Long glass fiber compression molding method and system

Abstract

The invention discloses a long glass fiber compression molding method and system, and the method comprises the steps: laying a continuous glass fiber plate and a long glass fiber material ball in a mold cavity in a partitioned manner, and obtaining a positioned composite green body; local prepressing is applied to the interlocking area of the composite blank, a mechanical meshing structure is formed, and a prefabricated body with an interface interlocking structure is obtained; establishing a pressure distribution model according to the geometric characteristics of the product to obtain differentiated pressure parameter combinations; placing the preform in a heating mold, and applying pressure in stages according to the pressure parameter combination to obtain a semi-finished product without glass fiber exposure; and for the semi-finished product, under the state that the second pressure value is kept, rapid cooling is conducted on the flange and the corner area, slow cooling is conducted on the large face area, and a final product with the balanced internal stress is obtained. By utilizing the embodiment of the invention, the efficient synergy of the continuous glass fiber plate and the long glass fiber material block can be realized, and the mechanical property and the structural integrity of the product are improved.

Description

Technical Field

[0001] The invention belongs to the technical field of compression molding, and in particular to a long glass fiber compression molding method and system. Background Art

[0002] Traditional long glass fiber compression molding processes typically utilize a single material layup or uniform pressure molding, resulting in uneven product performance across large surfaces and in complex structural areas such as flanges and corners. Continuous glass fiber sheets offer excellent mechanical properties across large surfaces, but are susceptible to fiber breakage or insufficient resin filling in complex geometric areas. While long glass fiber clumps can adapt to complex shapes, they struggle to achieve adequate overall strength when used alone. Furthermore, the single pressure distribution in existing technologies can easily lead to insufficient resin flow or uneven fiber distribution, which can lead to stress concentration within the product and exposed glass fibers. Summary of the Invention

[0003] The purpose of the present invention is to provide a long glass fiber molding method and system to address the deficiencies in the prior art, achieve efficient synergy between continuous glass fiber sheets and long glass fiber lumps, and improve the mechanical properties and structural integrity of the products.

[0004] One embodiment of the present application provides a long glass fiber compression molding method, the method comprising: Continuous glass fiber sheets and long glass fiber balls are laid in the mold cavity in different areas. The continuous glass fiber sheets cover the large surface area of ​​the product, and the long glass fiber balls are pre-placed in the flange and corner areas. The edges of the two overlap to form an interlocking area to obtain a fully positioned composite body. Applying local pre-compression to the interlocking area of ​​the composite body so that the long glass fiber balls are embedded in the fiber grid gaps of the continuous glass fiber sheet to form a mechanical bite structure, thereby obtaining a preform with an interface interlocking structure; A pressure distribution model is established based on the product's geometric characteristics. A first pressure value is generated for the large surface area, and a second pressure value is generated for the flange and corner areas. The second pressure value is greater than the first pressure value, resulting in a differentiated pressure parameter combination. Placing the preform in a heated mold and applying pressure in stages according to the pressure parameter combination, wherein a first pressure value is used to press the large surface area to allow the resin to melt and flow, and then a second pressure value is used to press the interlocking area to allow the molten resin to fill the fiber gaps, thereby obtaining a semi-finished product without exposed glass fibers; For the semi-finished product, while maintaining the second pressure value, the flange and corner areas are rapidly cooled, and the large surface area is slowly cooled to obtain a final product with balanced internal stress.

[0005] Optionally, the continuous glass fiber plate and the long glass fiber material group are laid in the mold cavity in a partitioned manner, wherein the continuous glass fiber plate covers the large area of the product, and the long glass fiber material group is prepositioned in the flange and corner area, and the edges of the two overlap to form an interlocking area, thereby obtaining a positioning completed composite blank, comprising: The curvature distribution characteristics of the product CAD model are extracted, the cavity surface is divided into a large surface low curvature area and a high stress flange / corner area by a curvature gradient algorithm, and a region boundary coordinate set is output; Based on the region boundary coordinate set, a dynamic laser grid is projected on the mold cavity surface to guide the mechanical arm to accurately lay the continuous glass fiber plate in the large area marked by the laser, and a plate positioning completion signal is generated; The fiber arrangement image of the plate surface is obtained by a micron-level optical scanner, the direction gradient histogram algorithm is used to calculate the average gap size of the fiber grid, and a gap distribution thermal map is output; According to the gap distribution thermal map, the long glass fiber material group is overlaid and laid at the edge of the plate in the flange and corner area, so that the material group covers the edge width of the plate by 1.5 times the average size of the gap, and a composite blank positioning report with an overlapping interlocking area is generated.

[0006] Optionally, the interlocking area of the composite blank is subjected to local pre-pressing, so that the long glass fiber material group is embedded in the fiber grid gap of the continuous glass fiber plate to form a mechanical interlocking structure, thereby obtaining a preform with an interfacial interlocking structure, comprising: Based on the three-dimensional topological graph of the composite blank, an infrared focusing heating device is deployed in the interlocking area to soften the surface layer resin of the material group to a viscous flow state, and a resin viscosity qualification signal is output; The resin viscosity qualification signal is received, a micro-press head array with a diameter matching the fiber gap is started, and a local pressure of 0.5 MPa is applied, and a real-time pressure distribution cloud map is output; The press head stroke is adjusted through the real-time pressure distribution cloud map, and when the fiber sensor detects that the embedding depth of the material group reaches 70% of the fiber diameter, an interlocking depth authentication code is output; Based on the interlocking depth authentication code, a micro-focus CT scan is triggered, a three-dimensional model of the interlocking area is reconstructed, and an interfacial interlocking structure authentication file is output.

[0007] Optionally, the pressure distribution model is established according to the geometric characteristics of the product, a first pressure value is generated for the large area, a second pressure value is generated for the flange and corner area, and the second pressure value is greater than the first pressure value, thereby obtaining a differentiated pressure parameter combination, comprising: The product geometric characteristic parameters are imported, a standard forming pressure is applied for nonlinear contact analysis, and an equivalent stress distribution matrix of each node in the cavity is output; The equivalent stress distribution matrix is subjected to regional clustering analysis, the flange and corner area with a stress value exceeding 85% of the material yield strength is screened, and a high pressure demand area coordinate set is generated; According to the material rheological database, the minimum forming pressure value of the large area region is solved as the first pressure reference value as a constraint condition of the resin critical flow shear rate; In the high pressure demand area coordinate set, the compensation pressure coefficient is dynamically calculated by the deep reinforcement learning model with the first pressure reference value as the input, and the second pressure value optimization scheme is output; Based on the first pressure reference value and the second pressure value optimization scheme, a spatially continuous pressure distribution function is constructed to generate a parameter combination configuration file containing a pressure gradient curve.

[0008] Optionally, the preform is placed in a heating mold, and the pressure is applied in stages according to the pressure parameter combination, wherein the first pressure value is used to press the large area region to make the resin melt and flow, and the second pressure value is used to press the interlocking area to make the molten resin fill the fiber gap, thereby obtaining a semi-finished product without glass fiber exposure, including: Read the parameter combination configuration file, set a 180℃ constant temperature field in the large area region of the heating mold, and set a 200℃ high temperature field in the interlocking area, and output a temperature field balance ready signal; After receiving the temperature field ready signal, the first pressure value is used to press the large area region, and when the resin front reaches the boundary of the interlocking area, a phase switching instruction is triggered through a melt flow sensor; Based on the phase switching instruction, the pressure is raised to the second pressure value within 0.2 seconds and maintained for 10 seconds, and the high pressure is used to force the molten resin to penetrate into the fiber gap, and a resin penetration completion marker is output; The surface of the product is scanned by a laser confocal microscope to verify that there is no glass fiber exposure and the porosity is less than 0.5%, and a semi-finished product quality certification code is generated.

[0009] Optionally, for the semi-finished product, the flange and corner regions are rapidly cooled and the large area region is slowly cooled under the condition of maintaining the second pressure value, thereby obtaining a final product with balanced internal stress, including: Under the trigger of the semi-finished product quality certification code, a liquid nitrogen injection module is deployed in the flange area of the mold, and a warm water circulation pipeline is enabled in the large area region, thereby generating a cooling system start instruction; According to the cooling system start instruction, the flange area is cooled to 80℃ at a rate of 50℃ / s and the large area region is cooled to 110℃ at a rate of 5℃ / s under the condition of maintaining the second pressure value, and an real-time temperature gradient map is output; According to the real-time temperature gradient map, the cooling rate is adjusted, and when the acoustic emission sensor detects that the characteristic frequency peak value decreases to a safety threshold value, a stress balance confirmation code is output; Based on the stress balance confirmation code, the ejection mechanism is controlled to uniformly unload the pressure and take out the product within 5 seconds, thereby obtaining a balanced product with a residual stress difference of less than 10MPa.

[0010] Another embodiment of the present application provides a long glass fiber compression molding system, the system comprising: The laying module is used to lay continuous glass fiber sheets and long glass fiber balls in the mold cavity in different areas. The continuous glass fiber sheets cover the large surface area of ​​the product, and the long glass fiber balls are pre-placed in the flange and corner areas. The edges of the two overlap to form an interlocking area to obtain a fully positioned composite body. An embedding module is used to apply local pre-compression to the interlocking area of ​​the composite body so that the long glass fiber material balls are embedded in the fiber grid gaps of the continuous glass fiber sheet to form a mechanical bite structure and obtain a preform with an interface interlocking structure; Establish a module for establishing a pressure distribution model based on the geometric characteristics of the product, generating a first pressure value for the large surface area and a second pressure value for the flange and corner area, and the second pressure value is greater than the first pressure value, thereby obtaining a differentiated pressure parameter combination; a pressurizing module, configured to place the preform in a heated mold and apply pressure in stages according to the pressure parameter combination, wherein a first pressure value is applied to press the large surface area to allow the resin to melt and flow, and a second pressure value is applied to press the interlocking area to allow the molten resin to fill the fiber gaps, thereby obtaining a semi-finished product without exposed glass fibers; The cooling module is used to quickly cool the flange and corner areas of the semi-finished product and slowly cool the large area while maintaining the second pressure value, so as to obtain a final product with balanced internal stress.

[0011] Yet another embodiment of the present application provides a storage medium, wherein the storage medium stores a computer program, wherein the computer program is configured to execute any of the above methods when run.

[0012] Yet another embodiment of the present application provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute any of the above methods.

[0013] Compared with the prior art, the present invention provides a long glass fiber compression molding method, which lays continuous glass fiber sheets and long glass fiber balls in a mold cavity in partitions to obtain a positioned composite blank; applies local pre-pressure to the interlocking areas of the composite blank to form a mechanical bite structure to obtain a preform with an interface interlocking structure; establishes a pressure distribution model based on the geometric characteristics of the product to obtain a differentiated pressure parameter combination; places the preform in a heated mold, and applies pressure in stages according to the pressure parameter combination to obtain a semi-finished product without exposed glass fiber; for the semi-finished product, while maintaining the second pressure value, the flange and corner areas are rapidly cooled, and the large surface area is slowly cooled to obtain a final product with balanced internal stress, thereby achieving efficient coordination between the continuous glass fiber sheets and the long glass fiber balls, and improving the mechanical properties and structural integrity of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A hardware structure block diagram of a computer terminal for a long glass fiber molding method provided by an embodiment of the present invention; Figure 2 A schematic diagram of a process for molding long glass fibers according to an embodiment of the present invention; Figure 3 A schematic structural diagram of a long glass fiber compression molding system provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and are not to be construed as limiting the present invention.

[0016] The embodiment of the present invention first provides a long glass fiber compression molding method, which can be applied to electronic equipment, such as computer terminals, specifically ordinary computers, etc.

[0017] The following describes it in detail by taking running on a computer terminal as an example. Figure 1 The hardware structure block diagram of a computer terminal for a long glass fiber molding method provided by an embodiment of the present invention. Figure 1 As shown, the computer device includes a processor, a memory, and a network interface connected via a system bus, wherein the memory may include a non-volatile storage medium and an internal memory.

[0018] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any long glass fiber molding method.

[0019] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0020] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any long glass fiber molding method.

