A lazy sofa filling density rebound self-adaptive control method and system

Abstract

The application discloses a lazy sofa filling density rebound adaptive control method and system. The method divides the soft filling product cavity into head and neck support area, waist and back support area, sitting and lying cushioning area and edge stability area, obtains the weighing data, negative pressure adsorption data, visual inflation profile data and pressure head loading rebound curve of each filling area, calculates the corresponding filling density, local hardness, rebound recovery rate, rebound hysteresis and shape retention capacity, and compares with the target sitting feeling curve. The controller adjusts the filling amount and mixing ratio of particles, sponge fragments and fibers according to the comparison result, performs supplement filling or material extraction on the abnormal collapse, bulge or rebound abnormal area before sealing, and generates quality records after each filling area meets the target range. The application can improve the partition support consistency and quality controllability before sealing of the soft filling product.

Description

Technical Field

[0001] This invention relates to the field of automated production control technology for soft-filled products, and in particular to an adaptive control method and system for the filling density rebound of beanbag chairs. Background Technology

[0002] Beanbag chairs, beanbag sofas, pet mats, and similar soft-filled products typically use foam granules, sponge scraps, fibers, or mixtures thereof as filling materials. They rely on the relative flow of the flexible outer shell and the filling material to create support and cushioning. During the production process, the filling status of these products is generally judged by total weight, filling time, or manual feel, which makes it difficult to accurately reflect the actual differences in support for different parts such as the head and neck, back, sitting / lying areas, and edges.

[0003] Existing soft-filled products can restrict filler migration through layers, seams, or localized restraint structures, and can also improve the material's resilience by modifying the filler formulation. However, these methods typically focus on product structure or material properties and cannot comprehensively test the density, hardness, rebound hysteresis, and shape retention of individual product areas before sealing. If an area shows insufficient filling, excessive bulging, slow rebound, or edge collapse, manual re-inspection is often required, and rework after sealing is quite difficult.

[0004] Therefore, the production of soft-filled products still requires a technical solution that can control the filling according to the functional areas of the product, and combine weighing, negative pressure, visual contour and compression rebound data to make closed-loop corrections for each area, so as to improve the consistency of support status and batch feel in different areas. Summary of the Invention

[0005] One aspect of this invention provides an adaptive control method for the rebound density of beanbag chair filling, addressing issues such as uneven distribution of filling material, lack of rebound performance detection before sealing, and difficulty in timely correction of localized collapses or bulges during the production of soft-filled products. This method does not use the total weight of the entire product as the sole control object. Instead, it divides the product's internal cavity into multiple filling zones according to their support functions, and establishes a correspondence between the target seating comfort curve, actual detection parameters, and compensation control actions in each filling zone, enabling regional-level calibration to be completed before sealing during the filling process.

[0006] The method includes the following steps: Obtain the product model and internal cavity partitioning scheme of the soft filling product to be filled. Divide the internal cavity of the soft filling product into a head and neck support area, a back support area, a sitting / lying buffer area, and an edge stabilization area, and apply the corresponding target sitting comfort curve to each filling area. Position and match the filling port, weighing detection position, negative pressure adsorption detection position, visual detection position, and pressure head loading detection position of each filling area. Fill each filling area with a mixture of granules, sponge fragments, and fibers according to the target sitting comfort curve. Collect the weighing data, negative pressure adsorption data, visual expansion profile data, and pressure head loading rebound curve of each filling area. Calculate the filling density, local hardness, rebound recovery rate, rebound hysteresis, and shape retention capability of each filling area based on the collected data. Compare the calculated actual parameters with the target parameters corresponding to the target sitting comfort curve to obtain the regional comprehensive deviation. Adjust the filling amount and material mixing ratio of the corresponding filling area according to the regional comprehensive deviation, and repeat the short-cycle compression rebound test before sealing. When an abnormal collapse, bulge, or rebound occurs in a filling zone, the filling zone is replenished, material is extracted, or the material ratio is corrected until all filling zones meet the sealing criteria. Then the zones are sealed and a quality record is generated.

[0007] Specifically, the internal cavity zoning scheme is determined based on the product's laid-out contour, the bearing direction of the user's posture, and the outer garment seam boundary. Each filling zone is separated by a breathable partition, flexible limiting seams, localized material guide channels, or controllable clamping areas, forming a relatively independent filling control space. The head and neck support zone corresponds to the upper part of the product and provides support for the head and neck; the back support zone corresponds to the backrest or upper-middle bearing area and provides continuous back support; the sitting / lying buffer zone corresponds to the main contact bearing areas of the human body or pet and provides compressible cushioning; and the edge stabilization zone is set along the outer perimeter of the product to limit the disorderly migration of filler to the outer perimeter. The volume reference, filling port position, sensor detection position, pressure head loading position, and allowable contour offset of each filling zone are written into the controller's process data table according to the product model.

[0008] Preferably, the target seating comfort curve includes a target parameter set T_i corresponding to each filling zone, T_i=[rho_i , H_i , R_i , L_i , S_i ], where i is the filling zone number, rho_i For the target fill density, H_i For the target local hardness, R_i For the target rebound recovery rate, Li For the target rebound hysteresis, S_i To maintain the target shape, the controller calls up the target parameter set based on the product model, the weight class of the target user, and the target sitting or lying posture. It then adjusts the target parameter set based on the elongation of the outer fabric, the batch loose density of the filling, and the ambient temperature and humidity, so that the target parameters can adapt to the looseness of different batches of materials and the deformation capacity of the outer garment.

[0009] Furthermore, visual expansion contour data is obtained through visual acquisition devices positioned above and to the side of the product. The controller estimates the effective volume V_i of the filling area based on the calibrated outer contour height, projected area, and local curvature. The effective volume V_i can be calculated as V_i = V_i0 × gamma_i, and the contour correction coefficient gamma_i can be calculated as gamma_i = 1 + k_a × (A_i - A_i) ) / A_i +k_h×(h_i-h_i ) / h_i Calculate, where V_i0 is the reference volume of the i-th filling zone, A_i is the measured projected area, and A_i h_i is the reference projected area, and h_i is the measured profile height. The baseline contour height is defined by k_a and k_h, which are the area correction weight and height correction weight, respectively. The fill density rho_i of the i-th filling region is calculated as M_i / V_i, where M_i is the region fill quality after deducting the effects of the fixture and outer jacket. When V_i is zero or lower than the preset effective volume lower limit, the controller does not use the current calculation result for closed-loop adjustment, but instead re-acquires visual expansion contour data or outputs a verification command.

[0010] Furthermore, the negative pressure adsorption data includes adsorption pressure P_i, holding time t_i, and contour rebound displacement d_i within the holding time. The controller determines the fit and migration trend of the filler within a local area based on the adsorption pressure, holding time, and contour rebound displacement. When a filling area exhibits insufficient contour height recovery, edge expansion, or abrupt height changes in adjacent areas after negative pressure release, the controller marks this filling area as a region with abnormal shape retention and, in subsequent compensation control, increases the proportion of edge-stabilizing material in this area, reduces the proportion of large-diameter particles, or adjusts the material guide path. The shape retention capability S_i can be determined based on the contour offset ΔC_i after negative pressure release and the allowable contour offset C_i for that area. The relationship between them is determined, for example, S_i = 1 - ΔC_i / C_i Where ΔC_i is the contour offset after negative pressure release, C_i This is the preset allowable contour offset in the target seating curve. When C_i When the value is zero or lower than the preset offset lower limit, the controller uses the preset offset lower limit as the denominator for normalization calculation.

