1%. The data server records equipment operation logs, process parameter adjustment records and material quality information, with a traceability period of not less than 1 year.

9. The glass bottle lightweight molding process optimization system according to claim 8, characterized in that, The supply chain collaboration interface unit is not limited to connecting to the enterprise ERP system to obtain order information. For new bottle type orders, the platform conducts virtual trial production and production change simulation in advance. The full lifecycle quality traceability unit assigns a unique QR code to each product batch, linking it to the entire chain of data from raw material formula, process parameters, test results to outgoing logistics, forming an immutable quality blockchain. The carbon footprint and cost accounting unit automatically calculates the carbon emissions and production costs per unit of product based on real-time energy consumption and material data.

10. The glass bottle lightweight molding process optimization system according to claim 9, characterized in that, The AR operation and maintenance unit is equipped with AR glasses that can be worn by operators, which can view the internal status of the equipment, parameter deviations and maintenance instructions in real time; The remote monitoring unit allows managers to view production line operation data in real time via PC or mobile device, and managers can remotely issue adjustment instructions. When production efficiency or pass rate declines, the operator can trigger one-click optimization. The system will automatically call the digital twin and AI algorithm, output the optimal parameters and perform adjustments. The hierarchical access control unit sets three levels of permissions: operators, administrators, and R&D personnel. Operators can only view operation-related parameters, administrators can view all data, and R&D personnel can only access the simulation model to prevent the leakage of core technologies. The data security protection unit performs regular data backups and security audits, in compliance with relevant data security regulations.

Images