Automobile seat back plate mold forming state analysis system based on finite element simulation

Abstract

The application discloses a kind of based on finite element simulation automobile seat back plate mould forming state analysis system, it is related to injection molding simulation technical field;The system is used to solve the problem that original convergence mark is deviated and the position field record of weld mark is inaccurate due to local flow stagnation after automobile seat back plate reinforcing rib root portion suddenly advances.The system includes rib root timing extraction module, pseudo-convergence phase discrimination module, rollback and re-push analysis module and forming state write-back module;Rib root timing extraction module forms rib root feature node record;Pseudo-convergence phase discrimination module carries out timing sudden advance analysis and direction phase comparison to the node corresponding to original convergence mark, obtains pseudo-convergence disposal mark;Rollback and re-push analysis module generates rollback filling factor trajectory, updates the local filling state in the range of affected control volume and executes front re-push, obtains the modified front arrival sequence and convergence node sequence;Forming state write-back module regenerates rib root weld mark position field and writes into forming state database.

Description

Technical Field

[0001] This invention relates to the field of injection molding simulation technology, and more specifically, to a system for analyzing the molding state of automotive seat back panel molds based on finite element simulation. Background Technology

[0002] Automotive seat back panels are typically manufactured using injection molding. Their structure includes a thin-walled main body, reinforcing ribs, rib root fillets, snap fasteners, screw posts, and edge flanges. To evaluate the molding state during mold development, existing technologies often utilize injection filling simulation to analyze melt front advancement, filling time, merging nodes, and weld line locations during the injection process. This type of simulation can record the node filling state based on finite element meshes, control volume advancement, or a combination of both. The weld line distribution results obtained from the simulation help determine whether the mold structure, gate arrangement, and local flow state meet design requirements. This type of simulation generally solves the pressure field based on finite element meshes and advances the melt front according to the node filling factor in the control volume. Merging nodes are formed when two melt fronts meet or adjacent control volumes merge. The weld line location field is then reconstructed based on the sequence of merging nodes. The above content corresponds to the descriptions of automotive seat back panel structure, finite element-control volume hybrid filling simulation, node filling factor, merging nodes, and weld line location field in this application specification.

[0003] However, there are often significant abrupt changes in wall thickness at the intersection of the thin-walled main body and reinforcing ribs in automotive seat back panels. Furthermore, the fillet radius at the rib root and the initial section of the rib sidewall are prone to forming localized high aspect ratio units. When the melt advances through these rib root areas, it may exhibit a localized filling phenomenon of initial stagnation followed by sudden acceleration. Existing methods that directly generate the location of the merging node and weld line based on the filling factor advancement results are prone to misidentifying this localized acceleration process as a true merging of the melt front. This results in the weld line location being recorded upstream of the rib root or on the thin-walled main body side, leading to an inaccurate reflection of the molding state of the true rib sidewall merging area. Consequently, existing molding state analysis results suffer from inaccurate merging node location offsets and weld line location fields in localized rib root areas, making it difficult to provide a reliable basis for subsequent mold verification, localized structural adjustments, and molding risk assessment. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a system for analyzing the forming state of automotive seat back panel molds based on finite element simulation, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] The automotive seat back panel mold forming state analysis system based on finite element simulation includes the following modules: rib root time sequence extraction module, which is used to read finite element mesh files and filling simulation result files containing node filling states, determine rib root characteristic nodes based on wall thickness mutation ratio, aspect factor and structural region labels, and form node records containing filling factor time series and original confluence markers.

[0007] The pseudo-merging phase discrimination module is used to perform temporal advance analysis and directional phase comparison on the root feature nodes corresponding to the original merging markers to obtain pseudo-merging disposal markers including candidates to be cleared, candidates to be delayed, and candidates to be retained.

[0008] The backtracking and re-pushing analysis module is used to determine the phase lag step and upstream unidirectional advance rate based on the pseudo-merging disposal marker and node records, and generate a backtracking filling factor trajectory. The backtracking filling factor trajectory starts from the filling factor of the time step before the rate inflection point, advances linearly according to the upstream unidirectional advance rate, and replaces the original filling factor from the rate inflection point. The affected control volume range is determined based on the adjacency range of the pseudo-merging node and the original arrival step. The node state outside the range is fixed, and the local filling state within the affected control volume range is updated with the replaced filling factor. Based on the updated local filling state, the frontier re-pushing is performed to obtain the corrected frontier arrival sequence and the corrected merging node sequence.

[0009] The forming state write-back module is used to regenerate the weld line position field of the rib root according to the corrected confluence node sequence and write it into the forming state database.

[0010] In a preferred embodiment, the rib root timing extraction module determines the rib root feature node by: calculating the wall thickness mutation ratio and the aspect factor of the adjacent unit of the node, wherein the wall thickness mutation ratio is the ratio of the minimum to the maximum nominal wall thickness of all adjacent units of the node, and the aspect factor of the adjacent unit is the ratio of the longest side length to the shortest feature height among the adjacent units of the node; when the wall thickness mutation ratio of the node is less than the wall thickness mutation threshold, the aspect factor of at least one adjacent unit is greater than the aspect threshold, and the node is located within the structural adjacency range of the thin-walled main body region, the stiffener sidewall region, or the stiffener root region, the node is recorded as a rib root feature node; the filling factor time series, node arrival step, upstream neighbor chain, opposite side branch chain, and original convergence mark are written into the node record of the rib root feature node; wherein the upstream neighbor chain is a sequence of consecutive adjacent nodes obtained by tracing back from the rib root feature node to the earlier arrival direction of the melt, and the opposite side branch chain is a sequence of consecutive adjacent nodes obtained by searching along the opposite side flow branch from the rib root feature node.

[0011] In a preferred embodiment, updating the local fill state within the affected control volume range with the replaced fill factor, and performing a frontier re-push based on the updated local fill state, includes:

[0012] Using the local filling state at the time step before the rate inflection point as the restart state, the arrival step, filling factor, pressure boundary, and source direction label of the nodes outside the affected control volume range are fixed as the original simulation results.

[0013] For pseudo-merging nodes within the affected control volume range, the original filling factor is replaced by a backfilling filling factor trajectory starting from the rate inflection point.

[0014] For nodes to be updated within the affected control volume, excluding pseudo-merging nodes, a source neighbor set is established at each time step. The source neighbor set includes nodes that have reached the reaching factor threshold in the previous time step and have a grid adjacency relationship with the node to be updated.

[0015] When the source neighbor set is empty, the correction filling factor of the node to be updated in the previous time step is maintained. When the source neighbor set is not empty, the source neighbor node with the largest correction filling factor in the previous time step is selected first. If there are multiple source neighbor nodes with the same correction filling factor, the source neighbor node that arrives earliest after correction is selected. If there are still multiple candidate source neighbor nodes, the source neighbor node with the smallest angle between its source direction and the original source direction of the node to be updated is selected. If the angle is still the same, a unique master source node is determined in ascending order of node number, and the correction filling factor of the node to be updated is updated based on the correction filling factor of the master source node in the previous time step and the non-negative value of the filling factor increment of the node to be updated in the original filling simulation result of adjacent time steps.

[0016] When the correction filling factor of the node to be updated reaches or exceeds the arrival factor threshold for the first time, record the correction arrival step of the node to be updated and the source direction label corresponding to the main source node.

[0017] When two different source direction labels appear in the same node or adjacent control volumes within a time step not greater than the true merging phase threshold, and both source directions reach the arrival factor threshold, they are re-recorded as true merging nodes.

[0018] In a preferred embodiment, the affected control volume range is determined by the number of affected layers and the original arrival time sequence, including: starting from the pseudo-merging node, searching for candidate control volumes layer by layer according to the mesh adjacency relationship along the direction of the melt front advancing from early to late in the original filling simulation; when there is a shared node or shared edge between two control volumes, it is recorded as one layer of adjacency relationship; for any candidate control volume, when the shortest adjacency layer between it and the pseudo-merging node does not exceed the number of affected layers, and its original arrival step is not earlier than the rate inflection point, the candidate control volume is included in the affected control volume range; control volumes whose original arrival step is earlier than the rate inflection point are recorded as boundary fixed control volumes. Boundary fixed control volumes do not participate in the filling factor rewriting, but their original filling factor, original arrival step, and source direction are used as boundary inputs for local front re-push; wherein, the number of affected layers is determined according to the local flow transition range of the rib root, and the affected control volume range covers the local range of the rib root inlet, the starting section of the rib sidewall, and the adjacent thin-walled body.

