A method, system, and medium for production parameter optimization of lightweight glass bottles

Abstract

The present application relates to the technical field of glass product manufacturing, and particularly relates to a production parameter optimization method and system for lightweight glass bottles and a medium. The method comprises the following steps: obtaining a glass bottle structure drawing, and identifying a bottle body structure and a bottle bottom structure; constructing a lightweight glass bottle model according to the bottle body structure and the bottle bottom structure; performing impact load simulation according to the lightweight glass bottle model to obtain impact load data; identifying structural weak points based on the impact load data; performing fatigue crack evolution according to the structural weak points to generate glass crack data; identifying crack shapes based on the glass crack data to obtain jagged glass crack data and radial glass crack data; and performing grain boundary detection based on the jagged glass crack data to obtain grain boundary data. The present application improves the strength retention rate and parameter adjustment accuracy rate in the lightweight production process of glass bottles based on glass product manufacturing technology.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of glass product manufacturing, in particular to a production parameter optimization method and system for lightweight glass bottles and a medium. BACKGROUND

[0002] Traditional lightweight glass bottle production fails to combine glass bottle structure characteristics for systematic lightweight modeling analysis, often relying on experience to adjust the bottle body or bottom design, lacking standardized processing and structure reconstruction process, resulting in limited lightweight effect; in the impact performance evaluation process, only static strength simulation is usually based on, ignoring local stress concentration, crack propagation path and fatigue life evolution caused by impact load, and the structure weak point and crack type cannot be accurately identified; in the crack data identification aspect, the fine identification of crack geometric features (such as jagged undulation, radial path, etc.) is lacking, and different forms of crack structure cannot be effectively distinguished, resulting in subsequent optimization difficult to target; in the grain boundary and bubble impurity analysis aspect, there is a lack of comprehensive identification means for microcrystal structure and glass optical properties, and impurity detection and parameter back calculation cannot be carried out through grain boundary identification and light spot blur characteristics; in the melting and annealing process parameter optimization process, a multi-dimensional data feedback mechanism based on crack morphology, density intersection and impact damage evaluation is not established, resulting in parameter optimization lacking in pertinence, low production process adjustment efficiency and uncontrollable effect, and it is difficult to meet the high-performance production demand of modern glass bottles with both strength and lightweight. SUMMARY

[0003] Therefore, it is necessary to provide a production parameter optimization method, system and medium for lightweight glass bottles to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a production parameter optimization method for lightweight glass bottles comprises the following steps:

[0005] Step S1: Obtain a glass bottle structure drawing and identify the bottle body structure and the bottle bottom structure; construct a lightweight glass bottle model according to the bottle body structure and the bottle bottom structure;

[0006] Step S2: Perform impact load simulation according to the lightweight glass bottle model to obtain impact load data; identify the structure weak point based on the impact load data; perform fatigue crack evolution according to the structure weak point to generate glass crack data; identify the crack shape based on the glass crack data to obtain jagged glass crack data and radial glass crack data;

[0007] Step S3: Perform grain boundary detection based on the jagged glass crack data to obtain grain boundary data; perform bubble impurity identification according to the grain boundary data to obtain bubble impurity data; optimize the glass bottle melting production parameters according to the bubble impurity data;

[0008] Step S4: determining annealing parameters according to the radial glass crack data; controlling the glass bottle production program according to the glass bottle melting production parameters and the annealing parameters.

[0009] The production of conventional lightweight glass bottles fails to combine the structural characteristics of glass bottles for systematic lightweight modeling analysis, often relying on experience to adjust the bottle body or bottom design, lacking standardized processing and structural reconstruction process, resulting in limited lightweight effect; in the impact performance evaluation process, only static strength simulation is usually based on, ignoring the local stress concentration, crack propagation path and fatigue life evolution caused by impact load, and the structural weak points and crack types cannot be accurately identified; in the crack data identification aspect, the fine identification of crack geometric features (such as dentiform undulation, radial path, etc.) is lacking, and different forms of crack structure cannot be effectively distinguished, resulting in subsequent optimization difficult to target; in the grain boundary and bubble impurity analysis aspect, there is a lack of comprehensive identification means for microcrystal structure and glass optical properties, and impurity detection and parameter backstepping cannot be carried out through grain boundary identification and light spot blur characteristics; in the melting and annealing process parameter optimization process, a multi-dimensional data feedback mechanism based on crack morphology, density intersection and impact damage evaluation is not established, resulting in parameter optimization lacking in pertinence, low production process adjustment efficiency and uncontrollable effect, which is difficult to meet the modern high-performance production demand for glass bottles with both strength and lightweight.

[0010] Preferably, step S1 is specifically:

[0011] Step S11: obtaining a glass bottle structure drawing and performing standardized processing to obtain a standardized glass bottle structure drawing;

[0012] Step S12: identifying a bottle body edge curve according to the standardized glass bottle structure drawing, and constructing a bottle body structure according to the bottle body edge curve;

[0013] Step S13: identifying a bottle bottom structure according to the standardized glass bottle structure drawing;

[0014] Step S14: identifying a high-redundancy material area based on the bottle body structure, and performing wall thickness distribution optimization on the high-redundancy material area to generate bottle body wall thickness optimization data;

[0015] Step S15: identifying a main stress area based on the bottle bottom structure, and performing support rib structure reconstruction on the main stress area to generate a bottle bottom support rib optimization structure;

[0016] Step S16: constructing a lightweight glass bottle model according to the bottle body wall thickness optimization data and the bottle bottom support rib optimization structure.

[0017] This invention standardizes glass bottle structural drawings, ensuring that the original design information conforms to unified specifications in terms of format, dimensional units, and structural annotations. This facilitates high-precision operation of subsequent automated modeling and recognition algorithms, guaranteeing modeling accuracy from the outset. Identification and structural reconstruction of the bottle's edge curves allow for the complete restoration of the glass bottle's outer contour features, providing an accurate geometric basis for wall thickness distribution analysis. Identification of the bottle bottom structure enables accurate extraction of the stress-concentrated bottom region, facilitating the identification of key load-bearing features and structural optimization. Identification of high-redundancy material areas, combined with wall thickness distribution optimization, removes excess material from non-critical areas without affecting overall mechanical properties, reducing bottle mass and achieving raw material conservation and lightweighting goals. Identification of main load-bearing areas, combined with support rib structure reconstruction, further enhances the stability of the bottle bottom under impact and stacking loads, improving local strength and crack resistance. The structural optimization data of the bottle body and bottom together constitute a lightweight glass bottle model, enabling the overall bottle structure to achieve minimal material configuration while maintaining necessary mechanical properties, providing accurate structural input for subsequent strength simulation, impact testing, and manufacturing processes. The entire process, from structural drawings to lightweight modeling, is streamlined, avoiding local over-enhancement or weakness caused by experience-based design. This enhances the controllability and data-driven capabilities of the design, providing a structural foundation and model support for subsequent process parameter optimization and high-performance glass bottle production.

[0018] Preferably, step S2 specifically includes:

[0019] Step S21: Simulate the impact load based on the lightweight glass bottle model to obtain the impact load data;

[0020] Step S22: Based on the impact load data, identify the stress concentration areas of the glass bottle; calculate the thickness gradient based on the stress concentration areas of the glass bottle, and determine the structural weak points based on the thickness gradient;

[0021] Step S23: Simulate crack propagation based on the weak points of the structure to obtain crack data;

[0022] Step S24: Predict fatigue life based on crack data to obtain fatigue life data;

[0023] Step S25: Identify low-life cracks based on fatigue life data to obtain glass crack data;

[0024] Step S26: Identify the crack shape based on the glass crack data to obtain serrated glass crack data and radial glass crack data.

[0025] This invention, through impact load simulation of a lightweight glass bottle model, accurately reflects the instantaneous stress response of the bottle under actual drop and collision scenarios, quantifying the temporal and spatial distribution characteristics of the impact force. Furthermore, by combining stress concentration area information from stress data statistics, it helps identify the location and distribution trend of stress extrema, providing a scientific basis for subsequent structural risk assessment. Using thickness gradients to calculate weak points effectively identifies localized mechanically disadvantageous areas caused by uneven wall thickness, achieving precise location of structural weaknesses. Subsequently, by simulating the crack propagation process in these weak areas and combining the crack driving mechanism of brittle glass materials under different stress states, the complete crack path from initiation to fracture can be reconstructed, obtaining physically based crack characteristic data. Based on this, fatigue life prediction can be carried out by combining crack length, propagation rate, and stress intensity factor changes to quantify the sustainable life of the crack under periodic impact, providing a quantitative basis for predicting the timing of fatigue failure. Further identification of low-life cracks based on fatigue life data helps to screen out crack features on the bottle surface or structure that are more likely to fail in the short term, providing data support for strengthening key areas and design modifications. In the crack shape identification stage, image processing and geometric feature matching technology can accurately distinguish between sawtooth cracks with periodic sharp corners and radial cracks with concentrated radial features in complex crack morphologies, realizing the classification and labeling of crack types. This effectively supports subsequent targeted structural optimization or process adjustment operations for different crack morphologies, building a precise feedback path from micro-crack evolution to macro-structural optimization, and improving the targeting and closed-loop capability of the entire production optimization system.

[0026] Preferably, step S23 specifically includes:

[0027] Step S231: Import the structural weak points into the crack propagation simulation software;

[0028] Step S232: In the crack propagation simulation software, set the crack length range to 0.05mm-1mm, the crack angle range to 0°-90°, and the crack tip radius range to 0.005mm-0.1mm;

[0029] Step S233: In the crack propagation simulation software, set the cyclic load range to 0.1MPa-2MPa, the impact load range to 10N-200N, and the simulation step range to 100-10000 steps;

[0030] Step S234: Set the fracture toughness range of the glass material to 0.5 MPa·m in the crack propagation simulation software. 0.5 -1.5MPa·m 0.5 The critical stress intensity factor ranges from 0.7 MPa·m. 0.5 -2.0MPa·m0.5 ;

[0031] Step S235: Run the crack simulation program and extract crack evolution characteristic data based on the crack propagation path and propagation rate to obtain crack data.