[0011] Furthermore, the indenter loading and rebound curve is obtained by pressing the indenter into the corresponding filling zone at a preset loading speed to the target displacement or target pressure and then unloading. The controller calculates the local hardness H_i, rebound recovery rate R_i, and rebound hysteresis L_i based on the maximum indentation displacement x_i,max, the residual displacement x_i,res after unloading, the loading stage energy E_load, and the unloading stage energy E_unload. Where H_i = F_0 / x_i,max, R_i = 1 - x_i,res / x_i,max, and L_i = (E_load - E_unload) / E_load, and F_0 is the reference load during the loading process. The loading stage energy E_load and the unloading stage energy E_unload are obtained from the areas under the loading curve and unloading curve, respectively. When E_load is zero or lower than the effective energy lower limit, the controller determines the current curve is invalid and reloads for testing. When L_i is higher than the target range and R_i is lower than the target range, the controller prioritizes increasing the proportion of sponge fragments or high-resilience fibers; when H_i is lower than the target range and rho_i is lower than the target range, the controller prioritizes filling; when H_i is higher than the target range and the contour height is higher than the target range, the controller prioritizes material extraction or reducing the particle ratio.

[0012] Furthermore, the regional comprehensive deviation E_i is calculated according to E_i = w_rho × |rho_i - rho_i | / rho_i +w_H×|H_i-H_i | / H_i +w_R×|R_i-R_i | / R_i +w_L×|L_i-L_i | / L_i +w_S×|S_i-S_i | / S_i The calculation is performed, where rho_i, H_i, R_i, L_i, and S_i represent the actual fill density, actual local hardness, actual rebound recovery rate, actual rebound hysteresis, and actual shape retention capability of the i-th filling zone, respectively, and w_rho, w_H, w_R, w_L, and w_S represent the weights of the corresponding indicators. These weights are determined based on the functional requirements of different filling zones. For example, w_H and w_R for the lumbar support zone can be higher than those for the sitting / lying buffer zone, and w_S for the edge stabilization zone can be higher than those for the sitting / lying buffer zone. When a target parameter is zero or lower than a preset normalization lower limit, a preset safety lower limit is used to replace the denominator to avoid distortion of the compensation control results due to abnormal denominators.

[0013] Optionally, the material mixing ratio is represented by P_i=[p_i1,p_i2,p_i3], where p_i1 is the particle ratio, p_i2 is the sponge fragment ratio, p_i3 is the fiber ratio, and p_i1+p_i2+p_i3=1. The controller generates an initial mixing ratio based on the target hardness and target rebound hysteresis of each filling zone, and adjusts the mixing ratio according to the regional comprehensive deviation during the closed-loop correction process. When bulging occurs in the filling zone and the local hardness is higher than the target range, the controller reduces the particle ratio or performs material extraction; when the filling zone collapses and the rebound recovery rate is lower than the target range, the controller increases the sponge fragment or fiber ratio; when the edge stability zone shows an outward expansion trend, the controller increases the ratio of materials with supporting and stabilizing effects and reduces the material guiding speed of adjacent areas.

[0014] Another aspect of the present invention provides an adaptive control system for the filling density and rebound of a beanbag chair. This system includes a zoning modeling module, a filling execution module, a weighing detection module, a negative pressure adsorption detection module, a visual contour detection module, a pressure head rebound detection module, a parameter calculation module, a closed-loop control module, a sealing judgment module, and a quality recording module. The zoning modeling module is used to acquire the product model and internal cavity zoning scheme, and divide the internal cavity of the soft-filled product to be filled into a head and neck support area, a back support area, a sitting / lying buffer area, and an edge stabilization area. The filling execution module is used to fill each filling area with a mixture of granules, sponge fragments, and fibers according to the target sitting comfort curve. The weighing detection module is used to collect weighing data for each filling area. The negative pressure adsorption detection module is used to collect negative pressure adsorption data for each filling area. The visual contour detection module is used to collect visual expansion contour data for each filling area. The pressure head rebound detection module is used to collect the pressure head loading rebound curve for each filling area. The parameter calculation module is used to calculate the filling density, local hardness, rebound recovery rate, rebound hysteresis, and shape retention capability. The closed-loop control module adjusts the filling volume and material mixing ratio based on the comparison between actual parameters and the target seating curve. The sealing judgment module issues a sealing command after each filling zone meets the sealing judgment conditions. The quality record module saves product model, zone parameters, test data, compensation actions, and sealing results.

[0015] Specifically, the filling execution module includes a granule feeding unit, a sponge fragment feeding unit, a fiber feeding unit, a mixing and metering unit, and a zoned material guiding unit. The zoned material guiding unit is connected to the filling ports of the head and neck support area, the back support area, the sitting and lying buffer area, and the edge stabilization area, respectively. The closed-loop control module is communicatively connected to the mixing and metering unit, the zoned material guiding unit, the negative pressure adsorption detection module, and the pressure head rebound detection module. It is used to control the replenishment amount, extraction amount, mixing ratio, and number of test retryes for the corresponding filling area based on the comprehensive deviation of the area. The quality recording module associates and stores the target sitting comfort curve of the same product, the actual parameters of each filling area, the comprehensive deviation of the area, the anomaly type, the compensation action, the material batch, and the sealing time to form a traceable production quality record.

[0016] The technical solution of this application transforms the flexible inner cavity of soft-filled products such as beanbag chairs, beanbag sofas, and pet cushions into multiple filling zones corresponding to their support functions. A target seating comfort curve and a set of regional parameters are established for each filling zone. The actual state of the zone is obtained through weighing, negative pressure adsorption, visual expansion contour analysis, and pressure head loading rebound curve analysis. Filling density, local hardness, rebound recovery rate, rebound hysteresis, and shape retention capability are calculated. The overall deviation of the zone is used as the basis for closed-loop control. Before sealing, abnormal areas are refilled, material is extracted, and material ratios are corrected. This solution transforms the filling process from overall weight control to regional performance control, and integrates the detection, judgment, and correction before sealing into a continuous control process.

[0017] Compared to existing technologies, this application can detect and address issues such as insufficient local filling, excessive local bulging, excessive rebound hysteresis, and unstable edge morphology before product sealing. Because each filling zone is configured with target density, target hardness, target rebound recovery rate, target rebound hysteresis, and target morphology retention capability, the controller can make differentiated corrections based on the functional differences of different areas, rather than uniformly replenishing the entire product or adjusting by hand. By jointly estimating the effective volume of the area using weighing data and visual contour data, the impact of irregular deformation of the flexible outer jacket on density judgment can be reduced. Identifying the rebound recovery rate and rebound hysteresis through the springback curve of the pressure head loading allows for determination of whether the filling material ratio is suitable for the target seating feel before sealing. Negative pressure adsorption detection identifies local fit and migration trends, improving issues such as edge collapse, lateral expansion, and abrupt changes in adjacent areas. The quality recording module saves target curves, detection data, compensation actions, and sealing results, providing data for subsequent production verification and batch parameter adjustments. Attached Figure Description

[0018] To more clearly illustrate the technical solution of this application, the accompanying drawings are briefly described below.

[0019] Figure 1 This is a schematic diagram of the overall process of the method of the present invention.

[0020] Figure 2 This is a block diagram of the system structure of the present invention.

[0021] Figure 3 This is a schematic diagram showing the correspondence between the internal cavity partitions and target parameters of the present invention.

[0022] Figure 4 This is a schematic diagram of the multi-source detection data stream of the present invention.

[0023] Figure 5 This is a schematic diagram of the compression rebound curve characteristics of the present invention.

