A Digital Traceability Method and System for the Quality of Lightweight Glass Bottles

Abstract

The present invention relates to the field of digital quality management technology, and in particular to a digital traceability method and system for lightweight glass bottle quality. The method comprises the following steps: obtaining glass bottle design data and constructing a lightweight glass bottle model; detecting the structural strength of the bottle body based on the lightweight glass bottle model; detecting surface defects of the bottle body based on the structural strength of the bottle body and generating bottle surface defect data; tracing defects based on the surface defect data of the bottle body and obtaining process data of bottle production defects; detecting uneven wall thickness anomalies based on the structural strength of the bottle body and obtaining uneven wall thickness data; evaluating the core-mold cavity alignment offset based on the uneven wall thickness data; detecting locating pin fractures based on the core-mold cavity alignment offsets and obtaining locating pin fracture data. The present invention optimizes the glass bottle production process based on digital quality management technology, effectively improving the quality recognition rate, abnormal response speed and production process transparency.

Description

Technical Field

[0001] The present invention relates to the technical field of digital quality management, and in particular to a method and system for digitally tracing the quality of lightweight glass bottles. Background Art

[0002] Traditional digital traceability of glass bottle quality usually relies on data from a single production link (such as the molding process or cold end inspection) to record quality, and fails to cover the multi-source data involved in the entire life cycle of the glass bottle, resulting in missing or incomplete traceability information. In addition, the system generally uses static barcodes or batch numbers as identification methods, which cannot uniquely identify each glass bottle. As a result, when quality abnormalities occur, they can only be located to a certain batch rather than a specific product, affecting the efficiency of accurate traceability and responsibility tracing. At the same time, traditional systems do not have sufficient guarantees for the real-time collection of data and the continuity of transmission. Data is prone to lags, omissions or breakpoints, which seriously affect traceability. The reliability of the source system; in terms of data processing methods, most systems are still based on manual judgment or single-dimensional indicator analysis, lacking the ability to conduct multi-dimensional fusion analysis and trend prediction of key process parameters such as temperature, pressure, and thickness, which limits the early perception and proactive intervention of quality risks; in addition, in terms of system integration, traditional solutions are often limited to data sharing within local equipment or local area networks, lacking cross-section and cross-system collaborative linkage mechanisms, making it difficult to achieve end-to-end full-process quality traceability, affecting the transparency and intelligence level of the entire manufacturing chain, and ultimately hindering lightweight glass bottle manufacturers from efficiently identifying quality risks, quickly responding to them, and holding them accountable throughout the process. Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide a method and system for digitally tracing the quality of lightweight glass bottles to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a method for digitally tracing the quality of lightweight glass bottles includes the following steps:

[0005] Step S1: Obtaining glass bottle design data and constructing a lightweight glass bottle model; testing the bottle body structural strength based on the lightweight glass bottle model;

[0006] Step S2: Detecting bottle surface defects based on the bottle structural strength to generate bottle surface defect data; tracing defects based on the bottle surface defect data to obtain bottle production defect process data;

[0007] Step S3: performing wall thickness unevenness detection based on the bottle body structural strength to obtain wall thickness unevenness data; evaluating the core-mold cavity alignment offset based on the wall thickness unevenness data; performing positioning pin fracture detection based on the core-mold cavity alignment offset to obtain positioning pin fracture data;

[0008] Step S4: Generate a unique glass number based on the lightweight glass bottle model; identify abnormal batches of lightweight glass bottles based on the bottle body production defect process data and the positioning pin breakage data to obtain abnormal batch data; map the abnormal batch data to the abnormal glass number based on the unique glass number to obtain the abnormal glass number, and transmit it to the production equipment system to perform the glass quality traceability task.

[0009] This invention addresses the problems of missing data, insufficient unique identification, insufficient real-time performance, and data analysis limitations found in traditional systems by enabling digital traceability of lightweight glass bottle quality. During the design phase, lightweight design optimizes the bottle structure, improving strength while reducing raw material consumption, thereby lowering production costs and meeting environmental protection requirements. Through lightweight model-based bottle structural strength testing, structural weaknesses are identified in advance, improving the overall safety and durability of the glass bottles and providing strong data support for subsequent quality control. During bottle surface defect detection, defect data is generated, enabling accurate identification of surface defects and traceability based on this data, thereby clarifying process issues during production and promoting optimization and improvement of the production process. Furthermore, wall thickness anomaly detection effectively identifies uneven wall thickness during bottle production. Based on this data, the accuracy of mold core and mold cavity alignment can be assessed, further enabling detection and prediction of locating pin breakage, preventing damage due to insufficient strength and ensuring the stability of the entire production process. Through this series of steps, a unique number can be generated for each bottle using the lightweight glass bottle model, ensuring that each glass bottle has a clear identity and providing an accurate basis for subsequent quality traceability. During the abnormal batch identification and glass unique number mapping process, the production defect process data and positioning pin breakage data are combined to accurately locate each abnormal product and transmit it to the production equipment system for processing in a timely manner. This not only improves the accuracy of quality traceability, but also strengthens the rapid response and processing capabilities of abnormal products, thereby improving the intelligence level of the entire production system. This method can ultimately ensure quality traceability and effective management throughout the entire process, thereby enabling lightweight glass bottle manufacturers to efficiently identify, quickly respond to, and hold accountable for quality risks throughout their life cycle.

[0010] Preferably, this specification also provides a digital traceability system for lightweight glass bottles, which is used to execute the digital traceability method for lightweight glass bottles as described above. The digital traceability system for lightweight glass bottles includes:

[0011] The bottle body structural strength detection module is used to obtain glass bottle design data and build a lightweight glass bottle model; based on the lightweight glass bottle model, the bottle body structural strength is detected;

[0012] The defect tracing module is used to detect bottle surface defects based on the bottle structural strength and generate bottle surface defect data; based on the bottle surface defect data, defect tracing is performed to obtain bottle production defect process data;

[0013] The positioning pin fracture detection module is used to detect uneven wall thickness based on the structural strength of the bottle body and obtain uneven wall thickness data; evaluate the core-cavity alignment offset based on the uneven wall thickness data; and perform positioning pin fracture detection based on the core-cavity alignment offset to obtain positioning pin fracture data;

[0014] The glass quality traceability module is used to generate a unique glass number based on the lightweight glass bottle model; identify abnormal batches of lightweight glass bottles based on the bottle body production defect process data and the positioning pin breakage data to obtain abnormal batch data; map the abnormal batch data to the abnormal glass number based on the unique glass number to obtain the abnormal glass number, and transmit it to the production equipment system to perform the glass quality traceability task.

[0015] The digital traceability system for lightweight glass bottle quality of the present invention can implement any one of the digital traceability methods for lightweight glass bottle quality of the present invention, and is used to combine the operation and signal transmission medium between various modules to complete the digital traceability method for lightweight glass bottle quality. The internal modules of the system cooperate with each other to optimize the glass bottle production process, effectively improving the quality recognition rate, abnormal response speed and production process transparency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0017] Figure 1 This is a schematic flow chart of the steps of a method for digitally tracing the quality of lightweight glass bottles according to the present invention;

[0018] Figure 2 Detailed step flow diagram of step S1 in the present invention;

[0019] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0020] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative work are within the scope of protection of the present invention.

[0021] In addition, the accompanying drawings are merely schematic illustrations of the present invention and are not necessarily drawn to scale. Identical reference numerals in the figures denote identical or similar parts, and thus repetitive descriptions thereof will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities that do not necessarily correspond to physically or logically separate entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor and / or microcontroller approaches.

[0022] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0023] To achieve this, please refer to Figures 1 to 2 The present invention provides a digital traceability method for the quality of lightweight glass bottles, which includes the following steps:

[0024] Step S1: Obtaining glass bottle design data and constructing a lightweight glass bottle model; testing the bottle body structural strength based on the lightweight glass bottle model;

[0025] In this example, the "Parametric Modeling" module in the 3D mechanical design software SolidWorks was used to obtain historical structural design drawings of the target glass bottle. After importing the design drawings, initial parameters were set, including a body diameter of 70 mm, a neck inner diameter of 28 mm, a height of 200 mm, and a base thickness of 5.5 mm. To achieve lightweighting goals and taking into account existing molding process limitations, the base thickness was set to 4.0 mm, the wall thickness to 1.6 mm, and the target bottle weight to be within 240 g. After completing the parameter update, a new lightweight structural design drawing was generated. Subsequently, a static strength analysis of the new structure was performed using the ANSYS structural analysis module. Input parameters included an internal pressure of 0.5 MPa, a Young's modulus of 62 GPa for the glass material, a Poisson's ratio of 0.23, and a maximum allowable stress of 18 MPa. Finite element meshing was performed on the entire bottle under internal pressure loading, using tetrahedral elements with a mesh density of 50 elements per cubic millimeter. After completing the loading and solving, the stress cloud diagram and displacement results are extracted. If the equivalent stress value at any position in the structure exceeds 18 MPa, the bottle body geometric parameters are returned and readjusted until the stress at all parts is less than the set threshold.

[0026] Step S2: Detecting bottle surface defects based on the bottle structural strength to generate bottle surface defect data; tracing defects based on the bottle surface defect data to obtain bottle production defect process data;

[0027] In this example, lightweight glass bottle samples, after strength verification, were submitted to a cold-end visual inspection device for surface defect detection. The inspection equipment utilizes a double-sided annular LED lighting system and a Basler acA1920-40gm industrial camera for image acquisition. The resolution is 1920×1200 pixels, the exposure time is set to 800 microseconds, and the frame rate is set to 25 frames per second. After image acquisition, the image enhancement module in Halcon Vision Software performs grayscale normalization and edge extraction using the edge gradient method. During this extraction process, the edge gradient threshold is set to 28 grayscale units. Abnormal areas such as dents, bubbles, and cracks detected in the image are classified and identified using an area threshold method. Cracks are identified as having a minimum curvature radius less than 0.3 mm and a length greater than 2.0 mm; bubbles greater than 1.5 mm in diameter; and dents greater than 0.2 mm in depth. The identified defect type, location coordinates, and dimensions constitute the bottle surface defect data. Subsequently, the production process records corresponding to the defective area on the bottle's surface are compared, including the pressing temperature (for example, the mold temperature for the bottle bottom area is recorded as 1180°C), cooling rate, and mold opening and closing time. This is mapped and compared using a time synchronization mechanism and the bottle's positioning code. If the defect location and process parameters within a specific time period are abnormal (such as a cooling rate below 40°C / s), the production process parameters for that period are marked as bottle production defect process data, forming a process traceability chain.