[0019] In a preferred embodiment, after the candidate control volume is included in the affected control volume range, when the backtracking and re-pushing analysis module performs frontier re-pushing: when the corrected frontier leaves the affected control volume range, or when all nodes within the affected control volume range reach the filling completion state, the current local frontier re-pushing is stopped; when the affected control volume ranges corresponding to multiple pseudo-merging nodes overlap, local frontier re-pushing is performed sequentially according to the rate inflection points of each pseudo-merging node in ascending order; the corrected filling factor, corrected arrival step, and source direction label in the overlapping control volume that has completed re-pushing first are used as the boundary states when subsequent pseudo-merging nodes perform local frontier re-pushing.

[0020] In a preferred embodiment, the pseudo-convergence phase discrimination module performs temporal advance analysis and directional phase comparison, including: performing adjacent difference processing on the filling factor time series of the root feature node to determine the time step when the filling factor increment reaches its maximum value as the rate inflection point; determining the advance ratio based on the ratio of the peak advance amount at the rate inflection point to the upstream unidirectional advance rate, and determining an abnormal advance state when the advance ratio is greater than the advance ratio threshold and the number of stagnation steps before the rate inflection point is not less than the number of stagnation steps threshold; reading the arrival steps of each node in the upstream neighbor chain according to the direction from upstream to the root feature node, and determining a stable single advance state when the arrival steps of two or more consecutive upstream nodes gradually increase along this direction. The advancing front is determined based on the maximum fill factor and earliest arrival step of the opposite branch chain before the rate inflection point. The opposite branch state includes missing advancing front, misphase advancing front, and valid advancing front. When a stable unidirectional advancing front exists, the opposite branch state is missing advancing front, and there is an abnormal advance state, the root feature node is marked as a clearance candidate. When a stable unidirectional advancing front exists, the opposite branch state is misphase advancing front, and there is an abnormal advance state, the root feature node is marked as a delay candidate. When the opposite branch state is valid advancing front, or there is no abnormal advance state, or there is no stable unidirectional advancing front, the root feature node is marked as a retention candidate.

[0021] In a preferred embodiment, the rollback and re-push analysis module determines the final action based on the pseudo-merge handling flag and the phase lag step number. The final action includes clearing, delaying, and retaining.

[0022] When a pseudo-merging is marked as a candidate for removal and the phase lag step is not less than the large lag step threshold, the final action is determined to be removal, the merging mark of the node in the original merging node sequence is revoked, and a local frontier push is performed.

[0023] When a pseudo-merging is marked as a candidate to be retained, or when the phase lag step number for clearing a candidate is less than the large lag step threshold, the final action is determined to be retention, and the original merge mark remains unchanged.

[0024] When a pseudo-merging action is marked as a delayed candidate, the final action is determined to be delayed, and a merge delay verification window is constructed based on the time range between the rate inflection point and the rate inflection point plus the phase lag step. Within the merge delay verification window, the original merge mark of the node is not confirmed, and the unidirectional leading edge is maintained according to the backfill factor trajectory. After the window ends, based on whether the opposite branch chain reaches the effective threshold of the opposite branch, and whether the difference between the arrival step of the opposite branch and the arrival step after the correction of the node is not greater than the true merge phase threshold, a new true merge node is generated or the original merge mark is revoked.

[0025] In a preferred embodiment, after the forming state write-back module regenerates the weld line position field of the rib root, it also calculates the weld line migration record, including: the migration amount of the weld line migration record, calculated based on the distance between the center lines of the corresponding weld line segments in the original weld line position field and the corrected weld line position field; the migration direction of the weld line migration record, determined by pointing from the structural region where the original position is located to the structural region where the corrected position is located; when the migration amount is not less than the effective migration threshold, the migration direction is recorded; when the migration amount is less than the effective migration threshold, it is recorded as no significant migration.

[0026] In a preferred embodiment, after writing the corrected weld line position field, the molding state write-back module generates a molding risk level based on the corrected weld line position, structural region label, the junction angle at the actual junction node, the number of pushes, and the migration amount, and generates cause location information for areas with high molding risk levels.

[0027] The confluence angle is the angle between the propagation directions of the two melt fronts at the actual confluence node. When the corrected weld line is located on the sidewall of the reinforcing rib or in the transition zone of the rib root fillet, and the confluence angle is less than the risk threshold for small confluence angles or the number of re-pushes is not less than the threshold for the number of re-pushes, the area is determined to be of a high forming risk level. When the corrected weld line is still located in the thin-walled main body area and the migration amount is less than the effective migration threshold, the area is determined to be of a medium forming risk level. When the corrected weld line is far away from the stress concentration areas of the rib root, snap-fit ​​seat, and screw column structure, the area is determined to be of a low forming risk level.

[0028] The cause location information includes the original confluence mark forward movement record corresponding to the sudden advance after the rib root stagnation, the flow path record from the nearest gate to the rib root, the record of the sudden change in local resistance at the rib root fillet, and the corresponding mold verification position.

[0029] In a preferred embodiment, the backtracking and re-pushing analysis module generates a backtracking filling factor trajectory, including: reading the rate inflection point of the root feature node and the arrival step of the nearest upstream neighbor, and determining the time step difference between the two as the phase lag step number; reading the average filling factor increment of the upstream neighbor chain within a preset statistical window before the rate inflection point, and determining the average filling factor increment as the upstream unidirectional propagation rate; taking the original filling factor of the time step before the rate inflection point as the backtracking starting point, generating a backtracking filling factor step by step from the rate inflection point according to the upstream unidirectional propagation rate; limiting the backtracking filling factor to a value range of 0 to 1, and replacing the original filling factor from the rate inflection point with the backtracking filling factor.

[0030] The technical effects and advantages of the automotive seat back panel mold forming state analysis system based on finite element simulation of this invention are as follows: This invention determines the rib root feature nodes from the finite element mesh file and the filling simulation result file through the rib root time sequence extraction module, and forms a node record containing the filling factor time sequence and the original convergence mark. This allows the forming state analysis object to focus on the structural adjacent areas of the thin-walled body, the sidewall of the reinforcing rib, and the root of the reinforcing rib, avoiding rough judgment based solely on the overall weld line results. Furthermore, the pseudo-convergence phase discrimination module performs time sequence advance analysis and directional phase comparison on the rib root feature nodes corresponding to the original convergence mark. This can distinguish between the true convergence of the frontal edges and the pseudo-convergence formed by the advance of a single frontal edge after stagnation at the root of the rib, providing a basis for subsequent processing such as clearing candidates, delaying candidates, and retaining candidates.

[0031] Furthermore, the system utilizes a backtracking and re-pushing analysis module to convert pseudo-merging disposal markers into backtracking filling factor trajectories. It then updates the local filling state within the affected control volume range with the replaced filling factor, and performs front-end re-pushing to re-obtain the corrected front-end arrival sequence and corrected merging node sequence. Therefore, the system does not simply delete or shift the original weld line results; instead, it regenerates the rib root weld line location field based on the corrected local filling state. This improves the accuracy of weld line location recording in the rib root region and the reliability of molding state analysis results, providing stable data for subsequent mold verification, local structural adjustments, and molding risk assessment. Attached Figure Description

[0032] Figure 1 This is a flowchart illustrating the automotive seat back panel mold forming state analysis system based on finite element simulation according to the present invention.

[0033] Figure 2 This is a schematic diagram illustrating the extraction of rib root feature nodes and adjacent chains in this invention; Figure 3 This is a schematic diagram of the temporal surge analysis of the filling factor of the rib root node in this invention; Figure 4 This is a schematic diagram illustrating the formation of the direction phase comparison and pseudo-merging handling markers of the present invention; Figure 5 This is a schematic diagram of the trajectory generation of the backfill factor in this invention; Figure 6 This is a schematic diagram of the affected control volume range and local frontal pushback of the present invention. Detailed Implementation

[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0035] Example

[0036] This embodiment discloses a finite element simulation-based molding state analysis system for automotive seat back panel molds, used to analyze the molding state of the melt front, merging nodes, and weld lines formed during the injection molding filling simulation of automotive seat back panels. The automotive seat back panel includes a thin-walled main body, reinforcing ribs, rib root fillets, snap fasteners, screw posts, and edge flanges. A significant thickness abrupt change occurs at the intersection of the thin-walled main body and the reinforcing ribs, where the melt is prone to local stagnation followed by a sudden surge, causing the original merging mark in the filling simulation result file to shift upstream of the rib root. Finite element mesh nodes, or simply nodes, have spatial coordinates and are connected to at least one finite element element, used to carry state quantities such as filling factor, arrival time, pressure, temperature, and merging marks.

[0037] Please see Figure 1 This invention discloses a system for analyzing the forming state of automotive seat back panel molds based on finite element simulation, comprising the following modules:

[0038] Please see Figure 2 The reinforcement root time series extraction module is used to read the finite element mesh file and the filling simulation result file, determine the reinforcement root feature nodes based on the wall thickness mutation ratio, aspect factor and structural region label, and form a node record containing the filling factor time series and the original confluence marker.