[0024] Figure 6 This is a schematic diagram of the closed-loop compensation control process of the present invention.

[0025] Figure 7 This is a schematic diagram of the device structure of the present invention.

[0026] Figure 8 This is a schematic diagram comparing the application scenarios and technical effects of the present invention. Detailed Implementation

[0027] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. The described embodiments are for illustrative purposes only and do not constitute a limitation on the scope of protection of the present invention. Where there is no conflict, the technical features of the following embodiments can be combined with each other. Example 1

[0028] This embodiment provides a method for adaptive control of the filling density and rebound of a beanbag chair. The overall process of this method is as follows: Figure 1 As shown. Figure 1 The data processing sequence in the production process is illustrated, including product model reading, internal cavity partitioning, target seating curve retrieval, initial partition filling, multi-source detection, parameter calculation, deviation judgment, closed-loop compensation, short-cycle compression and rebound retest, sealing judgment, and quality recording. Each step forms a closed-loop control relationship driven by detection data, enabling the controller to correct localized abnormal areas before sealing.

[0029] In step S101, the controller reads the product model, outer cover specifications, stitching structure, preset usage posture, and target user weight class of the soft stuffing product to be filled. The product model can be obtained through barcodes, RFID tags, production work orders, or input via the user interface. The controller calls up the corresponding internal cavity partitioning scheme based on the product model. For beanbag chairs, the internal cavity partitioning scheme may include a head and neck support area, a lumbar support area, a sitting / lying buffer zone, and an edge stabilization area; for beanbag chairs or pet mats, the partition names can be adjusted according to the actual structure, but each partition still corresponds to a different support function during product use.

[0030] In step S102, the controller determines the spatial range of each filling zone based on the product's laid-out outline, outer garment seam boundary, interlayer position, and predetermined load-bearing direction. Figure 3 The correspondence between the internal cavity partitions and target parameters is shown, where the head and neck support area is located at the upper part of the product or the upper end of the backrest, the lumbar support area is located in the main back support range, the sitting and lying buffer zone is located in the main pressure area, and the edge stabilization area is set along the outer periphery of the product. Figure 3 Each filling zone corresponds to the target density, target hardness, target rebound recovery rate, target rebound hysteresis, and target shape retention capability, making the partitioning not only a geometric division but also directly related to production control parameters.

[0031] In step S103, the controller positions and matches the filling port, weighing detection position, negative pressure adsorption detection position, visual detection position, and pressure head loading detection position of each filling area. This positioning and matching can be achieved through fixture positioning holes, outer casing stitching marks, visual recognition marks, or a tooling coordinate system. For flexible outer casings, the controller can first use vacuum adsorption or clamping of the frame to put the product in a repeatable detection state, and then map each detection position to the tooling coordinate system. After this processing, the pressure head loading position, visual acquisition area, and negative pressure adsorption area can be kept consistent with the same filling area, avoiding misjudgment of local parameters due to detection position offset.

[0032] In step S104, the controller invokes the target seating comfort curve. The target seating comfort curve includes the target parameter set T_i for each filling zone, T_i=[rho_i , H_i , R_i , L_i , S_i ]. Where i is the filling zone number, rho_i For the target fill density, H_i R_i represents the target local hardness. For the target rebound recovery rate, Li For the target rebound hysteresis, S_i To maintain the target shape, the target seating comfort curve can be established through prototype testing, ergonomic targets, pet pad load targets, or historical qualified batch statistics. The controller can also adjust the target parameters based on the elongation of the outer fabric, the batch loose packing density of the filling, and the ambient temperature and humidity, so that the target parameters match the current production conditions.

[0033] In step S105, the filling actuator fills the mixed filler into each filling zone according to the target seating comfort curve. The mixed filler includes granules, sponge fragments, and fibers. Granules provide flowability and volume support, sponge fragments provide compression rebound and a certain degree of damping, and fibers improve the overlap and shape retention between fillers. The controller generates an initial mixing ratio P_i=[p_i1,p_i2,p_i3] based on the target parameter set, where p_i1 is the granule ratio, p_i2 is the sponge fragment ratio, p_i3 is the fiber ratio, and p_i1+p_i2+p_i3=1. The zoned material guiding mechanism transports the mixed filler to the corresponding filling port and controls the initial filling amount through weighing feedback or volumetric metering feedback.

[0034] In step S106, the controller collects weighing data from each filling zone. The weighing data can be obtained jointly by the whole product support platform and the zone clamping mechanism, or it can be obtained by combining the difference before and after filling, zone material guide mass measurement, and whole-piece weighing verification. For flexible products where zone weighing is difficult to implement directly, the controller can use the material guide mass measurement value as the main source of the zone filling mass M_i, and verify the sum of the masses of each zone using the whole-piece weighing data. If the difference between the sum of the zone masses and the whole-piece weighing exceeds a set threshold, the controller pauses the sealing process and requires re-weighing or checking for residue in the material guide tube.

[0035] In step S107, the controller acquires negative pressure adsorption data and visual expansion contour data. The negative pressure adsorption detection position can be set on the surface of each filling area or near the edge stability zone. The negative pressure adsorption device holds the product under a predetermined pressure P_i for a time t_i and records the contour rebound displacement d_i after the negative pressure is released. The visual contour detection device acquires the product's outer contour from above and the side, obtaining the projected area A_i, contour height h_i, edge curvature, and height difference between adjacent areas. Figure 4 The multi-source detection data stream is shown, in which weighing data, negative pressure data, visual contour data, and indenter loading rebound curve are respectively entered into the parameter calculation module, and after being associated with a unified timestamp and area number, they form area status data.

[0036] In step S108, the controller performs a pressure head loading and rebound test. The pressure head rebound detection mechanism aligns the pressure head with the loading position of the corresponding filling area, presses it into the product surface at a preset loading speed, holds it for a short time after reaching the target displacement or target load, and then returns at a preset unloading speed. The controller records the force and displacement data during the loading and unloading phases to form a pressure head loading and rebound curve. Figure 5The main characteristics of the curves are shown, where the area difference between the loading and unloading curves characterizes rebound hysteresis, the maximum indentation displacement is used to calculate local hardness, and the residual displacement after unloading is used to calculate the rebound recovery rate. To avoid over-testing and affecting the condition of flexible products, the loading depth and holding time of the short-cycle compression rebound test should be set within a range that will not cause permanent deformation.

[0037] In step S109, the controller calculates the actual parameters of each filling region. The filling density rho_i can be calculated according to rho_i = M_i / V_i, where M_i is the region filling mass of the i-th filling region, and V_i is the effective volume estimated based on the visual expansion contour. The effective volume V_i can be obtained from the reference volume V_i0 and the contour correction coefficient gamma_i, i.e., V_i = V_i0 × gamma_i. The contour correction coefficient gamma_i can be calculated according to gamma_i = 1 + k_a × (A_i - A_i) ) / A_i +k_h×(h_i-h_i ) / h_i Calculation. If V_i is zero, negative, or below the preset effective volume lower limit, the controller will not output the density judgment result, but will instead re-acquire visual data or trigger manual review.