[0028] Step S3: performing wall thickness unevenness detection based on the bottle body structural strength to obtain wall thickness unevenness data; evaluating the core-mold cavity alignment offset based on the wall thickness unevenness data; performing positioning pin fracture detection based on the core-mold cavity alignment offset to obtain positioning pin fracture data;

[0029] In this example, a non-contact ultrasonic wall thickness tester (such as the GE CL5) was used to perform a full 360-degree rotational measurement on glass bottle samples that had passed strength testing. The measurement resolution was set to 0.05 mm, and the measurement path covered the vertical axis of the bottle body from the shoulder to the bottom. Measurements were taken every 10 mm, with 360 angle points collected per rotation. The acquired wall thickness data was processed using a Matlab data processing script to calculate the standard deviation σ and maximum deviation Δt of each measurement point. If σ exceeded 0.15 mm or Δt exceeded 0.25 mm, it was considered to be uneven wall thickness. The azimuthal angle difference between the maximum and minimum wall thickness values in that area was recorded. Combined with the mold structure data, the core-cavity alignment offset Δθ was calculated using the formula Δθ = atan[(t_max - t_min) / r], where r is the bottle body radius and t_max and t_min are the maximum and minimum wall thickness values for that section. If Δθ exceeds 3.5 degrees, an abnormal core-cavity alignment is identified. When an abnormal alignment offset occurs, the locating pins of the pressing mold are further inspected for fractures. A structured light 3D imaging device (such as the Gocator 2400 series) is used to obtain the topographic image of the locating pin area during mold pressing. The contour breakpoint method is used to determine the integrity of the locating pin. If the continuity interruption length of the scanned section exceeds 1.0 mm or the locating pin depth is less than 50% of the designed depth (such as 8.0 mm), the locating pin is determined to be broken. The workstation number, failure time, and corresponding mold number are recorded to generate locating pin fracture data.

[0030] Step S4: Generate a unique glass number based on the lightweight glass bottle model; identify abnormal batches of lightweight glass bottles based on the bottle body production defect process data and the positioning pin breakage data to obtain abnormal batch data; map the abnormal batch data to the abnormal glass number based on the unique glass number to obtain the abnormal glass number, and transmit it to the production equipment system to perform the glass quality traceability task.

[0031] In this example, a laser coding device (such as the KEYENCE ML-Z series) is used to etch a 12-digit unique identification number on the outer ring of the base of lightweight glass bottles immediately after they are ejected from the mold. The number format is YYMMDD + mold number + serial number. For example, "250514A1327" represents bottle number 1327 of mold A, dated May 14, 2025. The number etching depth is set to 0.05 mm, the character width is 0.4 mm, the scanning laser power is set to 20 W, and the scanning speed is set to 180 mm / s, ensuring that the code is completed within 1.5 seconds of ejection. All coded numbers are registered and archived in real time in the MES system. Subsequently, the bottle production defect process data from step S2 and the locating pin breakage data from step S3 are combined to cross-screen the glass bottle numbers within the same time window. If more than 90% of the bottles in a batch are associated with the same production defect parameter group or locating pin breakage data from the same workstation, the batch is marked as abnormal. The data indexing module generates a collection of abnormal glass serial numbers for each batch of abnormal bottles, such as the number range "250514A1300–250514A1350." This serial number is transmitted via industrial Ethernet to the hot-end annealing equipment, packaging equipment, and final inspection and sorting equipment for identification and rejection of the batch of bottles, and for digital quality traceability. This serial number data is also uploaded to the company's cloud-based traceability platform for subsequent quality event tracking.

[0032] It is particularly important that step S41 includes the following steps:

[0033] Step S41: Generate a unique glass number based on the lightweight glass bottle model;

[0034] In this embodiment, real-time equipment data from the glass bottle molding process serves as the primary data source. Based on the timestamps from the start of molding to the end of cooling for each lightweight glass bottle, combined with the mold cavity number, core number, locating pin number, team number, production line number, and specific machine number, a hashing algorithm (such as SHA-256) is used to perform field concatenation and encryption. The field concatenation sequence is fixed to "machine number + cavity number + core number + timestamp + locating pin number," with timestamps accurate to the millisecond level to ensure no duplicate codes. This method generates a globally unique number for each glass bottle. This number is stored in hexadecimal format in the production database and laser-coded as a QR code on the outer ring of the bottle base, ensuring it does not interfere with the functional structure of the bottle. The number generation process is integrated into the cold-end inspection equipment control system and is synchronously triggered by a PLC controller and an industrial vision recognition module, ensuring that bottles are uniquely numbered upon exiting the cooling process.

[0035] Step S42: performing normalization processing on the bottle production defect process data to obtain normalized production defect process data;

[0036] In this embodiment, surface defect data for bottle surfaces, such as cracks, bubbles, weld lines, cold spots, and wrinkles, is extracted from both the hot-end infrared inspection system and the cold-end visual inspection system. The data structure includes fields such as defect type, defect location (based on the bottle coordinate axis), defect area, defect grayscale value range, corresponding mold cavity number, and timestamp. The defect area in the data is normalized using mm² values, with a maximum value set to 50 mm² and a minimum value set to 0.5 mm². Min-max normalization is used for mapping. Defect locations are normalized using the bottle coordinate system (Z-axis vertical height, θ is the angular dimension). The Z-axis originates from the bottle base, with a maximum height set to 300 mm and discretized into 10 mm intervals. The θ angular dimension is mapped into 24 segments (15° each) divided into 360°. The normalized defect process data is organized chronologically and stored in a structured database table to facilitate subsequent time series analysis.

[0037] Step S43: marking the mold cavity number based on the positioning pin fracture data;

[0038] In this embodiment, the data on the fracture of the locating pins is derived from the bottle mold maintenance station and the mold monitoring system. During the replacement of the bottle mold core or the maintenance of the mold cavity, the mold cavity identification instrument equipped on the maintenance engineering station records the fracture or replacement event and identifies the mold cavity number through the sensor. After each identification, the built-in data processor registers the mold cavity number, the time of fracture occurrence, and the fracture position in three fields. The mold cavity number is a fixed structure, such as "M-12" represents mold cavity No. 12, and the fracture timestamp is accurate to the second level. The data is uploaded to the quality control master server through the Modbus TCP protocol and associated with the bottle body number and the production line station number. The mold cavity number and the unique number of the glass bottle are compared through the timestamp and calibrated with the production sequence number to achieve unique identification and binding of the fracture data and the specific bottle body data, forming a data chain that can be used for abnormal analysis.

[0039] Step S44: performing time correlation analysis based on the normalized production defect process data and the mold cavity number to obtain defect process-mold cavity number data;

[0040] In this embodiment, time association analysis is achieved by constructing a time window algorithm. First, the timestamp field in the normalized production defect process data is extracted and grouped and clustered with the cavity number field. A fixed time window is defined as 30 minutes, and each set of data is divided into subsets according to the cavity number and the timestamp window. In each subset, if there are 3 or more consecutively numbered glass bottles with the same type of defects, and the severity of their defects (based on the defect area exceeding 20mm²) accounts for more than 70%, then the window is defined as an abnormal time window. The cavity number corresponding to the time window is extracted synchronously with the unique glass number to form a defect process-cavity number data pair. The size of the time window can be obtained by fitting the historical abnormal bottle data, and a multiple that can cover the average bottle molding cycle is selected. For example, when the average bottle molding cycle is 30 seconds, the window length is set to 60 times the cycle, that is, 30 minutes, to ensure complete data association coverage.

[0041] Step S45: Identify abnormal production batches based on the defective process-cavity number data to obtain abnormal batch data;

[0042] In this embodiment, an anomaly identification algorithm is constructed based on the defective process-cavity number data pair. The cavity number is used as the clustering basis, and the proportion of defective glass bottles appearing in each cavity number within a certain time range (for example, 2 hours) is calculated. If the proportion is greater than 10%, it is considered that the cavity is abnormal within this time period. The record of the locating pin fracture corresponding to the cavity number is further compared. If the occurrence time is within 1 hour before the time period when the defects occur, it is considered that the fracture is causally related to the defect. All bottles within this time period are sorted by unique numbers, and the batches are designated as abnormal batches. The abnormal batch data generates a unique identifier with the structure of "date + machine number + cavity number + time period" and is stored in the abnormal batch database for subsequent mapping processing and docking with production data.

[0043] Step S46: Mapping the abnormal glass number to the abnormal batch data according to the unique glass number to obtain the abnormal glass number, and transmitting it to the production equipment system to perform the glass quality traceability task.

[0044] In this embodiment, after the abnormal batch data is generated, the number field in the glass unique number database is called to match and map it with the abnormal batch identifier. The matching method is: based on the cavity number and time period recorded in the abnormal batch data, all number records within the corresponding time range are screened in the unique number table, and the timestamp range is accurately matched with the cavity number. The number set that is successfully matched is the abnormal glass number. The number set is uploaded to the MES (manufacturing execution system) platform via industrial Ethernet and synchronized to the equipment control terminal via the OPC UA interface for alarm prompts, screening and recall control. At the same time, the abnormal glass number record is uploaded to the traceability data platform to support subsequent quality audits and responsibility determination. The abnormal number transmission uses JSON structure for data encapsulation, and the fields include "number ID", "defect type", "cavity number", "locating pin status", and "abnormal batch ID" to ensure data integrity and consistency.

[0045] Preferably, step S1 is specifically as follows:

[0046] Step S11: obtaining glass bottle design data and extracting glass bottle three-dimensional structure data;

[0047] In this embodiment, the process of acquiring glass bottle design data begins with processing using Pro / E or SolidWorks platforms within standard CAD (Computer-Aided Design) systems based on 2D engineering drawings and design specifications from a product structure database. Input data includes structural dimensional parameters such as mouth diameter, shoulder slope, body height, base thickness, overall weight, body axial symmetry, and wall thickness distribution, in accordance with the national glass container design standard GB / T 4544-2008, regarding body thickness and curvature radius. The "Reverse Modeling Tool" within SolidWorks is used to convert raw point cloud data into 3D glass bottle structural data. A 3D laser scanner (such as the FARO Edge ScanArm HD) is used to acquire point clouds of glass bottle samples with an accuracy set to ±0.01 mm. After registering the scanned data, Geomagic DesignX is used to perform point cloud reconstruction. Through Boolean operations, boundary repair, and closed surface generation, a 3D geometric model of the entire bottle is obtained. The final 3D structural data output is available in both STL and IGES formats for subsequent geometric feature recognition and optimization calculations.

[0048] Step S12: identifying key optimization areas based on the three-dimensional structural data of the glass bottle to obtain the bottle body area and the bottle shoulder area;

[0049] In this embodiment, the normal vector distribution of the bottle's outer surface is scanned and classified into typical regions, such as the bottle mouth, shoulder, body, and base, based on their changing trends. The body region is defined as the portion of the bottle's midsection from the shoulder's starting point to the upper end of the bottom's constricted section, while the shoulder region is defined as the transition structure from the lower edge of the bottle mouth to the starting point of the body. A curvature analysis algorithm based on second-order derivatives is used to detect local differences in the three-dimensional surface model. A critical curvature threshold of 0.002 mm⁻¹ is set, and shoulder regions with prominent curvature variations are identified based on this threshold. Spatial cross-sectional analysis is then used to extract the central continuous surface within an axial height range of 20 mm to 80 mm, which is identified as the bottle body region. A custom curvature analysis script developed in MATLAB, combined with an STL geometry mesh file, is used to calculate the Gaussian curvature and mean curvature matrices for each triangular facet. The extracted results are exported in CSV format, providing basic data for subsequent structural optimization.