[0039] First, the finite element mesh file of the car seat back panel and the filling simulation result file containing the node filling status are read. The finite element mesh file includes node coordinates, element numbers, element connection relationships, nominal element wall thickness, and structural region labels. The node filling status refers to a data set characterizing the time-progression process of the melt within the node's neighborhood, including at least the node filling factor, node arrival time sequence, original convergence marker, and original weld line location field for each time step. It may also include at least one auxiliary state data from pressure boundaries, temperature states, or source direction labels. The node filling factor represents the filling degree of the corresponding local filling region, ranging from 0 to 1. The node arrival step is the time step at which the node filling factor first reaches or exceeds the arrival factor threshold. The original convergence marker is the initial determination result recorded in the filling simulation result file when two melt fronts meet at the same node or adjacent local filling regions. The weld line location field is the spatial distribution result of the weld lines formed by spatial connection, region merging, and structural mapping of the merging node sequence.

[0040] In this embodiment, the control volume is a local filling calculation region constructed around the node; the source direction label is used to indicate the main propagation source direction of the melt front when the node reaches the filling state. It can be directly read from the filling simulation result file, or it can be generated according to the node adjacency relationship, the node arrival time sequence and the spatial position relationship of adjacent nodes.

[0041] It should be noted that the filling simulation result file is not limited to a specific solver, specific commercial simulation software or specific filling solution method. As long as it can correspond to the node number, element connection relationship and structural region label in the finite element mesh file, and can support the screening of root feature nodes, time-series advance analysis, direction phase comparison, determination of the affected control volume range and local front re-pushing, it can be used as the input file of this system.

[0042] Structural region labels are used to identify the location of a node or unit within the seat back panel structure, including the thin-walled main body area, stiffener sidewall area, stiffener root area, snap-fit ​​area, screw post area, and edge flange area. Structural region labels can be generated from feature face names, mesh partition names, or abrupt wall thickness boundaries in the product's 3D model. Adjacent transition nodes between the thin-walled main body area and the stiffener sidewall area are recorded as stiffener root areas, while snap-fit ​​areas, screw post areas, and edge flange areas exclude non-stiffener root areas.

[0043] Based on the nominal wall thickness of elements, element connectivity, and structural region labels in the finite element mesh file, stiffener features are used to screen each node. For any node, the system first calculates its wall thickness abrupt change ratio, which is the ratio of the minimum to the maximum nominal wall thickness among all adjacent elements of that node. This ratio characterizes whether there is a thickness abrupt change in the local region where the node is located, resulting from the transition from a thin-walled structure to a stiffened structure. The wall thickness abrupt change threshold is denoted as... Its initial value is set to 0.55; wall thickness mutation threshold. The initial value was obtained by calibrating a historical mold model of a similar seat back panel: a historical simulation model with a thin-walled main body thickness of 2.3 mm to 2.8 mm and a reinforcing rib thickness of 1.0 mm to 1.4 mm was selected, and the distribution of wall thickness abrupt change ratios of known rib root nodes was statistically analyzed. The upper boundary that could cover more than 95% of the known rib root nodes and exclude the smooth wall thickness transition area was taken as the initial value. In this embodiment, the thickness of the thin-walled main body is about 2.5 mm, and the thickness of the reinforcing rib is about 1.0 mm to 1.3 mm. Therefore, the wall thickness abrupt change threshold was set as follows. The initial value is set to 0.55.

[0044] Furthermore, the system calculates the aspect factor of the adjacent elements of a node. The aspect factor of an adjacent element is defined as the ratio of the longest side length to the shortest feature height in the adjacent elements of a node, and is used to characterize whether there is a slender mesh shape in the local control volume. The shortest feature height refers to the minimum projected height of the adjacent element in the direction perpendicular to its longest side. When the element is a triangular or tetrahedral element, the shortest feature height is the minimum distance from any vertex of the element to the opposite side or the opposite face. When the element is a quadrilateral or hexahedral element, the shortest feature height is the minimum value of the length of each side or the height of each local side of the element.

[0045] The reinforcement root area, constrained by the demolding direction, the radius of the rib root fillet, and the thin-walled dimensions, easily forms high aspect ratio elements. These elements tend to amplify local fill factor jumps during volume control propagation; therefore, the aspect ratio threshold is denoted as... Its initial value is set to 6.0; the span threshold. The initial value was obtained by calibrating historical misalignment samples. Specifically, the simulation results of seat back panels with known misalignment of rib weld lines were selected, the distribution of chord factor of rib unit that caused filling factor protrusion was statistically analyzed, the chord factor of the 80th percentile in the misalignment sample was taken as the initial value, and the initial value was confirmed by verifying the weld line position of the trial mold to ensure that the initial value would not mistakenly include large areas of normal thin-walled main body units in the rib protrusion inspection range.

[0046] When the wall thickness mutation ratio of a certain node is less than the wall thickness mutation threshold And at least one of its adjacent units has an aspect factor greater than the aspect threshold. Furthermore, the node is located within the structural adjacency range of the thin-walled main body region, the stiffener sidewall region, or the stiffener root region; the system records this node as a stiffener root feature node. The structural adjacency range includes the structural region boundary between the thin-walled main body region and the stiffener sidewall region, the stiffener root rounded transition area, and the local node range formed by extending one layer of adjacent units from the structural region boundary into the thin-walled main body region or the stiffener sidewall region; the aforementioned wall thickness abrupt change ratio, aspect factor, and structural region label are used together to define the geometric position, mesh shape, and structural region attributes of the stiffener root feature node. When all three are satisfied, the system uses this node as the main analysis node for subsequent pseudo-merging discrimination.

[0047] For nodes that do not simultaneously meet the wall thickness abrupt change ratio and aspect factor conditions, but are located within the adjacent range of the root feature nodes, if their structural region label belongs to the root fillet transition zone, the starting segment of the root sidewall, or the transition side of the thin-walled main body, the system can record them as root candidate extension nodes. Root candidate extension nodes do not trigger pseudo-merging disposal markers independently; they only participate in subsequent calculations as auxiliary nodes when associated with the original merge marker, the affected control volume range, or the opposite-side branch chain. This ensures that the screening boundaries for root feature nodes are clear while avoiding the omission of local filling changes in the root fillet transition zone due to overly narrow node screening.

[0048] For each root feature node, the system reads its complete fill factor time series; where the fill factor of node i at time step n is denoted as . , The value range is 0 to 1; the system uses the reach factor threshold. Determine the node arrival step, node arrival step Defined as the node filling factor first reaching or exceeding the reaching factor threshold. Time step; The initial value is set to 0.50. This initial value is calibrated based on the physical meaning that the leading edge usually crosses the center of the node control area when the control volume is half-filled. Combined with the position of the leading edge marked manually in the historical simulation animation, the filling factor value that minimizes the error between the position of the artificial leading edge and the arrival step of the node is selected as the initial value.

[0049] Furthermore, based on grid adjacency relationships and node arrival steps, the system determines the upstream neighbor chain and opposite-side branch chain for each root feature node. The upstream neighbor chain is a sequence of consecutive adjacent nodes obtained by tracing back from the root feature node towards the earlier arrival direction of the melt, representing the propagation path of a single melt front before entering the root. The opposite-side branch chain is a sequence of consecutive adjacent nodes obtained by searching along the opposite-side flow branch from the root feature node, used to determine whether another melt front participates in the actual convergence. It should be noted that the upstream neighbor chain provides a unidirectional propagation reference, while the opposite-side branch chain provides a reference for opposing fronts. The subsequent pseudo-convergence phase discrimination module determines, based on the arrival phase difference and filling factor state of the two, whether the original convergence marker is formed by a true opposing front or by a sudden advance after local stagnation at the root.

[0050] In this embodiment, the adjacency chain search is implemented based on the node adjacency list of the finite element mesh. The node adjacency list is generated by the element connection relationship. If two nodes belong to the same finite element element, they are adjacent nodes. When searching the upstream neighbor chain, the system starts from the root feature node and traverses its adjacent nodes, prioritizing the adjacent node whose arrival step is earlier than the current node and whose arrival step difference is the smallest. When there are multiple candidate nodes, the node with the smallest angle with the gate direction is selected as the next node, and the system continues to backtrack towards the direction where the melt arrives earlier until the set number of upstream chain layers is reached. Here, the gate direction refers to the direction vector from the current node to the nearest gate node on the filling path to which the node belongs. For the multi-gate model, the system selects the gate with the earliest arrival time of the melt front at the current node as the nearest effective gate of the node, and uses the direction from the current node to the nearest effective gate as the gate direction.