[0032] This invention imports structural weak points into crack propagation simulation software, allowing the simulation analysis to focus on the most stress-sensitive areas and ensuring a close correspondence between the simulation results and the actual failure mechanisms. By setting ranges for parameters such as crack length, angle, and tip radius, it comprehensively covers various combinations of initial microcrack states, enhancing the model's adaptability and reliability at the microscale. By setting the ranges for cyclic and impact loads and coordinating with reasonable simulation step control, the evolution path of crack propagation behavior under alternating quasi-static and dynamic loads is fully revealed, thereby capturing the dynamic evolution trend of cracks during high-frequency impact and fatigue processes. Finally, it sets the fracture toughness and critical stress intensity of the glass material. By combining the factor range with the intrinsic properties of materials to construct a critical criterion for cracking, the simulation process can dynamically determine whether the crack has reached the conditions for propagation or instability, thus improving the physical accuracy of the crack-driven model. In the process of running the crack simulation program and extracting crack evolution characteristic data, not only can the complete trajectory information of crack initiation, propagation and termination be obtained, but the response curve of the propagation rate with the changes of load, angle and material parameters can also be output. This provides a strong data foundation for subsequent fatigue life prediction, crack morphology classification and high-risk structural adjustment, thereby establishing a precise micro-macro failure mechanism chain and improving the strength control and production process feedback capability of glass bottles under the premise of lightweighting.

[0033] Preferably, step S26 specifically includes:

[0034] Step S261: Collect crack coordinate points based on glass crack data, and reconstruct the curve based on the crack coordinate points to obtain crack boundary data;

[0035] Step S262: Identify the inner edge contour of the crack based on the crack boundary data; calculate the number of contour undulations based on the inner edge contour of the crack; identify the undulating tooth segments of the crack edge based on the number of contour undulations.

[0036] Step S263: Calculate the fluctuation amplitude based on the fluctuating tooth segment at the crack edge; collect data on the sawtooth glass crack based on the fluctuation amplitude to obtain sawtooth glass crack data;

[0037] Step S264: Identify crack branch intersections based on glass crack data; identify crack propagation paths based on crack branch intersections; calculate the crack path initiation angle based on the crack propagation path;

[0038] Step S265: Calculate the uniformity of the orientation angle distribution based on the crack path initiation angle; identify radial glass cracks based on the uniformity of the orientation angle distribution to obtain radial glass crack data.

[0039] This invention significantly improves the accuracy and resolution of crack morphology identification through refined identification and type classification analysis of the geometric features of glass cracks, thus providing precise data support for glass bottle structure optimization. Collecting crack coordinate points and reconstructing curves restores the true boundary morphology of the crack, forming a quantifiable and analyzable boundary data foundation, which is helpful for subsequent geometric feature extraction and pattern recognition. By identifying the inner edge contour and calculating its undulation count, the fluctuation characteristics of the crack edge can be quantitatively described, thereby accurately identifying crack structures with obvious tooth-like features, providing a judgment standard for extracting sawtooth cracks. The calculation of fluctuation amplitude further strengthens the geometric identification logic of crack types, giving sawtooth crack data clear thresholds and identifiable boundary criteria. Identifying crack branch intersections and their extension paths allows for the reconstruction of multi-branch crack propagation. This method captures the spatial distribution of direction changes and path intersections during crack propagation. Based on the initial path angle, it calculates the uniformity of the directional angle distribution, revealing the radial symmetry or concentrated characteristics of crack propagation and enabling accurate identification of radial cracks. Ultimately, it generates refined classification data for two typical crack types, providing structural basis and geometric model support for subsequent directional reinforcement design, molding die optimization, and annealing process adjustments based on crack type. It also lays the foundation for establishing a structure-performance-morphology feedback closed-loop mechanism for glass bottle molding processes, improving the production system's ability to identify and respond to different crack types.

[0040] Preferably, the grain interface detection in step S3 specifically involves:

[0041] Identifying sharp turning areas based on serrated glass crack data;

[0042] Identifying abrupt edge fluctuation regions based on serrated glass crack data;

[0043] Grain region intersection operations are performed based on sharp transition regions and abrupt edge fluctuation regions to identify grain regions;

[0044] Electron backscattering diffraction data were obtained by performing electron backscattering diffraction based on the grain region.

[0045] Crystal orientation patterns were constructed based on electron backscatter diffraction data;

[0046] Calculate the orientation difference angle between adjacent points based on the crystal orientation map;

[0047] Grain boundaries are identified by the orientation difference angle between adjacent points, and grain interface data is obtained.

[0048] This invention identifies sharp turning regions and abrupt edge fluctuations based on serrated crack data, which helps extract potential microstructural change clues from crack morphology and enhances the spatial positioning accuracy of the relationship between cracks and internal grains. Through intersection operations of grain regions, it achieves precise mapping between complex crack features and actual grain structures, providing a high-confidence reference region for microstructure analysis. Electron backscatter diffraction testing based on these regions effectively collects the arrangement of microcrystals within the glass, providing a data foundation for further revealing stress-induced grain rearrangement and crystal orientation changes. After constructing a crystal orientation map, it can realize the crystal orientation at various local points of the material. Visualizing the orientation distribution provides reliable support for assessing the anisotropy of materials and the formation mechanism of local micro-defects. Further calculation of the orientation difference angle between adjacent points can quantify the intensity of crystal orientation changes, which helps to accurately identify the range of grain boundaries and determine important parameters such as grain size and distribution density. The resulting grain interface data not only reveals the internal structural characteristics of glass materials, but also provides microscopic-scale targeted parameter support for optimizing processes such as glass composition blending, melting temperature control, annealing homogenization, and strengthening treatment. This fundamentally makes up for the uncertainty brought about by traditional reliance on macroscopic experience judgment, and achieves multi-objective synergistic improvement in the strength, reliability, and process adaptability of lightweight glass bottles.

[0049] Preferably, the bubble impurity identification in step S3 specifically involves:

[0050] Calculate the interface curvature based on the grain interface data;

[0051] Identifying abrupt changes in interface curvature based on interface curvature;

[0052] Calculate the glass refractive index based on the abrupt change point of the interface curvature;

[0053] The refractive index discontinuity region is identified based on the glass refractive index, and spot blur detection is performed on the refractive index discontinuity region to obtain spot blur data; the spot blur boundary is identified based on the spot blur data.

[0054] Cavity structure data is obtained by identifying the blurred boundaries of the light spot.

[0055] Bubble impurity detection is performed based on cavity structure data to obtain bubble impurity data.

[0056] This invention quantifies the degree of geometric deformation at grain boundaries through interface curvature calculation, revealing potential structural stress concentration areas. Identification of interface curvature abrupt change points further pinpoints interface distortion regions caused by abnormal grain growth or uneven crystallization, providing structural basis for subsequent optical behavior anomalies. Calculating the glass refractive index based on these abrupt change points helps establish a mapping relationship between grain structure and refractive index distribution, improving the local resolution capability of optical property assessment. Identification of refractive index discontinuities and detection of light spot blurring in these regions capture the actual impact of microscopic defects within the glass on light propagation behavior, thus constructing a response mechanism between refractive index anomalies and visible blurring regions. Through the blurring of light spot boundaries... The identification of optical anomalies can clearly outline the distribution contours of optical anomalies, further assisting in locating the spatial range and morphological characteristics of cavity structures, and providing boundary constraints for accurate defect detection. The identification of cavity structures establishes a structural basis for the existence and diffusion mechanism of bubble impurities, which helps to combine cavity morphology and bubble behavior for physical model fitting and process source tracing. The final bubble impurity data can effectively support the purity assessment, composition improvement and melting parameter adjustment of glass products, thereby making up for the problems of lagging optical anomaly identification and unsystematic impurity detection in traditional processes. It realizes a comprehensive modeling and analysis from grain structure and interface characteristics to optical performance and bubble defects, effectively improving the quality stability of glass bottles and the controllability of lightweight production.

[0057] Preferably, the optimization of glass bottle melting production parameters in step S3 specifically involves:

[0058] The density of bubble impurities is calculated based on bubble impurity data.

[0059] Calculate the diameter of the bubble impurities based on the bubble impurity data;

[0060] The melting temperature adjustment data is obtained by adjusting the melting temperature based on the density of bubble impurities;

[0061] By increasing the stirrer speed based on the diameter of bubble impurities, optimized stirrer speed data was obtained.

[0062] By integrating melting temperature adjustment data and stirrer speed optimization data, dynamic homogenization parameters of molten glass are obtained;

[0063] The raw material ratio is adjusted based on the dynamic homogenization parameters of the molten glass to obtain the production parameters for glass bottle melting.

[0064] This invention helps quantify the overall distribution level of impurities within the glass by statistically analyzing bubble impurity density, thereby assessing overall melting homogeneity and determining the systematic nature of impurity sources. Calculating bubble impurity diameter provides quantitative data on impurity size characteristics, aiding in identifying localized melting disturbances or uneven stirring. Adjusting the melting temperature based on bubble density helps optimize the fluidity of the molten glass and bubble precipitation conditions, promoting bubble rise and dissipation from a thermodynamic perspective and enhancing the uniformity of raw material melting. Increasing the stirrer speed based on bubble diameter improves the internal mixing effect of the molten glass in a kinetic dimension, causing large-sized impurities to disperse rapidly and weaken their impact on the final product. Impact: Integrating key process data such as melting temperature and stirrer speed into dynamic homogenization parameters enables dynamic control and real-time evaluation of the glass melt state, promoting the formation of a multi-factor coordinated control strategy that takes into account both heat supply and flow field disturbance. Ultimately, adjusting the raw material ratio based on this dynamic homogenization parameter not only allows for adjustments to the proportions of various components according to the current glass melt state, but also further improves impurity tolerance and optimizes melting uniformity. This effectively avoids problems such as lack of basis for raw material adjustment and large fluctuations in melting efficiency in traditional processes, achieving precise control of the glass bottle melting process and a high degree of consistency in lightweight quality, meeting the intelligent manufacturing needs of modern high-performance glass products.

[0065] Preferably, this specification also provides a production parameter optimization system for lightweight glass bottles, used to execute the production parameter optimization method for lightweight glass bottles as described above, the production parameter optimization system for lightweight glass bottles comprising:

[0066] The lightweight glass bottle model building module is used to obtain glass bottle structural drawings and identify the bottle body and bottom structures; and to build a lightweight glass bottle model based on the bottle body and bottom structures.