[0021] The network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art will understand that Figure 1 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0022] It should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0023] See also Figure 2 , an embodiment of the present invention provides a long glass fiber molding method, which may include the following steps: S201, laying continuous glass fiber sheets and long glass fiber balls in the mold cavity in different areas, wherein the continuous glass fiber sheets cover the large surface area of ​​the product, and the long glass fiber balls are pre-placed in the flange and corner areas, and the edges of the two overlap to form an interlocking area, thereby obtaining a fully positioned composite body; Specifically, the curvature distribution characteristics of the product CAD model can be extracted, and the cavity surface can be divided into large-surface low-curvature areas and high-stress flange / corner areas using a curvature gradient algorithm, and the regional boundary coordinate set can be output; The system first imports the product's CAD model (Computer-Aided Design model), which contains the product's 3D geometric data (such as surface coordinates, normal vectors, and topological relationships). The core processing module uses the Curvature Gradient Algorithm, which, based on the principles of differential geometry, calculates the principal curvature value (PCV) of each tiny area on the model's surface. For example, for a flat area, the PCV is close to zero; for a circular corner with a radius of 5 mm, the PCV is as high as 0.2 mm. -1 The algorithm scans the entire model surface through a sliding window and calculates the curvature change rate (i.e., Curvature Gradient, CG) between adjacent windows. When CG exceeds a preset threshold (such as 0.05 mm), -2 ), it is marked as a high gradient transition area. Finally, the system divides the cavity surface into two functional areas: large surface low curvature area (such as the middle plane of the body shell, PCV < 0.01 mm) -1 ) and high stress flange / corner areas (such as bolted flanges or 90° corners, PCV>0.1mm -1The partitioning results are output as a set of three-dimensional spatial coordinates (such as an XYZ point cloud), forming a Regional Boundary Coordinate Set (RBCS). This file contains the coordinates and type labels of the contour boundary points of each region.

[0024] The implementation of the curvature gradient algorithm requires optimization based on engineering experience. For example, in an automotive fender mold, the algorithm identifies the side flange (the point of curvature change) as a high-stress area, while the central flat area is a low-curvature area. During the segmentation process, the system automatically filters out noise points (such as minor jagged edges caused by CAD modeling) and uses B-Spline curve fitting to smooth the boundary lines to ensure smoothness of the region boundaries. The algorithm also integrates a material mechanics knowledge base: based on historical data, when the local curvature radius is less than 8 mm, the area is prone to fiber breakage or resin accumulation defects during the molding process, and is therefore forcibly classified as a high-stress area. The final output, the Region Boundary Coordinate Set (RBCS), contains not only geometric coordinates but also metadata about the region attributes (such as "Flange Area - Maximum Stress Predicted 120 MPa"), providing a basis for subsequent material placement. This coordinate set is stored in a central database in a standardized format (such as STEP or IGES) for access by downstream equipment.

[0025] To ensure the accuracy of division, the system adopts a multi-level verification mechanism: Level 1 verification: Finite element pre-simulation (FEP) is used to apply virtual loads to the divided areas to verify whether the high stress areas match the actual stress locations.

[0026] Secondary Verification: The RBCS is superimposed on the physical mold cavity surface using Augmented Reality Projection (ARP), and the operator visually confirms that the boundaries match the mold geometry.

[0027] Level 3 Verification: If a fuzzy boundary area (such as a curvature gradient) is detected, the system activates the Manual Assistance Decision Interface (MADI), allowing process engineers to manually fine-tune the coordinate points. Once verification is passed, the RBCS is marked as "approved," triggering the next step.

[0028] Based on the region boundary coordinate set, a dynamic laser grid is projected onto the mold cavity surface, guiding the robotic arm to precisely lay the continuous fiberglass sheet on the large laser-marked area, generating a sheet positioning completion signal. After the mold is in place, the High-Precision Laser Projector System (HPLPS) reads the approved set of Region Boundary Coordinate Sets (RBCS). The system integrates multiple Galvanometer Laser Heads (GLHs) to project a Dynamic Laser Grid (DLG) onto the mold cavity surface according to the coordinate data in the RBCS. The grid lines are 0.1 mm wide and 50 mm apart, forming a chessboard-like positioning reference system. The key innovation lies in the dynamic adaptability of the grid: when the mold surface is curved, the laser beam maintains projection clarity through real-time focal length adjustment (FLA); if a local cavity reflection is detected (such as a polished steel surface), the system automatically switches to green laser (wavelength 532 nm) to enhance contrast. After projection is complete, the DLG precisely covers all large-area region boundaries with an error of less than ±0.05 mm.

[0029] The Robotic Manipulator (RM) is equipped with a Vacuum End-Effector (VEE) at the end of the arm, and its control system receives real-time spatial coordinate data from the DLG. The motion planning of the robotic arm uses Visual Servo Closed-Loop Control (VSCLC): First, the Industrial Camera (IC) mounted on the arm captures an image of the laser grid on the mold surface.

[0030] Through the Image Registration Algorithm (IRA), the captured image is matched with the theoretical coordinates of the RBCS, and the current position deviation of the robotic arm is calculated (e.g., X-axis deviation +0.3 mm).

[0031] Based on the deviation value, the Motion Controller (MC) dynamically adjusts the joint angular velocity to drive the vacuum end-effector to move the Continuous Glass Fiber Sheet (CGFS) to the target position. During placement, the vacuum end-effector adsorbs the sheet with a negative pressure of 0.01 MPa and moves at a constant speed 5 mm above the mold surface to avoid fiber disturbance. When the edge of the sheet coincides with the DLG boundary to 99%, the vacuum end-effector releases the negative pressure, and the sheet naturally adheres to the cavity.

[0032] After placement is complete, the system performs triple confirmation: Contour comparison: The arm-mounted laser displacement sensor (LDS) scans the plate contour and compares it with the theoretical contour of the large area in the RBCS. The overlap tolerance must be ≤0.1 mm.

[0033] Wrinkle detection: Structured Light 3D Scanning (SL3DS) is used to check the surface flatness of the sheet material. If wrinkles greater than 0.2 mm are detected, rework is triggered.

[0034] Fiber orientation verification: A near-infrared polarimeter (NIRP) is used to detect the fiber orientation of the board to ensure that the deviation from the process requirements (such as the 0° direction) is less than 2°.

[0035] After all are qualified, the central controller generates a sheet positioning completion signal (SPCS), which contains the actual placement coordinate record and quality data packet.

[0036] The fiber arrangement image on the plate surface is acquired through a micron-level optical scanner, and the average gap size of the fiber grid is calculated using the directional gradient histogram algorithm, and a gap distribution heat map is output; The sheet positioning signal (SPCS) triggers the activation of a micron-level optical scanner (MOS). This device integrates a confocal microscopic lens (CML) and a high-speed complementary metal-oxide-semiconductor (CMOS) sensor to scan the surface of continuous fiberglass sheets at a resolution of 5 microns. The scanning process follows a pre-set path (an S-shaped path covering the entire large surface area), capturing 1,000 local images per square centimeter. A key innovation lies in multi-spectral illumination (MSI): alternating 450nm blue and 650nm red light illumination enhances the contrast between the glass fibers (highly reflective) and the resin matrix (highly absorbent), ensuring clear visibility of fiber edges.

[0037] The original image is transmitted to the image processing unit, and the fiber arrangement is analyzed using the Histogram of Oriented Gradients Algorithm (HOGA): Step 1: Divide the image into 8×8 pixel detection cells (DC).

[0038] Step 2: Calculate the gradient orientation (GO) and gradient magnitude (GM) of each pixel. For example, the GO at the edge of a fiber is perpendicular to the fiber axis, and the GM is greater than 100 gray levels.

[0039] Step 3: Count the cumulative sum of the gradient amplitudes of the 36 directional intervals (one interval every 10°) in each DC to generate a directional gradient histogram.

[0040] Step 4: Use Peak Detection (PD) to identify the main fiber direction and calculate the normal spacing (NS) between adjacent fibers. Finally, calculate the average value (e.g., 0.25 mm) and standard deviation (e.g., 0.03 mm) of the NS for the entire sheet.

[0041] The algorithm outputs the Gap Distribution Heatmap (GDH): The fiber gap size is visualized using pseudo color coding, with blue indicating dense areas (gaps < 0.2 mm) and red indicating loose areas (gaps > 0.3 mm).

[0042] The thermal map is overlaid with spatial coordinate information to mark the location of the high gap area (e.g., the average gap is 0.35 mm at X=120mm, Y=80mm).

[0043] Additional data reports include: Maximum Gap (MG), Minimum Gap (MiG), and 90th Percentile Gap (90PG). This file is pushed to the control console and robotic arm system via the shop-level Industrial Internet of Things (IIoT).

[0044] Based on the gap distribution heat map, long glass fiber balls are stacked and laid at the edges of the panels in the flange and corner areas, so that the ball covers the panel edge width to 1.5 times the average gap size, and a composite blank positioning report with overlapping interlocking areas is generated.

[0045] The system analyzes the gap distribution heat map (GDH) to extract key parameters: Average Gap Size (AGS, for example, 0.25 mm). Based on process rules, the theoretical overlap width (TOW) of the long glass fiber charge (LGFC) required to cover the edge of the sheet is calculated: TOW = AGS × 1.5 = 0.25 × 1.5 = 0.375 mm.

[0046] The Regional Bounding Coordinate Set (RBCS) is also retrieved to determine the spatial extent of the flange and corner areas (e.g., X = 150-200mm, Y = 50-100mm). The Robotic Path Planning Module (RPPM) uses this information to generate a ball placement trajectory: the ball centerline is 0.1875 mm (TOW / 2) from the edge of the sheet to ensure symmetrical coverage.

[0047] The laying process uses hot-air assisted positioning technology (HAPT): The end of the robotic arm is switched to the Charge Extrusion Gun (CEG), and the muzzle temperature is set to 80°C to make the material mass (containing thermoplastic resin) semi-molten.

[0048] During extrusion, an annular hot air nozzle (AHAN) sprays 120°C hot air at a pressure of 0.3 MPa to press the material dough onto the surface of the sheet.

[0049] Real-time monitoring of coverage width: The Laser Triangulation Sensor (LTS) measures the boundary position of the material ball and dynamically adjusts the extrusion rate (e.g., 0.5 g / s) and the robot arm movement speed (e.g., 10 mm / s) to ensure that the deviation between the actual coverage width (AOW) and the TOW is ≤±0.05 mm.

[0050] After laying is completed, a Composite Blank Positioning Report (CBPR) is generated: Geometric data: The three-dimensional morphology of the interlocking area was measured using a white light interferometer (WLI) to confirm that the overlap width met the requirements.

[0051] Material data: record the mass ratio of the dough to the board (e.g. 1:4) and the volume compression rate (e.g. 30%).

[0052] Quality Mark: If the gap between the material ball and the sheet is greater than 0.1 mm, the report will be marked as "Manual Repair Required." Otherwise, an "Interlock Zone Certification Code" (IZCC) will be generated. The report will be automatically filed in the Manufacturing Execution System (MES), indicating that the composite body positioning is complete.

[0053] This step achieves optimized material performance through a zoned laying strategy. Continuous glass fiber sheets provide high strength and flatness over large areas, while long glass fiber clumps enhance the impact resistance of complex structures. The interlocking zones formed by overlapping edges lay the foundation for subsequent interface bonding, while precise positioning ensures the rationality of material distribution. This zoned laying method effectively combines the advantages of glass fiber materials with different forms, ensuring both mechanical properties over large areas and enhancing fatigue resistance in key areas. The interlocking zone design creates favorable interface bonding conditions for subsequent processes, significantly improving the performance degradation caused by material mutations in traditional processes.