[0051] When searching for opposite-side branch chains, the system starts with the rib root feature node, selects adjacent nodes that form a reverse fronting direction with the upstream chain's first segment as candidate nodes, and continues to expand outward according to the arrival step and spatial connectivity. The reverse fronting angle threshold is used to limit the minimum angle between the opposite-side branch chain and the upstream chain's first segment direction, thereby distinguishing between the single-strand fronting extension direction and the potential merging fronting direction. In this embodiment, the initial value of the reverse fronting angle threshold is set to 90 degrees. This initial value is obtained through historical simulation results: simulation samples of similar seat back panels that have been confirmed to have real rib sidewall merging are selected, the angle distribution of the real opposite-side fronting relative to the upstream chain's first segment direction is statistically analyzed, and the lower boundary that can cover most real opposite-side frontings and exclude single-strand advancing branches in the same direction is taken as the initial value.

[0052] The number of upstream chain layers is preset, with an initial value of 5 adjacent nodes. The initial value of the preset number of upstream chain layers is obtained by calibrating the local flow transition length of the rib root: the number of local transition unit layers from the thin-walled main body to the root of the reinforcing rib in similar seat back panels is counted, and the number of adjacent layers that can cover the rib root rounded corner, rib root entrance and the starting section of the rib sidewall is used as the initial value.

[0053] The opposite branch chain is searched in the opposite direction with the same number of layers as the upstream chain, and the reverse front angle threshold is used as the reverse front screening condition to identify whether there are opposing fronts participating in the real convergence.

[0054] For example, the fill factors of a certain root node in steps 40, 41, 42, and 43 are 0.18, 0.19, 0.86, and 0.91, respectively. The system calculates the wall thickness mutation ratio of this node to be 0.48 and the aspect factor to be 7.2, satisfying the recording conditions for root characteristic nodes. The time step in which the fill factor of this node first reaches 0.50 is step 42, therefore its arrival step is step 42; the system continues to read its upstream neighbor chain and finds that the nearest upstream neighbor has reached 0.80 in step 41, while the fill factors of the opposite branch chain before step 42 are all below 0.20.

[0055] After the root timing extraction module completes the processing, the system writes the filling factor time series, node arrival step, upstream neighbor chain, opposite branch chain, wall thickness mutation ratio, aspect factor and original convergence marker into the node record of each root feature node; the subsequent pseudo-convergence phase discrimination module continues processing based on this node record.

[0056] The pseudo-merging phase discrimination module is used to perform temporal advance analysis and directional phase comparison on the root feature nodes corresponding to the original merging markers, and obtain pseudo-merging disposal markers including candidates to be cleared, candidates to be delayed, and candidates to be retained.

[0057] Please see Figure 3 The system first performs adjacent difference processing on the time series of the filling factor of the root feature nodes; wherein, adjacent difference Define as a node The increment of the filling factor between adjacent time steps represents the change in the local filling propulsion speed of that node at the corresponding time step. The system sets a surge check window before and after each node's arrival step. ; rush into the inspection window Defined as a local temporal range formed by taking several time steps forward and backward around the node arrival step, used to capture the post-stagnant sudden jump of the root node; the post-stagnant sudden jump refers to the state where the filling factor of the root node increases slowly over several consecutive time steps, and then the increase in the filling factor in adjacent time steps is significantly higher than the upstream unidirectional propagation rate; sudden jump detection window The initial value is set to 3 steps before the arrival step and 2 steps after the arrival step. This initial value is obtained by calibrating historical misaligned samples: the number of time steps that the filling factor takes to advance from a low value to a high value in the pseudo-merging samples of seat backrest rib roots is statistically analyzed, and the window length that can cover most of the advancement process and does not cross the complete filling stage is taken as the initial value.

[0058] The system determines the rate inflection point. The rate inflection point is defined as the increment of the fill factor within the surge inspection window. The time step at which the maximum value is obtained is used to represent the critical time when a node transitions from stagnation to surge.

[0059] Furthermore, determine the peak surge amount. and number of lag steps Peak surge indicates the rate inflection point. The maximum fill factor increment at the point; the number of laminar flow steps represents the rate inflection point. Previously, the absolute value of the filling factor increment was continuously less than the stagnation increment threshold. Number of time steps, threshold for stagnation increment The initial value is set to 0.03. This initial value is obtained by calibrating normal smooth filling samples: the increment of filling factor in adjacent time steps of normal thin-walled propulsion zone is statistically analyzed, and the 95th percentile increment is used as the initial value reference so that continuous increments smaller than this value can represent local stagnation state.

[0060] To identify whether there is abnormal protrusion at the root node, the system calculates the protrusion ratio. , rush ratio Used to characterize the degree of amplification of the filling increment at the root feature node relative to the upstream unidirectional propagation process at the rate inflection point; upstream unidirectional propagation rate Defined as the upstream neighbor chain at the rate inflection point The previous average filling factor increment was calculated using three time steps as a rate statistical window. This statistical window was obtained by calibrating the advancing samples at the forefront of the normal thin-walled main body region, which can smooth random fluctuations and does not cross the complete filling stage. To facilitate a unified comparison of the degree of abnormal advance, the system calculates the advance ratio according to the following formula:

[0061] In the formula, To prevent the use of tiny constants with a denominator of zero, their initial value is set to 0.01. The initial value is determined according to the minimum effective increment of the filling factor. In this embodiment, increments below this value are considered numerical noise.

[0062] Please see Figure 4After completing the temporal surge characteristic analysis, the system performs direction phase comparison. Direction phase comparison refers to determining, based on the arrival order of the upstream neighbor chain and the filling arrival state of the opposite branch chain, whether the filling arrival at the root feature node is formed by the actual convergence of two opposing melt fronts or by a single incoming front that stagnates at the root and then surges forward.

[0063] The system first checks the advancement sequence of the upstream neighbor chain; the upstream neighbor chain represents the single-lead propagation path of the melt before it enters the root feature node from the gate direction; the system sequentially reads the arrival step of each node in the upstream neighbor chain according to the direction from upstream to the root feature node; when the arrival step of two or more consecutive upstream nodes gradually increases along this direction, the system determines that the root feature node has a stable unidirectional advancement front. A stable unidirectional advancement front indicates that the main filling source of this node comes from a single upstream direction;

[0064] Next, the opposing branch chains are evaluated for their opposing leading edge states; the opposing branch chains represent potential flow branches from the upstream to the other side relative to the root feature nodes; the system reads the velocity inflection points of the opposing branch chains. The previous maximum fill factor was used, and the earliest time step to reach the state in the opposite branch chain was read; when the opposite branch chain is at the rate inflection point The previous maximum fill factor was lower than the effective threshold of the opposite branch. When this happens, the system records the state of the opposite branch as missing the leading edge of the opposite direction;

[0065] When the opposite branch chain is at the rate inflection point The previous maximum filling factor reached the effective threshold of the opposite branch. However, the opposite branch reaches the step and rate inflection point earliest. The difference between them is greater than the true convergence phase threshold. When the opposite branch is at its inflection point, the system records the state of the opposite branch as a phase-leading phase misalignment; when the opposite branch chain is at its rate inflection point... The previous maximum filling factor reached the effective threshold of the opposite branch. Furthermore, the earliest arrival point and rate inflection point of the opposite branch. The difference between them is not greater than the true convergence phase threshold. When this occurs, the system records the state of the opposite branch as valid at the forward leading edge; where the valid threshold for the opposite branch is... This is used to determine whether an effective front has been formed in the opposite branch. Its initial value is set to 0.35, determined based on the physical meaning that when the control volume fill factor is below 0.35, the front has not yet covered the center of the node's neighborhood and it is difficult to form a stable merging surface; the true merging phase threshold is used to determine whether an effective front has been formed in the opposite branch. It is used to determine whether the arrival times of two fronts are close enough. Its initial value is set to 2 time steps. The initial value is determined based on the historical simulation playback results where the actual arrival time difference of the two fronts under the same time step discretization accuracy is usually no more than 2 steps.

[0066] The system completes the surge ratio and number of lag steps After calculation, the two are used as a joint criterion for abnormal advance state; when the advance ratio is... Greater than the surge ratio threshold And the number of stagnant steps When the ratio is not less than 2, the system determines that the characteristic node of the tendon root has an abnormal surge state; when the surge ratio is not less than 2, the system determines that the characteristic node has an abnormal surge state. Not greater than the surge ratio threshold Or the number of stagnant steps When the value is less than 2, the system determines that the root feature node does not exhibit an abnormal surge state; where the surge ratio threshold is... Used to distinguish between smooth advancement and abnormal sudden advance, its initial value is set to 5.0. The initial value is obtained by calibrating the pseudo-merging sample of the root: the ratio of the peak advance of the pseudo-merging node to the upstream unidirectional advance rate is statistically analyzed, and the lower boundary that can distinguish between smooth advance and abnormal advance is taken as the initial value.