[0067] The crack shape recognition module is used to simulate impact loads based on a lightweight glass bottle model to obtain impact load data; identify structural weak points based on the impact load data; perform fatigue crack evolution based on the structural weak points to generate glass crack data; and identify crack shapes based on the glass crack data to obtain serrated glass crack data and radial glass crack data.

[0068] The melting production parameter optimization module is used to detect grain interfaces based on serrated glass crack data to obtain grain interface data; identify bubbles and impurities based on the grain interface data to obtain bubble and impurity data; and optimize glass bottle melting production parameters based on the bubble and impurity data.

[0069] The annealing production parameter optimization module is used to determine annealing parameters based on radial glass crack data; and to control the glass bottle production process based on glass bottle melting production parameters and annealing parameters.

[0070] The present invention relates to a production parameter optimization system for lightweight glass bottles. This system can implement any of the production parameter optimization methods for lightweight glass bottles of the present invention. It serves as a medium for coordinating the operation and signal transmission between various modules to complete the production parameter optimization method for lightweight glass bottles. The internal modules of the system cooperate with each other to improve the strength retention rate and parameter adjustment accuracy during the lightweight glass bottle production process.

[0071] Optionally, this specification also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements any one of the methods for optimizing production parameters for lightweight glass bottles. Attached Figure Description

[0072] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0073] Fig. 1 This is a schematic diagram of the steps in a method for optimizing production parameters for lightweight glass bottles according to the present invention.

[0074] Fig. 2 This is a detailed flowchart of step S1 in the present invention;

[0075] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0076] The technical method of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0077] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0078] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0079] To achieve the above objectives, please refer to Figs. 1-2 This invention provides a method for optimizing production parameters for lightweight glass bottles, the method comprising the following steps:

[0080] Step S1: Obtain the glass bottle structure drawings and identify the bottle body and bottom structure; construct a lightweight glass bottle model based on the bottle body and bottom structure;

[0081] In this embodiment, two-dimensional structural drawings of the glass bottle product are imported into a CAD software system to obtain the bottle's geometric contour and bottom thickness distribution. Using 3D modeling tools such as SolidWorks, boundary features of the bottle body and bottom structures in the drawings are extracted, including structural parameters such as the inner diameter of the bottle mouth (e.g., 26mm), the maximum outer diameter of the bottle body (e.g., 65mm), the bottle height (e.g., 220mm), and the bottom thickness distribution (e.g., gradually decreasing from 8mm at the center to 3mm at the edge). A 3D solid model is then constructed using extrusion and rotation methods. During the modeling process, in accordance with national lightweight glass bottle standards (e.g., GB / T 4544-2021), the maximum wall thickness is constrained to not exceed 80% of the original thickness, and the bottom structure adopts a central reinforcing ring and a uniform slope structure to ensure that the 3D model possesses lightweight design characteristics, facilitating subsequent mechanical simulation and crack tracking calculations.

[0082] Step S2: Simulate impact loads based on the lightweight glass bottle model to obtain impact load data; identify structural weak points based on the impact load data; perform fatigue crack evolution based on the structural weak points to generate glass crack data; identify crack shapes based on the glass crack data to obtain serrated glass crack data and radial glass crack data.

[0083] In this embodiment, based on the constructed 3D model, static and dynamic impact load simulation modules are introduced using finite element analysis software (such as ANSYS). Before applying the load, the model is divided into a high-density tetrahedral mesh with a minimum element side length of 0.2 mm to enhance the simulation accuracy of stress concentration areas. The load application methods include top-down free-fall impact tests (weight of 0.5 kg, free fall height of 1 m, corresponding to an initial impact velocity of 4.43 m / s), applied to the widest part of the bottle body and the center region of the bottom. During the simulation, the impact response time, the location of the maximum stress, the equivalent stress contour map, and the nodal strain rate are recorded. The stress extrema are identified using the Von Mises criterion. If the stress exceeds the glass ultimate strength value (e.g., 40 MPa), the location is marked as a structural weak point. The Paris-Erdogan fatigue crack propagation model is applied to the weak point region, with the crack propagation coefficient set to 1.2 × 10⁻¹¹ and the crack propagation exponent set to 3.1. The crack path is traced based on the stress intensity factor variation curve to derive the crack propagation length and direction. Image processing algorithms (such as Canny edge detection combined with Hough transform) are used to extract the geometric features of crack images. The presence of periodic tooth-like fluctuations (period spacing < 0.8 mm, depth > 0.2 mm) is used to classify them as sawtooth cracks. Cracks exhibiting a linear radial divergent structure with a center density > 6 cracks / 100 mm are further classified as sawtooth cracks. 2 The regions are classified as radial cracks, and their location information and direction vectors are recorded.

[0084] Step S3: Detect grain interface based on serrated glass crack data to obtain grain interface data; identify bubble impurities based on grain interface data to obtain bubble impurity data; optimize glass bottle melting production parameters based on bubble impurity data.

[0085] In this embodiment, for the identified sawtooth glass crack region, a laser confocal microscope is used to scan the crack cross-section at a resolution of 100 nm to obtain a grayscale variation map of the grain boundary. The scanned image is input into a boundary detection algorithm based on threshold segmentation (the threshold is set to the average grayscale value plus 1.5 times the standard deviation) to extract the grain interface map. Structural regions with a grain boundary length exceeding 20 μm and a closure degree of over 90% are identified as complete grain regions, and the average grain diameter and distribution density are statistically analyzed. Local phase abrupt change regions are detected in the grain interface map using a digital image phase recognition algorithm, and bubble impurity regions are identified by refractive index anomalies. The refractive index identification standard is that circular or elliptical dark spot regions with a size greater than 5 μm and a deviation of more than 15% from the average refractive index of the background are identified as bubble impurities, generating bubble location and density data (e.g., a high-density region has more than 5 impurities per square millimeter). Map the bubble impurity area and its corresponding coordinate position to the glass melting process history database, retrieve the temperature control trajectory corresponding to the position, and identify whether the temperature fluctuation in the temperature control area exceeds the set threshold (e.g., ±15℃). If it does, adjust the temperature control setting value of the main melting zone temperature control point in the melting stage (e.g., reduce from 1450℃ to 1435℃), and at the same time adjust the stirring frequency (e.g., increase from 20 times per minute to 30 times per minute) to reduce bubble aggregation.

[0086] Step S4: Determine the annealing parameters based on the radial glass crack data; control the glass bottle production process based on the glass bottle melting production parameters and the annealing parameters.

[0087] In this embodiment, the location, length, density, and propagation angle of the crack center are quantitatively analyzed. In the radial crack areas of the bottle body and bottom, the three-dimensional spatial coordinates of the cracks are obtained using high-resolution industrial CT scanning. The scanning resolution is set to 50 μm. After volume reconstruction, the origin coordinates, direction vector, and length information of each crack are exported. Cracks with a length exceeding 10 mm and a crack density exceeding 6 cracks / 100 mm are considered as... 2The area was divided into high-stress concentration zones, and annealing temperature control points were set in these zones. The annealing temperature control points were measured using multi-point thermocouples: 550℃ at the center of the bottle body, 545℃ at the shoulder area, and 530℃ at the center of the bottom, with the temperature difference controlled within ±2℃. The holding time was set to 90 minutes. The annealing slow cooling rate was controlled through air duct adjustment and radiant cooling, reducing the temperature by 50℃ per hour. A temperature gradient control was implemented between the bottle body and bottom, with the cooling rate at the outer perimeter of the bottom controlled at 45℃ per hour and the main body at 50℃ per hour. Uniform cooling was achieved by controlling the distance between the cooling plate and the bottle body to within 15cm. The airflow in the annealing furnace was monitored in real time using an anemometer, maintained between 3.5m / s and 4.0m / s to ensure uniform heat transfer. During the annealing process, the temperature curves of six key measuring points were recorded every 15 minutes to adjust the heating tube voltage, ensuring that local temperature fluctuations did not exceed ±2℃. The melting temperature, furnace temperature control point, blowing die closing pressure, blowing speed, and annealing parameters are integrated into an automatic control command sequence. During the melting stage, the main melting zone temperature is set at 1435℃, and the auxiliary melting zone temperature is set at 1420℃. Temperature values ​​are recorded every 5 seconds using platinum-rhodium thermocouples. Temperature control points with deviations exceeding ±3℃ are automatically adjusted to 105% or 95% of the rated power to correct the temperature difference. During the forming stage, the die closing pressure is set at 5.2MPa, the bottle mouth inner diameter is controlled within the range of 26mm ± 0.1mm, the maximum outer diameter of the bottle body is controlled within 65mm ± 0.2mm, the blowing gas flow rate is set at 12L / s, the blowing time is 3.5 seconds, and the die is kept closed for 1.2 seconds after blowing to ensure the stability of the bottle bottom structure. The annealing stage is performed according to the aforementioned annealing parameters. After annealing, the bottles are automatically conveyed into the slow cooling zone. The slow cooling air velocity is maintained at 3.8 m / s through an adjustable air valve, and the cooling time is controlled at 180 minutes until the bottle temperature drops to room temperature (25°C). The entire production process is executed by a PLC controller, with a program cycle set for 500 bottles per batch. Temperature, pressure, time, and flow parameters are recorded in real time in the database, enabling precise parameter control and traceability of the batch production process.

[0088] Preferably, step S1 specifically includes:

[0089] Step S11: Obtain the glass bottle structure drawing and perform standardization processing to obtain a standardized glass bottle structure drawing;

[0090] In this embodiment, the original glass bottle structural drawings are imported into the CAD standard processing system in DWG format. The drawings originate from production design drawings provided by a glass container manufacturing company. First, the 2D drawing preprocessing module is used to check the structural conformity of the drawings, identifying the boundary lines of areas such as the bottle body outline, neck, shoulder, and bottom. Through boundary topology analysis, errors such as duplicate lines, broken lines, and suspended points are eliminated, and element segments are merged into continuous paths to ensure the integrity of the geometric topology. All annotation units are standardized to millimeters, the default scale is set to 1:1, and layer attributes are standardized, with the bottle body outline set to the "Structural Outline Layer" and the bottom structure set to the "Bottom Structure Layer." If the drawings contain multiple view structures (front view, top view, sectional view, etc.), a projection alignment algorithm is used for graphic registration to ensure consistency of 3D information. After standardization, the output is a standardized glass bottle structural drawing containing only standard layers, unified units, and no topological errors (output format: STEP or IGES file).