[0038] In step S110, the controller calculates the local hardness H_i, springback rate R_i, and springback hysteresis L_i based on the indenter loading springback curve. The local hardness H_i can be calculated as H_i = F_0 / x_i,max, where F_0 is the reference load during loading, and x_i,max is the maximum indentation displacement when the reference load is reached. The springback rate R_i can be calculated as R_i = 1 - x_i,res / x_i,max, where x_i,res is the residual displacement after unloading. The springback hysteresis L_i can be calculated as L_i = (E_load - E_unload) / E_load, where E_load is the energy during loading, and E_unload is the energy during unloading. The shape retention capability S_i can be calculated as S_i = 1 - ΔC_i / C_i Calculate, where ΔC_i is the contour offset after negative pressure release, C_i To allow for contour offsets, the denominators used in the above formulas all have valid lower limits; if the denominator is invalid, closed-loop correction is not performed.

[0039] In step S111, the controller calculates the regional comprehensive deviation E_i based on the actual parameters and the target parameters. The regional comprehensive deviation E_i can be expressed as E_i = w_rho × |rho_i - rho_i | / rho_i +w_H×|H_i-H_i | / H_i +w_R×|R_i-R_i | / R_i +w_L×|L_i-L_i | / L_i +w_S×|S_i-S_i | / S_i Calculation. The controller compares the contribution of each deviation item and determines the compensation action based on the anomaly type. When rho_i and H_i are both below the target range, the controller performs supplementary filling in the corresponding filling area; when the contour height is above the target range and H_i is above the target range, the controller performs material extraction or reduces the particle ratio; when L_i is above the target range and R_i is below the target range, the controller increases the proportion of sponge fragments or fibers; when S_i is below the target range and the edge expansion is obvious, the controller increases the proportion of edge stabilizing material or re-executes negative pressure shaping.

[0040] Figure 6 The closed-loop compensation control process is illustrated. After obtaining the comprehensive deviation of a region, the controller first determines whether the deviation exceeds the region's threshold. If it does not exceed the threshold, the region enters the sealing candidate state. If it exceeds the threshold, the controller identifies the anomaly type based on the deviation item and generates parameters such as replenishment amount, material extraction amount, mixing ratio adjustment amount, or negative pressure shaping parameters. After the compensation action is executed, the controller re-collects multi-source data and recalculates the actual parameters. This loop can be set with a maximum number of iterations. If the sealing judgment condition is still not met after reaching the maximum number of iterations, a verification command is output and sealing is paused to prevent abnormal products from entering subsequent processes.

[0041] During the sealing judgment stage, the controller checks whether each filling zone simultaneously meets the requirements for density error, hardness error, springback recovery rate error, springback hysteresis error, shape retention error, and appearance contour error. When all filling zones meet the requirements, the sealing judgment module sends a sealing command to the sealing mechanism, which then heat-seals, stitches, or zips the product filling opening. The quality record module saves the product model, zone target parameters, zone actual parameters, each compensation action, material batch, inspection time, sealing time, and anomaly handling results. This quality record is used for subsequent production traceability and can also serve as a data source for updating the target sitting comfort curve. Example 2

[0042] This embodiment describes the specific implementation methods for cavity partitioning, target seating curve establishment, and multi-source detection data fusion. This embodiment can be used in conjunction with the method described in Embodiment 1, or it can be used as a method for generating partitioning parameters and processing detection data in a soft filling product production line.

[0043] like Figure 3As shown, the product's inner cavity is fixed to the unfolding fixture before entering the filling station. The unfolding fixture may include a circumferential clamping frame, a filling port positioning clamp, a layer positioning pressure plate, and visual calibration points. The controller establishes a tooling coordinate system based on the visual calibration points and converts the product's laid-out outline into a two-dimensional area boundary. For beanbag chairs, the head and neck support area can be set within the upper third of the backrest, the lumbar support area can be set within the transition range from the middle of the backrest to the seat surface, the sitting and lying buffer zone can be set within the main pressure area of ​​the seat or lying position, and the edge stabilization area can be set within an annular area with a predetermined outer circumference. For pet mats, the head and neck support area can correspond to the backrest of the mat, the sitting and lying buffer zone can correspond to the central support part of the mat, and the edge stabilization area can correspond to the annular rim.

[0044] Zone boundaries do not necessarily have to be formed by rigid partitions. For products requiring flexible deformation capabilities, breathable partitions, flexible restraining sutures, dotted stitching connections, or temporary clamping during the filling stage can be used to create relatively independent filling spaces between filling zones. Breathable partitions can reduce the large-scale migration of filler from different areas in a short period of time, while allowing air to pass through, avoiding the influence of local air pockets on contour judgment during filling and negative pressure testing. Flexible restraining sutures allow for some deformation during use, but provide repeatable boundary constraints during production testing. Local material guide channels can be closed after filling or restricted by sealing sutures, controlling filler migration during subsequent use.

[0045] The controller establishes a partition parameter record for each filling zone. The partition parameter record includes zone number i, zone name, reference volume V_i0, and reference projected area A_i. Reference profile height h_i , filling port coordinates, weighing associated channel, negative pressure adsorption coordinates, visual inspection area, pressure head loading coordinates, allowable contour offset C_i The parameters include the regional comprehensive deviation threshold E_i,lim and the maximum number of compensation iterations N_i,max. These parameters can be obtained through measurement of prototypes when establishing the product model, or automatically generated through calibration prototypes during model changeover. The controller reads the partition parameter records during production to ensure that different batches of products execute consistent detection and control processes.

[0046] The target seating comfort curve can be established through sample testing. Specifically, samples that meet the target feel are first selected, and then fixed on the testing fixture according to a zoned scheme. Weighing data, visual contour data, negative pressure adsorption data, and indenter loading rebound curves for each filling zone are collected. The controller extracts the target filling density rho_i from the sample data. Target local hardness H_i Target rebound recovery rate R_i Target rebound hysteresis L_i The ability to maintain the target shape S_i When different hardness levels exist for the same model, multiple target parameter groups can be created, and the corresponding level can be selected through the product work order. For products such as pet mats, different target parameter groups can also be created based on the target load-bearing weight level.

[0047] The target seating comfort curve can also be updated based on historical production data. The controller uses the data of sealed, inspected, and qualified products as candidate samples, and removes samples with excessive anomaly compensation times, material batch abnormalities, or excessive test retries. The actual parameters of each zone of the retained samples are statistically analyzed to obtain the center value and allowable range of the target parameters. When updating the target parameters, the old version curve, update date, sample quantity, and applicable material batch should be retained to ensure the traceability of production records. This update process does not change the closed-loop control logic of this invention, but rather provides a data source that is closer to production conditions for the target parameters.

[0048] like Figure 4 As shown, multi-source detection data are aligned within the controller according to region number and timestamp. The weighing detection module outputs the region filling mass M_i or an estimated region filling mass; the negative pressure adsorption detection module outputs the adsorption pressure P_i, holding time t_i, contour rebound displacement d_i after negative pressure release, and pressure relief recovery curve; the visual contour detection module outputs the projected area A_i, contour height h_i, local curvature, edge expansion, and height difference between adjacent regions; the indenter rebound detection module outputs the loading curve, unloading curve, maximum indentation displacement x_i,max, residual displacement x_i,res, loading stage energy E_load, and unloading stage energy E_unload. The parameter calculation module writes the above data into the region status table and marks whether the data is valid.

[0049] Visual contour detection can be achieved using structured light, binocular cameras, time-of-flight cameras, or multi-camera contour stitching. Before detection, the controller determines the transformation relationship between the camera coordinate system and the tooling coordinate system using a calibration plate or tooling fixing points. During detection, the product surface may have wrinkles, local shadows, or fabric texture interference. The controller can use regional average height, contour boundary smoothing, and outlier removal processing. If the number of removed outliers exceeds a preset proportion, the controller will not use the current visual result and will prompt the user to rearrange the product cover or re-acquire the image. This avoids mistaking local wrinkles for bulges or collapses.