[0054] S202, applying local pre-compression to the interlocking region of the composite body so that the long glass fiber balls are embedded in the fiber grid gaps of the continuous glass fiber sheet to form a mechanical interlocking structure, thereby obtaining a preform having an interlocking interface structure; Specifically, based on the three-dimensional topological map of the composite body, an infrared focused heating device can be deployed in the interlocking area to soften the surface resin of the material mass to a viscous flow state and output a signal indicating that the resin viscosity meets the standard; After the system receives the composite billet positioning report generated in the previous step, it first calls the built-in composite billet three-dimensional topology map (3D Topology Map). This topology map is obtained by full-size digital reconstruction of the superimposed structure of the continuous fiber sheet and long glass fiber charge in the mold cavity by a high-precision optical scanning system (such as a laser triangulation instrument), which accurately calibrates the spatial coordinate range, geometric shape, and overlapping thickness of the interlock zone (Interlock Zone) and the material and sheet. Based on this topology map, the control system starts the infrared focused heating device (Infrared Focused Heating Device, IFHD). The device is composed of multiple independently controllable infrared emitter modules (Infrared Emitter Module, IEM), each module is equipped with a parabolic mirror and a precision optical lens system. The control system dynamically calculates the optimal irradiation angle, focal point position, and energy density distribution of each IEM according to the three-dimensional shape of the interlock zone. For example, for a curved interlock zone, the system will adjust the tilt angle (Tilt Angle, TA) of the IEM within the range of 0-30 degrees, and control the focal depth (Focal Depth, FD) between 5-15 millimeters (mm), to ensure that the infrared energy is accurately focused on the surface area (thickness about 2-3mm) of the long glass fiber charge (Long Glass Fiber Charge, LGFC), avoiding excessive heat penetration affecting the internal fiber performance or heating adjacent continuous fiber sheets (Continuous Fiber Sheet, CFS).

[0055] Precise control of the heating process relies on real-time temperature feedback and analysis of the resin's rheological state. A non-contact infrared thermal camera (ITC) is deployed directly above the infrared irradiation area, measuring temperature with an accuracy of ±1°C and a spatial resolution of 0.1 mm / pixel. The ITC captures a temperature distribution cloud map (TDCM) of the interlocking zone surface at a rate of 10 frames per second (FPS). Simultaneously, a microthermocouple sensor (TC) is embedded within the material mass at a depth of one-third of the mass thickness (e.g., 2 mm for a 6 mm mass thickness) to directly monitor the resin's core temperature (CT). The control system integrates the TDCM and CT data with the epoxy resin's dynamic viscosity-temperature curve (VRTC) from the material database to calculate the resin's apparent viscosity (AV) in real time. When the AV drops to the target viscosity threshold (e.g., 100 Pa·s), the resin has fully softened and reached the optimal viscosity flow state (VFS). At this point, the system automatically generates a Resin Viscosity Qualified Signal (RVQS), which contains the viscosity verification results and a timestamp (TS) for all heated zones.

[0056] To ensure heating uniformity and process repeatability, the system utilizes a Multi-Zone Cooperative Control Strategy (MZCCS). If the TDCM indicates a temperature deviation in a subzone exceeding ±5°C (e.g., low temperatures at the edge due to heat loss), the control system dynamically increases the emission power (EP) of the corresponding IEM. This increase is calculated using a Proportional-Integral Control Algorithm (PICA). For example, the EP could be increased from the default 800 watts (W) to 950 watts (W), with the temperature difference brought to within ±2°C within 30 seconds. Furthermore, to prevent resin degradation due to overheating (e.g., thermal decomposition at temperatures exceeding 180°C), the system implements a hard temperature limit (HTL). If the temperature at any monitoring point approaches the HTL (e.g., 175°C), the power cut-off mechanism (PCM) is triggered, shutting down the corresponding IEM and issuing an alarm. The entire heating process is usually completed within 60-120 seconds (s). The formation of RVQS indicates that the surface resin of the dough has ideal fluidity and embedding ability, creating the necessary conditions for subsequent mechanical biting.

[0057] Receive the signal that the resin viscosity reaches the standard, activate the micro-indenter array with a diameter matching the fiber gap, apply 0.5MPa local pressure, and output a real-time pressure distribution cloud map; When the control system receives the resin viscosity qualification signal (RVQS), it immediately activates the micro indenter array (MIA) pre-deployed above the mold. The array is composed of hundreds of independently driven indenter units (IU), each containing an indenter pin (IP) and a high-precision piezoelectric ceramic actuator (PCA). The key design is that the diameter (DIA) of each IP strictly matches the average gap size (AGS) of the fiber grid of the interlock area continuous fiber sheet (CFS). For example, if the gap distribution heat map (GDHM) generated in the previous step shows that the AGS is 0.3 millimeters (mm), the IP diameter is designed to be 0.28 millimeters (mm), leaving an allowance (ALW) of 0.02 millimeters (mm) to ensure that the fibers are not crushed and broken. The layout of the indenter array is not uniform, but is adaptively arranged according to the profile of the interlock area and the fiber orientation (FO) in the three-dimensional topological map: the IU density is increased in the dense fiber gap area (such as 25 per square centimeter), and the density is reduced in the sparse gap area (such as 10 per square centimeter), ensuring that each IP can accurately align with a fiber gap.

[0058] The pressure application process utilizes a stepwise progressive loading (SPL) strategy. Upon receiving the start command, all IUs descend synchronously, initially contacting the dough surface at a low speed (0.1 mm / s). Once the IP tip detects a contact force of 0.05 MPa (MPa) via a contact force sensor (CFS, accuracy ±0.01 Newton), contact is determined. The system then instructs all PCAs to simultaneously apply a target pressure (TPV) of 0.5 MPa. This pressure value, verified in preliminary experiments, effectively promotes the flow of softened resin while preventing breakage of long glass fibers (which typically have a compressive strength exceeding 1000 MPa). During the pressure application process, a micro pressure film sensor (MPFS, resolution 0.01 MPa) at the base of each IU collects the local pressure value (LPV) at the point of application in real time. The central control system collects LPV data from all IUs at a sampling frequency of 100 Hz and generates a real-time pressure distribution cloud map (RPDCM) covering the entire interlocking area using a spatial interpolation algorithm (SIA). The cloud map is displayed in pseudo-color, with red indicating high pressure areas (>0.55 MPa) and blue indicating low pressure areas (<0.45 MPa).

[0059] Dynamic control of pressure uniformity is a core technology. The control system continuously analyzes the RPDCM. If a pressure deviation exceeds ±10% (e.g., a region consistently displays 0.4 MPa against a target of 0.5 MPa), the pressure compensation mechanism (PCM) is activated. For example, for an IU in a low-pressure area, the PCA is instructed to increase the displacement increment (DI) by 5 microns (μm), thereby increasing local pressure through increased compression. For high-pressure areas, the DI is reduced by 5 μm. Furthermore, for abnormally high pressure points (e.g., >0.6 MPa) caused by uneven dough thickness or fiber resistance, the system triggers a single-point unloading command (SUC), temporarily retracting the corresponding IU by 2-3 μm to relieve pressure. The entire pressure application phase lasts 20 seconds, during which RPDCM data is recorded in real time and transmitted to the process database. The final output of the real-time pressure distribution cloud map not only contains the pressure value matrix, but also the Pressure Uniformity Index (PUI). This index must be ≥90% to be considered qualified to ensure the stability of the subsequent embedding process.

[0060] The pressure head stroke is adjusted through the real-time pressure distribution cloud map. When the fiber optic sensor detects that the material mass embedding depth reaches 70% of the fiber diameter, it outputs the bite depth authentication code. During the pressurization process, the control system dynamically adjusts the stroke (ST) of each indenter unit (IU) in the micro-indenter array (MIA) based on data from the real-time pressure distribution cloud map (RPDCM). This stroke adjustment is implemented using the Adaptive Embedding Control Algorithm (AECA). If the RPDCM indicates that the pressure in a certain area is below the target lower limit (e.g., <0.45 MPa), and the corresponding resin flow resistance model (RFRM) predicts that the embedding speed is too slow, the system increases the indentation speed (IS) of the IU in that area, for example, from 0.05 mm / s to 0.08 mm / s, and allows its maximum stroke (MS) to increase by 10% (e.g., from 3 mm to 3.3 mm). Conversely, if the pressure in a certain area is too high (>0.55 MPa) and the fiber optic sensor (FOS) feedback indicates that the embedding depth is increasing too quickly, the IS is reduced to 0.03 mm / s to avoid excessive compression.

[0061] Real-time monitoring of embedding depth (ED) is crucial for forming a reliable mechanical interlocking structure (MIS). A reflective fiber optic sensor (RFOS) is embedded at the bottom of the fiber bundle gap (FBG) in continuous glass fiber sheet (CFS). Its operating principle: The sensor's emitter (EM) emits near-infrared light (wavelength 850 nm). When the fibers (FB) in the long glass fiber bundle (LGFC) are pressed into the gap, they block the light path, causing a decrease in the reflected light intensity (RLI) detected by the receiver (RV). Using the RLI-ED conversion curve (RECC) established through preliminary calibration experiments, the RLI value can be converted into embedment depth (unit: microns) in real time. For example, when the RLI decreases from an initial value of 1000 units (U) to 650 units (U), the corresponding ED reaches 70% of the single glass fiber diameter (e.g., 17 μm) (i.e., 11.9 μm). This depth, verified by finite element analysis, generates sufficient mechanical anchoring force (MAF) while preventing fiber damage from excessive bending.

[0062] The Zone-based Group Authentication Strategy (ZGAS) is used to determine compliance with interlocking depth standards. The interlocking zone is divided into 1 cm x 1 cm grid cells (GCs), each containing at least three RFOS monitoring points. When the ED of all RFOS measurements within a GC is ≥ the target value (70% of the fiber diameter) and remains stable for more than 3 seconds, the GC is marked as meeting the interlocking depth standard. When more than 95% of the GCs (area-weighted) within the interlocking zone meet the standard, the control system generates an Interlocking Depth Certification Code (IDCC). The authentication code is a 128-bit encrypted string containing the following key information: Certification Timestamp (CTS), Qualified Area Ratio (QAR, e.g., 96.3%), Minimum Embedding Depth (MED, e.g., 11.2 μm), Maximum Embedding Depth (MXED, e.g., 13.5 μm), and Depth Uniformity Coefficient (DUC, e.g., 0.92). The generation of the IDCC indicates that the mechanical occlusal structure has been stabilized, triggering subsequent CT scan verification.

[0063] Based on the occlusal depth authentication code, microfocus CT scanning is triggered to reconstruct the three-dimensional model of the interlocking area and output the interface interlocking structure authentication file.

[0064] Once the interlocking depth verification code (IDCC) is verified by the system, the microfocus computed tomography system (μCT) mounted on the side of the mold is immediately activated. The core component of this system is a microfocus X-ray source (MXS), with a focal spot size less than 5 microns (μm). The voltage (V) and current (C) are preset based on the material density (e.g., 100 kilovolts (kV) and 150 microamperes (μA) for epoxy-glass fiber composites). Before scanning, the mold's interlocking zone observation window (IZOW) (made of highly X-ray-transparent silicon carbide ceramic) automatically opens. The scanning platform (SP) carrying the preform (PF) rotates 360 degrees (°), capturing a projection image (PI) every 0.5 degrees (°), for a total of 720 frames (FR). To reduce motion artifact (MA), the rotation speed was controlled at 0.5 degrees per second (° / s), and the total scanning time was approximately 12 minutes (min).

[0065] The collected projection images were reconstructed into three dimensions using the Filtered Back Projection Algorithm (FBPA). The reconstruction process was completed on an industrial graphics workstation (IGW). First, the original images were subjected to ring artifact correction (RAC) and beam hardening compensation (BHC) to improve data quality. The reconstructed three-dimensional volume data (3DVD) contains millions of voxels (VX), each of which is 10 microns × 10 microns × 10 microns (μm). 3), the Gray Value (GV) reflects the local Density (D) of the material. The system applies a Machine Learning-based Segmentation Algorithm (MLSA) to automatically identify different materials: high GV (>2500) for Glass Fiber (GF), medium GV (800<GV<2000) for Epoxy Resin (ER), and low GV (<500) for Pore (PO). Finally, a 3D Interlock Zone Model (3DIZM) is generated, which clearly shows how the long glass fiber clusters (LGFC-FB) embed into the grid gaps (CFS-Gap) of the continuous fiber sheet (CFS) and form hook-shaped interlocks (HSI).