[0091] Step S12: Identify the bottle body edge curves according to the standardized glass bottle structure drawings, and construct the bottle body structure based on the bottle body edge curves;

[0092] In this embodiment, the bottle edge curves are identified using a geometric contour extraction algorithm in the standardized glass bottle structural drawings. The edge detection path is set to scan from the bottle mouth towards the bottom, with a scanning interval of 0.2 mm. Spline fitting technology is used in edge extraction to reconstruct the continuous bottle contour curve from the extracted discrete edge points using cubic B-spline curves. The main features of the bottle edge curves include the rate of change of the bottle shoulder radius, the bottle taper, and the curvature of the transition section between the bottle mouth and bottom. All curves are converted into parametric curve expressions to support subsequent construction of a rotating solid based on the contour rotation axis. The complete bottle structure is constructed by rotating the edge curve 360° around the central axis (z-axis) using solid rotation modeling technology. The initial bottle cross-sectional thickness is defined as 3.5 mm, and a wall thickness attribute label is embedded in the modeling for subsequent wall thickness optimization calculations.

[0093] Step S13: Identify the bottom structure of the bottle according to the standardized glass bottle structure drawing;

[0094] In this embodiment, all geometric boundaries of the bottle bottom are extracted by identifying the "bottom structure layer" in the standardized drawings. A boundary grouping algorithm is used to divide the bottle bottom into three parts: the bottom center, the support ring, and the bottom outer edge. For the bottom center region, the symmetry of the center is checked to ensure it meets design requirements; if the deviation exceeds 1mm, the center point is refitted. When identifying the geometric features of the support ribs, a boundary contour closure detection method is used, requiring each rib structure to be a closed path and to have an evenly spaced angle with the center of the bottle bottom (e.g., 60° for six ribs). Bottle bottom thickness data is extracted using a thickness extraction probe technique, projecting the profile line vertically to the bottom surface to obtain thickness data at every 10° angle, with a thickness measurement accuracy of 0.1mm. The final output is a bottle bottom geometric structure file, including the bottom center diameter (e.g., 12mm), the support ring width (e.g., 5mm), and the number and distribution information of the rib structures.

[0095] Step S14: Identify high-redundancy material regions based on the bottle structure, optimize the wall thickness distribution of high-redundancy material regions, and generate bottle wall thickness optimization data.

[0096] In this embodiment, a finite element mesh is generated on the constructed bottle structure. A 0.3mm mesh size is used to mesh the entire bottle area, generating a structure with uniform mesh density. A volumetric thickness distribution analysis algorithm is used to calculate the wall thickness data for every 2mm section along the height of the bottle. Local extremum analysis is performed on the wall thickness data to identify areas with wall thicknesses greater than the average wall thickness (based on 3.5mm) + 1.0mm as high-redundancy material regions. These redundant regions are reconstructed as parametric surfaces. Under the premise that the minimum wall thickness is not less than 2.0mm and the local wall thickness gradient is not greater than 0.2mm / mm, the wall thickness is gradually optimized using geometric stretching and shrinkage operations. During the wall thickness adjustment process, Bezier surface control point transformation technology is used to adjust the curvature, ensuring a smooth transition during wall thickness shrinkage. After optimization, a new wall thickness distribution data file is generated, recording the wall thickness value corresponding to each height position of the bottle, and saved in CSV format for subsequent model calls.

[0097] Step S15: Identify the main stress area based on the bottle bottom structure, and reconstruct the support rib structure of the main stress area to generate an optimized bottle bottom support rib structure.

[0098] In this embodiment, a stress analysis is performed on the support rib area defined in the bottle bottom structure. Vertically upward internal pressure (e.g., 0.5 MPa) and vertically downward external force (e.g., 10 N impact force) are simulated to extract stress concentration areas. An equivalent stress distribution diagram is used to determine if there are uneven stress areas in the rib structure. Areas with stress greater than 35 MPa are identified as the main stress areas. The rib shape in this area is optimized using a vector optimization-based topology reconstruction method. The original support rib design is changed from rectangular ribs to arc-shaped ribs, with the arc radius set to half the rib width, and the bottom of the rib transitioning to the bottle bottom using a R0.5 rounded corner. After reconstruction, the number of ribs remains unchanged, and the optimized rib cross-sectional shape is a trapezoidal structure (top width 1.5 mm, bottom width 3.0 mm, height 2.5 mm) to improve the rib's support stability to the bottle bottom. The output rib structure optimization file is in IGES format and is used to merge into the final glass bottle model.

[0099] Step S16: Construct a lightweight glass bottle model based on the optimized data of the bottle wall thickness and the optimized structure of the bottle bottom support ribs.

[0100] In this embodiment, the optimized wall thickness data of the bottle body and the optimized structure of the bottle bottom support ribs are imported into the modeling module. Within the existing rotating bottle body structure, the wall thickness value corresponding to each cross-sectional height is replaced with the optimized value by adjusting the coordinates of the control points. A three-dimensional surface offset algorithm is used to achieve precise wall thickness construction. A Boolean operation is performed on the bottle bottom structure to merge the optimized rib structure into the central region of the bottle bottom. During the merging process, the model topology consistency is checked, overlapping lines in intersecting areas are eliminated, and the material properties are unified as soda-lime glass (density 2500 kg / m³). 3 (Poisson's ratio 0.23, Young's modulus 70 GPa). The final lightweight glass bottle model is generated, and the model output format is .STEP and .STL, which are used for numerical simulation verification and 3D printing manufacturing verification. The model structure includes optimized wall thickness attribute labels and support rib structure description labels.

[0101] Preferably, step S2 specifically includes:

[0102] Step S21: Simulate the impact load based on the lightweight glass bottle model to obtain the impact load data;

[0103] In this embodiment, after obtaining the lightweight glass bottle model, the three-dimensional model is imported into the finite element analysis tool ABAQUS 2023, and the impact load simulation is performed using the Explicit dynamics solver provided by the software. The loading condition is set to vertical impact, and an instantaneous impact load is applied to the bottom of the glass bottle using a rigid plate. The impact velocity is set to 3.5 m / s, corresponding to a drop height of approximately 0.65 meters during transportation. The glass bottle material properties are set to typical soda-lime glass with a density of 2.5 g / cm³. 3The elastic modulus was 70 GPa, Poisson's ratio was 0.22, and the fracture strength was set to 55 MPa. The mesh was generated using C3D10 elements, with element size controlled below 1.5 mm to ensure computational accuracy. By applying a load and performing a dynamic analysis with a time step of 0.001 s, stress-strain data at each node of the bottle during the impact process were extracted. The output format was ODB file, which was then converted to CSV format and named "impact_stress_data.csv".

[0104] Step S22: Based on the impact load data, identify the stress concentration areas of the glass bottle; calculate the thickness gradient based on the stress concentration areas of the glass bottle, and determine the structural weak points based on the thickness gradient;

[0105] In this embodiment, the impact load data obtained in step S21 is imported into MATLAB R2022b for stress analysis. First, the maximum principal stress is extracted from the mesh nodes of the bottle using a custom function `stress_cluster_region.m`, and the spatial clustering algorithm DBSCAN (distance threshold eps = 2.5 mm, minimum neighbor number minPts = 10) is used to identify areas of significant stress concentration. For each clustered stress concentration area, wall thickness information is sampled within a 0.5 cm radius around it, using the formula: gradient G = |T_max - T_min| / D, where T_max and T_min represent the maximum and minimum wall thickness values ​​within the radius, respectively, and D is the sampling diameter. Points with a thickness gradient greater than 0.12 mm / cm are defined as structural weak points, and their three-dimensional coordinates are recorded in the structural weak point database (named `weak_point_set.json`) as input parameters for subsequent crack simulation analysis.

[0106] Step S23: Simulate crack propagation based on the weak points of the structure to obtain crack data;

[0107] In this embodiment, the Workbench module of ANSYS 2022R2, a crack propagation simulation tool, is used to simulate crack propagation. The three-dimensional coordinates of the structural weak points in step S22 are read, and initial crack defects are introduced sequentially at these locations. The defect length is set to 0.3 mm, and the crack type is an open Mode-I crack. The Smeared Crack model is selected for crack propagation analysis, with the material fracture toughness set to 0.8 MPa·m^0.5 and the mesh locally refined to a 0.3 mm element size. The load application method is consistent with step S21, using an instantaneous impact condition, and the simulation duration is set to 0.005 s. The crack front displacement, propagation length, and principal stress direction at each time step during the simulation are extracted and exported as a crack data file "crack_propagation_results.csv". Parameters such as the crack initiation time, propagation path, crack surface angle, and final crack length corresponding to each weak point are recorded.

[0108] Step S24: Predict fatigue life based on crack data to obtain fatigue life data;

[0109] In this embodiment, fatigue life prediction is performed using the material fatigue analysis standard ASTM E739. The crack data output from step S23 is used as input parameters, and the crack propagation rate is calculated using the Paris law model. The model expression is: da / dN=C*(ΔK)^m, where C is 1.5×10^-10, m is 3.1, and ΔK is the range of the stress intensity factor at the crack tip. The crack propagation rate is calculated based on the stress intensity factor at each crack point in the simulation process. Then, combining the initial crack length of 0.3mm and the critical failure length of 5.0mm, the number of fatigue load cycles N_f required for crack propagation to failure is predicted using a numerical integration method. Finally, the fatigue life values ​​corresponding to all weak points are recorded in the fatigue life data table “fatigue_life_output.csv”. Each record in the table includes fields such as crack number, location coordinates, and fatigue life N_f (unit: cycles).

[0110] Step S25: Identify low-life cracks based on fatigue life data to obtain glass crack data;

[0111] In this embodiment, the fatigue life data from step S24 is imported into a Python environment, and pandas is used for data processing. A life threshold N_thresh is set to 8000 iterations, and crack data below this threshold are used as the screening criterion for low-life cracks. Using the conditional filtering method: df[df['fatigue_life']<8000], all low-life cracks are extracted, and their coordinates, initial length, and final length are calculated and output to "glass_low_life_cracks.csv". Furthermore, the positions of the corresponding cracks on the 3D glass bottle model are spatially labeled using the Open3D library, and a PLY format visualization file "low_life_cracks_model.ply" is generated, with the label color set to red RGB values ​​(255,0,0).