[0050] Negative pressure adsorption testing is used to evaluate the adhesion between the flexible jacket and the filler, as well as the filler migration trend. During testing, the negative pressure adsorption head is attached to the surface or edge area of ​​the corresponding filling area. The negative pressure source adjusts the adsorption pressure to a preset pressure P_i and maintains it for a preset time t_i. If the contour near the adsorption head drops too quickly during the holding period, it indicates that the local filler may be too loose or there may be cavities. If the contour height does not recover sufficiently after the negative pressure is released, it indicates that there may be insufficient rebound or uneven filler accumulation in that area. If the edge area expands outward after release, it indicates that the edge stability zone does not adequately restrict the filler in adjacent areas. The controller classifies the anomalies into collapse tendency, bulging tendency, edge expansion tendency, or insufficient adhesion tendency based on the above phenomena.

[0051] When fusing multi-source detection data, the controller first performs individual validity checks, followed by regional consistency checks. Individual validity checks include whether the weighing difference exceeds the weighing range, whether the negative pressure reaches the set pressure, whether the visual data is complete, and whether the indenter curve shows sensor saturation or data interruption. Regional consistency checks include whether the density matches the contour height, whether the local hardness matches the indentation displacement, and whether the contour offset after negative pressure release matches the shape retention capability. If the weighing indicates sufficient filling but the visual contour still shows obvious collapse, the controller can determine that there is local accumulation or insufficient expansion of filler within the zone, and prioritize low-intensity vibration conditioning or negative pressure shaping instead of immediately adding material.

[0052] This embodiment is illustrated by... Figure 3 and Figure 4 The illustrated regional parameter records and multi-source data streams enable irregular flexible products to be transformed into controllable objects with regional numbers, target parameters, detection locations, and compensation actions. Since each filling zone has independent target parameters and detection data, the controller can distinguish the sources of anomalies in different zones. For example, low density in the sitting / lying buffer zone usually corresponds to a refill action; low shape retention in the edge stability zone usually corresponds to material ratio or edge limit adjustments; and low hardness and high rebound hysteresis in the back support zone may correspond to insufficient sponge fragments or overly loose filler distribution. This distinction helps reduce indiscriminate overall refilling. Example 3

[0053] This embodiment describes the specific implementation method of compression-rebound curve feature extraction and closed-loop compensation control. This embodiment is related to... Figure 5 and Figure 6 Correspondingly, the focus is on explaining how the controller determines the actions of replenishment, material extraction, proportion correction, and retesting based on the pressure head loading springback curve and the overall deviation of the area.

[0054] like Figure 5As shown, the indenter loading and rebound curve includes a loading phase, a holding phase, and an unloading phase. During the loading phase, the indenter moves towards the product surface at a preset speed. A force sensor records the load F after the indenter contacts the product, and a displacement sensor records the indentation displacement x. During the holding phase, the indenter is held near the target displacement or target load for a short time to observe the rearrangement of particles, sponge fragments, and fibers within the filler. During the unloading phase, the indenter returns to its initial position, and the controller records the recovery state of the product surface during the unloading process. The controller can extract the maximum load, maximum indentation displacement, residual displacement, loading curve slope, unloading curve slope, loading energy, and unloading energy from this curve.

[0055] See further Figure 5 The three curves in the diagram represent the relationship between the indenter displacement and the loading force during the loading phase, and are used to determine the indentation displacement x_i,max under the reference load F_0. The springback curve during the release phase represents the decay of residual indentation over time after unloading, and is used to determine the residual displacement x_i,res after the preset recovery time. The loading and unloading energy difference curve represents the area difference between the loading curve E_load and the unloading curve E_unload, which serves as the basis for calculating the springback hysteresis L_i. These three curves correspond to the calculation of local hardness, springback recovery rate, and springback hysteresis, respectively, enabling... Figure 5 The curve characteristics in the curve correspond to the subsequent parameter calculations of H_i, R_i, and L_i.

[0056] Local hardness H_i is used to characterize the compressive resistance of the corresponding filling area under a reference load. If calculated using H_i = F_0 / x_i,max, where F_0 is the reference load during loading and x_i,max is the indentation displacement when the reference load is reached. A larger x_i,max indicates a softer area or insufficient filling; a smaller x_i,max indicates a harder area or localized bulging. For different product models, F_0 can be set according to the weight class of the target user. For example, the reference load for an adult beanbag chair can be higher than that for a pet mat. To avoid the influence of noise at the moment of contact of the indenter, the controller can take the average displacement as x_i,max within a stable sampling range near the load reaching F_0.

[0057] The rebound recovery rate R_i characterizes the ability of the product surface to return to its original shape after unloading. R_i = 1 - x_i,res / x_i,max, where x_i,res is the residual displacement that remains after a preset recovery time following unloading. The smaller x_i,res is, the closer R_i is to 1, indicating sufficient recovery in the area; the larger x_i,res is, the lower R_i is, indicating insufficient rebound or obstructed filler redistribution in the area. If x_i,max is zero or below the effective displacement lower limit, the controller does not calculate R_i and re-executes the loading detection or prompts for pressure head position calibration.

[0058] Rebound hysteresis L_i characterizes the difference between the input energy during the loading phase and the energy recovered during the unloading phase. L_i = (E_load - E_unload) / E_load, where E_load is the area under the loading curve and E_unload is the area under the unloading curve. E_load and E_unload can be obtained by discrete integration of the force-displacement curve. A large L_i indicates significant energy loss in the filling area during compression and recovery, which may be caused by material friction, filler interlocking, localized over-density, or insufficient rebound of sponge fragments. For sitting / lying buffer zones, a certain degree of rebound hysteresis is permissible to improve cushioning; however, for head and neck support areas and lumbar and back support areas, excessive rebound hysteresis may lead to slow support recovery, therefore it needs to be controlled within the target range.

[0059] Closed-loop compensation control such as Figure 6 As shown. The controller first determines whether the regional comprehensive deviation E_i exceeds the regional comprehensive deviation threshold E_i,lim. ​​If E_i does not exceed the threshold, the controller marks the region as qualified and awaiting sealing. If E_i exceeds the threshold, the controller calculates the contribution rate of each deviation item. The contribution rate can be understood as the proportion of a single deviation in E_i. The controller determines the primary anomaly type based on the deviation item with the highest contribution rate and the secondary anomaly type based on the second highest deviation item. For example, when rho_i is low and H_i is low, the primary anomaly is insufficient filling; when H_i is high and visual height is high, the primary anomaly is local bulging; when R_i is low and L_i is high, the primary anomaly is insufficient rebound; when S_i is low and edge expansion is large, the primary anomaly is insufficient shape retention.

[0060] Combination Figure 6 The deviation determination section first receives the actual parameters and target parameters of each filling zone, calculates the comprehensive deviation E_i of the region, and then identifies the deviation type based on the deviation direction of rho_i, H_i, R_i, L_i and S_i. Figure 6Insufficient filling, excessive bulging, delayed rebound, and edge collapse correspond to the actions of replenishing filling, material extraction, adjusting material ratio, and improving edge shape retention, respectively. After the actions are completed, the retest and sealing sections re-collect test data and repeat the compression and rebound test; if the retest results still do not meet the sealing conditions, then... Figure 6 The feedback path returns to the deviation judgment section for further correction; if the retest result meets the sealing conditions, the sealing process begins and the quality record is saved.