[0050] Step S13: performing curvature transition optimization design based on the bottle body area to obtain the optimized curvature of the bottle body;

[0051] In this embodiment, a Bezier curve fitting approach was used to optimize the curvature of the bottle body, achieving a continuous transition. Specifically, longitudinal cross-sections within the bottle body were selected, and the cross-sectional curves were extracted. A fifth-order Bezier curve function was used to fit the curves, achieving uniform curvature distribution by optimizing the positions of control points. The curvature change rate metric, K′ (dK / ds), where s is the arc length of the curve, was used as the control variable of the optimization objective function, with the target maximum value of K′ set to no more than 0.005 mm². Curve reconstruction was performed using the ANSYS SpaceClaim Geometry Modeling tool. Twenty cross-sections along the height of the bottle body were processed using the Curve Reconstruction module (Curves → FitCurve to Edge). The control point movement range for each cross-sectional curve was limited to ±0.3 mm, and the distance difference between control points was maintained at no more than 0.8 mm to avoid the generation of sharp or concave areas. The mean square error (MSE) of the original and fitted curvatures was compared to ensure that the fitting error was less than 0.02 mm. Each optimized cross-sectional curve was then projected back onto the 3D model to generate the updated bottle surface.

[0052] Step S14: performing wall thickness thinning design based on the bottle shoulder area to obtain wall thickness thinning data;

[0053] In this example, a method based on modifying finite element mesh thickness parameters was used to implement wall thinning in the shoulder area. During the analysis, the shoulder area was first cut out of the overall model and imported into HyperMesh for local meshing. The target element size was set to a 1 mm tetrahedral mesh, and the total mesh count was kept below 50,000. The initial wall thickness was set to 3 mm. Based on the compressive strength of the glass material, σ_c (set to 45 MPa), and a safety factor (set to 2), and the relationship between thickness and stress, σ = F / A, where A is the load-bearing area, the ultimate wall thickness was deduced to be approximately 1.6 mm. In practice, the target wall thinning value was set to 2.0 mm, resulting in a thinning rate of 33.3%. To avoid stress concentration, a 5 mm transition zone was created at the junction of the shoulder and the body, using a linear thickness transition strategy from the original 3 mm to 2.0 mm. After the wall thickness was updated, a single-point compression simulation of the shoulder stress-bearing area was performed using Altair OptiStruct to verify that the maximum equivalent stress in the thinned area remained below σ_c. The wall thinning data is exported to TXT format through the mesh thickness attribute and written into the design file for subsequent modeling.

[0054] It is particularly important that step S14 includes the following steps:

[0055] Step S141: identifying the bottle shoulder structure contour based on the bottle shoulder area;

[0056] In this embodiment, a high-precision 3D laser scanner is used to acquire 3D point cloud data of the glass bottle shoulder. The scanner's resolution should be set to no less than 0.05mm to accurately capture every detail of the shoulder area. The scanning process requires 360-degree multi-angle coverage of the shoulder area to ensure complete 3D data acquisition. After scanning, data processing software is used to denoise and filter the point cloud data to remove irrelevant stray data. After processing the point cloud data, a boundary extraction algorithm (such as the RANSAC algorithm) is used to identify the primary structural contours of the shoulder area, defining its edge lines and transition surfaces. These extracted boundary lines are used to construct a precise shoulder geometry model, which serves as the basis for subsequent wall thickness measurement and optimization.

[0057] Step S142: measuring the thickness of the bottle shoulder based on the bottle shoulder structure profile to obtain bottle shoulder thickness data;

[0058] In this embodiment, after identifying the shoulder contour, the shoulder thickness is measured using an automated wall thickness measurement system. This system uses a LiDAR sensor for precise measurement, ensuring an error of no more than 0.1mm at each measurement point. A wraparound measurement strategy is employed, covering the entire geometric surface of the shoulder area. At each circular measurement point, the LiDAR sensor precisely scans the shoulder surface perpendicular to the shoulder, acquiring wall thickness data from the surface to the interior. The resulting thickness data is recorded based on the spatial location of the measurement point (i.e., distance from the bottle base) and converted into a two-dimensional or three-dimensional format. This data provides detailed thickness distribution across the shoulder area, providing a basis for subsequent stress testing and wall thickness optimization.

[0059] Step S143: Based on the bottle shoulder thickness data, standard simulated internal pressure is applied by industrial stress testing equipment and connected to a strain bridge array for real-time stress response monitoring, thereby obtaining bottle shoulder stress data;

[0060] In this embodiment, a standard simulated internal pressure is applied using industrial stress testing equipment. The equipment should be able to simulate an operating environment with an internal pressure of 0.8 MPa. This pressure value is set according to the design standards for glass bottles and is typically a typical value for the pressure the bottle body can withstand. The pressure sensor of the stress testing equipment must have a measurement accuracy of at least 0.1 MPa to ensure that the applied pressure meets the design requirements. After applying the internal pressure, the stress response of the bottle shoulder area is monitored in real time using a strain bridge array. The strain bridge array consists of multiple strain sensors evenly distributed across the bottle shoulder area to monitor stress changes at different locations. Each strain bridge array is connected to a data acquisition system with a data acquisition frequency of at least 100 Hz to ensure timely recording of stress changes. By processing and analyzing the strain signals, stress distribution data for the bottle shoulder area can be obtained. This stress data not only reflects the load-bearing capacity of the bottle shoulder area but also reveals stress concentration issues in certain areas of the bottle shoulder, providing important information for subsequent wall thickness adjustments.

[0061] Step S144: adjusting the wall thickness point by point based on the bottle shoulder stress data, limiting the minimum wall thickness to no less than 2.1 mm, and obtaining wall thickness thinning data.

[0062] In this embodiment, the stress concentration in different areas of the bottle shoulder is analyzed based on stress data to determine the key locations of stress concentration. For example, certain areas of the bottle shoulder will have higher stress values under standard simulated internal pressure, exceeding the safety threshold of the glass material. For these areas, the wall thickness is adjusted point by point. During the adjustment process, based on the design requirements, the minimum wall thickness of each adjustment point is ensured to be no less than 2.1 mm. If the thickness of certain areas is insufficient, adjustments can be made using a thickness increase and decrease algorithm to thicken areas that are too thin and thin areas that are thick enough, so that the stress distribution of the bottle shoulder is more uniform when subjected to internal pressure. The specific wall thickness change value of each adjustment point is calculated by the difference with the original thickness data to obtain the wall thickness thinning data for each point. The final wall thickness thinning data will be used to guide the processing or reprocessing of the bottle shoulder mold to ensure that the structure of the bottle shoulder meets both lightweight requirements and the strength and safety of the glass bottle.

[0063] Step S15: Integrate the bottle body optimized curvature and wall thickness reduction data, and construct a lightweight glass bottle model;

[0064] In this embodiment, the integration process is implemented using the 3D reconstruction module of the CATIA platform. First, the optimized curvature of the bottle body obtained in step S13 is imported as an independent geometric surface object. Using CATIA's "Multi-section Surface" tool, the new bottle body surface is constructed by setting the continuity between each section to G2 (curvature continuity). Subsequently, the shoulder thinning mesh data obtained in step S14 is imported as a thickness reference. The original shoulder surface is offset equidistantly using the Surface Offset module to construct an updated, thinned-wall solid. The Join module is then used to connect the optimized body surface, the shoulder thinning surface, and the base and mouth areas to form a closed body. The Part Design module is used for solidification, generating the complete 3D structure of the lightweight glass bottle. The model is ultimately output as a STEP file, containing complete solid information and wall thickness annotations for structural strength simulation analysis.

[0065] Step S16: Detecting the structural strength of the bottle body based on the lightweight glass bottle model.

[0066] In this embodiment, the structural strength test was performed using finite element analysis using ANSYS Workbench. First, the lightweight glass bottle model was imported into ANSYS for meshing, using tetrahedral structural units with a unit size of 0.8 mm, generating approximately 180,000 units for the entire bottle. The material properties were set to silicate glass, with an elastic modulus of 70 GPa, a Poisson's ratio of 0.23, and a density of 2500 kg / m³. The loading conditions were set as follows: a 5 kg pressure was applied vertically to the bottle body to simulate a stacking load, and a fixed constraint was set at the bottom of the bottle; a 20 N axial impact force was applied to the bottle shoulder area to simulate the bottle cap packaging force; and a 10 N lateral pressure was applied to the side wall of the bottle body to simulate lateral impact during transportation. The calculation objectives include the maximum principal stress, Von Mises equivalent stress, and deformation, and the limit judgment criteria were set separately: the maximum principal stress must not exceed 45 MPa, the equivalent stress must not exceed 40 MPa, and the total deformation must not exceed 0.3 mm. After the calculation is complete, the stress distribution diagram (JPG format) and stress value file (CSV format) are exported and recorded in the quality traceability system for subsequent defect correlation analysis. All simulation data is archived synchronously with a timestamp and unique design number.

[0067] Preferably, step S16 is specifically as follows:

[0068] Step S161: importing the lightweight glass bottle model into the simulation software;

[0069] In this embodiment, the constructed three-dimensional structure file of the lightweight glass bottle is exported in STEP format, and the "Static Structural" module in ANSYS Workbench 2021 R2 is used to perform the structural strength analysis import operation. In the ANSYS environment, select the "Engineering Data" module to load the new material property library and configure the simulation project file, and then load the glass bottle structure file through the "Import Geometry" command in the "Geometry" module. During the import process, the default coordinate system is used for three-dimensional position recognition to ensure that the model does not have geometric anomalies such as topological gaps and discontinuous surfaces. At the same time, the contact areas are merged with the help of the "Share Topology" tool to avoid discontinuous boundary conditions during simulation analysis. After importing, the "Mesh" module is used to perform model meshing operations. The mesh type is set to tetrahedral element (Tetrahedral Element), and the mesh size is initially set to 1.0mm. Local encryption is used according to high stress areas such as bottlenecks and bottle bottoms. The minimum mesh size is controlled to be no less than 0.2mm and the maximum size does not exceed 3mm. The mesh quality standard is Skewness < 0.9 to ensure calculation accuracy.

[0070] Step S162: setting the bottle body wall thickness range to 0.3mm-3mm, the bottle neck thickness range to 0.2mm-1.5mm, and the bottle bottom thickness range to 0.5mm-4mm in the simulation software;

[0071] In this example, after importing and meshing the glass bottle, we entered the ANSYS "Model" module and used the "Named Selections" feature to group the different regions of the bottle structure: Body, Neck, and Bottom. We then used the "Thickness" command to set the initial wall thickness parameters for each of these three groups of regions. The bottle body wall thickness range is controlled to be 0.3mm-3mm. The setting strategy is to set the wall thickness change in layers every 10mm in the axial direction of the bottle body height, using the distribution function Wall_Thickness_Body(z)=0.3+2.7*sin(πz / H), where z is the axial height and H is the total height of the bottle body. The bottleneck thickness setting range is 0.2mm-1.5mm, using a tapered gradient structure. The thickness along the axis follows the linear function Neck_Thickness(z)=0.2+1.3(z / L), where L is the bottleneck height. The bottom area of the bottle is set to have a thickness range of 0.5mm-4mm, and the thickness transition is performed according to the fillet radius r of the connection with the bottle body. The thickness function is Bottom_Thickness(r)=0.5+3.5(1-exp(-r / 5)), where r is the radial distance from the center to the edge of the bottle bottom (unit: mm), which is used to control the thickness change of the bottom thickened area to adapt to the force difference.