[0112] Step S26: Identify the crack shape based on the glass crack data to obtain serrated glass crack data and radial glass crack data.

[0113] In this embodiment, crack shape recognition is performed on the crack image data in step S25. Image morphology analysis is used to process the microcrack image, with the image resolution set to 1200×1200 pixels and the input format being TIFF. First, using...

[0114] The Canny edge detection algorithm in OpenCV (with thresholds of 80 and 150) extracts crack boundaries, and then applies Hough transform to identify linear features. Cracks with a linear feature density higher than 40% are classified as radial cracks. Subsequently, Fourier transform is performed on the crack contour to analyze the contour frequency features, and cracks with frequency peak distribution intervals less than 10° are identified as sawtooth cracks. After identification, radial cracks are output as "radial_cracks_info.json", and sawtooth cracks are output as "ated_cracks_info.json". Each file includes detailed parameter information such as crack number, type, boundary point set, average crack width, and direction distribution, and is displayed in the 3D model using different colors (blue for radial and green for sawtooth) for layered annotation.

[0115] Preferably, step S23 specifically includes:

[0116] Step S231: Import the structural weak points into the crack propagation simulation software;

[0117] In this embodiment, after identifying structural weak points, their three-dimensional coordinate data is exported in TXT format. Each record contains X, Y, and Z axis coordinate information and the corresponding thickness gradient value. This file is named weak_zone_coords.txt. This coordinate data is imported into the FractureToolkit module of the crack propagation simulation platform ANSYS Workbench 2022R2. Using the "Initial Flaw Definition" function of this module, defect points are inserted point by point on the finite element model of the lightweight glass bottle. The method for inserting defect points is to manually define the initial position of the defect and use the volume method to cut small-sized cavities into the solid structure. The local region corresponding to each structural weak point is extracted as a sub-model. Each sub-model retains a boundary range of 20mm×20mm×5mm. All models use equivalent refined meshes with SOLID186 element type and a mesh size controlled at 0.3mm to meet the accuracy requirements of crack propagation calculation. Finally, the local sub-models containing structural weak point defects are saved as the analysis file submodel_with_flaws.wbpj.

[0118] Step S232: In the crack propagation simulation software, set the crack length range to 0.05mm-1mm, the crack angle range to 0°-90°, and the crack tip radius range to 0.005mm-0.1mm;

[0119] In this embodiment, initial crack geometry parameters are set for the structural weak points within each sub-model in the ANSYS Fracture Toolkit. The crack length is set in the "Initial Flaw Geometry" interface to a range of 0.05mm to 1mm, with a crack length interval of 0.05mm, generating 20 initial defect models of different lengths. The crack angle is set with the X-axis as the reference axis, with a rotation angle range of 0° to 90°, generating an angle direction every 15°, resulting in 7 direction configurations. The crack tip radius is set to 0.005mm, 0.02mm, 0.05mm, 0.08mm, and 0.1mm, for a total of 5 levels. Each combination generates one simulation configuration, requiring a total of 700 simulation scenarios (20×7×5). The crack tip uses virtual crack propagation technology to simulate an open crack tip; all cracks are set to three-dimensional ellipses, and the crack depth is set to 80% of the crack length. Each set of crack geometry configurations is used as an independent case to create an analysis scenario and saved with a specific naming format.

[0120] The analysis input file is “Flaw_L{length}_A{angle}_R{radius}.inp”.

[0121] Step S233: In the crack propagation simulation software, set the cyclic load range to 0.1MPa-2MPa, the impact load range to 10N-200N, and the simulation step range to 100-10000 steps;

[0122] In this embodiment, based on the above simulation model, cyclic loads and impact loads are defined sequentially in the Workbench load definition module. Cyclic loads are applied to the boundary nodes at the mouth of the glass bottle, with the load direction perpendicular to the bottle's axis. The applied load range is 0.1 MPa to 2 MPa, the load type is a sine function, the frequency is set to 20 Hz, and the load amplitude starts at 0.1 MPa, increasing in increments of 0.1 MPa to generate 20 configurations. The impact load simulates the bottom of the bottle contacting the ground, using a concentrated force loading method. The load range is 10 N to 200 N, increasing in increments of 10 N, generating a total of 20 impact scenarios. Each cyclic + impact combination scenario defines a simulation step range from 100 to 10000 steps, a simulation time step of 0.0001 s, and a maximum simulation time of 1 second. The load definitions are imported into ANSYS Mechanical via the FORTRAN user subroutine interface, generating the simulation load configuration script "load_sequence_config.f90", and all load configurations are recorded in the database "load_profile_database.db".

[0123] Step S234: Set the fracture toughness range of the glass material to 0.5 MPa·m in the crack propagation simulation software. 0.5 -1.5MPa·m 0.5 The critical stress intensity factor ranges from 0.7 MPa·m. 0.5 -2.0MPa·m 0.5 ;

[0124] In this embodiment, in the crack propagation analysis settings, based on the typical performance range of glass materials, five sets of parameters are set for fracture toughness: 0.5 MPa·m^0.5, 0.8 MPa·m^0.5, 1.0 MPa·m^0.5, 1.2 MPa·m^0.5, and 1.5 MPa·m^0.5, to simulate different material processing possibilities. Four sets of parameters are set for the critical stress intensity factor K_IC: 0.7 MPa·m^0.5, 1.0 MPa·m^0.5, 1.5 MPa·m^0.5, and 2.0 MPa·m^0.5. All crack front point calculations use the VCCT crack propagation criterion based on J-integral. If the equivalent K value corresponding to the J-integral exceeds the set K_IC, it is considered a crack propagation trigger condition. This setting is completed in the "Material Property Table" and specified as a nonlinear material behavior mode in the simulation solver control file. All parameter configurations are generated and nested within the aforementioned crack geometry and load configurations, resulting in a total of 700 (geometry) × 20 (load) × 5 (toughness) × 4 (K_IC) = 280,000 combinations.

[0125] Step S235: Run the crack simulation program and extract crack evolution characteristic data based on the crack propagation path and propagation rate to obtain crack data.

[0126] In this embodiment, after configuring all the above parameters, the ANSYS BatchManager batch simulation platform is invoked to execute all 280,000 crack propagation simulation tasks using a 64-core parallel cluster server. After each simulation run, the crack propagation path is exported as a 3D trajectory file (in .vtk format), and the crack propagation rate is obtained by removing the step size at each step, stored as a CSV file. The Python program crack_feature_extractor.py is used to read all simulation result files and extract the following feature parameters: maximum crack length L_max, average crack propagation rate V_avg, crack propagation direction vectors Δx, Δy, and Δz, crack path deflection angle θ_deflect, and crack morphology change rate R_shape. Finally, all data is summarized and output into the database "crack_growth_features.sqlite", which contains fields including geometric configuration, load configuration, material parameter configuration, and various crack evolution characteristics, generating more than 2 million records. To facilitate analysis and visualization, Paraview was used to generate crack path animations on the .vtk dataset, and the output was a video sequence with a frame rate of 30fps for subsequent analysis and evaluation.

[0127] Preferably, step S26 specifically includes:

[0128] Step S261: Collect crack coordinate points based on glass crack data, and reconstruct the curve based on the crack coordinate points to obtain crack boundary data;

[0129] In this embodiment, during the analysis of crack evolution data in lightweight glass bottles, the coordinate point set of the crack in three-dimensional space is first extracted based on the crack data obtained in step S235. Specifically, laser confocal microscopy is used to acquire the surface morphology of the crack. A Z-axis focusing layer scan with a precision of 0.1 μm is used to ensure the continuity of the surface contour. The image resolution is set to 2048×2048 pixels, and the image sampling precision is set to 1 μm / pixel. The crack edge point contour is extracted using an edge detection method based on the Sobel operator, converting the two-dimensional image coordinates into actual coordinate points in a three-dimensional geometric coordinate system. Subsequently, a continuous curve reconstruction of the crack point set is performed using a Catmull-Rom spline interpolation algorithm to avoid abrupt angle changes between discrete points. The curve reconstruction accuracy is controlled to be less than 0.5%. The continuous surface contour of the crack boundary is generated from the reconstructed curve data, and the spatial geometric position information of all boundary points is exported as crack boundary data, with the unit uniformly set to millimeters (mm).

[0130] Step S262: Identify the inner edge contour of the crack based on the crack boundary data; calculate the number of contour undulations based on the inner edge contour of the crack; identify the undulating tooth segments of the crack edge based on the number of contour undulations.

[0131] In this embodiment, after obtaining the crack boundary data, curvature analysis is used to further identify the contour lines of the crack's internal edges. Specifically, a method combining Gaussian curvature (K) and average curvature (H) is used to determine the local morphology of the surface, with a threshold set to K > 0.05 mm. -2 And H>0.1mm -1 The region exhibiting significant undulations was identified. By calculating the first and second derivatives of the edge curve within each crack segment, all segments with alternating positive and negative curvature changes were extracted as profile undulation segments. A minimum angle difference of 10° and a minimum length difference of 0.1 mm were set between adjacent undulation segments to eliminate low-frequency or weakly disturbed profiles. The number of undulations was determined by numerical integration to obtain the total number of curvature extrema, thus counting the number of undulations for each profile segment. Based on the curvature reversal of adjacent boundaries of continuous undulation segments, local oscillation abrupt change segments, i.e., wave tooth segments, were identified at the crack edge, and their spatial length, quantity, density, and other parameters were extracted, with units uniformly set to mm.