[0061] When the primary anomaly is insufficient filling, the controller calculates the refill amount based on the density difference and effective volume. The refill amount can be calculated according to ΔM_i=(rho_i) The formula is determined by ΔM_i × V_i × η_m, where ΔM_i is the filler mass and η_m is the compensation coefficient. η_m can be less than or equal to 1 to avoid overfilling at once. After filling, the controller can perform a short-term low-intensity shaping to ensure uniform distribution of the filler within the area, and then re-perform visual contour detection and indenter loading rebound test. If the density reaches the target but the hardness is still lower than the target after filling, the controller can increase the particle ratio or add more supportive sponge fragments.

[0062] When the main anomaly is a localized bulge, the controller first determines whether the bulge is caused by excessive filling. If rho_i is higher than the target range and the contour height h_i is higher than the reference height h_i... The controller executes the material extraction action. This extraction can be accomplished through the reverse suction channel of the partitioned material guiding unit or by connecting the extraction pipe to the filling port. The extraction amount can be determined based on density difference, contour height difference, and extraction coefficient. If rho_i is not higher than the target range but the local contour height is too high, the controller can determine that the filler is unevenly piled and prioritize negative pressure shaping, light clamping, or guide path adjustment instead of direct extraction.

[0063] When the main anomaly is insufficient rebound, the controller determines the direction of material ratio adjustment based on the deviation of R_i and L_i. If R_i is below the target range and L_i is above the target range, it indicates that the recovery after release in this area is slow and the energy loss is large. The controller can increase the proportion of sponge fragments or high-resilience fibers and reduce the proportion of excessively flowing large-diameter particles. If H_i is below the target range at the same time, it indicates that insufficient material rebound and insufficient filling may coexist. The controller first performs a small amount of replenishment and then adjusts the material ratio. If H_i is above the target range, it indicates that the material is too densely packed or the particle ratio is too high. The controller can first remove material or reduce the particle ratio and then retest the rebound curve.

[0064] When the primary anomaly is insufficient shape retention, the controller determines a compensation strategy based on negative pressure adsorption data and visual contour data. If the outward expansion of the edge stabilization zone exceeds the target range, the controller can increase the proportion of fiber or supporting sponge fragments in the edge stabilization zone and reduce the material conveying speed from adjacent sitting / lying buffer zones to the edge area. If the local contour recovery is insufficient after negative pressure release but the density is normal, the controller can adjust the negative pressure shaping path to redistribute the filler near the interlayer. If there is a sudden change in height between adjacent areas, the controller can perform sequential compensation between the two areas, first stabilizing the edge or lumbar support area, and then adjusting the sitting / lying buffer zone.

[0065] During the compensation control process, the controller sets iteration termination conditions. The first termination condition is when all individual parameters fall within the target range and the overall regional deviation E_i does not exceed the threshold. The second termination condition is when the number of compensation iterations reaches the maximum number N_i,max, at which point the controller pauses sealing and outputs a verification command. The third termination condition is when the number of invalid sensor data exceeds a preset number, at which point the controller prompts a check of the sensor or product clamping status. By setting termination conditions, the controller can avoid continuous compensation under abnormal data conditions and can also prevent overfilling or over-extraction.

[0066] This embodiment can also employ a zone-priority approach for compensation sorting. For beanbag chairs, the lumbar support area and edge stabilization area are typically reviewed before the sitting / lying buffer zone, because the lumbar support area has a greater impact on support continuity, while the edge stabilization area has a greater impact on maintaining the overall shape. For pet mats, the edge stabilization area and sitting / lying buffer zone can have higher priority. The controller executes compensation sequentially according to priority. If the compensation action in one area might affect adjacent areas, then a linked retest is performed in the adjacent areas. This reduces the disturbance of single-area compensation to other areas.

[0067] Figure 5 and Figure 6 The proposed solution transforms compression rebound testing from a mere quality sampling method into a direct component of filling control. The controller determines regional hardness, rebound recovery, and hysteresis based on curve characteristics, and then uses density and visual profile data to determine compensation actions. Since short-cycle compression rebound testing is re-executed after compensation, the control results can be verified before sealing, thus reducing the likelihood of discovering localized collapses or insufficient rebound only after sealing. Example 4

[0068] This embodiment provides a beanbag chair filling density rebound adaptive control system. The module relationship of the system is as follows: Figure 2 As shown, a specific arrangement of the device structure is as follows: Figure 7 As shown. Figure 2From the perspective of functional modules, the data flow and control flow between the partition modeling module, filling execution module, weighing detection module, negative pressure adsorption detection module, visual contour detection module, pressure head rebound detection module, parameter calculation module, closed-loop control module, sealing judgment module and quality recording module are shown. Figure 7 The spatial relationship between the feeding mechanism, zoned guide pipe, weighing platform, negative pressure adsorption head, vision acquisition device, pressure head detection mechanism, clamping fixture, sealing mechanism and control cabinet is shown from the perspective of the production station.

[0069] The zoning modeling module is used to obtain the product model and internal cavity zoning scheme, and divides the internal cavity of the soft filler product to be filled into a head and neck support zone, a back support zone, a sitting / lying buffer zone, and an edge stabilization zone. This module can consist of a process database in an industrial control computer, a product model identification unit, and a zoning calculation program. After the product model identification unit reads the production work order or product label, the process database outputs the corresponding zoning boundaries, filling port positions, detection positions, and target sitting comfort curves. The zoning calculation program converts the above data into region numbers and control parameters in the tooling coordinate system and sends them to the filling execution module, detection module, and closed-loop control module.

[0070] The filling execution module is used to fill each filling zone with a mixture of granules, sponge fragments, and fibers according to the target seating comfort curve. This module may include a granule feeding unit, a sponge fragment feeding unit, a fiber feeding unit, a mixing and metering unit, and a zoned feeding unit. The granule feeding unit can use screw conveying, pneumatic conveying, or vibratory feeding. The sponge fragment feeding unit may include an anti-bridging hopper and a quantitative feeding mechanism. The fiber feeding unit may include a loosening mechanism and a metering mechanism. The mixing and metering unit performs proportional metering of the three materials based on the controller output P_i=[p_i1,p_i2,p_i3]. The zoned feeding unit delivers the mixture to the corresponding filling zone through a switching valve or a multi-channel feed pipe.

[0071] The weighing and detection module is used to collect weighing data from each filling zone. This module may include a whole-piece weighing platform, zoned material feeding and metering sensors, and a unit for calculating the mass difference before and after filling. For products that can be relatively separated by fixtures, the weighing and detection module can obtain the estimated regional mass value through the local support platform; for products that are difficult to weigh directly by zone, the weighing and detection module can use the metering values ​​of each feeding unit as the source of zoned mass and verify the total mass using the whole-piece weighing platform. The weighing and detection module outputs M_i and a data validity marker to the parameter calculation module.

[0072] The negative pressure adsorption detection module is used to collect negative pressure adsorption data from each filling zone. This module may include a negative pressure source, a pressure regulating valve, a pressure sensor, a negative pressure adsorption head, a displacement detection unit, and a pressure relief control valve. The negative pressure adsorption head can be mounted on a movable crossbeam or on a multi-point detection frame. During detection, the negative pressure adsorption head moves to the target area, applies a preset adsorption pressure P_i, and holds it for a time t_i. The displacement detection unit records the contour change. The negative pressure adsorption detection module outputs the adsorption pressure, holding time, contour rebound displacement after release, and a valid negative pressure detection mark to the parameter calculation module.