[0067] It should be noted that a high surge ratio alone may be caused by time step discretization or local grid fluctuations, and a high number of stagnation steps alone may be caused by normal deceleration filling. Therefore, this implementation adopts a joint criterion of surge ratio exceeding the limit and stagnation step number meeting the lower limit, in order to reduce the situation of misjudging normal smooth advancement as surge after stagnation at the root.

[0068] After determining the stable unidirectional advancing front, the status of the opposite branch, and the abnormal surge status, the system generates pseudo-merging handling markers. These markers include cleared candidates, delayed candidates, and retained candidates. Cleared candidates indicate that the original merge marker lacks support from the opposite advancing front and is more likely formed by a single-stream stagnant front followed by a surge. Delayed candidates indicate that the opposite branch has appeared, but its arrival phase and velocity inflection point do not match. Retained candidates indicate that the original merge marker has genuine merge support or that the surge characteristics are insufficient to trigger pseudo-merging handling.

[0069] Specifically, when a root feature node has a stable unidirectional advancing front, the opposite branch state is a missing opposing leading edge, and the node exhibits an abnormal surge state, the system marks the node as a removal candidate; when a root feature node has a stable unidirectional advancing front, the opposite branch state is a phase misalignment of opposing leading edges, and the node exhibits an abnormal surge state, the system marks the node as a delay candidate; when the opposite branch state is a valid opposing leading edge, or the node does not exhibit an abnormal surge state, or the node does not have a stable unidirectional advancing front, the system marks the node as a retention candidate. It should be noted that being a retention candidate does not necessarily mean that the node is free from local filling fluctuations, but rather that the current node does not meet the handling conditions for a false convergence caused by a surge after a single leading edge stagnation, and therefore does not trigger subsequent regression of the filling factor trajectory and local leading edge re-push.

[0070] The above-mentioned candidate for removal indicates that the original convergence marker of the node is more likely caused by a single frontal flow stagnating at the root and then suddenly advancing. It is necessary to remove the original convergence marker and perform a back-up and re-push of the filling factor. The candidate for delay indicates that there is a branch on the opposite side near the node, but the arrival phase of the branch on the opposite side does not match that of the root node. It is necessary to suspend the convergence determination within the delay window and verify whether a true convergence has been formed. The candidate for retention indicates that the node has a true opposing frontal support, or its sudden advance characteristics are insufficient to support the handling of false convergence. The original convergence marker is kept unchanged.

[0071] For example, the fill factors of a certain root node in steps 40 to 43 are 0.18, 0.19, 0.86, and 0.91 respectively, with the velocity inflection point at step 42 and a peak advance of 0.67. There are two consecutive stagnation steps before the velocity inflection point. The upstream neighbor chain of this node arrives in steps 39, 40, and 41 respectively, indicating that the node has a stable unidirectional advancing front. The maximum fill factor of the opposite branch chain before step 42 is 0.20, which does not reach the effective threshold for the opposite branch; therefore, the opposite branch state is recorded as missing opposing fronts. If the upstream unidirectional advancing velocity of this node is 0.01, the advance ratio is 33.5, which is greater than the advance ratio threshold, and the number of stagnation steps is not less than 2. The system determines that the node has an abnormal advance state. Since this node simultaneously satisfies the conditions of a stable unidirectional advancing front, missing opposing fronts, and abnormal advance state, the system marks this node as a candidate for removal.

[0072] After the pseudo-merging phase discrimination module completes its processing, the system supplements the original merging node record with pseudo-merging disposal markers, velocity inflection points, surge ratios, stagnation steps, upstream arrival steps, and opposite branch status. The pseudo-merging disposal markers include cleared candidates, delayed candidates, and retained candidates. Among them, cleared candidates are used to trigger subsequent filling factor rollback and local front pushback, delayed candidates are used to trigger subsequent merging verification within the delayed window, and retained candidates are used to keep the original merging markers unchanged. The above discrimination results are only used as the basis for the subsequent rollback and pushback analysis module to determine the final disposal action and are not used as the final formed state output.

[0073] The backtracking and re-pushing analysis module is used to determine the phase lag step and upstream unidirectional advance rate based on the pseudo-merging disposal markers and node records, and generate a backtracking filling factor trajectory. The backtracking filling factor trajectory starts from the filling factor of the time step before the rate inflection point, advances linearly according to the upstream unidirectional advance rate, and replaces the original filling factor from the rate inflection point. The affected control volume range is determined based on the adjacency range of the pseudo-merging node and the original arrival step. The node state outside the range is fixed, and the local filling state within the affected control volume range is updated with the replaced filling factor. Based on the updated local filling state, the frontier re-pushing is performed to obtain the corrected frontier arrival sequence and the corrected merging node sequence.

[0074] Please see Figure 5 First, calculate the phase lag steps. Phase lag steps are defined as the time step difference between the velocity inflection point of a root node and the arrival step of its nearest upstream neighbor, used to describe the degree of delay in the sudden advance of the node after lag relative to the single-stream advance; the system reads the node rate inflection point And read the arrival steps of its nearest upstream neighbor. The difference between the two is used as the phase lag step. .

[0075] To determine the lag of the pseudo-merging node relative to the upstream leading edge, the system calculates the phase lag steps according to the following formula: In the formula, the number of phase lag steps is... A value greater than 0 indicates that the sudden movement of the root node occurs after the arrival of the upstream neighbor, exhibiting a post-lag sudden movement characteristic; phase lag steps A value of 0 or less than 0 indicates that the node arrived at the same time as or ahead of its upstream neighbor, and will not be subject to rollback.

[0076] Based on pseudo-convergence handling markers and phase lag steps The final action is determined; actions include clearing, delaying, and retaining. Clearing indicates that the original convergence marker was caused by a false convergence, and the convergence marker needs to be revoked and the affected control volume needs to be backtracked and re-pushed from the rate inflection point. Delaying indicates that the original convergence marker has uncertainty, and it is not directly deleted, but the convergence determination is suspended within the delay window and the leading edge is kept to advance unidirectionally. At the end of the window, it is checked again to see if a true convergence has occurred. Retaining indicates that the node meets the conditions for a true convergence, and the original convergence marker is not rewritten.

[0077] When a node's pseudo-merge is marked as a candidate for removal, and the phase lag steps are... Greater than or equal to the large lag step threshold When a node's pseudo-merger is marked as a delayed candidate, the system will determine the final action as "clear"; when a node's pseudo-merger is marked as a retention candidate, or the phase lag steps of the candidate node are cleared. Less than the large lag step threshold When this happens, the system will determine the final action as "reserved"; among which, the large lag step threshold... The initial value is set to 1 time step, based on the fact that when the root protrusion is at least one time step later than the nearest upstream neighbor, it can indicate that the node has an identifiable lag relative to the propagation of the single-strand front.

[0078] For root feature nodes whose final disposal action is determined to be removal or postponement, the system generates a backfill factor trajectory. The backfill factor trajectory is defined as the corrected timing sequence used to replace the original advance segment fill factor. Its function is to restore the node from the jump fill state to the single-lead continuous advance state. Specifically, the system uses the original fill factor of node i at the time step before the rate inflection point. As the starting point for retreat, and with an upstream unidirectional advance rate As the continuous incremental advancement of subsequent time steps, the rollback fill factor trajectory is generated according to the following formula:

[0079] In the formula, This represents the backoff fill factor of node i at time step n. This indicates that node i is at the rate inflection point. The original fill factor of the previous time step, This represents the upstream unidirectional propulsion rate, obtained from the average fill factor increment of the upstream neighbor chain before the rate inflection point. This means that the calculation result is limited to the range of 0 to 1; This indicates the time step that needs correction, and n is not less than the rate inflection point. .

[0080] It should be noted that when the time step n is less than the rate inflection point... When node i retains its original fill factor, the time step n is not less than the rate inflection point. At that time, the system adopted Replace the original fill factor of node i. As a result, the leap from low fill factor to high fill factor in the original simulation results is replaced with a corrected trajectory based on the continuous increase of the upstream unidirectional propulsion rate, thus providing a consistent initial fill state for the subsequent local frontal re-push within the affected control volume.

[0081] Please see Figure 6 The system then determines the affected control volume range, which refers to the set of control volumes that need to participate in the front re-push within the local mesh adjacency range, starting from the pseudo-merging node i. The system performs a layer-by-layer search along the direction of the melt front advancing from early to late in the original filling simulation; when there are shared nodes or shared edges between two control volumes, they are recorded as one layer of adjacency; for any candidate control volume, if the shortest adjacency layer between it and the pseudo-merging node i does not exceed the number of affected layers. Furthermore, the initial arrival step of the candidate control volume is no earlier than the rate inflection point. If so, the candidate control volume is included in the range of affected control volumes; where the number of affected layers is... The initial value for defining the spatial influence range of the pseudo-merger of the reinforcing ribs on the downstream local filling state is set as 3 adjacent control volumes. This initial value is obtained by calibrating historical samples of pseudo-merger of the reinforcing ribs in the seat back panel. The spatial expansion range of the influence of the erroneous weld line position after the occurrence of the pseudo-merger node is statistically analyzed. The error is usually limited to the three control volumes of the reinforcing rib inlet, the starting section of the reinforcing rib sidewall, and the adjacent thin-walled body. Therefore, 3 layers are used as the initial value.