[0132] Step S263: Calculate the fluctuation amplitude based on the fluctuating tooth segment at the crack edge; collect data on the sawtooth glass crack based on the fluctuation amplitude to obtain sawtooth glass crack data;

[0133] In this embodiment, geometric features are quantified for the extracted crack edge undulating segments. First, a vertical projection algorithm is used to calculate the maximum and minimum edge height difference within each undulating segment, defining this value as the undulating amplitude of that segment. The calculation accuracy of the undulating amplitude is required to be no less than 0.01 mm, and edge segments with undulating amplitudes greater than 0.08 mm are defined as serrated feature regions. Combining the number and location markings of the edge undulating segments, crack regions meeting this amplitude standard are collected as serrated glass cracks. The data structure includes the spatial coordinate range of each crack serration, the maximum amplitude value, the tooth density (unit: pieces / mm), and the average tilt angle of each segment. By establishing a structured data form for this type of crack data, serrated glass crack data is generated for subsequent correlation analysis with parameters such as processing technology, cooling curves, and mold forming temperature.

[0134] Step S264: Identify crack branch intersections based on glass crack data; identify crack propagation paths based on crack branch intersections; calculate the crack path initiation angle based on the crack propagation path;

[0135] In this embodiment, crack branches and intersections are identified in the crack data structure based on spatial curve geometric topology analysis. First, the spatial trajectory of each crack path is parameterized, and the path is reconstructed using a cubic Bezier curve function. Intersections are identified based on the shortest distance between each path being less than 0.05 mm and the path angle being between 5° and 60°. All intersections are labeled as point sets, and their spatial location, number of connecting paths, and intersection angle are retained as structured attributes. Subsequently, a multi-path tracing operation is performed based on the intersections. The tracing logic is based on the principle of directional consistency, extending from the intersection to the starting direction of each branch path. Each path is traced to its endpoint where the curvature gradient change is less than 0.01, and this endpoint is defined as the crack extension path. When calculating the starting angle of the crack path, the angle between the vector direction formed by the first 10 coordinate points of the crack from the intersection and the normal to the bottle is taken as the starting angle. The calculation accuracy requirement is 0.1°. All angle values ​​are stored in the crack structure dataset for path structure analysis.

[0136] Step S265: Calculate the uniformity of the orientation angle distribution based on the crack path initiation angle; identify radial glass cracks based on the uniformity of the orientation angle distribution to obtain radial glass crack data.

[0137] In this embodiment, based on existing crack propagation path and initial angle data, the uniformity of directional angle distribution is calculated. First, all crack path initial angle values ​​are mapped to a spatial angle range of [0°, 180°]. This range is divided into 18 directional intervals with a 10° increment, and the number of cracks in each interval is counted and normalized. The Shannon information entropy formula H = -∑p(i)log2p(i) is used to calculate the distribution entropy value, and the distribution uniformity U = H / H_max is calculated based on the maximum uniform entropy H_max = log218≈4.17. Crack sets with U greater than 0.85 are defined as having highly uniform directional distribution. Combining the crack trajectory morphology and origin centrality within these sets, sets of cracks with uniform direction and radiating outwards from the center are identified and marked as radial glass cracks. The data records parameters such as the center point coordinates, number of propagation directions, minimum and maximum angle range, directional entropy value, and average path length of each radial crack, ultimately forming a radial glass crack dataset for subsequent correlation analysis with the bottle cooling rate and thickness gradient.

[0138] Preferably, the grain interface detection in step S3 specifically includes:

[0139] Identifying sharp turning areas based on serrated glass crack data;

[0140] In this embodiment, sawtooth glass crack data is first selected as the input dataset, and a method based on vector angle analysis is used to identify sharp turning regions in the crack trajectory. Each sawtooth path is discretized, and equidistant crack coordinate points are extracted with a step size of 1 μm. The angle θ formed by three consecutive points is calculated using the formula: θ = arccos[(v1·v2) / (|v1||v2|)], where v1 and v2 are two adjacent vector segments. An traversal analysis of the angles of all points is performed, and the angle threshold for identifying sharp turning regions is set to θ < 90°. A set of coordinate points continuously satisfying this condition with a length greater than 10 μm is defined as a valid turning segment. To eliminate false turning segments, a curvature jump index ΔK is introduced, and the average rate of curvature change between consecutive turning points is calculated, with ΔK > 0.15 mm. -1 This is one of the judgment conditions. All regions that meet the above two conditions are marked as "sharp turning regions", and their spatial location, start and end point indices, and curvature change sequence are output as data structures, with units of mm and rad (radians).

[0141] Identifying abrupt edge fluctuation regions based on serrated glass crack data;

[0142] In this embodiment, the same serrated glass crack data is used as a basis, and the first derivative rate of change method is employed to identify edge fluctuation abrupt change regions. First, based on each crack boundary curve, a function expression f(x) is constructed. The first derivative f′(x) of the curve is calculated, followed by the second derivative f″(x) to obtain the acceleration of boundary shape change. The fluctuation abrupt change criterion is set as follows: if the first derivative of the curve changes by more than 0.4 within 1 μm and the value of f″(x) is continuously greater than 0.3, then an abrupt change edge is considered to exist at that location. To identify continuous regions, a sliding window analysis method is introduced, with a window length of 20 μm and a window step size of 1 μm. The proportion of points satisfying the abrupt change condition within each window segment is analyzed; if the proportion exceeds 50%, the entire window is marked as an "edge fluctuation abrupt change region". The output includes a region index, curve slope distribution map, number of abrupt change points, maximum abrupt change rate, and spatial location, used for subsequent grain region determination calculations.

[0143] Grain region intersection operations are performed based on sharp transition regions and abrupt edge fluctuation regions to identify grain regions;

[0144] In this embodiment, the data from the two regions mentioned above are imported into the spatial Boolean operation module, and the spatial geometry intersection algorithm is used to identify the grain regions. Specifically, firstly, polygonal envelope contours are constructed based on the transition region and the fluctuating abrupt change region, respectively, and a two-dimensional projected mesh model is constructed using Delaunay triangulation. The two region meshes are then intersected, with the following rule: if the overlap area of ​​the two regions is greater than 15 μm... 2 If the overlapping portion contains at least 30 consecutive coordinate points, it is considered a single grain region. For each grain region, output its geometric attributes, including the set of boundary coordinate points, center point coordinates, area value, perimeter, and boundary curvature distribution, in units of μm. 2 This forms a standard grain region data table.

[0145] Electron backscattering diffraction data were obtained by performing electron backscattering diffraction based on the grain region.

[0146] In this embodiment, a commercial scanning electron microscope (SEM) equipped with an EBSD detector is used to acquire high-resolution diffraction data of the grain region. The preparation process includes: ion polishing the crack region sample to ensure a surface roughness of less than 30 nm to guarantee clear EBSD imaging signals. The SEM accelerating voltage is set to 20 kV, the sample tilt angle to 70°, the scanning step size to 100 nm, and the EBSD screen sampling frequency to 100 Hz. A matrix scan is performed on each grain region to acquire the Kikuchi pattern at each scan point, with an image size of 640 × 480 pixels. The system automatically performs image analysis, extracting the crystal orientation angle (Euler angle) and Bravais point identifier (Miller index) for each point, generating electron backscatter diffraction data, and saving it in both .csv and .tiff formats, containing spatial coordinates, orientation matrix, crystal structure identifiers, etc., with units uniformly in ° and μm.

[0147] Crystal orientation patterns were constructed based on electron backscatter diffraction data;

[0148] In this embodiment, a crystal orientation map is constructed using a color-coded mapping method based on the orientation matrix in the electron backscatter diffraction data. The map construction uses an Inverse Pole Figure (IPF) mapping scheme, with RGB color mapping applied to the

[001] ,

[101] , and

[111] directions, corresponding to R, G, and B values ​​ranging from 0 to 255, respectively. After mapping the crystal principal axis direction of each measurement point to the RGB color space, a two-dimensional image is drawn with a resolution of 300 dpi and an image area equal to the EBSD acquisition area. Each pixel in the map corresponds to a measurement point location, containing color, orientation angle, and spatial coordinate information. In addition to the image, the map data is also output in the form of a data matrix, recording the RGB value, Euler angle set, crystal orientation index, etc., for subsequent angle difference calculations.

[0149] Calculate the orientation difference angle between adjacent points based on the crystal orientation map;

[0150] In this embodiment, each measurement point in the crystal orientation pattern is traversed, and its four adjacent points (top, bottom, left, and right) are selected to calculate the Euler angle difference. The orientation difference angle is calculated using the principal axis difference between the orientation tensors, specifically using the Rodrigues transformation formula Δθ=arccos[(Trace(G1G2)] . -1 [-1) / 2], where G1 and G2 are the orientation tensors of the two measurement points. A lower threshold for valid differences is set to Δθ≥5° to exclude minor intracrystalline fluctuations. All point pairs meeting the condition are marked with coordinates, and a difference angle heatmap is constructed. Grayscale is used to represent the angle difference in the image, with black representing 0° and white representing a maximum of 90°. The data is also output as a spatial mapping matrix for grain boundary determination.

[0151] Grain boundaries are identified by the orientation difference angle between adjacent points, and grain interface data is obtained.

[0152] In this embodiment, based on the heatmap of adjacent point difference angles, boundary points with continuous orientation difference angles Δθ ≥ 10° are extracted and defined as high-angle grain boundaries. A region growth method based on edge connectivity is used to connect all high-angle difference points to form a closed path, extracting the grain boundary lines. To avoid noise interference, a minimum boundary length threshold of 20 μm is set; boundary lines smaller than this value are discarded. The grain interface data structure includes fields such as boundary segment start and end coordinates, boundary length, maximum orientation difference, adjacent grain index, and average curvature of the grain boundary. All interface data is output in tabular form for subsequent analysis of glass bottle stress direction heterogeneity and backtracking of production molding areas.

[0153] Preferably, the bubble impurity identification in step S3 specifically involves:

[0154] Calculate the interface curvature based on the grain interface data;

[0155] In this embodiment, after obtaining the grain interface data, a curvature reconstruction method based on discrete point sets is used to calculate the curvature of the interface contour point by point. Specifically, each continuous curve segment in the grain interface data is selected, and a least-squares arc is constructed at each center point using the arc fitting technique in surface differential geometry. The curvature κ is calculated using the formula κ = 1 / R, where R is the radius of the fitted circle. To ensure calculation accuracy, the number of points used for each fitting segment is set to 7 (the first 3 points, the current point, and the last 3 points), and the curvature value of each point is smoothed using a weighted moving average to reduce the error caused by abnormal fluctuations. For abnormal boundary regions, a curvature change gradient constraint is used, with a gradient threshold Δκ = 0.05 mm. -1 This is done to identify discontinuous regions and eliminate their interference with the overall curvature field. The curvature calculation operation is performed in a two-dimensional interface image based on coordinate space, with an image resolution of 1 μm / pixel.