[0066] Quality evaluation and certification based on the 3D model is the last step. The system automatically calculates the following key indicators: Interlock Area Ratio (IAR): the percentage of the area actually forming mechanical connections in the occlusal interface, which is required to be ≥90%; Fiber Crossing Angle (FCA): the average angle between the embedded fibers and the matrix fibers, which is ideally 60-90 degrees (°) and affects the anchoring strength; Resin Filling Ratio (RFR): the volume percentage of resin in the gap, which is required to be ≥95%; Defect Density (DD): the number density of pores or un-infiltrated areas, which is required to be <5 per cubic millimeter (mm 3 ).

[0067] If all indicators meet the preset process standards (e.g., IAR=92.5%, FCA=78°, RFR=97.2%, DD=3.2 per mm 3The system generates an Interface Interlock Structure Certification File (IISCF). This file includes a 3D model screenshot, a quantitative performance report, a Deviation Analysis (DA) from the design target, and a unique QR Code (QRC). The generation of the IISCF signifies that the preform (PF) has a qualified interface interlock structure and can be safely transferred to the main molding process. Simultaneously, all raw data (CT images, reconstructed models, and certification results) are archived to the Process Big Data Platform (PBDP) for quality traceability and continuous optimization.

[0068] Through a precisely controlled, localized pre-pressing process, the long glass fiber clumps and continuous glass fiber sheet form a mechanical interlock at the microscopic level. This physical bonding avoids the limitations of relying solely on resin bonding, establishing a more robust connection network at the fiber level. The mechanical interlocking structure significantly enhances the bond strength at the material interface and ensures more uniform stress transfer. This pre-treatment effectively prevents delamination that can occur during the molding process, providing structural support for subsequent high-pressure molding.

[0069] S203: Establishing a pressure distribution model based on the product's geometric characteristics, generating a first pressure value for the large surface area and a second pressure value for the flange and corner areas, wherein the second pressure value is greater than the first pressure value, thereby obtaining a differentiated pressure parameter combination; Specifically, the product geometric characteristic parameters can be imported, standard molding pressure can be applied to perform nonlinear contact analysis, and the equivalent stress distribution matrix of each node in the cavity can be output; The system first extracts key geometric feature parameters from the product design database, including 3D surface curvature, thickness gradients, and chamfer radii. For example, the flange corner radius of a certain automotive door panel part is 5 millimeters (mm), the central large surface thickness is 2.5 millimeters (mm), and the edge reinforcement rib height is 8 millimeters (mm). These parameters are imported into the finite element analysis engine via standardized interfaces (such as the STEP file format). The system then discretizes the mold cavity and blank, generating a computational model containing hundreds of thousands of mesh nodes. In the initial analysis phase, the system applies a preset standard molding pressure (for example, 10 megapascals (MPa)) to simulate the contact process between the blank and the mold. This process uses nonlinear contact algorithms (such as the augmented Lagrangian method) to account for material elastic-plastic deformation, the friction coefficient (set to 0.25), and the effect of temperature softening. The computation engine traverses all mesh nodes, calculating the stress tensor for each node in three-dimensional space. Ultimately, using the von Mises Criterion, it simplifies the complex stress state into scalar values, forming an Equivalent Stress Distribution Matrix (ESDM) covering the entire cavity. This matrix is ​​stored as a two-dimensional array, with row numbers corresponding to mesh IDs and columns containing stress values ​​(in MPa), coordinate positions, and normal vector directions.

[0070] The accuracy of nonlinear contact analysis relies on the convergence mechanism of multiple iterations. The system sets a convergence tolerance of 0.5% (i.e., the calculation is terminated if the stress change rate between two consecutive iterations is less than 0.5%). During the simulation, the rigid boundary conditions of the mold are dynamically coupled with the viscoelastic constitutive model of the composite body (using the generalized Maxwell model). For example, when stress concentration occurs in the flange area due to geometric changes, the algorithm automatically refines the mesh there (with a minimum mesh size of 0.1 millimeters) and inserts virtual damping elements at the contact interface to suppress numerical oscillations. After each iteration, the system verifies the rationality of stresses by accessing glass fiber-resin interface peel strength data (e.g., 35 megapascals (MPa)) from the material database. The final output ESDM matrix is ​​converted to continuous field data using spatial interpolation techniques, and the coordinates of stress hotspots are annotated (e.g., the stress peak at the flange root node reaches 120 MPa), providing a quantitative basis for subsequent region delineation.

[0071] To verify the reliability of the analysis, the system simultaneously starts the digital twin verification process: micro piezoelectric stress sensors (range 0-200 MPa, accuracy ±0.5%) are embedded in typical locations of the physical mold (such as the corner R5 area) to collect stress data during the actual pressing process. The measured value is compared with the predicted value of the corresponding node in the ESDM matrix. If the deviation exceeds 8%, the parameter adaptive correction module is triggered. For example, when it is detected that the predicted stress in a flange area is 110 MPa and the measured value is 125 MPa, the system automatically increases the friction coefficient of the area by 0.1 and re-iterates the calculation until the error converges. The ESDM matrix that has completed the verification will be marked as "verified" and output to the next processing module.

[0072] Perform regional cluster analysis on the equivalent stress distribution matrix, screen flanges and corner areas where stress values ​​exceed 85% of the material yield strength, and generate a coordinate set for the high-pressure demand area; The system calls the pre-stored yield strength threshold for glass fiber reinforced polypropylene (GFRP) in the material library (e.g., 80 megapascals (MPa)) and calculates its 85% critical value (i.e., 68 MPa). All nodes in the ESDM matrix with stress values ​​above this critical value are marked as "high-risk nodes." A density clustering algorithm (DBSCAN) is then used to spatially cluster high-risk nodes: the neighborhood search radius (Eps) is set to 3 millimeters (mm), and the minimum number of clustered nodes (MinPts) is set to 50. The algorithm automatically identifies high-density stress clusters at flanges and corners (e.g., a flange region with 1,200 continuous high-risk nodes). Discrete high-risk points in large areas (fewer than MinPts) are considered noise points and excluded. The clustering results are output as multiple independent high-pressure demand regions (HPDRs), each containing boundary contour coordinates and a list of internal node IDs.

[0073] To precisely define the region boundaries, the system initiates topology optimization post-processing: a morphological dilation operation (structural element radius 1.5 millimeters) is performed on each HPDR to fill internal voids. Edge detection algorithms (such as the Canny operator) are then used to extract smooth contours. Stress statistics are also calculated for each region: for example, the maximum stress in a particular engine hood corner is 142 megapascals (MPa), the average stress is 92 MPa, and the stress gradient is 15 MPa / mm. This data is then linked to a coordinate set to produce a structured report. The system also links to a geometric feature library and automatically annotates the HPDR type attributes (such as "flange - convex" or "corner - groove") for subsequent development of differentiated pressure strategies.

[0074] The resulting High-Pressure Zone Coordinate Set (HPZ_CS) uses a hierarchical storage structure: the top layer represents the total number of zones (e.g., five independent zones). Each layer contains a sequence of 3D coordinate points (with an accuracy of ±0.01 mm), bounding box dimensions (e.g., Zone A: 30 mm long, 15 mm wide, and 8 mm high), and a stress level label (e.g., Level_3 corresponds to a peak stress of 120-150 MPa). This coordinate set is encapsulated in JSON format and includes a timestamp and version number for real-time access by downstream modules.

[0075] According to the material rheology database, with the resin critical flow shear rate as the constraint condition, the minimum molding pressure value of the large surface area is solved as the first pressure reference value; The system accesses the Material Rheology Database (MRD) to extract key parameters of the current resin system (such as PA66-GF30): melt viscosity-shear rate curve (viscosity at 190°C from 1000 Pa·s (Pa·s) @ 0.1 / s (s -1 ) is reduced to 80 Pa·s @ 100 / s (s -1 )), critical shear rate γ_crit (the limit value to prevent glass fiber breakage, such as 50 / second (s -1 Based on this, a pressure-flow velocity model is established: the typical flow channel length of the large surface area is set to 200 mm and the thickness is 2 mm. The relationship between pressure and shear rate is inversely deduced by the generalized Herschel-Bulkley equation. The constraint condition is that the shear rate at the melt front must not exceed γ_crit (50 / s (s -1 ), the objective function is to minimize the molding pressure.

[0076] The solution process uses the sequential quadratic programming algorithm (SQP): starting from the initial pressure estimate of 8 MPa, iteratively calculates the melt flow front velocity V_front (unit: millimeter / second (mm / s)) and the shear rate γ_max (unit: / second (s -1 )). When it is detected that γ_max approaches 50 / second (s -1 ) (e.g. 48.5 / s (s -1 ), recording the current pressure value as a feasible solution. By verifying multiple boundary conditions (e.g., thickness fluctuation of ±0.3 millimeters (mm), the first pressure baseline value P_base (e.g., 7.2 megapascals (MPa)) is ultimately determined. This value must meet dual constraints: ensuring that the resin fully fills the large surface area (flow time < 15 seconds) while avoiding shear damage to the glass fibers.

[0077] The system performs process robustness verification: simulate ±10% material viscosity fluctuations in a virtual environment (e.g., retrieve three batches of resin data in MRD) to test the stability of P_base. If a batch of resin is found to have a γ_max exceeding the standard of 52 / s at 7.2 MPa, the system will automatically verify the stability of P_base. -1 ), triggering a conservative strategy and raising P_base to 7.5 MPa. The final locked P_base is stored in association with the material batch number and annotated with the applicable temperature range (e.g., 180-190°C).

[0078] In the high-pressure demand area coordinate set, the first pressure reference value is input, the compensation pressure coefficient is dynamically calculated through the deep reinforcement learning model, and the second pressure value optimization solution is output; The deep reinforcement learning (DRL) model takes as input the coordinate set of the high-pressure demand zone (HPZ_CS) and the first pressure baseline value P_base (7.5 megapascals (MPa)). The model architecture adopts an actor-critic framework: the actor network is a four-layer fully connected neural network (with 256-128-64-32 neurons). The input dimensions include regional geometric features (bounding box size, curvature), stress statistics (peak value, mean value, gradient), and P_base; the output is the compensation pressure coefficient K_comp (defined as the second pressure value P_high = K_comp × P_base). The critic network evaluates the value of the action, and the reward function R is designed as: Positive reward: Porosity of the pressed area <0.3% (+5); Negative reward: glass fiber breakage rate > 1% (-10) or energy consumption exceeds the standard (-3).

[0079] The model is iteratively optimized in a virtual training environment. 100 sets of high-pressure area pressing data (such as flange dimensions, pressing pressure, and defect indicators) are loaded from a historical case library, and the network weights are updated using a policy gradient algorithm. For example, in the corner area of ​​a luggage rack (25 mm x 20 mm), an initial K_comp of 1.6 resulted in a 1.2% glass fiber fracture rate. Adjusting the model to K_comp of 1.45 reduced the fracture rate to 0.7% and the porosity to 0.25%, resulting in a positive reward. Training is terminated when reward fluctuations remain below 2% for 20 consecutive iterations. The final deployed model outputs a recommended K_comp value (e.g., in the range of 1.35-1.8) in real time.

[0080] The system generates a second pressure value optimization solution: for each independent zone in HPZ_CS, the DRL model outputs a customized K_comp. For example: Region A (flange with small curvature): K_comp = 1.42 → P_high_A = 10.65 MPa; Region B (acute corner): K_comp = 1.68 → P_high_B = 12.6 MPa.

[0081] The solution is attached with a confidence score (e.g., 92% confidence for Region B) and linked to historical similar cases (e.g., "Solution Reference 2023-07 Order #2056 Corner Compression Log").

[0082] Based on the first pressure reference value and the second pressure value, the solution is optimized to construct a spatially continuous pressure distribution function, generating a parameter combination configuration file containing a pressure gradient curve.

[0083] The system converts discrete pressure values (large area P_base = 7.5 MPa, each high pressure region P_high_A = 10.65 MPa, P_high_B = 12.6 MPa) into a continuous field. Radial basis function interpolation (RBF) is used to construct the pressure distribution function: take the center of the high pressure region as the control point (e.g., Region A center coordinates (x1, y1, z1)), set the influence radius R_infl = 15 mm. The function expression weight is determined by the Gaussian kernel function, ensuring smooth transition of pressure values at the region boundary (e.g., the pressure gradient from the flange to the adjacent large area is controlled within 0.8 MPa / mm).