[0082] For the initial arrival step to be earlier than the rate inflection point The control volume is recorded by the system as the boundary fixed control volume; the boundary fixed control volume does not participate in the filling factor rewriting, but its original filling factor, original arrival step, and source direction can be used as the boundary input for local frontier re-pushing; for the shortest adjacent layer number between the pseudo-merging node i and the pseudo-merging node i exceeds the number of affected layers. The control volume, even if its initial arrival step is not earlier than the rate inflection point. Furthermore, it is not included in the affected control volume range. Therefore, the system limits the frontal pushback to the control volume range directly affected by the sudden advance after local stagnation at the root, avoiding unnecessary rewriting of the normal filling area outside the range.

[0083] The system performs local front re-pushing on the affected control volume. Local front re-pushing is implemented based on the filling factor advancement subroutine of the original finite element-control volume hybrid solver, without changing the material parameters, gate pressure curve, mold temperature boundary and time step. Its function is to make consistent corrections to the filling factor timing of the pseudo-merging node of the rib root and its downstream local control volume while keeping the global filling boundary unchanged.

[0084] In specific implementation, the system uses The local filling state at time t is taken as the restart state, and the arrival step, filling factor, and pressure boundary of nodes outside the affected control volume are fixed to the original simulation results; for pseudo-merging node i within the affected control volume, the system self-rate inflection point The rollback fill factor trajectory was adopted. Replace its original fill factor; for nodes other than pseudo-merging node i within the affected control volume, the system updates its modified fill factor step by step according to the original time step sequence.

[0085] Furthermore, for any node p to be updated within the affected control volume, the system establishes a source neighbor set Sp(n) at time step n. The source neighbor set Sp(n) includes nodes that have reached the arrival factor threshold at time step n-1 and have a grid adjacency relationship with node p. If the source neighbor set Sp(n) is empty, node p retains the modified filling factor of the previous time step at time step n. If the source neighbor set Sp(n) is not empty, the system selects a unique primary source node q from the source neighbor set Sp(n) according to deterministic rules.

[0086] Specifically, the system first compares the correction fill factors of each source neighbor node in the previous time step and selects the source neighbor node with the largest correction fill factor as the primary source node q. If multiple source neighbor nodes have the same correction fill factor, the source neighbor node with the earliest arrival step after correction is selected. If the arrival steps after correction are still the same, the source neighbor node with the smallest angle between its source direction label and the original source direction of the node to be updated is selected. If the angles are still the same, a unique primary source node q is determined according to the node number in ascending order. Through the above sequential selection, the primary source node in each time step is unique, and the node to be updated p can obtain a unique source direction label.

[0087] The system reads the local propulsion increment of node p at the corresponding time step in the original filling simulation results. The non-negative local advance increment is used to inherit the local flow advance amplitude in the original filling simulation, but not the erroneous merging state caused by sudden advances at pseudo-merging nodes. For the node p to be updated, the system reads the difference in its fill factor between adjacent time steps in the original filling simulation results, and sets the differences that are less than zero to zero, thus obtaining the non-negative local advance increment. This non-negative local advance increment is only used for local front re-push within the affected control volume range and does not change the original filling state of nodes outside the affected control volume range. The corrected fill factor of node p at time step n Update according to the following formula:

[0088] In the formula, This represents the correction fill factor of node p at time step n; This represents the corrected fill factor of node p in the previous time step; This represents the correction fill factor of the primary source node q at the previous time step; This represents the non-negative local advance increment of node p in the original filling simulation result; clip[·,0,1] indicates that the calculation result is limited to the filling factor value range of 0 to 1. It should be noted that this update rule does not regenerate the global filling result, but uses the local advance increment of the original simulation to locally rearrange the node filling timing within the affected control volume range, so that the advance segment of the pseudo-merging node will not continue to propagate downstream as an incorrect merging result.

[0089] When the modified fill factor of node p First time reaching or exceeding the reaching factor threshold At that time, the system records time step n as the corrected arrival step of node p, and records the source direction label corresponding to the main source node q; if the control volume of the same node or adjacent nodes is not greater than the true convergence phase threshold... Within the time step, two labels from different source directions appeared, and both source directions reached the factor threshold. If the system records the node or the adjacent control volume as a true rendezvous node, then the system will re-record that node or the adjacent control volume as a true rendezvous node; if there is only a single source direction label, or the arrival step difference between different source directions is greater than 1, the system will re-record that node or the adjacent control volume as a true rendezvous node. If so, it will not be recorded as a real rendezvous node;

[0090] If the correction front has left the affected control volume range within a certain time step, or if all nodes within the affected control volume range have reached the filling completion state, the system stops the current local front re-push. When the affected control volume ranges corresponding to multiple pseudo-merging nodes overlap, the system executes the local front re-push in sequence according to the rate inflection points of each pseudo-merging node from small to large. The overlapping control volumes that are re-pushed first are used as the boundary states for subsequent pseudo-merging node re-pushes, with their correction filling factor, corrected arrival step, and source direction label as the boundary states.

[0091] For nodes whose final action is clearing, the system writes a backfill factor trajectory starting from the rate inflection point, removes the node's merge marker in the original merge node sequence, and performs a local front re-push within the affected control volume until local filling is completed or the corrected front leaves the affected control volume. For nodes whose final action is delayed, the system constructs a delayed merge review window, with the starting time step of the delayed merge review window being the rate inflection point. The termination time step is Within this window, the system temporarily does not confirm the original merge marker of the node and follows the backfill factor trajectory. Maintain continuous unidirectional forward propagation. After the window ends, the system verifies whether the opposite-side branch chain has reached the effective threshold of the opposite-side branch. And determine whether the difference between the arrival step of the opposite branch and the corrected arrival step of the node is not greater than the true merging phase threshold. If both conditions are met, a new real merge node is generated; if neither condition is met, the original merge mark of that node is revoked.

[0092] For example, a certain root node is marked as a candidate for removal, and the phase lag step is 1. The system determines the final action for this node as removal. The original filling factor of this node jumps from 0.19 to 0.86 in step 42; the system calculates using an upstream unidirectional advance rate of 0.01 per time step, obtaining... Starting from step 42, the affected control volume is locally pushed back along the front edge. After the push back, the original confluence mark of the thin-walled body upstream of the rib root is deleted, the front edge continues to spread along the sidewall of the reinforcing rib, and a new real confluence node is formed on the sidewall of the rib and the opposite branch, so that the original weld line position is corrected to the real confluence area of ​​the sidewall of the rib.

[0093] After the backtracking and re-pushing analysis module completes its processing, the system generates the handling result for each pseudo-merging node and writes the corresponding backtracking filling factor trajectory into the affected control volume. After local front re-pushing, the system obtains the corrected front arrival sequence, the corrected merging node sequence, and the corrected weld line position field, which are then used by the subsequent forming state write-back module.

[0094] The forming state write-back module is used to regenerate the weld line position field of the rib root according to the corrected confluence node sequence and write it into the forming state database.

[0095] The molding state write-back module calls the corrected leading edge arrival sequence, corrected merging node sequence, and corrected weld line location field generated by the backtracking and re-analysis module, writes them into the molding state database, and regenerates the weld line migration record, risk level, and cause location information of the automotive seat back panel reinforcement root based on the written-back molding state results. The molding state write-back module generates a risk level based on the corrected leading edge arrival sequence, merging node sequence, and weld line location field, ensuring that the risk level originates from the molded state results after the backtracking and re-analysis.

[0096] The forming state database refers to a database or result file used to store the finite element simulation results of seat back panel mold forming. This forming state database can be implemented using relational data tables, simulation result files, or result fields within CAE software plugins. Taking a relational data table as an example, it includes at least a node status table, a merging node table, a weld line segment table, and a risk record table. The node status table records the node number, node coordinates, original arrival step, corrected arrival step, original filling factor sequence, and corrected filling factor sequence. The merging node table records the merging node number, merging time step, source of the associated frontier, and handling status. The weld line segment table records the weld line segment number, spatial location, structural region, and migration amount. The risk record table records the risk level, merging angle, number of re-pushing attempts, and cause location information.