[0156] Identifying abrupt changes in interface curvature based on interface curvature;

[0157] In this embodiment, after arranging the obtained curvature value sequence according to spatial position, the curvature change rate is calculated using the first-order difference method, with the formula Δκi=κi+1-κi. Based on this, a mutation recognition threshold θ_c=0.15mm is set. -1All points satisfying |Δκi|≥θ_c are identified as curvature abrupt change points. During abrupt change identification, a continuous difference strategy is used to filter out isolated fluctuation points, retaining only abrupt change segments with three or more consecutive segments meeting the threshold condition to avoid false identification. The above abrupt change identification process is implemented using an edge curvature abrupt change detection script written in Python. The script calls the standard NumPy numerical processing library for sequence calculations, and the point coordinates are synchronously returned to the original spatial coordinates of the image for subsequent localization and annotation.

[0158] Calculate the glass refractive index based on the abrupt change point of the interface curvature;

[0159] In this embodiment, the locations of the interface curvature abrupt change points are projected onto the refraction experimental platform of the glass bottle sample along the glass thickness direction. The actual change in the refraction angle at each abrupt change point under illumination is measured using a laser interferometry device. The experimental setup is as follows: laser wavelength λ = 632.8 nm, incident angle α = 45°, air as the medium, and refractive index n0 = 1. According to Snell's law: n = n0·sin(α) / sin(β), where β is the refraction angle measured from the high frame rate camera system. This angle is sampled in 0.1° increments, with the measurement error controlled within ±0.05°. Each abrupt change point is measured at least three times, and the average value is taken to ensure stability. The obtained refractive index is mapped onto the coordinates of the curvature abrupt change point to generate a refractive index spatial distribution matrix.

[0160] The refractive index discontinuity region is identified based on the glass refractive index, and spot blur detection is performed on the refractive index discontinuity region to obtain spot blur data; the spot blur boundary is identified based on the spot blur data.

[0161] In this embodiment, a two-dimensional difference analysis is performed on the refractive index spatial distribution matrix. A refractive index abrupt change threshold η = 0.12 is set. Whether a region is discontinuous is determined based on the difference |n(i+1,j)-n(i,j)|≥η or |n(i,j+1)-n(i,j)|≥η. Within discontinuous regions, a local window size of 11×11 pixels is set. The gradient distribution of the spot image is analyzed using the structure tensor method to determine the edge blurriness of the spot. The spot image is acquired using a high frame rate industrial camera, with a grayscale value range of [0,255]. The gradient is calculated using the Sobel operator. The blurriness index is compared between the total gradient magnitude σ_g and the blur threshold γ_g = 25. If σ_g < γ_g, it is determined to be a blurred spot region. The spatial location of the blurred region and its corresponding gradient value are output as the blurred spot data. The region growing method is used to extract the boundary contours of the blurred spot data. The specific operation is as follows: In the blurred region image, the blurred center point with the smallest grayscale change is selected as the seed point. The growth condition is set to a blur gradient change of less than Δσ_g = 5. The blurred region is expanded pixel by pixel until all pixels meeting the condition are added to the region. After completion, the Canny edge detection algorithm is used to generate boundary lines at the edges of the blurred region. A dual threshold processing is used during edge detection, with a low threshold of T_low = 20 and a high threshold of T_high = 40 to reduce false edge interference. The final output boundary coordinate set constitutes the blurred boundary data of the light spot.

[0162] Cavity structure data is obtained by identifying the blurred boundaries of the light spot.

[0163] In this embodiment, the blurred boundary data of the light spot is mapped onto the three-dimensional coordinate system of the glass bottle structure, and the formation of a closed region is determined by the boundary closure degree. If there is no identifiable refractive structure in the closed region, and its internal average gray value is lower than the threshold μ = 50 (based on a grayscale standard of 0-255), it is defined as a cavity structure. The three-dimensional reconstruction operation uses stereo vision matching technology, employing dual industrial cameras (resolution 2048×2048 pixels, baseline length 100mm) to perform depth modeling of the blurred boundary. All identified cavity structures are output as point cloud data, including information parameters such as their three-dimensional geometric center, maximum inscribed circle diameter, and boundary roughness.

[0164] Bubble impurity detection is performed based on cavity structure data to obtain bubble impurity data.

[0165] In this embodiment, high-resolution X-ray images are used to perform further internal imaging analysis on the identified cavity structures. Imaging parameters are set as follows: voltage 60kV, current 0.5mA, exposure time 100ms, and image resolution 5μm / pixel. Texture analysis is performed on each cavity structure region to calculate its internal grayscale variance σ. 2The energy term in the gray-level co-occurrence matrix is ​​used. Typical gray-level variance for bubbles is between 150 and 300. Energy values ​​below 0.15 are defined as heterogeneous structural regions and identified as bubbles. All identified impurity bubble structures are labeled with their location, gray-level value, equivalent diameter, and shape factor (roundness), and the output is a standardized structured data table.

[0166] Preferably, the optimization of glass bottle melting production parameters in step S3 is as follows:

[0167] The density of bubble impurities is calculated based on bubble impurity data.

[0168] In this embodiment, using the spatial coordinates and volume data of the bubble impurities marked in the acquired cavity structure data, a three-dimensional spatial grid partitioning method is employed to divide the glass bottle sample space into multiple cubic units with a side length of 1 mm. An industrial-grade image recognition system (such as a FLIR high-definition infrared industrial camera combined with a structured light scanner) is used to obtain the spatial distribution and size of the bubbles at the center of the glass bottle wall or body through multi-angle scanning. The number of bubbles and the total volume contained in each cubic unit are counted. The density calculation formula is ρ = N / V, where N is the total number of bubbles per unit volume, and V is the unit volume (i.e., 1 mm²). 3 After multiple units have been calculated, the average bubble density of all units is taken as the bubble impurity density value inside the glass bottle, and this density value is recorded in the original structure database. The entire process uses a CNC motion platform and a precision positioning module to perform partitioned sampling and coordinate positioning, avoiding duplicate calculations and missed detections.

[0169] Calculate the diameter of the bubble impurities based on the bubble impurity data;

[0170] In this embodiment, based on each identified bubble structure region, a laser interferometer is used for high-precision contour mapping, and edge extraction is performed in conjunction with micro-focus microscopic images. The edge curve of each bubble is fitted using a circle fitting algorithm with least squares to obtain the diameter value d of the fitted circle; all d values ​​are stored in array form. The bubble diameter data is the maximum edge span of the bubble, obtained by multiplying the maximum distance between the boundary point and the center point of each bubble contour by 2. The edge extraction threshold used in the measurement is set to a grayscale difference ΔI ≥ 15, and the bubble boundary accuracy is controlled within ±0.01 mm. Data from no fewer than 1000 bubbles are collected, and the mean, maximum, and standard deviation of the calculated bubble diameters are saved as reference indicators for subsequent process adjustments.

[0171] The melting temperature adjustment data is obtained by adjusting the melting temperature based on the density of bubble impurities;

[0172] In this embodiment, the statistically obtained bubble density ρ is judged. If ρ is greater than the set process critical density threshold ρ0 (ρ0 is set to 10 bubbles / mm), the result is determined. 3 If ρ > ρ0, the melting temperature needs to be lowered; if ρ is less than ρ0, the current temperature is maintained. The specific temperature adjustment method is as follows: Assuming the current temperature T1 is 1450℃, when ρ > ρ0, the temperature is lowered by ΔT = k·(ρ - ρ0), where k is the temperature sensitivity coefficient, set to 2℃·mm. 3 / bubbles. Example: If ρ = 14 bubbles / mm 3 Therefore, ΔT = 2 × (14 - 10) = 8℃, and the adjusted melting temperature is T2 = 1442℃. The temperature adjustment process is executed by an automatic temperature control system, which connects the PLC system to the high-temperature zone electric heating module to achieve gradual adjustment, with an adjustment of no more than 2℃ per minute, and continuously samples the bubble density until it is below ρ0 for 30 consecutive minutes before stopping the adjustment. The adjustment result is stored as the melting temperature adjustment data T2 and linked to the current batch production parameter record.

[0173] By increasing the stirrer speed based on the diameter of bubble impurities, optimized stirrer speed data was obtained.

[0174] In this embodiment, the average bubble diameter D obtained in the aforementioned steps is... - Compare with the target bubble diameter threshold D0, if D - If D0 is greater than 0.6 mm, then the stirring speed of the glass melt needs to be increased. The adjustment coefficient α for increasing the speed is Δn = α·(D - -D0), where α is set to 50 rpm / mm. For example, if D - =0.75mm, then Δn = 50 × (0.75 - 0.6) = 7.5 rpm, rounded up to increase the speed to 8 rpm. The current speed of the mixer is set to n1 = 60 rpm, then the new speed n2 = 68 rpm. The mixer speed is adjusted in real time by the central drive controller, and a new frequency value is set in conjunction with the frequency converter (e.g., 0.95Hz corresponds to 60 rpm, then 1.08Hz corresponds to 68 rpm). The updated data is recorded as the optimized mixer speed data n2.

[0175] By integrating melting temperature adjustment data and stirrer speed optimization data, dynamic homogenization parameters of molten glass are obtained;

[0176] In this embodiment, the melting temperature adjustment data T2 and the stirrer speed optimization data n2 are jointly encoded to construct a dynamic homogenization parameter set {T2, n2} for molten glass. This parameter set must meet the stability criteria, namely, the molten glass temperature fluctuation range ΔT ≤ ±3℃ and the stirrer speed fluctuation range Δn ≤ ±2rpm within a continuous 3-hour monitoring period. If the fluctuation exceeds this range, the aforementioned adjustment steps must be repeated until the target is met. The dynamic homogenization parameter set is uploaded to the process parameter optimization module of the MES system, assigned a unique batch number and timestamp, to ensure that subsequent raw material ratio adjustments are linked with the actual process equipment. This parameter serves as the baseline input value for adjusting the ratio in the next stage.