[0072] Step S163: in the simulation software, set the external pressure range to 0.5 MPa-2 MPa, the internal pressure range to 0.3 MPa-1.5 MPa, and the transport impact force range to 10 N-200 N;

[0073] In this example, when configuring load conditions in the ANSYS "Static Structural" module, various external environmental stresses were applied to the model surface. The external pressure range was set to 0.5 MPa-2 MPa, with a constant surface pressure applied to the entire outer surface. A phased load was established, with four conditions set: 0.5 MPa, 1.0 MPa, 1.5 MPa, and 2.0 MPa. The internal pressure range was set to 0.3 MPa-1.5 MPa, applied outward to the inner surface of the bottle. Four pressure values were also set to establish a pressure influence curve. The transport impact force was set to a transient concentrated force (point load), acting on key areas of the bottle, including the shoulder, bottom edge, and waist. The corresponding impact forces were set to 10 N, 50 N, 100 N, and 200 N, with a step duration of less than 0.1 second. Coupled simulation was performed using the "Transient Structural" module, with impact directions including axial, radial, and vertical. A total of 16 simulation conditions are set in the form of load and boundary condition combinations to cover the pressure and impact conditions encountered in typical transportation and storage environments.

[0074] Step S164: setting the elastic modulus of the glass material to a range of 60 GPa-75 GPa and the Poisson's ratio to a range of 0.22-0.26 in the simulation software;

[0075] In this example, glass material properties are configured in the "Engineering Data" module, with silicate glass selected as the base material. The Young's Modulus (Young's Modulus) is set between 60 GPa and 75 GPa, with four values set at 60 GPa, 65 GPa, 70 GPa, and 75 GPa, with the input unit being Pa. The Poisson's Ratio (Poisson's Ratio) is set between 0.22 and 0.26, with corresponding values of 0.22, 0.24, and 0.26. Fracture toughness is set as a nonlinear material parameter, with three values of 0.3, 0.5, and 0.8 MPa·m^1 / 2 entered into the ANSYS material nonlinear fracture library. The fracture behavior function is loaded using the User Defined Material model, and the LEFM (Linear Elastic Fracture Mechanics) criterion is used to analyze the fracture stress distribution under various pressure and impact loading conditions. The various parameters are derived from the national glass product quality standard GB / T2828 and ASTM C158-02, and are calibrated by comparing with existing glass bottle physical property test data to ensure that the parameter settings in the material constitutive model are consistent with the physical properties of the glass.

[0076] Step S165: Run the structural strength analysis program in the simulation software and record the structural strength of the bottle.

[0077] In this example, after setting boundary conditions, loading conditions, and material properties, "Equivalent Stress (von-Mises)" and "Maximum Principal Stress" were selected as output variables in the ANSYS "Solution" module, and the simulation task was run using the "Solve" command. The simulation results record the location of structural stress concentration areas, the maximum stress value, and the corresponding coordinates. The output results are exported in tabular format for different thickness, pressure, and impact combinations. The stress threshold is controlled between 30 MPa and 90 MPa based on the ultimate strength of glass. If the equivalent stress in a local area exceeds the set material fracture stress value (e.g., 85 MPa), the design is considered to fail structural integrity requirements. The stress distribution data output from each simulation is stored in a .csv file by node number and spatial coordinates. The data fields include node ID, X / Y / Z coordinates, von-Mises stress value, principal stress value, and its orientation angle. All results are automatically classified and summarized in the Post-Processing module for subsequent bottle quality assessment and traceability comparative analysis.

[0078] Preferably, step S2 is specifically as follows:

[0079] Step S21: extracting bottle deformation characteristics and bottle pressure resistance characteristics based on the bottle structural strength to obtain bottle deformation data and bottle pressure resistance data;

[0080] In this embodiment, after completing the structural strength simulation analysis, the deformation and pressure resistance characteristics of the bottle structure were extracted based on the recorded stress distribution and displacement field data under different operating conditions. First, under the extreme load conditions of 2 MPa external pressure, 1.5 MPa internal pressure, and 200 N impact force, the total displacement of each unit node was derived from the finite element simulation results, and the displacement mutation interval was extracted using a three-dimensional point cloud interpolation method. The maximum deformation region of the bottle was identified by setting a displacement gradient threshold Δd = 0.4 mm. Any region with a gradient exceeding this threshold was defined as a high-risk deformation zone, and its spatial coordinates were recorded. To extract the pressure resistance characteristics of the bottle, the internal pressure was linearly increased from 0.3 MPa to 1.5 MPa, and the equivalent stress distribution was calculated in increments of 0.1 MPa. If the local maximum equivalent stress σ_eq exceeded the compressive limit of the glass material σ_c = 120 MPa, the corresponding pressure value was recorded as the instability pressure value. In this way, the maximum stable pressure resistance value of the bottle structure is recorded to obtain complete bottle deformation data (including node coordinates, total deformation value, stress concentration area) and bottle pressure resistance data (including critical internal pressure value, stress distribution matrix under various pressures).

[0081] Step S22: performing bottle body crack detection based on the bottle body deformation data to obtain bottle body crack data;

[0082] In this embodiment, crack detection is performed using digital image correlation (DIC) and 3D structured light scanning results based on the bottle deformation data obtained in step S21. Based on the displacement data extracted from stress concentration areas, 3D microscopic observations are performed on high-gradient regions. A confocal microscope with a 5μm precision scans the bottle surface layer by layer to determine crack depth and width. The crack detection thresholds are set as follows: crack depth d_crack ≥ 0.1mm, crack width w_crack ≥ 0.02mm. If the extracted region in the 3D scan image exhibits continuous grayscale changes greater than 50 units and exhibits a linear morphology extending for more than 1mm, the region is identified as a structural crack. Crack characteristics are stored as crack data in the form of parameters such as structural coordinates, crack length, crack depth, and crack direction. Combined with the initial simulated pressure conditions, crack growth trends can be further annotated to generate a bottle crack dataset with time node identifiers.

[0083] Step S23: performing a transportation simulation based on the bottle pressure resistance data, and identifying surface glass peeling during the bottle transportation process, and obtaining surface glass peeling data;

[0084] In this embodiment, the bottle pressure resistance data obtained in step S21 is imported into a transportation simulation environment. Based on the GB / T4857.5-92 transport packaging drop test standard, a drop height of 1.2 meters is set. Free-fall impacts the bottom, side, and shoulder of the glass bottle, and the surface stress distribution and damage characteristics after impact are recorded. A high-speed camera (frame rate >5000 fps) is used to record the surface glass fracture process at the moment of impact, and image processing algorithms are used to extract the spalling area. Spalling detection utilizes infrared thermal imaging combined with surface refractive index measurement. If the surface refractive index change Δn is greater than 0.05 or the surface spalling particle size is greater than 100 μm, the area is defined as a glass spalling zone. After each simulation, the spalling coordinates, particle size distribution, glass thickness change, and stress change are recorded together to generate surface glass spalling data including spalling location, area, and glass strength reduction.

[0085] Step S24: Integrate the bottle body crack data and the surface glass peeling data to obtain bottle body surface defect data;

[0086] In this embodiment, the crack data obtained in step S22 and the surface spalling data obtained in step S23 are integrated. A unified three-dimensional coordinate system is established through spatial overlap analysis, and all crack and spalling locations are spatially calibrated. The overlapping areas and damaged areas with adjacent boundary distances less than 1 mm are extracted as the core set of bottle surface defect data. Coding rules for structural defect blocks are defined, and a data table is constructed based on defect type (crack CR, spalling SP), location coordinates, size, area, direction, and other information. The volume loss rate of the defect area is calculated through 3D point cloud comparison. If it exceeds 0.5 cm³, it is marked as a severe defect point. The final bottle surface defect data is exported in CSV format, with fields including defect type, location coordinates (x, y, z), defect size (length, width, depth), loss volume, crack direction angle, and maximum spalling particle diameter.

[0087] Step S25: performing defect tracing based on the bottle surface defect data to obtain bottle production defect process data.

[0088] In this embodiment, after acquiring bottle surface defect data, defect traceability analysis is performed based on glass bottle production process parameter records. Each defect sample is matched against parameters in the production process database, including mold temperature (set range 1100°C-1400°C, acquisition interval 1 second), preform cooling time (set range 2s-5s), spray concentration (measurement range 0%-5%), and annealing temperature curve (set between 580°C and 620°C). A decision tree classification algorithm is used to construct association rules. If cracks are concentrated in the shoulder area, with a corresponding mold cooling time less than 2.5s and an annealing curve offset of ±10°C, they are identified as thermal stress concentration cracks caused by insufficient annealing. If the spalling area is concentrated in the concave portion of the bottle base, with a corresponding preform thickness less than 0.8mm and spray coverage less than 90%, spalling is attributed to cooling stress accumulation caused by uneven spraying. The final output is the bottle production defect process data, with fields including defect code, defect type, traceability process parameters (temperature, time, thickness, etc.), abnormal interval time point, matching probability score, etc., which serves as the historical tracking basis for quality control.

[0089] Preferably, step S22 is specifically as follows:

[0090] Step S221: performing bottle rotation control based on the bottle deformation data to obtain bottle rotation data;

[0091] In this embodiment, the bottle deformation data extracted in step S21 is used as basic data input into the control system. A six-degree-of-freedom rotation execution platform controls the stable rotation of the bottle along its vertical and horizontal axes. The bottle is secured in a pneumatic clamp at the center of the rotating platform, with a clamping pressure set to 0.5 MPa to ensure that the bottle does not slide or shift during rotation. The angular velocity is set to 15° / s, and the rotation angle range is from 0° to 360°. Angular position information and deformation vector direction are collected every 5°. An angle encoder (resolution 0.01°) and a laser displacement sensor (measurement accuracy 1μm) are used in conjunction to collect deformation distribution data at any angle during the rotation process, forming a complete three-dimensional rotational deformation vector field. The final output is the bottle rotation data, which includes parameters such as angular coordinates, rotational deformation vector value, axial torque distribution, and rotational inertia change.

[0092] Step S222: Calculating stress gradients based on the bottle rotation data, and extracting high stress gradient values based on the stress gradients;

[0093] In this embodiment, after the bottle rotation data obtained in step S221 is imported into the stress calculation system, the system first calculates the deformation stress distribution based on the displacement data at each angle along the rotation path. Using the elastic modulus of 72 GPa and Poisson's ratio of 0.22 for the glass material, these material parameters are used to calculate the stress distribution of the glass bottle at each rotation angle based on linear elastic body theory. During the calculation process, the system first obtains the displacement corresponding to each rotation angle using a displacement difference method and then calculates the local strain at each angle using a differential method. Based on the relationship between strain and stress, the stress value at each angle is calculated using basic formulas in elastic mechanics. Next, the system calculates the stress difference between each angle and adjacent angles. First, the stress values at adjacent angles are extracted, the difference between them is calculated, and then divided by the angle difference (for example, 5° per rotation) to obtain the stress gradient at each point. The stress gradient is a metric that describes the change in stress distribution; its magnitude represents the rate of stress change per unit angle change. After the calculation is complete, the system sets a threshold (for example, 3 MPa / °). If the stress gradient at a particular angle exceeds this threshold, the point is marked as a high stress gradient point. Finally, the system records all stress gradient values above the set threshold and their corresponding rotation angle positions, generating a high-stress gradient dataset that is exported and saved. This dataset contains the specific value and location of each high-stress gradient point, providing data support for subsequent crack identification and high-stress area location.