[0097] The system first writes the corrected frontier arrival sequence: for nodes not within the local re-push range, the system retains the original arrival step; for nodes participating in the local re-push, the system replaces the original arrival step with the re-push arrival step. Then, the system writes the corrected rendezvous node sequence: for nodes corresponding to clearing actions, the system removes their original rendezvous marker from the valid rendezvous node sequence; for true rendezvous nodes confirmed at the end of the delayed action window, the system writes the new rendezvous node number, rendezvous time step, and associated branch number into the valid rendezvous node sequence; for nodes with retained actions, the original rendezvous node number is retained.

[0098] The system further regenerates the weld line location field based on the corrected confluence node sequence. The generation process of the weld line location field adopts the same spatial restoration logic as the original CAE system: confluence nodes that are temporally continuous, spatially adjacent, and belong to the same structural region are merged into weld line segments; weld line segments spanning the thin-walled main body region and the reinforcing rib sidewall region are split according to structural region labels; and the merged weld line segments are mapped onto the two-dimensional projection of the seat back panel and the three-dimensional mold cavity surface. This processing allows the corrected weld line locations to be regenerated from the re-push confluence node sequence, rather than simply translating the original labels.

[0099] The system calculates weld line migration records, which describe the spatial relationship between the original weld line location field and the corrected weld line location field. The migration amount is calculated based on the distance between the centerlines of the corresponding weld line segments, and the migration direction is determined by pointing from the structural region where the original location is located to the structural region where the corrected location is located. The effective migration threshold is denoted as... The initial value is set to two local average cell side lengths, which are obtained through mesh scale error calibration: at the same mesh scale, weld line variations smaller than one cell side length are easily caused by node discretization errors. Taking two local average cell side lengths as the initial value can distinguish mesh noise from substantial weld line misalignment; if the migration amount is less than the effective migration threshold The system records no significant migration; if the migration amount is not less than the effective migration threshold... The system records the migration direction, for example, from the upstream of the thin-walled main body to the side wall of the reinforcing rib.

[0100] The system generates a weld line formation risk level based on the corrected weld line location field. This risk level includes A, B, C, and D levels; where A corresponds to a high risk level, B to a relatively high risk level, C to a medium risk level, and D to a low risk level. The formation risk level is determined based on the corrected weld line location, structural region label, the convergence angle at the actual fusion node, and the number of re-pushes in the corresponding rib region. The convergence angle is the angle between the propagation directions of the two melt fronts at the actual fusion node, used to characterize the degree of melt fusion at the weld line; the small convergence angle risk threshold is denoted as... The initial value was set to 40 degrees, which was obtained through trial molding backtracking calibration: Select the sample specimen of the side wall of the seat back panel rib, backtrack the samples with cracks near the weld line or the white line on the appearance, statistically analyze the distribution of the simulated convergence angle corresponding to the above samples, and take the upper bound angle of the abnormal sample concentration as the small convergence angle risk threshold. The initial value of .

[0101] It should be noted that the molding risk level is a result labeling after the molding state is written back. It is generated only based on the corrected weld line position field, the corrected merging node sequence, and the weld line migration record, and is used to provide a sorting basis for subsequent mold verification. The molding risk level does not participate in the formation of pseudo-merging disposal markers, nor is it used as a judgment condition for the generation of backfilling factor trajectory or the re-pushing of local frontier, thereby avoiding the risk labeling results from having a reverse impact on the aforementioned filling state correction process.

[0102] When the corrected weld line is located in the sidewall area of ​​the reinforcing rib or the transition area of ​​the rib root fillet, and the fusion angle is less than the small fusion angle risk threshold. At the same time, the number of retries is not less than the retries threshold. When the corrected weld line is located in the reinforcing rib sidewall area or the reinforcing rib fillet transition area, and only meets the risk threshold of a smaller confluence angle, the system classifies the weld line area as a Class A risk. Or the number of retries is not less than the retries threshold. If any of the following conditions are met, the system classifies the rib root area as a Class B risk; if the corrected weld line is located in the thin-walled main body area or near the starting segment of the rib sidewall, and the confluence angle is not less than the small confluence angle risk threshold. The number of retweets is less than the retweet threshold. When the corrected weld line is far from the fillet of the rib root, the sidewall of the reinforcing rib, the snap-fit ​​seat, and the stress concentration area of ​​the screw column structure, and the confluence angle is not less than the small confluence angle risk threshold, the system will classify the area as a Class C risk. The number of retweets is less than the retweet threshold. At that time, the system classifies the area as a Level D risk; the threshold for the number of re-pushing attempts. Its initial value is set to 2. Used to characterize the degree of flow instability that still forms concentrated weld lines after being pushed back through multiple pseudo-convergence nodes in the same root region;

[0103] Among them, Class A and Class B risks are both high molding risk levels, Class C risks are medium molding risk levels, and Class D risks are low molding risk levels. The distinction between Class A and Class B risks is only used to determine the priority of mold review and does not change the overall classification logic of high, medium, and low molding risk levels.

[0104] For example, the original weld line of the third reinforcing rib on the right side of a seat back panel was located upstream of the thin-walled main body. After being pushed again, the weld line migrated to the side wall of the reinforcing rib, with a migration amount of 3.4 local average unit side lengths and a confluence angle of 32 degrees. The total number of pushes in this rib root area was 2. The system marked this area as a level 1 high risk and located the cause as the rib root stagnating and then suddenly advancing, causing the original confluence mark to move forward. The system further recorded the verification and positioning information: the flow path from the nearest gate to this rib root was too long, and there was a significant change in local resistance at the rounded corner of the rib root. Subsequent mold verification focused on the rounded corner of the rib root, the rib side wall inlet, and the flow path of the corresponding gate.

[0105] After the molding state write-back module completes the processing, the system writes the corrected leading edge arrival sequence, merging node sequence, and weld line position field into the molding state database, and forms a rib weld line migration record, risk level, and cause location information; subsequent strength verification, appearance risk verification, and mold improvement positioning are all performed based on the written-back molding state record.

[0106] All the above formulas are calculated using dimensionless numerical values. The parameters in the formulas can be determined based on sample data and simulation calibration results, and can be adjusted by those skilled in the art in combination with specific application scenarios.

[0107] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or a combination thereof. When implemented in software, the above embodiments can be implemented in whole or in part as a computer program product.

[0108] Those skilled in the art will understand that the modules and algorithm steps disclosed in the embodiments herein can be implemented using electronic hardware or a combination of computer software and electronic hardware. The specific choice between hardware and software depends on the application scenario and constraints of the technical solution. Those skilled in the art can select the appropriate implementation method based on the specific application, and such implementation method should not be considered beyond the scope of this application.

[0109] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0110] In conclusion, the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A system for analyzing the forming state of automotive seat back panel molds based on finite element simulation, characterized in that, The system includes the following modules: a reinforcement root time series extraction module, which reads the finite element mesh file and the filling simulation result file containing the node filling status, determines the reinforcement root feature nodes based on the wall thickness mutation ratio, aspect factor, and structural region labels, and forms a node record containing the filling factor time series and the original convergence marker; and a pseudo-convergence phase discrimination module, which performs time series advance analysis and direction phase comparison on the reinforcement root feature nodes corresponding to the original convergence marker, and obtains pseudo-convergence disposal markers including candidates to be cleared, candidates to be delayed, and candidates to be retained. The backtracking and re-pushing analysis module is used to determine the phase lag step and upstream unidirectional advance rate based on the pseudo-merging disposal markers and node records, and generate a backtracking filling factor trajectory. The backtracking filling factor trajectory starts from the filling factor of the time step before the rate inflection point, advances linearly according to the upstream unidirectional advance rate, and replaces the original filling factor from the rate inflection point. The affected control volume range is determined based on the adjacency range of the pseudo-merging node and the original arrival step, the node state outside the range is fixed, the local filling state within the affected control volume range is updated with the replaced filling factor, and local front-end re-pushing is performed based on the source direction label to obtain the corrected front-end arrival sequence and the corrected merging node sequence. The forming state write-back module is used to regenerate the weld line position field of the rib root based on the corrected merging node sequence and write it into the forming state database.

2. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 1, characterized in that, The rib root timing extraction module determines rib root feature nodes, including: calculating the node's wall thickness mutation ratio and adjacent unit aspect factor, where the wall thickness mutation ratio is the ratio of the minimum to the maximum nominal wall thickness of all adjacent units of the node, and the adjacent unit aspect factor is the ratio of the longest side length to the shortest feature height among the node's adjacent units; when the node's wall thickness mutation ratio is less than the wall thickness mutation threshold, at least one adjacent unit's aspect factor is greater than the aspect threshold, and the node is located within the structural adjacency range of the thin-walled main body region, the stiffener sidewall region, or the stiffener root region, the node is recorded as a rib root feature node; the node record of the rib root feature node is written with the filling factor time series, node arrival step, upstream neighbor chain, opposite side branch chain, and original convergence marker; where the upstream neighbor chain is a sequence of consecutive adjacent nodes obtained by tracing back from the rib root feature node to the earlier arrival direction of the melt, and the opposite side branch chain is a sequence of consecutive adjacent nodes obtained by searching along the opposite side flow branch from the rib root feature node.

3. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 1, characterized in that, The local filling state within the affected control volume is updated with the replaced filling factor, and a frontier re-push is performed based on the updated local filling state. This includes: using the local filling state of the time step before the velocity inflection point as the restart state, fixing the arrival step, filling factor, pressure boundary, and source direction label of nodes outside the affected control volume as the original simulation results; for pseudo-merging nodes within the affected control volume, replacing their original filling factor with a backtracking filling factor trajectory starting from the velocity inflection point; for nodes to be updated within the affected control volume, excluding pseudo-merging nodes, establishing a source neighbor set at each time step, which includes nodes that have reached the arrival factor threshold in the previous time step and have a mesh adjacency relationship with the node to be updated; when the source neighbor set is empty, maintaining the corrected filling factor of the node to be updated in the previous time step; when the source neighbor set is not empty, first selecting the source neighbor node with the largest corrected filling factor in the previous time step; if there exists If multiple source neighbor nodes have the same corrected fill factor, the source neighbor node with the earliest arrival step after correction is selected. If multiple candidate source neighbor nodes still exist, the source neighbor node with the smallest angle between its source direction and the original source direction of the node to be updated is selected. If the angle is still the same, a unique master source node is determined in ascending order of node number. Based on the corrected fill factor of the master source node at the previous time step and the non-negative value of the fill factor increment of the node to be updated in the original filling simulation results of adjacent time steps, the corrected fill factor of the node to be updated is updated. When the corrected fill factor of the node to be updated reaches or exceeds the arrival factor threshold for the first time, the corrected arrival step of the node to be updated and the source direction label corresponding to the master source node are recorded. When two different source direction labels appear in the same node or adjacent control volumes within a time step difference not greater than the true merging phase threshold, and both source directions reach the arrival factor threshold, they are re-recorded as true merging nodes.

4. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 3, characterized in that, The affected control volume range is determined by the number of affected layers and the original arrival time sequence, including: starting from the pseudo-merging node, searching for candidate control volumes layer by layer according to the mesh adjacency relationship along the direction of the melt front advancing from early to late in the original filling simulation; when there are shared nodes or shared edges between two control volumes, they are recorded as one layer of adjacency relationship; for any candidate control volume, when the shortest adjacency layer between it and the pseudo-merging node does not exceed the number of affected layers, and its original arrival step is not earlier than the rate inflection point, the candidate control volume is included in the affected control volume range; control volumes whose original arrival step is earlier than the rate inflection point are recorded as boundary fixed control volumes. Boundary fixed control volumes do not participate in the filling factor rewriting, but their original filling factor, original arrival step, and source direction are used as boundary inputs for local front re-push; among them, the number of affected layers is determined according to the local flow transition range of the rib root, and the affected control volume range covers the local range of the rib root inlet, the starting section of the rib sidewall, and the adjacent thin-walled body.

5. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 4, characterized in that, After incorporating the candidate control volume into the affected control volume range, when the backtracking and re-pushing analysis module performs frontier re-pushing: when the corrected frontier leaves the affected control volume range, or when all nodes within the affected control volume range reach the filling completion state, the current local frontier re-pushing stops; when the affected control volume ranges corresponding to multiple pseudo-merging nodes overlap, local frontier re-pushing is performed sequentially according to the rate inflection points of each pseudo-merging node in ascending order; the corrected filling factor, corrected arrival step, and source direction label in the overlapping control volumes that have completed re-pushing first are used as the boundary states when subsequent pseudo-merging nodes perform local frontier re-pushing.

6. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 1, characterized in that, The pseudo-convergence phase discrimination module performs time-series advance analysis and directional phase comparison, including: performing adjacent difference processing on the time series of the filling factor of the root feature node to determine the time step when the increment of the filling factor reaches its maximum value as the rate inflection point; determining the advance ratio based on the ratio of the peak advance amount at the rate inflection point to the upstream unidirectional advance rate, and determining an abnormal advance state when the advance ratio is greater than the advance ratio threshold and the number of stagnation steps before the rate inflection point is not less than the number of stagnation steps threshold; reading the arrival steps of each node in the upstream neighbor chain according to the direction from upstream to the root feature node, and determining the existence of a stable unidirectional advance front when the arrival steps of two or more consecutive upstream nodes gradually increase along this direction; root The state of the opposite branch is determined based on the maximum fill factor and the earliest arrival step before the rate inflection point. The opposite branch states include missing front edges, misphased front edges, and valid front edges. When a stable unidirectional advancing front exists, the opposite branch state is missing front edges, and there is an abnormal advance state, the root feature node is marked as a clearance candidate. When a stable unidirectional advancing front exists, the opposite branch state is misphased front edges, and there is an abnormal advance state, the root feature node is marked as a delay candidate. When the opposite branch state is valid front edges, or there is no abnormal advance state, or there is no stable unidirectional advancing front, the root feature node is marked as a retention candidate.

7. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 6, characterized in that, The backtracking and re-pushing analysis module determines the final action based on the pseudo-merging disposal marker and the phase lag step count. The final action includes clearing, delaying, and retaining: when the pseudo-merging disposal marker is a clearing candidate and the phase lag step count is not less than the large lag step count threshold, the final action is clearing, the merge marker of the node in the original merge node sequence is removed, and local front re-pushing is performed; when the pseudo-merging disposal marker is a retaining candidate, or the phase lag step count of the clearing candidate is less than the large lag step count threshold, the final action is retaining, keeping the original merge marker unchanged. When a pseudo-merging action is marked as a delayed candidate, the final action is determined to be delayed, and a merge delay verification window is constructed based on the time range between the rate inflection point and the rate inflection point plus the phase lag step. Within the merge delay verification window, the original merge mark of the node is not confirmed, and the unidirectional leading edge is maintained according to the backfill factor trajectory. After the window ends, based on whether the opposite branch chain reaches the effective threshold of the opposite branch, and whether the difference between the arrival step of the opposite branch and the arrival step after the correction of the node is not greater than the true merge phase threshold, a new true merge node is generated or the original merge mark is revoked.

8. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 7, characterized in that, After the forming state write-back module regenerates the weld line position field of the rib root, it also calculates the weld line migration record, including: the migration amount of the weld line migration record, calculated based on the distance between the center lines of the corresponding weld line segments in the original weld line position field and the corrected weld line position field; the migration direction of the weld line migration record, determined by pointing from the structural area where the original position is located to the structural area where the corrected position is located; when the migration amount is not less than the effective migration threshold, the migration direction is recorded; when the migration amount is less than the effective migration threshold, it is recorded as no significant migration.

9. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 8, characterized in that, After writing the corrected weld line location field, the molding state write-back module generates a molding risk level based on the corrected weld line location, structural area label, confluence angle at the actual confluence node, number of re-pushes, and migration amount. It also generates causal location information for high-risk areas. The confluence angle is the angle between the propagation directions of the two melt fronts at the actual confluence node. When the corrected weld line is located on the sidewall of the reinforcing rib or in the transition zone of the rib root fillet, and the confluence angle is less than the small confluence angle risk threshold or the number of re-pushes is not less than the number of re-pushes threshold, the area is determined to be of high molding risk level. When the corrected weld line is still located in the thin-walled main body area and the migration amount is less than the effective migration threshold, the area is determined to be of medium molding risk level. When the corrected weld line is far from the stress concentration areas of the rib root, snap-fit ​​seat, and screw column structure, the area is determined to be of low molding risk level. The causal location information includes the original confluence mark forward movement record corresponding to the rib root's stagnant flow and subsequent surge, the flow path record from the nearest gate to the rib root, the record of sudden changes in local resistance at the rib root fillet, and the corresponding mold verification location.

10. The automotive seat back panel mold forming state analysis system based on finite element simulation according to claim 1, characterized in that, The backtracking and re-pushing analysis module generates a backtracking filling factor trajectory, including: reading the rate inflection point of the root feature node and the arrival step of the nearest upstream neighbor, and determining the time step difference between the two as the phase lag step number; reading the average filling factor increment of the upstream neighbor chain within a preset statistical window before the rate inflection point, and determining the average filling factor increment as the upstream unidirectional propagation rate; taking the original filling factor of the time step before the rate inflection point as the backtracking starting point, generating a backtracking filling factor step by step according to the upstream unidirectional propagation rate from the rate inflection point; limiting the backtracking filling factor to a value range of 0 to 1, and replacing the original filling factor from the rate inflection point with the backtracking filling factor.

Images