[0177] The raw material ratio is adjusted based on the dynamic homogenization parameters of the molten glass to obtain the production parameters for glass bottle melting.

[0178] In this embodiment, four main raw materials are used: silica sand, soda ash, limestone, and boric acid. The silica sand ratio is adjusted according to T2 and n2 to improve glass strength, with the basic ratio set as 60% silica sand, 18% soda ash, 10% limestone, and 12% boric acid. When T2 < 1450℃, the silica sand ratio is increased by 1% to compensate for the viscosity change caused by the decrease in melting temperature; when n2 > 65 rpm, to reduce the formation of stirring shear bubbles, the soda ash ratio is decreased by 1%, and the boric acid ratio is increased by 1%. Therefore, the final ratio is adjusted to: 61% silica sand, 17% soda ash, 10% limestone, and 12% boric acid. The ratio adjustment is performed by an automatic raw material batching system. The weighing module sets the batch feed mass (e.g., 1000 kg of raw material corresponds to 610 kg of silica sand and 170 kg of soda ash), and after the ratio calculation is completed, it is written into the control system and exported as glass bottle melting production parameters for archiving, used to control the feeding operation of the glass furnace. The data acquisition frequency for the entire process is once per minute, and it is automatically linked to the quality inspection results.

[0179] Preferably, this specification also provides a production parameter optimization system for lightweight glass bottles, used to execute the production parameter optimization method for lightweight glass bottles as described above, the production parameter optimization system for lightweight glass bottles comprising:

[0180] The lightweight glass bottle model building module is used to obtain glass bottle structural drawings and identify the bottle body and bottom structures; and to build a lightweight glass bottle model based on the bottle body and bottom structures.

[0181] The crack shape recognition module is used to simulate impact loads based on a lightweight glass bottle model to obtain impact load data; identify structural weak points based on the impact load data; perform fatigue crack evolution based on the structural weak points to generate glass crack data; and identify crack shapes based on the glass crack data to obtain serrated glass crack data and radial glass crack data.

[0182] The melting production parameter optimization module is used to detect grain interfaces based on serrated glass crack data to obtain grain interface data; identify bubbles and impurities based on the grain interface data to obtain bubble and impurity data; and optimize glass bottle melting production parameters based on the bubble and impurity data.

[0183] The annealing production parameter optimization module is used to determine annealing parameters based on radial glass crack data; and to control the glass bottle production process based on glass bottle melting production parameters and annealing parameters.

[0184] Optionally, this specification also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements any one of the methods for optimizing production parameters for lightweight glass bottles.

[0185] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0186] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for optimizing production parameters for lightweight glass bottles, characterized in that, Includes the following steps: Step S1: Obtain the glass bottle structure drawings and identify the bottle body and bottom structure; construct a lightweight glass bottle model based on the bottle body and bottom structure; Step S2: Simulate impact loads based on the lightweight glass bottle model to obtain impact load data; identify structural weak points based on the impact load data; perform fatigue crack evolution based on the structural weak points to generate glass crack data; Based on glass crack data, crack shapes are identified to obtain serrated glass crack data and radial glass crack data; Step S3: Based on the serrated glass crack data, perform grain interface detection to obtain grain interface data; Bubble impurities are identified based on grain interface data to obtain bubble impurity data; glass bottle melting production parameters are then optimized based on the bubble impurity data. Step S4: Determine the annealing parameters based on the radial glass crack data; control the glass bottle production process based on the glass bottle melting production parameters and the annealing parameters.

2. The method for optimizing production parameters for lightweight glass bottles according to claim 1, characterized in that, Step S1 is as follows: Step S11: Obtain the glass bottle structure drawing and perform standardization processing to obtain a standardized glass bottle structure drawing; Step S12: Identify the bottle body edge curves according to the standardized glass bottle structure drawings, and construct the bottle body structure based on the bottle body edge curves; Step S13: Identify the bottom structure of the bottle according to the standardized glass bottle structure drawing; Step S14: Identify high-redundancy material regions based on the bottle structure, optimize the wall thickness distribution of high-redundancy material regions, and generate bottle wall thickness optimization data. Step S15: Identify the main stress area based on the bottle bottom structure, and reconstruct the support rib structure of the main stress area to generate an optimized bottle bottom support rib structure. Step S16: Construct a lightweight glass bottle model based on the optimized data of the bottle wall thickness and the optimized structure of the bottle bottom support ribs.

3. The method for optimizing production parameters for lightweight glass bottles according to claim 1, characterized in that, Step S2 is as follows: Step S21: Simulate the impact load based on the lightweight glass bottle model to obtain the impact load data; Step S22: Based on the impact load data, identify the stress concentration areas of the glass bottle; calculate the thickness gradient based on the stress concentration areas of the glass bottle, and determine the structural weak points based on the thickness gradient; Step S23: Simulate crack propagation based on the weak points of the structure to obtain crack data; Step S24: Predict fatigue life based on crack data to obtain fatigue life data; Step S25: Identify low-life cracks based on fatigue life data to obtain glass crack data; Step S26: Identify the crack shape based on the glass crack data to obtain serrated glass crack data and radial glass crack data.

4. The method for optimizing production parameters for lightweight glass bottles according to claim 3, characterized in that, Step S23 is as follows: Step S231: Import the structural weak points into the crack propagation simulation software; Step S232: In the crack propagation simulation software, set the crack length range to 0.05mm-1mm, the crack angle range to 0°-90°, and the crack tip radius range to 0.005mm-0.1mm; Step S233: In the crack propagation simulation software, set the cyclic load range to 0.1MPa-2MPa, the impact load range to 10N-200N, and the simulation step range to 100-10000 steps; Step S234: Set the fracture toughness range of the glass material to 0.5 MPa·m in the crack propagation simulation software. 0.5 -1.5MPa·m 0.5 The critical stress intensity factor ranges from 0.7 MPa·m. 0.5 -2.0MPa·m 0.5 ; Step S235: Run the crack simulation program and extract crack evolution characteristic data based on the crack propagation path and propagation rate to obtain crack data.

5. The method for optimizing production parameters for lightweight glass bottles according to claim 3, characterized in that, Step S26 is as follows: Step S261: Collect crack coordinate points based on glass crack data, and reconstruct the curve based on the crack coordinate points to obtain crack boundary data; Step S262: Identify the inner edge contour of the crack based on the crack boundary data; calculate the number of contour undulations based on the inner edge contour of the crack; identify the undulating tooth segments of the crack edge based on the number of contour undulations. Step S263: Calculate the fluctuation amplitude based on the fluctuating tooth segment at the crack edge; The sawtooth glass crack data is obtained by collecting data on the sawtooth glass crack based on the fluctuation amplitude. Step S264: Identify crack branch intersections based on glass crack data; Identify the crack propagation path based on the intersection of crack branches; Calculate the crack path initiation angle based on the crack propagation path; Step S265: Calculate the uniformity of the direction angle distribution based on the crack path initiation angle; Radial glass cracks are identified based on the uniformity of directional angle distribution, and radial glass crack data are obtained.

6. The method for optimizing production parameters for lightweight glass bottles according to claim 1, characterized in that, The specific steps of grain interface detection in step S3 are as follows: Identifying sharp turning areas based on serrated glass crack data; Identifying abrupt edge fluctuation regions based on serrated glass crack data; Grain region intersection operations are performed based on sharp transition regions and abrupt edge fluctuation regions to identify grain regions; Electron backscattering diffraction data were obtained by performing electron backscattering diffraction based on the grain region. Crystal orientation patterns were constructed based on electron backscatter diffraction data; Calculate the orientation difference angle between adjacent points based on the crystal orientation map; Grain boundaries are identified by the orientation difference angle between adjacent points, and grain interface data is obtained.

7. The method for optimizing production parameters for lightweight glass bottles according to claim 1, characterized in that, The specific steps for identifying bubble impurities in step S3 are as follows: Calculate the interface curvature based on the grain interface data; Identifying abrupt changes in interface curvature based on interface curvature; Calculate the glass refractive index based on the abrupt change point of the interface curvature; The refractive index discontinuity region is identified based on the glass refractive index, and spot blur detection is performed on the refractive index discontinuity region to obtain spot blur data; the spot blur boundary is identified based on the spot blur data. Cavity structure data is obtained by identifying the blurred boundaries of the light spot. Bubble impurity detection is performed based on cavity structure data to obtain bubble impurity data.

8. The method for optimizing production parameters for lightweight glass bottles according to claim 1, characterized in that, The optimization of glass bottle melting production parameters in step S3 is as follows: The density of bubble impurities is calculated based on bubble impurity data. Calculate the diameter of the bubble impurities based on the bubble impurity data; The melting temperature adjustment data is obtained by adjusting the melting temperature based on the density of bubble impurities; By increasing the stirrer speed based on the diameter of bubble impurities, optimized stirrer speed data was obtained. By integrating melting temperature adjustment data and stirrer speed optimization data, dynamic homogenization parameters of molten glass are obtained; The raw material ratio is adjusted based on the dynamic homogenization parameters of the molten glass to obtain the production parameters for glass bottle melting.

9. A production parameter optimization system for lightweight glass bottles, characterized in that, For performing the production parameter optimization method for lightweight glass bottles as described in claim 1, the production parameter optimization system for lightweight glass bottles includes: The lightweight glass bottle model building module is used to obtain glass bottle structural drawings and identify the bottle body and bottom structures; and to build a lightweight glass bottle model based on the bottle body and bottom structures. The crack shape recognition module is used to simulate impact loads based on a lightweight glass bottle model to obtain impact load data; identify structural weak points based on the impact load data; perform fatigue crack evolution based on the structural weak points to generate glass crack data; and identify crack shapes based on the glass crack data to obtain serrated glass crack data and radial glass crack data. The melting production parameter optimization module is used to detect grain interfaces based on serrated glass crack data to obtain grain interface data; identify bubbles and impurities based on the grain interface data to obtain bubble and impurity data; and optimize glass bottle melting production parameters based on the bubble and impurity data. The annealing production parameter optimization module is used to determine annealing parameters based on radial glass crack data; and to control the glass bottle production process based on glass bottle melting production parameters and annealing parameters.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the production parameter optimization method for lightweight glass bottles as described in any one of claims 1 to 8.

Images