[0073] The visual contour detection module is used to acquire visual expansion contour data for each filling area. This module may include an upper camera, a side camera, a light source, a calibration plate, and an image processing unit. The upper camera is used to obtain the product's projected contour and the projected area A_i of each zone, while the side camera is used to obtain the contour height h_i and the edge expansion. The image processing unit converts the image coordinates to product coordinates based on the tooling calibration points and outputs the area, height, curvature, and height difference of each zone to the parameter calculation module. If the image is occluded or overexposed, the visual contour detection module outputs an invalid mark, and the controller re-acquires the image or prompts the user to adjust the product cover.

[0074] The indenter springback detection module is used to collect the indenter loading springback curves for each filling zone. This module may include a servo indenter, force sensor, displacement sensor, loading controller, and safety limit structure. The servo indenter performs loading and unloading according to the loading speed, target load, or target displacement issued by the controller. The force sensor and displacement sensor sample synchronously, and the loading controller generates force-displacement curves. The safety limit structure is used to prevent the indenter loading from exceeding the allowable range of the product or tooling. The indenter springback detection module outputs x_i,max, x_i,res, E_load, E_unload, and curve validity markers to the parameter calculation module.

[0075] The parameter calculation module is used to calculate fill density, local hardness, rebound rate, rebound hysteresis, and shape retention capability. This module can be implemented by an industrial computer, a computing unit in a programmable logic controller (PLC), or an edge computing device. After receiving multi-source detection data, the parameter calculation module first checks the timestamp, region number, and validity marker of each data point, and then calculates Vi, rho, H, R, Li, and Si. For invalid data, the parameter calculation module does not output closed-loop control results but instead sends a retest reason to the closed-loop control module. For valid data, the parameter calculation module outputs the actual parameters of the region and the deviations of each individual parameter.

[0076] The closed-loop control module adjusts the filling amount and material mixing ratio based on the comparison between actual parameters and the target seating curve. This module may include a deviation calculation unit, an anomaly identification unit, a compensation amount generation unit, a ratio correction unit, and an iterative control unit. The deviation calculation unit calculates the comprehensive deviation E_i for the region. The anomaly identification unit determines the anomaly type based on density deviation, hardness deviation, rebound recovery rate deviation, rebound hysteresis deviation, and shape retention deviation. The compensation amount generation unit outputs the filling amount, material extraction amount, or negative pressure shaping parameters. The ratio correction unit outputs the new material mixing ratio P_i. The iterative control unit records the number of compensations for the current region and outputs a verification command when the maximum number of compensations is reached.

[0077] The sealing judgment module issues a sealing command after all filling areas meet the sealing judgment conditions. These conditions may include: the overall regional deviation of all filling areas not exceeding a threshold; no abnormalities in individual parameters that would prohibit sealing; no obvious bulging or collapse in the visual contour; a valid indenter rebound test; passing the quality difference verification; and the number of compensation iterations not exceeding the allowable range. Once the sealing judgment module confirms that the conditions are met, it sends a sealing command to the sealing mechanism. The sealing mechanism can employ heat sealing, stitching, zipper locking, or a combination of sealing methods, depending on the product's outer casing structure.

[0078] The quality record module is used to store product model, zone parameters, test data, compensation actions, and sealing results. This module can be set up in an industrial control computer database, a local server, or a production management system. Quality records include production batch, material batch, target comfort curve version, target parameters for each filling zone, actual parameters for each filling zone, overall deviation of the area before and after each compensation, compensation action type, compensation amount, number of retests, sealing time, and operator or equipment number. Quality records can be exported as production reports or used for subsequent process parameter optimization.

[0079] Figure 7 The illustrated device structure may include a feeding station, a filling and inspection station, a compensation and retesting station, and a sealing station arranged along the production line. The feeding station is used to fix the outer casing to be filled onto a clamping fixture. The filling and inspection station is equipped with a feeding mechanism, zoned guide pipes, a weighing platform, and a vision acquisition device. The compensation and retesting station is equipped with a negative pressure suction head and a pressure head detection mechanism, and may also be combined with the filling and inspection station. The sealing station is equipped with a sealing mechanism and a finished product output mechanism. The control cabinet is communicatively connected to each actuator and inspection mechanism, and controls them via industrial Ethernet, fieldbus, or discrete signals. This device structure enables… Figure 2 The functional modules in the process receive hardware support on the production line.

[0080] Combination Figure 7The granular material silo, sponge fragment silo, and fiber silo each output a single material to the metering and weighing unit. The metering and weighing unit determines the amount of each material to be added based on the proportioning control signal issued by the controller. The mixing and conveying unit delivers the mixed filler to the switchable valve assembly. The switchable valve assembly selects the appropriate filling nozzle and zone guide pipe according to the zoned material guiding instructions, allowing the mixed filler to enter the corresponding filling area of ​​the soft product to be filled. The vision acquisition unit collects the contour data of the product to be filled, and the movable pressure head and force and displacement sensors collect loading test and curve data. The negative pressure adsorption channel performs directional adsorption or shaping on the designated area according to the negative pressure control signal output by the controller. After receiving the vision data and curve data, the controller generates proportioning control, negative pressure control, and sealing control instructions, enabling... Figure 7 The feeding, testing, compensation, and sealing mechanisms shown form a combination with... Figure 2 The closed-loop device corresponding to the system module.

[0081] In a feasible control sequence, the zoning modeling module first sends the initial filling targets for each region to the filling execution module. After the filling execution module completes the initial filling, the weighing detection module, visual contour detection module, negative pressure adsorption detection module, and pressure head rebound detection module collect data sequentially or in parallel. The parameter calculation module generates actual parameters. The closed-loop control module determines whether compensation is needed. If compensation is needed, the filling execution module performs supplementary filling, material extraction, or proportional correction. After compensation is completed, the detection module re-collects data. After all regions pass the test, the sealing judgment module sends a sealing command. The quality recording module saves the complete process data. This sequence corresponds to the steps in the method embodiment and can realize zoning filling and rebound closed-loop control in a systematic manner. Example 5

[0082] This embodiment illustrates the application of the invention using a production scenario of an adult beanbag chair. The beanbag chair includes a flexible outer cover and an internal mixed filling. When unfolded, the product is approximately 1600 mm long and 900 mm wide. The outer cover is made of fabric with a certain elongation, and the filling includes foam particles, sponge scraps, and fibers. This product requires stable support in the head and neck area, continuous support in the lumbar and back area, compressible cushioning in the sitting and reclining area, and that the edge areas do not exhibit significant outward expansion or collapse after transportation and use.

[0083] The production line configuration includes three types of material feeding hoppers, a mixing and metering unit, a four-channel zoned material guiding unit, a whole-piece weighing platform, an overhead vision camera, side vision cameras, a negative pressure adsorption detection head, a servo pressure head springback detection mechanism, flexible clamping fixtures, a sealing mechanism, and an industrial control computer. The three material feeding hoppers respectively store foam granules, sponge scraps, and fibers. The mixing and metering unit outputs the mixed filler according to the proportions issued by the controller. The zoned material guiding unit connects to the filling ports of the head and neck support area, back support area, sitting / lying buffer zone, and edge stabilization area. The vision cameras are installed above and to the side of the filling detection station, and the negative pressure adsorption detection head and servo pressure head are mounted on a movable crossbeam.

[0084] The technical requirements for this scenario can be set as follows: the target local hardness in the head and neck support area is higher than that in the sitting / lying buffer zone, and the target rebound hysteresis is lower than that in the sitting / lying buffer zone; the target rebound recovery rate in the lumbar and back support area is higher than that in the sitting / lying buffer zone, and the height difference between adjacent areas should not exceed a preset range; the sitting / lying buffer zone allows for a larger indentation displacement, but the rebound recovery rate must not be lower than the target lower limit; the target shape retention ability in the edge stability area is higher than that in other areas, and the edge expansion after negative pressure release should not exceed the allowable range. These requirements are written into the target sitting comfort curve, rather than relying on human touch for judgment.