[0084] For visualization verification, the system generates a pressure gradient curve: along a typical path (e.g., from the center of the large area to the end of the flange), sample the pressure value, and draw the distance-pressure curve. For example: 0-50 mm (large area): pressure stabilizes at 7.5 MPa; 50-70 mm (transition area): pressure linearly rises to 10.65 MPa; 70-90 mm (flange area): pressure maintains 10.65 MPa.

[0085] The curve must satisfy the monotonicity constraint (prohibit pressure sudden change), and the gradient extreme value <1.2 MPa / mm is confirmed by derivative analysis.

[0086] The final output parameter combination configuration file contains: Global parameters: P_base value, temperature setting (large area 180°C / interlocking area 200°C); High pressure region table: coordinate range, P_high value, K_comp source trace code; Pressure gradient curve graph: SVG format vector graph; Process constraints: maximum allowable pressure (e.g., mold limit 15 MPa).

[0087] After being digitally signed, the file is stored in the blockchain process library for direct call and execution by the pressing equipment.

[0088] A pressure distribution model based on product structural characteristics enables intelligent, zoned control of molding pressure. By identifying the mechanical requirements of different regions, optimal molding pressure parameters are matched to simple and complex areas. This differentiated combination of pressure parameters ensures adequate resin flow while avoiding fiber damage caused by localized overpressure. This precise control significantly improves material utilization while ensuring consistent molding quality across all regions.

[0089] S204, placing the preform in a heated mold and applying pressure in stages according to the pressure parameter combination, wherein a first pressure value is applied to the large surface area to allow the resin to melt and flow, and a second pressure value is applied to the interlocking area to allow the molten resin to fill the fiber gaps, thereby obtaining a semi-finished product without exposed glass fibers; Specifically, the parameter combination configuration file can be read to set a 180°C constant temperature field in the large surface area of ​​the heated mold and a 200°C high temperature field in the interlocking area, and output a temperature field balance ready signal; After receiving a Parameter Combination Configuration File (PCCF) generated by the pressure distribution model, which contains pressure and temperature settings for different mold zones, the control system first interprets the temperature instructions in the file: a 180°C constant temperature field (180CTF) is established in the large area region (LAR) of the part (i.e., the flat bulk of the part) and a 200°C high temperature field (200HTF) is established in the interlocking zone (IZ) (i.e., the interface between the continuous glass fiber sheet and the long glass fiber clumps). This process is executed by the Distributed Temperature Control Unit (DTCU). Based on the mold's physical partitioning, the DTCU precisely embeds a multi-zone independent heating rod array (MIHRA) on the backside of the mold plate corresponding to the large area region (usually a specific area of ​​the upper or lower mold). Each heater rod is equipped with a high-precision K-type thermocouple (KTC), a commonly used temperature sensor with a wide measurement range and low cost, as a temperature feedback element. The control system (typically a Programmable Logic Controller (PLC)) dynamically adjusts the current (CUR) flowing to each heater rod based on a set target value (TV) of 180°C using a proportional-integral-derivative control algorithm (PIDCA), ensuring a uniform and stable mold surface temperature of 180±1°C over a large area. Simultaneously, high-frequency induction heating coils (HFIHC) or high-power density ceramic heaters (HPDCH) are deployed in the mold locations corresponding to the interlocking zones. These heaters offer fast response times and can quickly reach and maintain the higher set point of 200°C in localized areas.The interlock area is also equipped with a dedicated infrared thermometer (IRTM) or embedded thermocouple for real-time temperature monitoring (RTM).

[0090] The key to achieving temperature field equilibrium (TFE) lies in eliminating thermal interference between zones and ensuring rapid setpoint stabilization. Because the interlock zone (200°C) is significantly hotter than the large area zone (180°C), heat conduction (HC) is easily generated at the interface between the two. To this end, a vacuum insulation barrier (VIB) was embedded at the physical boundary between the LAR and IZ during mold design to effectively prevent heat transfer. During the temperature ramp, the PLC system coordinates the heating rates (HR) of the two zones. For example, the HFIHC or HPDCH system initially heats the interlock zone to 195°C at a high rate (e.g., 30°C / minute), while the MIHRA system simultaneously heats the large area zone to 175°C at a moderate rate (e.g., 15°C / minute). Both systems then switch to fine-tuning mode: the interlock zone slowly heats to and stabilizes at 200±1°C, while the large area slowly heats to and stabilizes at 180±1°C. The PLC continuously compares the actual temperature value (ATV) from each zone's temperature sensor (KTC, IRTM) with the set target value (TV). Using PIDCA, it calculates a real-time power adjustment value (PAV) and sends it to the corresponding heater actuator. When the ATV for all zones remains within ±1°C of the set TV for 30 consecutive seconds (reaching a steady state, SS), and the maximum temperature difference between zones is less than 5°C (meeting the equilibrium requirement), the PLC system automatically generates a Temperature Field Equilibrium Ready Signal (TFERS). This signal is a high-level (HL) digital output, indicating that the mold is in precise temperature control and ready for the next pressurization step.

[0091] To ensure reliable and traceable temperature control, key data from the entire heating and holding process, including timestamps (TS), power output percentages (POPs) for each heater, ATVs for each sensor, calculated PAVs, and the final TFERS status, are recorded in a process historian (PHDL). Furthermore, thermal imaging cameras (TICs) may be installed at key locations on the mold to visually display the temperature distribution across the mold surface using non-contact infrared thermal images (NCITIs), supplementing thermocouple point monitoring. Operators can view the temperature curve (TC) and heat map (HM) in real time on the human-machine interface (HMI) to confirm the absence of local overheating (LO) or undercooling (UC). Only when TFERS is effectively emitted will the system unlock the subsequent pressure application step to prevent poor material solidification or defects caused by pressing when the temperature does not meet the standard.

[0092] After receiving the temperature field ready signal, the large surface area is pressed at the first pressure value, and the phase switching instruction is triggered when the resin front reaches the boundary of the interlocking zone through the melt flow sensor; Upon receiving a valid Temperature Field Equalization Ready Signal (TFERS), the PLC system immediately initiates the first stage of the pressing sequence. The core objective of this stage is to close the mold and compact the Large Area Rectangle (LAR) using the First Pressure Value (FPV). This ensures sufficient melt flow (MF) of the matrix resin (typically a thermoset resin such as epoxy or phenolic) in this area, preparing for subsequent filling of the interlocking zones. The FPV is derived from the Parameter Combination Profile (PCCF) and is a specific value (e.g., 5 MPa). The system controls the Main Hydraulic Cylinder (MHC) or a servo-driven Toggle Mechanism (SDTM) to smoothly lower the upper die. Pressure control utilizes a closed-loop servo valve control system (CSVCS). During the pressing process, high-precision pressure sensors (HPPS, such as piezoelectric or strain gauge) located at the four corners of the mold monitor the actual clamping force (ACF) in real time. The PLC compares the ACF with the FPV and dynamically adjusts the servo valve opening (VOD) through PIDCA, precisely controlling the amount of oil entering the hydraulic cylinder or the servo motor torque (TOR) to ensure that the average pressure applied to the large area of ​​the mold remains stable within the FPV ±0.2MPa range. During this stage, pressure must be applied smoothly to avoid shocks, and the pressure rise time (PRT) is typically set at 2-5 seconds.

[0093] Under the influence of FPV, the resin within the large, continuous glass fiber sheet begins to melt and flow. To accurately capture the location of the resin flow front (RFF) and its arrival at the interlocking zone boundary (IZB), the system pre-embeds a melt flow sensor array (MFSA) at key flow path locations within the mold (specifically the transition between the LAR and the IZ). These sensors typically utilize one of the following principles: a microthermocouple array (MTA), which generates a signal based on temperature changes caused by resin flow (the cold front); a dielectric sensor (DS), which detects capacitance changes caused by the passage of resin (insulating) and air (with different dielectric constants); or a fiber bragg grating sensor (FBGS), which utilizes the wavelength shift of the grating reflection caused by microstrain or temperature changes induced by resin flow. For example, six to eight FBGS sensors are evenly spaced near the IZB line. When molten resin flows past a sensor location, it causes slight changes in temperature and / or stress at that point, resulting in a shift (Wavelength Shift, WS) in its Bragg Wavelength (BW, the characteristic reflection wavelength of the FBGS). The Signal Processing Unit (SPU) analyzes the WS data of all FBGS in real time. When more than 50% (e.g., 3 out of 4) of the sensors located on the IZB line continuously monitor and detect significant wavelength shifts (e.g., exceeding 0.5 nanometers) within 0.5 seconds that are consistent with the resin flow characteristic (CFS), the system determines that the resin front has reached the interlocking zone boundary (RFF Reached IZB, RFF_RI).

[0094] Once the system confirms the RFF_RI state is established, the PLC immediately generates and issues a Phase Switching Instruction (PSI). The PSI is a critical trigger signal, typically a rising-edge (RE) pulse or the setting of a specific status bit (SB). This instruction contains two core messages: first, it instructs the main pressure system to cease maintaining the current FPV; second, it instructs the system to prepare to increase the pressure to a higher secondary pressure (SPV) within a very short period of time. To ensure timely and reliable switching, the RFF_RI judgment logic and PSI generation process are given extremely high priority (HP) and fast response (typically in milliseconds). The system records the exact time of PSI triggering, the sensor number that triggered it, and the specific signal value detected (such as wavelength offset) as important process evidence and stores it in the PHDL. Simultaneously, the simulated position of the resin flow front and the "Phase Switch Triggered" status prompt are highlighted on the HMI for easy operator monitoring. The accurate triggering of PSI is the core of the entire staged pressing process, which marks the end of the first stage (large area compaction and resin melt flow) and the beginning of the second stage.

[0095] Based on the phase switching instruction, the pressure is increased to the second pressure value within 0.2 seconds and maintained for 10 seconds, using high pressure to force the molten resin to penetrate into the fiber gaps, and a resin penetration completion mark is output; Upon receiving the Phase Switch Instruction (PSI) from the PLC, the pressure control system immediately enters high-pressure, fast-response mode. The Second Pressure Value (SPV), derived from the Parameter Combination Configuration File (PCCF), is significantly higher than the FPV (e.g., if the FPV is 5 MPa, the SPV may be 15 MPa or higher). This value is specifically used to compact the interlock zone (IZ) and ensure complete penetration of the molten resin into the fiber interstices (FI) within this area. The key to achieving "raising the pressure to the SPV within 0.2 seconds" lies in: 1) High-pressure accumulators (HPAs): Prior to pressing or during the first low-pressure pressing stage, one or more accumulators driven by high-pressure gas (e.g., nitrogen) are pre-charged with high-pressure hydraulic oil to store a large amount of energy. 2) High-speed servo valves (HSSVs): Their response time (RT) is typically in the millisecond range (<10 ms), enabling rapid opening of high-flow channels. 3) Low-compressibility Hydraulic Line (LCHL): Using special hydraulic fluid or designing short, rigid lines reduces pressure transmission delay. When the PSI reaches the desired value, the PLC instantly opens the High-pressure Spool (HSSV) to the preset Maximum Opening Degree (MOD). High-pressure oil surges from the HPA and is injected directly into the boost chamber of the Master Hydraulic Cylinder (MHC) through the LCHL. The system continuously monitors the ACF fed back by the HPPS. Using a preset Pressure Rise Slope Control Algorithm (PRSCA), the time it takes for the ACF to rise linearly from the FPV to the SPV is strictly controlled within 0.2±0.02 seconds. This process is accompanied by a noticeable pressure impact sound (PIS), but is kept within the process tolerance.