[0094] Step S223: identifying high deformation areas of the bottle body according to high stress gradient values;

[0095] In this embodiment, regions of high stress gradient value data in step S222 are identified. The spatial clustering method DBSCAN is used, with a neighborhood radius ε set to 3 mm and a minimum sample size MinPts set to 6, based on empirical values for the density of discrete points on the bottle surface. Input data includes the spatial coordinates (X, Y, Z) of the points and their angle θ labels. The clustering process identifies multiple high-density clusters, each containing several continuous stress gradient abrupt changes. Each cluster is fitted to an ellipsoid, and the ellipsoid's center coordinates C (x, y, z), major and minor axis lengths (a, b, c), and the number of cluster points N are recorded. The screening criteria are N ≥ 10, with a / b ≥ 1.5 as the criterion for high deformation regions. Output data includes the spatial boundary point set for each high deformation region, the deformation field (projected from the deformation vector field), the region number, and the rotation angle θ range, which guides subsequent X-ray scan path planning.

[0096] Step S224: performing an X-ray circular irradiation scan on the high deformation area of the bottle body to obtain an X-ray image of the bottle body;

[0097] In this example, a high-resolution X-ray scanner (such as the GE Phoenix V|Tome|XM) was used with a focal spot size of 2μm, an operating voltage of 100kV, a current of 250μA, and an exposure time of 50ms. The bottle was mounted on a 360° rotating platform with a rotation accuracy of 0.05°. The scanning path was limited to an angle range of ±10° within the high-deformation region identified in step S223 to ensure complete crack information. The X-ray detector used a 2000×2000 pixel flat-panel detector with a single pixel size of 50μm. Image acquisition captured one frame per 1° rotation, with the image number corresponding to the angle θ. The generated image file was named "Region_XX_Angle_XXX.tiff." A total of 21 images were collected, covering each high-stress region. All X-ray images were archived in an image management system, containing metadata such as the image matrix, exposure parameters, scanning angle, bottle number, and region number.

[0098] Step S225: extracting significant boundaries based on the bottle X-ray image to obtain significant boundary data;

[0099] In this embodiment, the X-ray images acquired in step S224 are processed frame by frame using the OpenCV image processing library. First, a Gaussian filter (kernel size 5×5, σ = 1.0) is applied to the original image to remove high-frequency noise. The Canny edge detection algorithm is then used to extract boundaries, with thresholds set to a low threshold T1 = 50 and a high threshold T2 = 100, respectively, based on the image's grayscale dynamic range and the crack edge gradient. After boundary extraction, contour tracing is performed using the OpenCV findContours function. Closed boundaries are numbered and a point set (minimum 100 points) is extracted for each boundary. Curvature analysis is performed on the boundary point set, and boundaries with a curvature change of <5% are defined as "significant boundaries." The output significant boundary data structure includes: boundary number, contour point coordinate sequence, boundary length, closedness flag (Boolean value), curvature change rate, image number, and angle θ label.

[0100] Step S226: Calculating the aspect ratio based on the significant boundary data to obtain the aspect ratio;

[0101] In this embodiment, for each set of significant boundary contour points, a minimum circumscribed rectangle fitting is performed (using the minAreaRect function). The major axis L and minor axis W of the rectangle are expressed in physical units (mm). The image calibration parameters are used to convert pixels (px) to actual size, and the calibration coefficient is 1px=0.05mm. The aspect ratio AR is calculated as L / W. If the number of boundary points is ≥150 and the contour is approximately a straight line structure (rectangular angle difference <10°), the area is determined to be a crack area. All boundary data record fields include: boundary number, circumscribed rectangle major axis L (mm), minor axis W (mm), aspect ratio AR (unitless), contour center point position, and rotation angle θ.

[0102] Step S227: identifying the crack profile according to the aspect ratio;

[0103] In this embodiment, based on the aspect ratio data calculated in step S226, crack contour identification is performed for boundaries that satisfy AR ≥ 5. The angle φ between the principal direction vector of the boundary and the principal direction of the local stress of the bottle is further calculated. The principal stress direction is obtained using the projected stress direction algorithm. If the angle φ ≤ 15°, it is calibrated as the crack path. The boundary point sequence is fitted using spline interpolation (B-spline), with a fitting step size of 0.2 mm. The crack contour is output as a continuous path line with a direction vector, recording the crack start and end points, path length, crack number, scanning angle θ, and fitting residual RMSE.

[0104] Step S228: Calculate the crack defect density based on the crack profile to obtain bottle body crack data.

[0105] In this embodiment, the bottle X-ray image area is projected onto a two-dimensional plane and divided into equally spaced 25 mm x 25 mm grid cells, generating M x N grid cells. Within each grid cell, the crack paths in the X-ray image are analyzed to calculate the total length (in mm) of the crack segments and the number of cracks within each grid cell. The path length of each crack within the corresponding grid cell is recorded and used to calculate the crack defect density. The specific crack defect density ρ is calculated as follows: For each grid cell (i, j), the sum of the path lengths L_k of all cracks within that cell is first calculated. Then, according to the formula ρ_i,j = Σ(L_k) / 625, where L_k is the path length of each crack in mm. The division by 625 is because the area of each grid cell is 25 mm x 25 mm, or 625 mm². Therefore, the calculated density is in mm / mm². This calculation yields the crack defect density value within each grid cell. Finally, the crack defect density of all mesh elements is output in matrix form. Mesh elements with density values greater than or equal to 0.4 are marked red, indicating high crack risk areas. The final output is a crack defect density map (CSV format matrix), an index table of corresponding crack numbers, and a crack structure file corresponding to the bottle number (JSON format) to facilitate subsequent traceability comparison and defect statistical analysis. During the processing process, all crack paths and density values are precisely calculated based on image processing technology and geometric analysis methods to ensure data accuracy and reliability.

[0106] Preferably, step S23 is specifically as follows:

[0107] Step S231: screening low-pressure bottles based on the bottle pressure resistance data, and performing transportation simulation to obtain low-pressure bottle transportation data;

[0108] In this embodiment, low-pressure bottles are screened out by collecting bottle pressure data. During the screening process, a pressure threshold of 30 MPa is first set, and the pressure data of each bottle is compared one by one. Bottles with pressure values below this threshold are considered low-pressure bottles. Next, a transportation simulation is performed on the screened low-pressure bottles. This transportation simulation utilizes dedicated transportation simulation software, which considers factors such as impact, vibration, and external pressure encountered by the bottles during transportation. By setting the maximum impact force during transportation to 200 N and the maximum vibration frequency to 50 Hz, and performing multiple simulations on the bottles, the external forces and deformation experienced by the low-pressure bottles in each simulation are recorded, ultimately generating low-pressure bottle transportation data.

[0109] Step S232: performing force analysis based on the low-pressure bottle transportation data to obtain bottle transportation force data;

[0110] In this example, finite element analysis, combined with transportation simulation data, meticulously calculated the stresses on low-pressure bottles. First, the bottle's geometry and material properties were determined (e.g., an elastic modulus of 72 GPa and a Poisson's ratio of 0.22). Within the simulation, external forces were distributed to various parts of the bottle, and the stress and strain at each location were calculated. Based on the force analysis results, a force distribution diagram was plotted, and the areas of maximum stress experienced by the bottle during transportation were recorded. By comparing stress values at different locations, the bottle's stress data was derived, including the maximum stress value, stress concentration, and areas of rupture.

[0111] Step S233: identifying stress concentration areas based on the bottle transportation force data and recording them as peeling risk areas;

[0112] In this embodiment, sensors (such as strain gauges and stress sensors) collect real-time stress data at multiple locations on the glass bottle during transportation. After processing, the acquired data can be generated into a stress distribution map, showing stress values at different locations. Next, analysis is performed based on a set stress concentration threshold, set at 150 MPa. When the stress value at a particular location on the bottle is greater than or equal to 150 MPa, it indicates excessive stress at that location, leading to damage or spalling risk. By comparing and analyzing the stress distribution map across the entire bottle, areas with stress values exceeding the threshold are identified and marked as spalling risk areas. These areas are typically located in irregularly shaped parts of the bottle that are subject to high stress, such as the bottle mouth, bottom, and sharp corners. After identifying spalling risk areas, subsequent testing and monitoring are carried out, with enhanced inspections of these high-risk areas being particularly strengthened to ensure that the glass bottles do not break or spall during transportation, thereby improving transportation safety and overall quality control of the glass bottles.

[0113] Step S234: acquiring a multispectral image based on the peeling risk area to obtain a multispectral image of the bottle body;

[0114] In this embodiment, a multispectral camera capable of providing reflectance data in multiple wavelength bands is first selected. This camera is capable of capturing images within a specific wavelength range. The spectral band range is set from 400nm to 1000nm, covering the spectral region from ultraviolet to near-infrared. This allows for the acquisition of rich spectral information about the bottle surface. During operation, the bottle is placed in a stable environment, ensuring that its surface is free of obstructions. An appropriate light source is used for illumination to avoid overexposure caused by strong direct sunlight. During the image acquisition process, the multispectral camera captures reflectance data from the bottle surface at different wavelengths, thereby producing images in multiple wavelength bands. These images represent reflectance information at different wavelengths, providing detailed information about the bottle surface material and condition. After image acquisition, data preprocessing is performed, first using a denoising algorithm to eliminate image interference caused by environmental or equipment noise. Common denoising methods include median filtering and mean filtering, which help remove random noise from the image, resulting in a clearer image. Next, illumination equalization is performed to ensure uniform brightness distribution across all image bands, minimizing image quality variations caused by uneven lighting conditions. Common techniques include histogram equalization. After preprocessing, high-resolution images are obtained for each band, containing information on the bottle's surface reflectivity. These images can be used to further analyze the bottle surface for defects, areas of stress concentration, and other issues, providing foundational data for subsequent risk assessment and quality analysis.

[0115] Step S235: extracting pixel point spectral feature vectors based on the multispectral image of the bottle, and constructing a pixel-level spectral feature matrix based on the pixel point spectral feature vectors;

[0116] In this embodiment, a multispectral image of the bottle is processed at the pixel level to extract a spectral feature vector from each pixel. The spectral feature vector for each pixel contains the reflectance values for that pixel in different wavelength bands. Specifically, for each pixel, reflectance data corresponding to each designated wavelength band (e.g., multiple wavelength bands between 400 nm and 1000 nm) is first read from the multispectral image. Assuming the image contains reflectance information for multiple wavelength bands, extracting the reflectance values for each pixel in these wavelength bands yields a multidimensional vector representing the spectral characteristics of that pixel. Next, the spectral data for each pixel in the image is converted into a vector, and these vectors together form a data set. For each row of pixels in the image, the extracted spectral feature vector represents the spectral feature data for that pixel. All extracted pixel spectral feature vectors are organized to construct a pixel-level spectral feature matrix. Each row of this matrix represents a pixel in the image, while each column corresponds to the reflectance value for a wavelength band. Each element in the matrix represents the reflectance value for that pixel in a specific wavelength band. This matrix contains spectral information for the entire image, reflecting the reflectance characteristics of different regions and types, providing a data foundation for subsequent analysis. To further process and analyze this matrix, cluster analysis methods such as K-means or hierarchical clustering can be used to group similar spectral features and identify different regions or feature types. This matrix is not only used for subsequent spectral anomaly detection but also provides essential data support for defect analysis. By processing this matrix, different material or structural issues on the bottle surface can be detected, providing a scientific basis for subsequent quality control and defect diagnosis.