[0096] Once the ACF reaches the target SPV value, the system immediately enters the Pressure Holding Phase (PHP). The PLC precisely controls the opening of the HSSV and performs fine pressure compensation adjustments to stabilize the ACF within the SPV ±0.5MPa range for 10 seconds (10 seconds). During these critical 10 seconds, the high pressure (SPV) applied to the mold interlocking area produces multiple effects: 1) Mechanical Compaction (MC): This further compacts the overlapping area between the long glass fiber mass and the continuous glass fiber sheet, eliminating any minor voids that may remain from the initial pre-compression. 2) Resin Penetration (RP): High pressure forces the molten resin (thanks to the 200°C high temperature in the interlocking zone) with reduced viscosity (VIS) to overcome capillary resistance (CR) and flow resistance (FR), forcing it into the fiber mesh gaps (FMG) at the edges of the continuous glass fiber sheet and between the fiber bundles within the long glass fiber clumps. 3) Venting (VEN): High pressure helps compress microbubbles (MB) trapped between the fiber gaps or within the clumps, allowing them to escape through the precise venting grooves (VG) designed into the mold. The high temperature (200°C) in the interlocking zone is also crucial at this stage, ensuring the resin maintains a sufficiently low viscosity (LV) and excellent flowability (FLW) to quickly and effectively fill all microscopic voids (MV) under high pressure.

[0097] After the 10-second high-pressure hold period expires, the system automatically generates a Resin Penetration Completion Mark (RPCM) based on a timer (TIM) or Pressure Holding Completion Logic (PHCL). The RPCM is a critical process status indicator, typically a digital signal or status bit. Its generation indicates that the objectives of the second high-pressure pressing stage have been achieved—that is, through the combined effects of high pressure and high temperature, the molten resin should theoretically fully penetrate the interfiber gaps within the interlocking zone, achieving densification (DEN) and good impregnation (GI) in this area. While generating the RPCM, the system also records the actual Pressure Holding Curve (PHC), including key parameters such as the pressure fluctuation range and actual hold time. The operator will see a clear "Resin Penetration Completion Mark" on the HMI, and the timer will reset to zero. The generation of the RPCM not only signals the achievement of the core objectives of this step but also provides a prerequisite signal for the subsequent cooling and final pressure relief. At this point, a green body has been formed in the mold with preliminary consolidation and the fibers fully wrapped by resin.

[0098] A laser confocal microscope is used to scan the surface of the product to verify that there is no exposed glass fiber and the porosity is less than 0.5%, and a semi-finished product quality certification code is generated.

[0099] After the resin penetration completion mark (RPCM) is generated, the pressing process ends, but the mold remains closed and under pressure from the SPV (preparing for subsequent cooling and holding). To quickly and non-destructively verify that the pressing results meet critical quality standards (especially the interlocking zone quality) in situ within the mold, the system uses an online laser confocal microscope (OLCM) to perform high-precision scanning of the surface of the newly formed semi-finished product (SFP), particularly the interlocking zone and its surrounding areas. The OLCM is typically integrated into the mold or implemented using a precision probe arm (PA) that can be inserted into the mold cavity. Its core principle is to focus a laser beam (LB) onto the sample surface, collect reflected light (RL) from the focal plane through a confocal optical path (COP), and acquire 3D topography information (3DTI) of the sample surface point by point and layer by layer using a point detector (PD, such as a photomultiplier tube (PMT) or avalanche photodiode (APD)) and a precision Z-axis scanner (PZS). The system controls the movement of the probe arm or the micro-motion of the mold, ensuring that the OLCM's scanning area (SA) covers the predefined critical interlock positions (CIP) and the adjacent large area boundary (LAB).

[0100] OLCM scans acquire high-resolution 3D point cloud data (3DPCD). The system processes this data using its built-in Surface Defect Intelligent Recognition Algorithm (SDIRA), focusing on two key indicators: 1) Fiber Exposure (FE): The algorithm analyzes the characteristics of each scan point (such as height, reflectivity, and color), identifying areas that are significantly higher than the surrounding resin matrix and exhibit typical fiberglass optical properties (such as high reflectivity and a specific texture). If any continuous exposed fiber area exceeding a preset threshold (e.g., 50 microns) in length or width is detected, the FE is deemed unqualified. 2) Surface Porosity (SP): The algorithm identifies points on the 3D topography that appear as pits (PT) with depths and diameters consistent with pore characteristics (e.g., diameter >5μm) and calculates the percentage (PCT) of these pores in the total scanned area. SDIRA outputs a detailed inspection report, including the presence of exposed glass (Boolean value: yes / no), the maximum exposed area detected (Maximum Exposure Area, MEA), the calculated surface porosity value (SP Value, SPV), and a visual heat map (Visualized Heat Map, VHM) marking the defect location.

[0101] The system automatically makes a determination based on the SDIRA analysis results. The criteria strictly adhere to pre-set process requirements: no exposed glass fiber (MEA = 0) and surface porosity < 0.5% (SPV < 0.5%). If both conditions are met, the PLC system automatically generates a Semi-finished Product Quality Certification Code (SPQCC). The SPQCC is typically a unique encrypted string (ES) or QR code (QRC) containing the batch number (BN), mold number (MN), timestamp (TS), key test results (SPV value, FE status), and the control system signature (SS) that generated the code. The SPQCC is immediately stored in the PHDL and product databases and can be displayed or printed via the HMI. This code indicates that the semi-finished product (SFP) in the current mold has passed critical quality verification during the pressing process and is eligible to proceed to the final cooling and finishing step. If the test results fail (e.g., exposed glass fiber or SPV ≥ 0.5%), the system will not generate an SPQCC. Instead, it will trigger a Quality Anomaly Alarm (QAA), locking the mold and prohibiting further operations until manual intervention is required. Regardless of pass / failure, the detailed OLCM scan data and SDIRA analysis report are fully preserved for quality traceability and process optimization analysis. A semi-finished product that passes the SPQCC is essentially a "semi-finished product without exposed glass fiber," as described in this step.

[0102] The phased pressure application process precisely controls the timing of pressure, ensuring sufficient resin melting and flow over large areas before applying high-pressure penetration to critical areas. This progressive molding approach optimizes resin distribution. Phased pressure application effectively eliminates the uneven resin distribution problem encountered in traditional processes, ensuring complete fiber impregnation. The resulting surface finish, free of exposed glass fibers, significantly improves the product's mechanical properties and aesthetic quality.

[0103] S205, for the semi-finished product, while maintaining the second pressure value, performing rapid cooling on the flange and corner areas, and performing slow cooling on the large surface area, to obtain a final product with balanced internal stress.

[0104] Specifically, under the trigger of the semi-finished product quality certification code, a liquid nitrogen injection module can be deployed in the mold flange area, and a warm water circulation pipeline can be activated in the large surface area to generate a cooling system startup instruction; Upon receiving the Semi-Finished Product Quality Certification Code (SPQCC) generated by laser confocal microscopy scanning—which verifies a porosity of less than 0.5% and the absence of exposed glass fibers—the system's Cooling Control Hub (CCH) immediately activates a differentiated cooling strategy. First, targeting the flange and corner regions (FCR), the system activates a Liquid Nitrogen Injection Module (LNIM) deployed on the mold sidewall. This module consists of a high-precision electromagnetic valve array (EVA), an insulated delivery line (IDL), and a micro-nozzle matrix (MNM). Each micro-nozzle has a diameter of approximately 0.3 mm and is tilted at a 15-degree angle toward the mold cavity surface. The nozzle spacing is optimized to 5 mm based on heat conduction simulations. The liquid nitrogen tank (LNT) maintains a low temperature of -196°C (-196°C), with the output pressure stabilized at 0.8 MPa via a PID pressure regulator (PR). When the CCH issues an activation command (AC), a solenoid valve opens within 10 milliseconds (ms), uniformly covering the FCR surface with a spray of liquid nitrogen (droplet size < 50 microns). Simultaneously, for the large surface region (LSR), the system activates a warm water circulation pipeline (WWCP) embedded within the mold. This pipeline utilizes a multi-channel serpentine design (MSD) with an 8 mm diameter. Warm water maintained at 80°C (TC) by a temperature controller (TC) is circulated at a flow rate of 3 meters per second (m / s) via a variable frequency pump (VFP). The switching of cooling media is monitored by the Safety Interlock System (SIS) to ensure that there is no risk of interference between the liquid nitrogen and warm water systems.All deployment statuses are mapped to the Virtual Mold Dashboard (VMD) in real time, and CCH ultimately generates a Cooling System Start Command (CSSC) containing device coordinates, media parameters, and timing logic.

[0105] Precise control of the liquid nitrogen injection module relies on a dynamic thermal feedback system (DTFS). Near each cooling unit in the flange area, embedded microthermocouples (MTCs) collect surface temperature (ST) at a frequency of 100 times per second (100 Hz). When the local temperature is detected to be above a preset threshold (e.g., 200°C), the DTFS calculates the liquid nitrogen flow compensation value (FCV) using a fuzzy control algorithm (FCA) and dynamically adjusts the valve opening (VO) of the corresponding solenoid valve. For example, if the temperature at a corner is 5°C higher due to material heat storage, the system increases the VO from 45% to 60%, and the liquid nitrogen flow rate from 0.5 liters / minute (L / min) to 0.7 liters / minute (L / min). The warm water circulation in the large area utilizes a zonal temperature control strategy (ZTS): the large area is divided into 20 independent control units (ICUs), each equipped with a separate temperature sensor (TS) and proportional valve (PV). If the temperature in a unit falls below the setpoint (e.g., 110°C), the PV opening is reduced to reduce the warm water flow; otherwise, the flow is increased. All temperature control data is transmitted to the CCH via Industrial Ethernet (IE) and integrated with the temperature data from the liquid nitrogen system to generate a global heat map (GHM). When the GHM indicates that the cooling medium is ready in all zones and the temperature fluctuation is less than ±1°C, the CCH sends a final confirmation signal (CS) to the actuator, completing the closed-loop verification of the cooling system start command.

[0106] To ensure repeatable cooling, the system performs medium flow field calibration (MFFC) during startup. The liquid nitrogen injection module verifies atomization uniformity using particle image velocimetry (PIV). Trace particles (TP) are injected into the injection area, and flow field images are captured with a high-speed camera (HSC). The liquid nitrogen coverage uniformity (CU) is analyzed to ensure it exceeds 95%. An ultrasonic flow meter (UFM) measures the flow rate (FR) of each branch pipe in the warm water circulation system, and a differential pressure sensor (DPS) verifies that the pipe is unobstructed (pressure drop < 0.05 MPa). After comparing the calibration data with preset process parameters (such as a liquid nitrogen pressure of 0.8 MPa and a warm water flow rate of 3 m / s), a System Health Report (SHR) is generated. Only when the report status is "Ready" (RD) is the cooling system start command marked as valid (VL). Otherwise, an alert is triggered and the process is suspended. This mechanism ensures precise and controllable subsequent cooling rates at the hardware level.

[0107] According to the cooling system startup instruction, the second pressure value is maintained, and the flange area is cooled to 80°C at a rate of 50°C / s, and the large surface area is cooled to 110°C at a rate of 5°C / s, and a real-time temperature gradient map is output; At the same time of the cooling system start command (CSSC) taking effect, the hydraulic power unit (HPU) strictly maintains the second pressure value (SPV, for example, 30 mega pascal (MPa)) to avoid the material from being debonded due to the shrinkage during cooling. At this time, the flange region (FR) liquid nitrogen injection module executes the rapid cooling program (RCP): through the temperature PID controller (TPC), the target cooling rate (TCR) is set to 50 degrees Celsius per second (50℃ / s), and the target termination temperature (TTT) is 80℃. To achieve this rate, the system uses feedforward-feedback compound control (FFCC). The feedforward control (FFC) is based on the material heat capacity database (MHCD) to pre-load the liquid nitrogen flow curve (FC), and the initial flow is set to 1.2 liters per minute (L / min); the feedback control (FBC) dynamically adjusts according to the real-time temperature difference (RTD = current temperature - target temperature): if the actual rate is lower than 50℃ / s, the liquid nitrogen supply is increased by the proportion of 0.1L / min flow per 1℃ / s deviation. When the temperature approaches 80℃ (such as 85℃), the system switches to slope decay control (SDC) to avoid overshoot (OS).

[0108] The warm water circulation system of the Large Surface Region (LSR) performs a Gradual Cooling Program (GCP): the target cooling rate (TCR) is set to 5 degrees Celsius per second (5°C / s) and the target termination temperature (TTT) is 110°C. Due to the large thermal inertia of this region, the system uses a Model Predictive Control (MPC) algorithm: based on a Phase Change Model (PCM), every 0.5 seconds the system predicts the Temperature Trend (TT) for the next 3 seconds and adjusts the warm water flow rate in advance. For example, if the prediction shows that the cooling rate will reach 6°C / s, the Variable Frequency Pump (VFP) rotation speed (RS) is immediately reduced, reducing the flow rate from 3 m / s to 2.5 m / s. At the same time, near the resin glass transition temperature (Tg) (e.g. 120°C-115°C), the system starts an Isothermal Holding Stage (IHS): the cooling is paused for 10 seconds (s) to allow the Molecular Chain Relaxation (MCR) and reduce internal stress generation. The whole process is monitored by a Distributed Temperature Sensing Network (DTSN) that contains 56 Platinum Resistance Thermometers (PRTs) with a spatial resolution of 2 millimeters (mm).

[0109] The temperature data during the cooling process is processed by a Spatio-temporal Fusion Engine (SFE) to generate a Real-time Temperature Gradient Map (RTGM). This map contains three layers of key information: Axial Gradient (AG): the temperature value every 0.1 millimeter (mm) along the thickness direction of the product, measured by a Micro Thermal Probe (MTP) implanted in the mold; Planar Gradient (PG): the temperature of every 5 millimeter (mm) grid point on the surface of the cavity, captured by an Infrared Thermal Imager (ITI) at 30 frames per second (30 fps); Temporal Gradient (TG): Temperature curve (TC) at key locations (such as the interface between the flange and the large surface), with a sampling frequency of 1 kHz.

[0110] The spectrum is visualized using color coding (CC): blue indicates low-temperature areas (e.g., 80°C for the flange), red indicates high-temperature areas (e.g., 110°C for the main surface), and a gradient color spectrum is used to display the transition zone (TZ). If the temperature in any area exceeds the process window (PW; ±3°C / s for the flange area and ±0.5°C / s for the main surface), the spectrum automatically marks the anomaly area (AA) and triggers an alarm (AL). The final spectrum is exported to the quality traceability system (QTS) in the form of a process data packet (PDP).

[0111] The cooling rate is adjusted according to the real-time temperature gradient map, and when the peak value of the characteristic frequency detected by the acoustic emission sensor drops to the safety threshold, a stress balance confirmation code is output; The real-time temperature gradient map (RTGM) is fed into the Cooling Rate Regulator (CRR), the core of which is the Gradient-Driven Control Algorithm (GDCA). The algorithm first identifies high-risk zones (HRZs) within the map, such as the temperature jump zone (>15°C / mm) at the interface between the flange and the main surface. To address the HRZ, the system initiates a Gradient Smoothing Program (GSP): This reduces the liquid nitrogen injection intensity in the flange area (e.g., from 1.0 L / min to 0.7 L / min) while simultaneously increasing the warm water temperature in the main surface area (e.g., from 80°C to 85°C), reducing the slope of the transition zone to <10°C / mm. For non-uniform shrinkage zones (NSZs), which appear as localized high-temperature islands, spot compensation cooling (SCC) is activated. An auxiliary micro-nozzle (AMN) deployed on the mold sprays 0.1-second pulses (Pulse Width, PW=100ms) of liquid nitrogen at these locations to eliminate heat buildup. After each adjustment, the system delays for 0.5 seconds to allow the thermal field to equilibrate. The RTGM is then re-acquired and iteratively optimized (IO) until the global temperature gradient meets the preset Equilibrium Index (EI>0.95).

[0112] The physical effects of cooling rate regulation are monitored in real time using an acoustic emission sensor array (AESA). Eight broadband sensors (frequency range: 50kHz-1MHz) are positioned in a star layout (SL) against the mold's outer wall, capturing elastic waves (EW) generated by stress release within the material. The raw waveform is converted to a spectrogram (SG) using a Fast Fourier Transform (FFT). The system focuses on monitoring two characteristic frequency peaks (CFP): Micro-crack Generation Peak (MGP): frequency band 300-400 kHz, reflecting micro defects caused by local stress concentration; Molecular Chain Slip Peak (MCSP): frequency band 100-150 kHz, characterizing the viscoelastic relaxation behavior of polymer chains.

[0113] When cooling nears the end point (flange area <100°C), the system initiates Peak Attenuation Tracking (PAT). If the MGP amplitude exceeds 5 volts (V), it indicates excessive stress and requires cooling to be slowed; if the MCSP amplitude is <0.5 volts (V), it indicates insufficient relaxation and requires extended holding. The Safety Threshold (ST) is defined as: MGP <2 volts (V) and MCSP >1 volt (V), indicating internal stress equilibrium (ISE).

[0114] When the acoustic emission system detects that the MGP of all sensor channels is less than 2 volts (V) for 10 seconds (s) and the MCSP is stable in the range of 1-1.5 volts (V), the acoustic emission analysis engine (AEAE) generates a spectrum compliance certificate (SCC). This certificate is input into the stress fusion arbiter (SFA) together with the real-time temperature gradient map (RTGM). The SFA performs dual verification (DV): Spatial Verification (SV): Check whether there is a residual gradient >10°C / mm in the RTGM; Frequency Domain Verification (FDV): Verify that the MGP / MCSP ratio is less than 1.5.

[0115] If both verifications pass (Pass, PS), a 32-bit encrypted Stress Equilibrium Confirmation Code (SECC) is output, containing a snapshot of key parameters (e.g., maximum temperature gradient 8.7°C / mm, average MCSP amplitude 1.2V). If verification fails (Fail, FL), a cooling parameter re-optimization cycle (RC) is triggered until a SECC is generated.

[0116] Based on the stress balance confirmation code, the ejection mechanism is controlled to release the pressure at a uniform speed within 5 seconds to remove the component, and finally a balanced product with a residual stress difference of less than 10MPa is obtained.

[0117] The Stress Balance Confirmation Code (SECC) unlocks the Ejection Mechanism Control Authority (EMCA). This authority consists of three parts: Hydraulic Pressure Relief Unit (HPRU): A servo cylinder (SC) with a linear displacement sensor (LDS) is responsible for uniform pressure relief. Ejector Pin Array (EPA): 12 sets of Silicon Nitride Ceramic Pins (SNCP), with a travel accuracy of ±0.01 mm; Vacuum Suction Cup Set (VSCS): Prevents products from falling freely.

[0118] The control center (CCH) sends a Pressure Relief Profile Command (PRPC): It requires a linear decrease in pressure from the secondary pressure value (SPV = 30 MPa) to 0.1 MPa over 5 seconds, with a constant slope (CS) of 6 MPa / s. The hydraulic unit (HPU) regulates the flow rate (FR) using a high-response proportional valve (HRPV), while the LDS provides real-time feedback on the pressure curve (PC). If a pressure fluctuation greater than ±0.5 MPa is detected, the system instantly activates the Pressure Compensation Algorithm (PCA) to ensure uniformity (UF > 99%).

[0119] During the depressurization process, ejection synchronization control (ESC) is executed simultaneously. When the pressure drops to 5 MPa (4.17 seconds after depressurization begins), the ejector pin array (EPA) pre-lifts (PL) 0.2 mm at a speed of 0.5 mm / s (mm / s) to eliminate static friction (SF) between the part and the mold. At the moment the pressure returns to zero (t=5s), the ejector pins eject the part from the cavity (Cavity Ejection Stroke, CES=15mm) at a speed of 2 mm / s (mm / s). Simultaneously, the vacuum suction cup (VSCS) builds negative pressure (Vacuum Pressure, VP=-80 kPa) within 0.1 seconds to hold the top surface of the part. The ejection trajectory is monitored by a laser interferometer (LI). If deviation (DEV>0.05mm) is detected, the ejector motion sequence (MS) is immediately adjusted to correct the deviation. After the product is demolded, the robotic arm (RA) transfers it to the buffer soft pad (BSP) to avoid collision damage.

[0120] The residual stress difference (RSD) of the final product is confirmed by the dual-channel verification method (DVM): Non-destructive Testing (NDT): Use a portable X-ray diffractometer (PXRD) to scan 10 key points on the product surface (such as the flange root and the center of the large surface) and calculate the stress range (SR). If the SR is less than 10 megapascals (MPa), the test is marked as NDT passed (PS).

[0121] Destructive Sampling Test (DST): One test specimen (TS) is randomly selected from every 50 units produced and subjected to 3D photoelastic scanning (3DPS). A full-field stress cloud map (FSCM) is generated. If the maximum stress difference (MSD) is less than 10 MPa, the test is marked as DST passed (PS).

[0122] These two verification data, along with the stress balance confirmation code (SECC) and cooling parameters, are combined to generate the Final Product Quality Dossier (FPQD), marking the achievement of a balanced product with a residual stress difference (RSD) <10MPa (10BP). This product can be directly assembled, meeting the service requirements of high-precision structural components.

[0123] A differentiated cooling strategy effectively controls stress distribution within the material by prioritizing cooling of high-stress areas under high pressure. This temperature gradient management technique achieves self-balancing stress within the part. The zoned cooling process significantly reduces internal stress levels in the part, avoiding deformation associated with traditional uniform cooling. This stress-balanced final part offers improved dimensional stability and long-term performance.

[0124] It can be seen that continuous glass fiber sheets and long glass fiber balls are laid in partitions in the mold cavity to obtain a composite blank with complete positioning; local pre-pressure is applied to the interlocking areas of the composite blank to form a mechanical bite structure to obtain a preform with an interface interlocking structure; a pressure distribution model is established according to the geometric characteristics of the product to obtain a differentiated pressure parameter combination; the preform is placed in a heated mold and pressure is applied in stages according to the pressure parameter combination to obtain a semi-finished product with no exposed glass fiber; for the semi-finished product, while maintaining the second pressure value, the flange and corner areas are rapidly cooled, and the large surface area is slowly cooled to obtain a final product with balanced internal stress, thereby achieving efficient coordination between the continuous glass fiber sheets and the long glass fiber balls and improving the mechanical properties and structural integrity of the product.

[0125] Another embodiment of the present invention provides a long glass fiber molding system, see Figure 3 , the system may include: Laying module 301 is used to lay continuous glass fiber sheets and long glass fiber balls in the mold cavity in different areas. The continuous glass fiber sheets cover the large surface area of ​​the product, and the long glass fiber balls are pre-placed in the flange and corner areas. The edges of the two overlap to form an interlocking area to obtain a fully positioned composite body. The embedding module 302 is used to apply local pre-compression to the interlocking area of ​​the composite body so that the long glass fiber balls are embedded in the fiber grid gaps of the continuous glass fiber sheet to form a mechanical interlocking structure, thereby obtaining a preform with an interlocking interface structure; Establishing module 303, for establishing a pressure distribution model based on the geometric characteristics of the product, generating a first pressure value for the large surface area, and generating a second pressure value for the flange and corner area, wherein the second pressure value is greater than the first pressure value, thereby obtaining a differentiated pressure parameter combination; A pressure module 304 is configured to place the preform in a heated mold and apply pressure in stages according to the pressure parameter combination, wherein a first pressure value is applied to press the large surface area to allow the resin to melt and flow, and a second pressure value is applied to press the interlocking area to allow the molten resin to fill the fiber gaps, thereby obtaining a semi-finished product without exposed glass fibers; The cooling module 305 is used to quickly cool the flange and corner areas of the semi-finished product and slowly cool the large surface area while maintaining the second pressure value, so as to obtain a final product with balanced internal stress.

[0126] An embodiment of the present invention further provides a storage medium storing a computer program, wherein the computer program is configured to execute the steps of any one of the above method embodiments when running.