[0117] Step S236: identifying abnormal spectral bands based on the pixel-level spectral feature matrix and calculating the band mutation coefficient;

[0118] In this embodiment, a statistical analysis is first performed on each band in the pixel-level spectral feature matrix. Specifically, for each band, the mean and standard deviation of the reflectance values of all pixels within it are calculated. The mean reflects the average reflectance of the band, while the standard deviation reflects the fluctuation in reflectance within the band. These statistics can be used to identify regions with significant reflectance variations across bands. To identify spectral anomalous bands, a standard deviation threshold is set (for example, a band with a standard deviation greater than 0.1). When the standard deviation of a band exceeds this threshold, it is considered to have a significant reflectance variation and is therefore considered an anomalous band. Next, a mutation coefficient is calculated for each anomalous band. The mutation coefficient is calculated by comparing the degree of reflectance variation between each pixel in the band. Specifically, the mutation coefficient can be calculated by comparing the reflectance differences between adjacent pixels and calculating the ratio of these differences, thereby quantifying the severity of the reflectance variation. A large reflectance difference, coupled with a large mutation coefficient, indicates a sudden change in reflectance in that region, indicating a surface defect. By calculating the mutation coefficient, abnormal spectral bands can be identified, and it can be further determined whether these bands correspond to defective areas on the bottle surface, providing a basis for subsequent quality analysis and defect location.

[0119] Step S237: Calculate the reflectivity gradient based on the multispectral image of the bottle; determine the surface glass peeling according to the band mutation coefficient and the reflectivity gradient, and obtain the surface glass peeling data.

[0120] In this embodiment, the reflectivity gradient of each pixel is first calculated—that is, the rate of change in reflectivity between adjacent pixels. Specifically, for each pair of adjacent pixels in the image, the difference in their reflectivity at the same wavelength is calculated and normalized by the distance between the two pixels to obtain the reflectivity gradient. Regions with large reflectivity gradients typically indicate significant reflectivity variation, indicating an uneven surface structure or defects. Next, by combining the mutation coefficient obtained in the previous step, regions with significant reflectivity variation and meeting the characteristics of spalling risk are further screened. Specifically, when a region has a large reflectivity gradient and its mutation coefficient exceeds a set threshold, it is considered to be at risk for surface glass spalling. Based on these criteria, these regions are marked as surface glass spalling areas. Finally, the coordinates of these spalling areas and corresponding defect information, including the extent and depth of the spalling, are recorded to obtain surface glass spalling data. This data will provide important information for subsequent quality analysis, defect remediation, and traceability, helping to analyze whether there are quality issues in the bottle production process and providing a reference for subsequent improvement measures.

[0121] Preferably, the abnormal detection of uneven wall thickness in step S3 is specifically as follows:

[0122] Identify low-strength areas of the glass bottle based on the structural strength of the bottle;

[0123] In this embodiment, in this step, it is first necessary to collect the material strength data and bottle structure data of the glass bottle. The structural strength threshold of the bottle body is set to 50MPa, and all areas below this strength value are considered low-strength areas. Finite element analysis (FEA) is used to model the structure of the glass bottle, set material properties (such as the elastic modulus of glass is 72GPa, and the Poisson's ratio is 0.22), and the shape and size of the bottle body are taken into account during the modeling process. By simulating the impact of external forces (such as impact, pressure, etc.) on the bottle body, the stress value of each area is calculated and compared with the set strength threshold to identify low-strength areas. During the calculation, the geometric dimensions of the glass bottle and the preset material parameters are used, and based on the stress distribution map, the location of the low-strength area is obtained as the basic data for subsequent operations.

[0124] According to the low-strength area of the glass bottle body, the wall thickness area is meshed to obtain the bottle body mesh area;

[0125] In this embodiment, the bottle surface is scanned, and a laser scanner is used to perform a three-dimensional modeling of the glass bottle to generate three-dimensional point cloud data of the bottle. This point cloud data is imported into the meshing software, and meshing is performed based on the geometric structure of the bottle and the distribution of low-intensity areas. The grid unit size is set to 25mm×25mm to ensure that each grid unit can accurately reflect the changes in wall thickness. The bottle is evenly meshed, and the meshed area after division includes all surfaces of the bottle. It is ensured that the low-intensity area is consistent with the mesh division to facilitate subsequent wall thickness measurement. Ultimately, a gridded data containing all surfaces of the bottle and low-intensity areas is obtained, which serves as the basis for wall thickness measurement.

[0126] Measure the wall thickness of the glass bottle based on the grid area of the bottle body to obtain the wall thickness data of the glass bottle;

[0127] In this embodiment, wall thickness measurements are performed on each grid cell using either X-ray tomography (CT) or ultrasonic measurement within the divided grid area. CT scanning generates a cross-sectional image of each grid cell, which is then analyzed using an image analysis algorithm to measure the wall thickness of each grid cell. For ultrasonic measurement, a sensor is placed on the bottle surface and the reflection time of ultrasonic pulses is measured, generating wall thickness data for each grid cell. All grid cells are measured one by one, ultimately generating wall thickness data for each grid area and recording this data as input for subsequent wall thickness deviation calculations.

[0128] Calculate the wall thickness deviation based on the glass bottle wall thickness data; identify the wall thickness deviation area based on the wall thickness deviation;

[0129] In this embodiment, first, a standard wall thickness value is set (for example, the designed wall thickness of the bottle body is 4mm), the actual wall thickness of each grid area is compared with the standard wall thickness, and the wall thickness deviation of each area is calculated. The calculation formula for wall thickness deviation is: wall thickness deviation = actual wall thickness - standard wall thickness. For each grid unit, the wall thickness deviation data of each grid unit is calculated by the difference with the standard wall thickness. The wall thickness deviations of all grid areas are recorded and summarized to form an overall wall thickness deviation data set, which is convenient for subsequent identification of wall thickness deviation areas. First, a threshold range for wall thickness deviation is set. Specifically, when the wall thickness deviation is greater than 0.5mm, the area is considered to have obvious wall thickness deviation and needs to be marked. By traversing all grid areas, the wall thickness deviation of each grid is calculated one by one, and the area where the deviation value exceeds the set threshold is identified and marked as a wall thickness deviation area. The identification of the wall thickness deviation area is based on the size of the deviation value, combined with the actual wall thickness data of each grid, so as to facilitate subsequent more in-depth analysis and classification.

[0130] According to the wall thickness deviation area, the wall thickness deviation type is divided to obtain the vertical wall thickness deviation data and the circumferential wall thickness deviation data;

[0131] In this example, by further analyzing the wall thickness deviation areas and based on the bottle's geometric structure, wall thickness deviations are classified according to the vertical and circumferential directions. In this step, the vertical and circumferential wall thickness deviation thresholds are set at 0.3 mm, and the wall thickness deviation data for each direction is extracted separately. By classifying the wall thickness deviations in each direction, vertical and circumferential wall thickness deviation data are obtained. Vertical deviations are analyzed on vertical cross-sections of the bottle, while circumferential deviations are analyzed on horizontal cross-sections. This data provides a basis for subsequent anomaly identification and eccentricity detection.

[0132] Based on the vertical wall thickness deviation data, gravity casting anomaly identification is performed to obtain gravity casting anomaly data;

[0133] In this embodiment, a gravity casting deviation threshold must first be set. This threshold can be adjusted based on experience or historical data to ensure the accuracy of the identification results. The set threshold, for example, 0.5 mm, serves as the standard for determining gravity casting anomalies. When the vertical wall thickness deviation value in a certain area exceeds this threshold, and the direction of this deviation aligns with the direction of gravity on the bottle body, that area is considered to have a gravity casting anomaly. During the analysis process, the direction of gravity on the bottle body must first be determined. This can be determined using the bottle body's geometric model and the gravity direction during the manufacturing process. Next, the vertical wall thickness deviation of each area is compared with the gravity direction to identify areas with large wall thickness deviation values and a direction that aligns with the gravity direction. These areas are more likely to exhibit gravity casting anomalies. In this way, defective areas in the bottle body manufacturing process can be accurately located. Finally, these areas identified as gravity casting anomalies are recorded and included in the anomaly dataset to facilitate subsequent analysis, correction, or optimization of the manufacturing process. This gravity casting anomaly data can provide valuable information for further quality control.

[0134] Wall thickness eccentricity detection is performed based on the circumferential wall thickness deviation data to obtain the vertical wall thickness eccentricity;

[0135] In this embodiment, it is first necessary to perform wall thickness eccentricity detection based on the circumferential wall thickness deviation data. Circumferential wall thickness deviation data is obtained by measuring the difference in wall thickness at different circumferential locations on the bottle body. Typically, wall thickness measurements are performed at multiple circumferential locations on the bottle body to obtain wall thickness data for each location. Next, a calculation formula for the wall thickness eccentricity is set. Specifically, the eccentricity value of each circumferential region can be obtained by calculating the difference between the maximum and minimum wall thickness values for that region. The formula can be expressed as: Eccentricity = Maximum Wall Thickness - Minimum Wall Thickness. The key to this step is to accurately extract the wall thickness data for each circumferential region and ensure that the wall thickness is measured at multiple points within each circumferential region to calculate an accurate eccentricity value. After calculating the eccentricity values, these values need to be compared with the circumferential uniformity of the bottle body to determine the eccentricity of each circumferential region. The eccentricity can be calculated using the following formula: Eccentricity = Eccentricity / Circumferential Average Wall Thickness. The circumferential average wall thickness is the average of the wall thickness values of all measurement points within each circumferential region. This calculation helps measure the wall thickness uniformity of each circumferential region. A larger eccentricity indicates a more uneven wall thickness and more pronounced eccentricity in that region. Finally, based on the eccentricity calculated for each circumferential region, the eccentricities of all circumferential regions are summed to obtain the overall vertical wall thickness eccentricity. The vertical wall thickness eccentricity can be calculated by taking a weighted average of the eccentricities of each circumferential region or directly calculating the standard deviation. This metric provides an overall picture of the bottle's wall thickness eccentricity, reflecting the uniformity of the bottle's geometry and wall thickness distribution, and providing a basis for subsequent quality analysis and optimization.

[0136] The wall thickness unevenness data was obtained by integrating the gravity casting anomaly data and the vertical wall thickness eccentricity.

[0137] In this example, gravity casting anomaly data and vertical wall thickness eccentricity are integrated to generate wall thickness unevenness data. Finally, the gravity casting anomaly data and vertical wall thickness eccentricity data are integrated and weighted averaging is used to generate final wall thickness unevenness data. This data demonstrates the wall thickness unevenness problem that exists in the glass bottle manufacturing process and provides a basis for subsequent quality control and defect tracing.