[0127] Specifically, in this embodiment, the above-mentioned storage medium may be configured to store a computer program for performing the following steps: S201, laying continuous glass fiber sheets and long glass fiber balls in the mold cavity in different areas, wherein the continuous glass fiber sheets cover the large surface area of ​​the product, and the long glass fiber balls are pre-placed in the flange and corner areas, and the edges of the two overlap to form an interlocking area, thereby obtaining a fully positioned composite body; S202, applying local pre-compression to the interlocking region of the composite body so that the long glass fiber balls are embedded in the fiber grid gaps of the continuous glass fiber sheet to form a mechanical interlocking structure, thereby obtaining a preform having an interlocking interface structure; S203: Establishing a pressure distribution model based on the product's geometric characteristics, generating a first pressure value for the large surface area and a second pressure value for the flange and corner areas, wherein the second pressure value is greater than the first pressure value, thereby obtaining a differentiated pressure parameter combination; S204, placing the preform in a heating mold, and pressing in stages according to the pressure parameter combination, wherein the first pressure value is used to press the large area region to make the resin melt and flow, and the second pressure value is used to press the interlocking region to make the molten resin fill the fiber gap, so as to obtain a semi-finished product without glass fiber exposure; S205, for the semi-finished product, maintaining the second pressure value, and implementing rapid cooling on the flange and corner regions and slow cooling on the large area region to obtain a final product with balanced internal stress.

[0128] The embodiment of the present application also provides an electronic device, including a memory and a processor, the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any of the above method embodiments.

[0129] Specifically, the above electronic device can further include a transmission device connected with the processor and an input and output device connected with the processor.

[0130] Specifically, in the embodiment, the processor can be configured to execute the following steps through the computer program: S201, laying continuous glass fiber plate and long glass fiber material in the mold cavity in zones, wherein the continuous glass fiber plate covers the large area region of the product, the long glass fiber material is prepositioned in the flange and corner regions, the edges of the two overlap to form an interlocking region, and a positioned composite blank is obtained; S202, applying local pre-pressure to the interlocking region of the composite blank to make the long glass fiber material embedded in the fiber grid gap of the continuous glass fiber plate to form a mechanical interlocking structure, and a preform with an interfacial interlocking structure is obtained; S203, establishing a pressure distribution model according to the geometric characteristics of the product, generating a first pressure value for the large area region, generating a second pressure value for the flange and corner regions, and the second pressure value is greater than the first pressure value, and obtaining a differentiated pressure parameter combination; S204, placing the preform in a heating mold, and pressing in stages according to the pressure parameter combination, wherein the first pressure value is used to press the large area region to make the resin melt and flow, and the second pressure value is used to press the interlocking region to make the molten resin fill the fiber gap, so as to obtain a semi-finished product without glass fiber exposure; S205, for the semi-finished product, maintaining the second pressure value, and implementing rapid cooling on the flange and corner regions and slow cooling on the large area region to obtain a final product with balanced internal stress.

[0131] The above describes in detail the structure, features and effects of the present invention based on the embodiments shown in the drawings. The above is only a preferred embodiment of the present invention, but the scope of implementation of the present invention is not limited to what is shown in the drawings. Any changes made in accordance with the concept of the present invention, or modifications to equivalent embodiments with equivalent changes, which do not exceed the spirit covered by the description and drawings, should be within the scope of protection of the present invention.

Claims

1. A long glass fiber compression molding method, characterized in that: The method comprises: Continuous glass fiber sheets and long glass fiber balls are laid in the mold cavity in different areas. The continuous glass fiber sheets cover the large surface area of ​​the product, and the long glass fiber balls are pre-placed in the flange and corner areas. The edges of the two overlap to form an interlocking area to obtain a fully positioned composite body. Applying local pre-compression to the interlocking area of ​​the composite body so that the long glass fiber balls are embedded in the fiber grid gaps of the continuous glass fiber sheet to form a mechanical bite structure, thereby obtaining a preform with an interface interlocking structure; A pressure distribution model is established based on the product's geometric characteristics. A first pressure value is generated for the large surface area, and a second pressure value is generated for the flange and corner areas. The second pressure value is greater than the first pressure value, resulting in a differentiated pressure parameter combination. Placing the preform in a heated mold and applying pressure in stages according to the pressure parameter combination, wherein a first pressure value is used to press the large surface area to allow the resin to melt and flow, and then a second pressure value is used to press the interlocking area to allow the molten resin to fill the fiber gaps, thereby obtaining a semi-finished product without exposed glass fibers; For the semi-finished product, while maintaining the second pressure value, the flange and corner areas are rapidly cooled, and the large surface area is slowly cooled to obtain a final product with balanced internal stress.

2. The method according to claim 1, characterized in that The continuous glass fiber sheet and the long glass fiber mass are laid in the mold cavity in different areas, wherein the continuous glass fiber sheet covers the large surface area of ​​the product, and the long glass fiber mass is pre-placed in the flange and corner areas, and the edges of the two overlap to form an interlocking area, thereby obtaining a composite body with complete positioning, including: Extract the curvature distribution characteristics of the product CAD model, divide the cavity surface into large-surface low-curvature areas and high-stress flange / corner areas using the curvature gradient algorithm, and output the regional boundary coordinate set; Based on the region boundary coordinate set, a dynamic laser grid is projected onto the mold cavity surface, guiding the robotic arm to precisely lay the continuous fiberglass sheet on the large laser-marked area, generating a sheet positioning completion signal. The fiber arrangement image on the plate surface is acquired through a micron-level optical scanner, and the average gap size of the fiber grid is calculated using the directional gradient histogram algorithm, and a gap distribution heat map is output; Based on the gap distribution heat map, long glass fiber balls are stacked and laid at the edges of the panels in the flange and corner areas, so that the ball covers the panel edge width to 1.5 times the average gap size, and a composite blank positioning report with overlapping interlocking areas is generated.

3. The method according to claim 2, characterized in that The method of applying local pre-compression to the interlocking area of ​​the composite body so that the long glass fiber balls are embedded in the fiber grid gaps of the continuous glass fiber sheet to form a mechanical interlocking structure, thereby obtaining a preform having an interface interlocking structure, comprises: Based on the three-dimensional topological map of the composite body, an infrared focused heating device is deployed in the interlocking area to soften the surface resin of the material mass to a viscous flow state and output a signal indicating that the resin viscosity meets the standard; Receive the signal that the resin viscosity reaches the standard, activate the micro-indenter array with a diameter matching the fiber gap, apply 0.5MPa local pressure, and output a real-time pressure distribution cloud map; The pressure head stroke is adjusted through the real-time pressure distribution cloud map. When the fiber optic sensor detects that the material mass embedding depth reaches 70% of the fiber diameter, it outputs the bite depth authentication code. Based on the occlusal depth authentication code, microfocus CT scanning is triggered to reconstruct the three-dimensional model of the interlocking area and output the interface interlocking structure authentication file.

4. The method according to claim 3, characterized in that The pressure distribution model is established based on the geometric characteristics of the product, a first pressure value is generated for the large surface area, and a second pressure value is generated for the flange and corner area, and the second pressure value is greater than the first pressure value, to obtain a differentiated pressure parameter combination, including: Import product geometric feature parameters, apply standard molding pressure to perform nonlinear contact analysis, and output the equivalent stress distribution matrix of each node in the cavity; Perform regional cluster analysis on the equivalent stress distribution matrix, screen flanges and corner areas where stress values ​​exceed 85% of the material yield strength, and generate a coordinate set for the high-pressure demand area; According to the material rheology database, with the resin critical flow shear rate as the constraint condition, the minimum molding pressure value of the large surface area is solved as the first pressure reference value; In the high-pressure demand area coordinate set, the first pressure reference value is input, the compensation pressure coefficient is dynamically calculated through the deep reinforcement learning model, and the second pressure value optimization solution is output; Based on the optimization scheme of the first pressure reference value and the second pressure value, a spatially continuous pressure distribution function is constructed, and a parameter combination configuration file containing a pressure gradient curve is generated.

5. The method according to claim 4, characterized in that The preform is placed in a heated mold and pressure is applied in stages according to the pressure parameter combination, wherein the large surface area is first pressed with a first pressure value to melt and flow the resin, and then the interlocking area is pressed with a second pressure value to allow the molten resin to fill the fiber gaps, thereby obtaining a semi-finished product without exposed glass fibers, including: Read the parameter combination configuration file, set a constant temperature field of 180℃ in the large area of ​​the heated mold, set a high temperature field of 200℃ in the interlocking area, and output a temperature field balance ready signal; After receiving the temperature field ready signal, the large surface area is pressed at the first pressure value, and the phase switching instruction is triggered when the resin front reaches the boundary of the interlocking zone through the melt flow sensor; Based on the phase switching instruction, the pressure is increased to the second pressure value within 0.2 seconds and maintained for 10 seconds, using high pressure to force the molten resin to penetrate into the fiber gaps, and a resin penetration completion mark is output; A laser confocal microscope is used to scan the surface of the product to verify that there is no exposed glass fiber and the porosity is less than 0.5%, and a semi-finished product quality certification code is generated.

6. The method according to claim 5, characterized in that The semi-finished product is subjected to rapid cooling on the flange and corner areas and slow cooling on the large surface area while maintaining the second pressure value, to obtain a final product with balanced internal stress, including: Triggered by the semi-finished product quality certification code, a liquid nitrogen injection module is deployed in the mold flange area, a warm water circulation pipeline is activated in the large surface area, and a cooling system startup instruction is generated; According to the cooling system startup instruction, the second pressure value is maintained, and the flange area is cooled to 80°C at a rate of 50°C / s, and the large surface area is cooled to 110°C at a rate of 5°C / s, and a real-time temperature gradient map is output; The cooling rate is adjusted according to the real-time temperature gradient map, and when the peak value of the characteristic frequency detected by the acoustic emission sensor drops to the safety threshold, a stress balance confirmation code is output; Based on the stress balance confirmation code, the ejection mechanism is controlled to release the pressure at a uniform speed within 5 seconds to remove the component, and finally a balanced product with a residual stress difference of less than 10MPa is obtained.

7. A long glass fiber molding system, characterized in that: The system comprises: The laying module is used to lay continuous glass fiber sheets and long glass fiber balls in the mold cavity in different areas. The continuous glass fiber sheets cover the large surface area of ​​the product, and the long glass fiber balls are pre-placed in the flange and corner areas. The edges of the two overlap to form an interlocking area to obtain a fully positioned composite body. An embedding module is used to apply local pre-compression to the interlocking area of ​​the composite body so that the long glass fiber material balls are embedded in the fiber grid gaps of the continuous glass fiber sheet to form a mechanical bite structure and obtain a preform with an interface interlocking structure; Establish a module for establishing a pressure distribution model based on the geometric characteristics of the product, generating a first pressure value for the large surface area and a second pressure value for the flange and corner area, and the second pressure value is greater than the first pressure value, thereby obtaining a differentiated pressure parameter combination; a pressurizing module, configured to place the preform in a heated mold and apply pressure in stages according to the pressure parameter combination, wherein a first pressure value is applied to press the large surface area to allow the resin to melt and flow, and a second pressure value is applied to press the interlocking area to allow the molten resin to fill the fiber gaps, thereby obtaining a semi-finished product without exposed glass fibers; The cooling module is used to quickly cool the flange and corner areas of the semi-finished product and slowly cool the large area while maintaining the second pressure value, so as to obtain a final product with balanced internal stress.

8. The system according to claim 7, characterized in that The laying module is specifically used for: Extract the curvature distribution characteristics of the product CAD model, divide the cavity surface into large-surface low-curvature areas and high-stress flange / corner areas using the curvature gradient algorithm, and output the regional boundary coordinate set; Based on the region boundary coordinate set, a dynamic laser grid is projected onto the mold cavity surface, guiding the robotic arm to precisely lay the continuous fiberglass sheet on the large laser-marked area, generating a sheet positioning completion signal. The fiber arrangement image on the plate surface is acquired through a micron-level optical scanner, and the average gap size of the fiber grid is calculated using the directional gradient histogram algorithm, and a gap distribution heat map is output; Based on the gap distribution heat map, long glass fiber balls are stacked and laid at the edges of the panels in the flange and corner areas, so that the ball covers the panel edge width to 1.5 times the average gap size, and a composite blank positioning report with overlapping interlocking areas is generated.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 6 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 6.

Images