[0138] Preferably, the evaluation of the core-cavity alignment offset in step S3 is specifically as follows:

[0139] Count the time periods of uneven wall thickness of glass bottles based on uneven wall thickness data;

[0140] In this embodiment, first, the wall thickness data of the glass bottles at various time points during the production process are collected and sorted. These data can be obtained through CT scanning, ultrasonic measurement or real-time monitoring by optical sensors. For each glass bottle, the wall thickness deviation during the production process is recorded, and the degree of uneven wall thickness of each bottle body is calculated. A standard threshold for uneven wall thickness is set, for example, the part with a deviation greater than 0.3 mm is considered uneven. Using a time series analysis method, based on the wall thickness unevenness data at each time point, the wall thickness unevenness within the production cycle is statistically analyzed, and time periods with significant wall thickness unevenness problems are identified. For example, if the deviation of the bottle wall thickness within a certain time period is generally greater than the set standard threshold, the time period is marked as an "uneven wall thickness time period." The significance of the uneven phenomenon is confirmed by calculating the mean value, standard deviation and other statistical quantities of the wall thickness deviation values within each time period. Record and output these statistical results as the basis for subsequent analysis.

[0141] Extract the operating parameters of the forming mold according to the time period of uneven wall thickness of the glass bottle;

[0142] In this embodiment, the mold operating parameters related to the molding process within the time period of uneven wall thickness obtained in the previous step are extracted. By monitoring the production system, the various operating parameters of the mold within these time periods, such as mold heating temperature, cooling time, injection pressure, injection speed, etc., are obtained. These parameters are usually collected in real time by sensors connected to the molding equipment and recorded in the production control system. The set parameter range includes mold temperature between 100°C and 200°C, injection pressure between 8MPa and 15MPa, and injection speed between 30mm / s and 60mm / s. The mold operating parameters within each time period are extracted, all relevant parameters are recorded, and they are screened according to the time period of uneven wall thickness. Finally, a set of mold operating parameters related to uneven wall thickness are obtained for subsequent thermal expansion analysis.

[0143] Identify the thermal expansion state based on the operating parameters of the molding die and obtain the thermal expansion data of the die;

[0144] In this embodiment, the thermal expansion state of the mold is identified by analyzing the operating parameters of the molding mold during operation. First, a thermal expansion coefficient is set (for example, the thermal expansion coefficient of the mold material is 12×10^-6 / °C). This coefficient describes the volumetric expansion of the mold during heating. Based on the extracted mold temperature change data, the temperature range of the mold during heating is calculated. For example, the mold temperature increases from 25°C to 180°C during the heating phase. Combined with the thermal expansion coefficient, the amount of mold expansion due to temperature changes during heating is calculated according to the formula ΔL = L0 * α * ΔT, where ΔL is the thermal expansion of the mold, L0 is the initial length, α is the thermal expansion coefficient, and ΔT is the temperature change. This process generates mold thermal expansion data, reflecting the degree of mold expansion caused by temperature changes. Recording this thermal expansion data provides the necessary input information for subsequent mold deformation analysis.

[0145] Calculate the mold thermal expansion deformation according to the mold thermal expansion data;

[0146] In this embodiment, the thermal expansion deformation of the mold is calculated based on thermal expansion data obtained from the forming mold. First, the mold's geometric dimensions and material properties, such as its initial length, width, height, and elastic modulus, are determined. By applying thermal expansion theory and combining it with the geometry of each mold component, the deformation of the mold during the heating process is calculated. Based on the mold's thermal expansion and material properties, the thermal expansion deformation of each mold component can be calculated, such as the deformation of a specific mold component along the X-axis and the deformation of a specific mold component along the Y-axis. The calculation formula is: Thermal expansion deformation = initial dimensions × thermal expansion coefficient × temperature change. Key parameters in this step include the mold's initial dimensions (e.g., a mold length of 1 meter and a width of 0.5 meter) and the thermal expansion coefficient (e.g., the mold material's thermal expansion coefficient is 12 × 10^-6 / °C). Using these parameters, the specific deformation of the mold during the heating phase is calculated, generating mold thermal expansion deformation data that serves as the basis for further prediction of mold deformation.

[0147] Predict the relative deformation probability of the core and cavity based on the thermal expansion deformation of the mold;

[0148] In this embodiment, the relative deformation probability of the core and the cavity is predicted based on the thermal expansion deformation of the mold. In this step, a geometric model between the core and the cavity is first established, and the relative displacement between the core and the cavity is calculated based on the thermal expansion data of the mold. A standard deformation value is set. For example, when the relative deformation value between the core and the cavity exceeds 0.5 mm, it is considered that significant deformation has occurred. The deformation of the core and the cavity during the heating process is obtained through the thermal expansion deformation calculation model, and the relative deformation probability of the core and the cavity is predicted using a probability theory method based on the distribution of the deformation. These probability data reflect the possibility of deformation between the core and the cavity under specific thermal expansion conditions, providing data support for mold design and adjustment.

[0149] The core-cavity alignment offset is determined based on the relative deformation probability of the core-cavity.

[0150] In this embodiment, the relative deformation probability data of the core and cavity are combined to analyze and calculate the offset between the core and cavity. First, based on the relative deformation probability calculated in the previous step, the areas with a greater possibility of deformation are determined, and the actual offset of the core and cavity is calculated in these areas. For example, if the relative deformation probability of a certain mold area is greater than 90%, there is a large offset in this area. Based on the tolerance requirements in the mold design, the maximum allowable offset is calculated (for example, the maximum allowable offset is 0.2mm). The relative deformation probability is combined with the allowable offset to obtain the actual offset of the core and cavity. These data can be used to adjust the mold design and reduce quality problems caused by deformation.

[0151] Preferably, the positioning pin fracture detection in step S3 is specifically as follows:

[0152] Evaluate mold assembly accuracy based on core-cavity alignment offset;

[0153] In this embodiment, first, the alignment offset data of the mold core and the mold cavity at different time points during the production process are collected. This data can be measured by a high-precision displacement sensor or a three-dimensional laser scanner to obtain the relative displacement information of the mold core and the mold cavity. Specifically, the measurement points are set and the displacement measurements are performed at the key positions of the mold (such as the contact area between the mold cavity and the mold core). Assuming that the set allowable error is 0.1mm, if the actual measured offset exceeds this threshold, it is considered that there is a problem with the mold assembly accuracy. By comparing the offset data in different time periods, the errors in the mold assembly process are identified and the assembly accuracy is calculated. For example, if the maximum alignment offset between the mold core and the mold cavity is 0.15mm, which exceeds the standard error range, it means that the mold assembly accuracy is unqualified.

[0154] Perform positioning pin load detection based on mold assembly accuracy to obtain positioning pin load data;

[0155] In this embodiment, the load condition of the locating pin is detected based on the assembly accuracy of the mold. A load sensor (such as a piezoelectric sensor or a strain gauge) is installed on the locating pin to monitor in real time the load borne by the locating pin during operation. A standard load value range is set. For example, the maximum load capacity of the locating pin during operation is 500N. By measuring the load data of the locating pin under different working conditions, the load values of the locating pin at different production stages are obtained. This data reflects the load status of the locating pin and provides a basis for subsequent durability analysis. If the load of the locating pin exceeds the set maximum load threshold of 500N in a certain measurement, it will be recorded as abnormal load data.

[0156] According to the positioning pin load data, the positioning pin static load data and positioning pin dynamic load data are obtained;

[0157] In this embodiment, the load data is analyzed and classified based on the temporal changes in the load of the positioning pin. First, the distinction criteria between static load and dynamic load are determined. The standard for static load is set as the load maintaining a constant state for a certain period of time, while the standard for dynamic load is the load changing rapidly over time. Through signal processing techniques, such as Fourier transform, the load signal is decomposed into low-frequency (static) and high-frequency (dynamic) parts. The load data is analyzed in the frequency domain to extract the data of static load and dynamic load. For example, if the load of the positioning pin does not change by more than 10N within a certain period of time, it is regarded as a static load; if the load changes by more than 10N and the frequency is high, it is regarded as a dynamic load. According to this method, static load data and dynamic load data are obtained respectively.

[0158] Perform material durability analysis based on the static load data of the positioning pin to obtain material durability data;

[0159] In this embodiment, first, the durability standard of the material is set. For example, the material used for the locating pin must have a durability of at least 1 million static load cycles. Based on the static load data (such as a single load of 200N and a load duration of 2 hours), the material fatigue test standard (such as the SN curve) is used for analysis to calculate the fatigue life of the material under different loads. For example, if the static load is maintained at 200N for 2 hours and the fatigue life of the material under this load is 500,000 times, the durability data of the material is 500,000 times. In addition, according to different load levels, the durability of the material can be further divided, such as 800,000 times under low load, 600,000 times under medium load, and 400,000 times under heavy load.

[0160] Detect vibration amplitude based on dynamic load data of positioning pins;

[0161] In this embodiment, a vibration sensor (such as an accelerometer or a vibration sensor) is used to detect the vibration amplitude of the locating pin based on the dynamic load data. First, the detection range of the vibration amplitude is set. For example, if the vibration amplitude exceeds 5 mm / s, it is an abnormal vibration. The vibration sensor is installed near the locating pin, and the vibration amplitude is calculated by collecting the vibration data of the locating pin in the working state in real time. The processing method of the vibration signal is set, including removing background noise and signal filtering. If the vibration amplitude data of the locating pin exceeds 5 mm / s during operation, it is considered that the vibration amplitude exceeds the normal range. By statistically analyzing the vibration data, the distribution of the vibration amplitude is obtained and used for subsequent impact tolerance evaluation.

[0162] Evaluate the impact tolerance of the positioning pin based on the vibration amplitude to obtain the impact tolerance data of the positioning pin;

[0163] In this embodiment, the impact tolerance of the locating pin is evaluated based on the vibration amplitude data. First, the impact tolerance standard is set. For example, the maximum impact acceleration that the locating pin can withstand is 1000m / s². If the vibration amplitude exceeds the set value of 5mm / s, the impact component in the vibration signal is further analyzed. Through high-speed sampling and fast Fourier transform, the vibration signal is decomposed into an impact signal and a normal vibration signal. The impact energy in the vibration signal is calculated using the impact energy formula (such as E = 1 / 2 * m * v²) to further evaluate the impact tolerance of the locating pin in the working state. If the calculated impact energy exceeds the set maximum impact energy standard, the impact tolerance data of the locating pin indicates that its tolerance is insufficient.

[0164] The positioning pin fracture prediction is performed based on the material durability data and the positioning pin impact tolerance data to obtain the positioning pin fracture data.

[0165] In this embodiment, the material durability data of static load and the impact tolerance data of dynamic load are combined to predict the fracture of the locating pin. A comprehensive indicator is set, for example, the fracture probability of the locating pin is related to factors such as the fatigue life of the material and the impact energy. According to the above analysis, if the durability of the material is less than 800,000 times and the impact energy exceeds the set threshold, the fracture risk of the locating pin increases. A stress analysis method (such as finite element analysis) is used to simulate the stress distribution of the locating pin under different loads to predict the probability of its fracture. For example, if the predicted fracture probability is greater than 5%, it is considered that the locating pin has a high fracture risk. Finally, the fracture data of the locating pin is obtained and used to optimize the design or provide early warning.

[0166] Preferably, this specification also provides a digital traceability system for lightweight glass bottles, which is used to execute the digital traceability method for lightweight glass bottles as described above. The digital traceability system for lightweight glass bottles includes:

[0167] The bottle body structural strength detection module is used to obtain glass bottle design data and build a lightweight glass bottle model; based on the lightweight glass bottle model, the bottle body structural strength is detected;

[0168] The defect tracing module is used to detect bottle surface defects based on the bottle structural strength and generate bottle surface defect data; based on the bottle surface defect data, defect tracing is performed to obtain bottle production defect process data;

[0169] The positioning pin fracture detection module is used to detect uneven wall thickness based on the structural strength of the bottle body and obtain uneven wall thickness data; evaluate the core-cavity alignment offset based on the uneven wall thickness data; and perform positioning pin fracture detection based on the core-cavity alignment offset to obtain positioning pin fracture data;

[0170] The glass quality traceability module is used to generate a unique glass number based on the lightweight glass bottle model; identify abnormal batches of lightweight glass bottles based on the bottle body production defect process data and the positioning pin breakage data to obtain abnormal batch data; map the abnormal batch data to the abnormal glass number based on the unique glass number to obtain the abnormal glass number, and transmit it to the production equipment system to perform the glass quality traceability task.

[0171] The present invention is therefore intended to be illustrative and non-restrictive in all respects, with the scope of the invention being defined by the appended claims rather than the foregoing description, and all changes that come within the meaning and range of equivalents of the application documents are intended to be embraced therein.

[0172] The foregoing description is intended only to provide specific embodiments of the present invention, which will enable those skilled in the art to understand and implement the present 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 present invention. Therefore, the present invention is not intended to be limited to the embodiments shown herein, but is to be construed in the widest possible manner consistent with the principles and novel features disclosed herein.

Claims

1. A digital traceability method for lightweight glass bottle quality, characterized in that: The following steps are involved: Step S1: Obtain glass bottle design data and build a lightweight glass bottle model; Detect the structural strength of the bottle based on the lightweight glass bottle model; Step S2: Detecting bottle surface defects based on the bottle structural strength to generate bottle surface defect data; Based on the bottle surface defect data, defect traceability is performed to obtain the bottle production defect process data; Step S3: performing wall thickness unevenness detection based on the bottle body structural strength to obtain wall thickness unevenness data; Evaluate the core-cavity alignment offset based on the wall thickness unevenness data; perform locating pin fracture detection based on the core-cavity alignment offset to obtain locating pin fracture data; Step S4: Generate a unique glass number based on the lightweight glass bottle model; identify abnormal batches of lightweight glass bottles based on the bottle body production defect process data and the positioning pin breakage data to obtain abnormal batch data; Mapping the abnormal glass number to the abnormal batch data according to the unique glass number to obtain the abnormal glass number, and transmitting it to the production equipment system to perform the glass quality traceability task. Step S4 is specifically as follows: Step S41: Generate a unique glass number based on the lightweight glass bottle model; Step S42: performing normalization processing on the bottle production defect process data to obtain normalized production defect process data; Step S43: marking the mold cavity number based on the positioning pin fracture data; Step S44: performing time correlation analysis based on the normalized production defect process data and the mold cavity number to obtain defect process-mold cavity number data; Step S45: Identify abnormal production batches based on the defective process-cavity number data to obtain abnormal batch data; Step S46: Mapping the abnormal glass number to the abnormal batch data according to the unique glass number to obtain the abnormal glass number, and transmitting it to the production equipment system to perform the glass quality traceability task.

2. The digital traceability method for lightweight glass bottle quality according to claim 1 is characterized in that: Step S1 is specifically as follows: Step S11: obtaining glass bottle design data and extracting glass bottle three-dimensional structure data; Step S12: identifying key optimization areas based on the three-dimensional structural data of the glass bottle to obtain the bottle body area and the bottle shoulder area; Step S13: performing curvature transition optimization design based on the bottle body area to obtain the optimized curvature of the bottle body; Step S14: performing wall thickness thinning design based on the bottle shoulder area to obtain wall thickness thinning data; Step S15: Integrate the bottle body optimized curvature and wall thickness reduction data, and construct a lightweight glass bottle model; Step S16: Detecting the structural strength of the bottle body based on the lightweight glass bottle model.

3. The digital traceability method for lightweight glass bottle quality according to claim 2 is characterized in that: Step S16 is specifically as follows: Step S161: importing the lightweight glass bottle model into the simulation software; Step S162: setting the bottle body wall thickness range to 0.3mm-3mm, the bottle neck thickness range to 0.2mm-1.5mm, and the bottle bottom thickness range to 0.5mm-4mm in the simulation software; Step S163: in the simulation software, set the external pressure range to 0.5 MPa-2 MPa, the internal pressure range to 0.3 MPa-1.5 MPa, and the transport impact force range to 10 N-200 N; Step S164: setting the elastic modulus of the glass material to a range of 60 GPa-75 GPa and the Poisson's ratio to a range of 0.22-0.26 in the simulation software; Step S165: Run the structural strength analysis program in the simulation software and record the structural strength of the bottle.

4. The digital traceability method for lightweight glass bottle quality according to claim 1 is characterized in that: Step S2 is specifically as follows: Step S21: extracting bottle deformation characteristics and bottle pressure resistance characteristics based on the bottle structural strength to obtain bottle deformation data and bottle pressure resistance data; Step S22: performing bottle body crack detection based on the bottle body deformation data to obtain bottle body crack data; Step S23: performing a transportation simulation based on the bottle pressure resistance data, and identifying surface glass peeling during the bottle transportation process, and obtaining surface glass peeling data; Step S24: Integrate the bottle body crack data and the surface glass peeling data to obtain bottle body surface defect data; Step S25: performing defect tracing based on the bottle surface defect data to obtain bottle production defect process data.

5. The digital traceability method for lightweight glass bottle quality according to claim 4 is characterized in that: Step S22 is specifically as follows: Step S221: performing bottle rotation control based on the bottle deformation data to obtain bottle rotation data; Step S222: Calculating stress gradients based on the bottle rotation data, and extracting high stress gradient values based on the stress gradients; Step S223: identifying high deformation areas of the bottle body according to high stress gradient values; Step S224: performing an X-ray circular irradiation scan on the high deformation area of the bottle body to obtain an X-ray image of the bottle body; Step S225: extracting significant boundaries based on the bottle X-ray image to obtain significant boundary data; Step S226: Calculating the aspect ratio based on the significant boundary data to obtain the aspect ratio; Step S227: identifying the crack profile according to the aspect ratio; Step S228: Calculate the crack defect density based on the crack profile to obtain bottle body crack data.

6. The digital traceability method for lightweight glass bottle quality according to claim 4 is characterized in that: Step S23 is specifically as follows: Step S231: screening low-pressure bottles based on the bottle pressure resistance data, and performing transportation simulation to obtain low-pressure bottle transportation data; Step S232: performing force analysis based on the low-pressure bottle transportation data to obtain bottle transportation force data; Step S233: identifying stress concentration areas based on the bottle transportation force data and recording them as peeling risk areas; Step S234: acquiring a multispectral image based on the peeling risk area to obtain a multispectral image of the bottle body; Step S235: extracting pixel point spectral feature vectors based on the multispectral image of the bottle, and constructing a pixel-level spectral feature matrix based on the pixel point spectral feature vectors; Step S236: identifying abnormal spectral bands based on the pixel-level spectral feature matrix and calculating the band mutation coefficient; Step S237: Calculate the reflectivity gradient based on the multispectral image of the bottle; determine the surface glass peeling according to the band mutation coefficient and the reflectivity gradient, and obtain the surface glass peeling data.

7. The digital traceability method for lightweight glass bottle quality according to claim 1 is characterized in that: The specific detection of uneven wall thickness abnormality in step S3 is as follows: Identify low-strength areas of the glass bottle based on the structural strength of the bottle; According to the low-strength area of the glass bottle body, the wall thickness area is meshed to obtain the bottle body mesh area; Measure the wall thickness of the glass bottle based on the grid area of the bottle body to obtain the wall thickness data of the glass bottle; Calculate the wall thickness deviation based on the glass bottle wall thickness data; identify the wall thickness deviation area based on the wall thickness deviation; According to the wall thickness deviation area, the wall thickness deviation type is divided to obtain the vertical wall thickness deviation data and the circumferential wall thickness deviation data; Based on the vertical wall thickness deviation data, gravity casting anomaly identification is performed to obtain gravity casting anomaly data; Wall thickness eccentricity detection is performed based on the circumferential wall thickness deviation data to obtain the vertical wall thickness eccentricity; The wall thickness unevenness data was obtained by integrating the gravity casting anomaly data and the vertical wall thickness eccentricity.

8. The digital traceability method for lightweight glass bottle quality according to claim 1 is characterized in that: The evaluation of the core-cavity alignment offset in step S3 is specifically as follows: Count the time periods of uneven wall thickness of glass bottles based on uneven wall thickness data; Extract the operating parameters of the forming mold according to the time period of uneven wall thickness of the glass bottle; Identify the thermal expansion state based on the operating parameters of the molding die and obtain the thermal expansion data of the die; Calculate the mold thermal expansion deformation according to the mold thermal expansion data; Predict the relative deformation probability of the core and cavity based on the thermal expansion deformation of the mold; The core-cavity alignment offset is determined based on the relative deformation probability of the core-cavity.

9. The digital traceability method for lightweight glass bottle quality according to claim 1 is characterized in that: The specific steps of detecting the fracture of the positioning pin in step S3 are as follows: Evaluate mold assembly accuracy based on core-cavity alignment offset; Perform positioning pin load detection based on mold assembly accuracy to obtain positioning pin load data; According to the positioning pin load data, the positioning pin static load data and positioning pin dynamic load data are obtained; Perform material durability analysis based on the static load data of the positioning pin to obtain material durability data; Detect vibration amplitude based on dynamic load data of positioning pins; Evaluate the impact tolerance of the positioning pin based on the vibration amplitude to obtain the impact tolerance data of the positioning pin; The positioning pin fracture prediction is performed based on the material durability data and the positioning pin impact tolerance data to obtain the positioning pin fracture data.

10. A digital traceability system for lightweight glass bottle quality, characterized in that: The method for digitally tracing the quality of lightweight glass bottles according to claim 1 is used to implement the method. The system for digitally tracing the quality of lightweight glass bottles comprises: The bottle body structural strength detection module is used to obtain glass bottle design data and build a lightweight glass bottle model; based on the lightweight glass bottle model, the bottle body structural strength is detected; The defect tracing module is used to detect bottle surface defects based on the bottle structural strength and generate bottle surface defect data; based on the bottle surface defect data, defect tracing is performed to obtain bottle production defect process data; The positioning pin fracture detection module is used to detect uneven wall thickness based on the structural strength of the bottle body and obtain uneven wall thickness data; evaluate the core-cavity alignment offset based on the uneven wall thickness data; and perform positioning pin fracture detection based on the core-cavity alignment offset to obtain positioning pin fracture data; The glass quality traceability module is used to generate a unique glass number based on the lightweight glass bottle model; identify abnormal batches of lightweight glass bottles based on the bottle body production defect process data and the positioning pin breakage data to obtain abnormal batch data; map the abnormal batch data to the abnormal glass number based on the unique glass number to obtain the abnormal glass number, and transmit it to the production equipment system to perform the glass quality traceability task.

Images