[0085] During one production process, the controller first reads the product work order and invokes the zoning scheme for the adult beanbag chair. A clamping fixture unfolds and secures the outer cover, while a vision acquisition device identifies the cover's outline and calibration points. The controller divides the product's interior into four filling zones and determines the fixture coordinates for each filling port and detection point. Subsequently, the controller generates the initial mixing ratio for each zone. The head and neck support zone and the lumbar support zone have a relatively high proportion of granules and supporting sponge fragments; the sitting and lying buffer zone has a relatively high proportion of sponge fragments and fibers; and the edge stabilization zone has a relatively high proportion of fibers and supporting materials.

[0086] After initial filling, the overall weighing platform outputs the total product mass, the zoned material feeding unit outputs the material supply mass for each zone, and the controller verifies the sum of the zone masses based on the total mass. The overhead vision camera obtains the projected area of ​​each zone, and the lateral vision camera obtains the contour height of each zone. The negative pressure adsorption detection head sequentially applies negative pressure to the edge stabilization zone, back support zone, and sitting / lying buffer zone, recording the contour changes after release. The servo pressure head sequentially performs short-cycle loading and rebound tests on the head and neck support zone, back support zone, and sitting / lying buffer zone. All data is entered into the parameter calculation module according to the zone number.

[0087] In this detection, the controller may find that the rho_i and H_i of the sitting / lying buffer zone are below the target range, and the visual contour height is also below the reference height, indicating insufficient filling in this area. The filler quantity is generated by calculating (-rho_i)×V_i×η_m, and a small amount of filler is performed while maintaining the original mixing ratio. After filler, the controller re-executes visual inspection and pressure head loading rebound test. If H_i reaches the target range after filler but L_i is still higher than the target range, the controller increases the proportion of sponge fragments or high-resilience fibers in the next compensation to reduce rebound hysteresis.

[0088] The controller may also detect that the density of the edge stabilization zone is within the target range, but the outward expansion after negative pressure release is too large, and S_i is lower than the target range. In this case, the controller does not directly add a large amount of material, but instead increases the fiber ratio in the edge stabilization zone, reduces the material guiding speed from the adjacent sitting and lying buffer zones towards the edge, and performs a negative pressure shaping. During the retest, if the outward expansion of the edge decreases and the height difference between adjacent areas enters the allowable range, the edge stabilization zone enters the sealing candidate state. This processing reflects the present invention's differentiation of abnormal types, that is, when the density is qualified but the shape is not maintained sufficiently, the compensation action should focus on the material ratio and shaping path.

[0089] For the back support area, if the rebound curve after indentation shows that both H_i and R_i are below the target range, the controller determines that the support is insufficient and that recovery is also insufficient. If the weighing data also shows that rho_i is below the target range, the controller first performs refilling; if rho_i is close to the target range, the controller adjusts the material ratio, increasing the proportion of supportive sponge fragments and decreasing the proportion of excessively flowing particles. During the retest after compensation, the controller compares whether H_i and R_i have entered the target range. If the target is reached, the back support area enters the sealing candidate state.

[0090] Figure 8 The application scenarios and technical effects of this embodiment are shown in comparison. Figure 8 The results include three sub-plots. Sub-plot A uses the number of detection and compensation rounds as the x-axis and the overall deviation E_i as the y-axis to compare the changes in the overall deviation before sealing under overall weight control and zoned closed-loop control. Sub-plot B uses the head and neck, back, sitting / lying, and edge regions as the x-axis and the rebound recovery rate R_i as the y-axis to compare the rebound recovery rate of the regions after the target lower limit, zoned closed-loop control, and overall weight control. Sub-plot C uses the head and neck, back, sitting / lying, and edge regions as columns and the density deviation after closed-loop control, original process density deviation, rebound deviation after closed-loop control, and original process rebound deviation as rows to form a region state matrix. Figure 8 The comparison conditions are the pre-sealing test results of the same model product, the same outer shell specification and the same batch of filling material. The evaluation objects include density, hardness, rebound recovery rate, rebound hysteresis and shape retention ability, rather than just the total weight as the evaluation basis.

[0091] In this embodiment, the overall weight control method and the zoned closed-loop control method can be compared as shown in the table below.

[0092]

[0093] As can be seen from the above application process, this embodiment maps the product's functional areas, detection data, and control actions. Insufficient filling of the sitting / lying buffer zone is identified through density, hardness, and contour height; the outward expansion of the edge stability zone is identified through contour changes after negative pressure release; and insufficient support in the lumbar support zone is identified through local hardness and rebound recovery rate. Different anomalies correspond to different compensation actions, avoiding uniform material replenishment based solely on total weight. Short-cycle compression and rebound testing before sealing verifies whether the compensated local parameters meet the target range, enabling regional-level quality confirmation of the product before it enters the sealing process.

[0094] This embodiment can also be extended to beanbag sofas and pet mats. For beanbag sofas, zones can be established based on the backrest, seat, and outer stable area, with the particle ratio as the primary adjustment target, and sponge fragments and fibers as the adjustment targets for rebound and shape retention. For pet mats, zones can be established based on the central lying area, headrest area, and side area, and different target parameters can be set according to the pet's weight class. Although the shape, size, and material ratio of different products vary, as long as zone target parameters can be established, area detection data can be collected, and compensation control can be implemented, the zoned filling density rebound adaptive control method and system of this invention can be used.

[0095] In the aforementioned production scenario for adult beanbag chairs, hardware selection can be adjusted based on product size and production cycle. The weighing platform should ideally have a range covering the maximum filling mass of a single product and a resolution capable of distinguishing changes in single filling volume. The visual acquisition device should ensure the complete outline of the product in its flat state is within the field of view; the lateral visual acquisition device should cover the height variation range of the edge stability zone and the back support zone. The contact area of ​​the negative pressure suction head should be smaller than the corresponding filling area but larger than the area of ​​a single local wrinkle, so that the negative pressure detection results reflect the overall fit of the area rather than a single wrinkle change. The pressure head end face can be circular or elliptical, with chamfered edges to reduce localized damage to the outer fabric. The pressure head loading speed, target displacement, and holding time should be set according to the fabric elongation and filling compression characteristics to ensure the test elicits a measurable rebound response without causing permanent indentations.

[0096] During parameter setting, the controller can initially set different initial mixing ratios for each region. For example, the head and neck support zone uses granules and supportive sponge fragments as the main components, while retaining a certain proportion of fibers to maintain shape; the back support zone increases the proportion of supportive sponge fragments to give the loading curve a more stable slope; the sitting / lying buffer zone appropriately increases the proportion of fibers and resilient fragments to make the recovery process after compression smoother; the edge stabilization zone increases the proportion of fibers or shape-maintaining materials and reduces the instantaneous flow rate during material feeding to reduce local accumulation. These ratios are not fixed values, but rather initial values ​​under the target sitting comfort curve. The controller corrects these ratios based on the overall deviation of the regions after detection. To avoid excessive single corrections causing reverse deviations, the controller can set single upper limits for the replenishment amount, extraction amount, and ratio change, and perform a retest after each correction.

[0097] In the execution process, the production line can operate with one quality record number per product. After the product enters the clamping fixture, the controller reads the number and creates a record file. After the initial filling is completed, the record file records the batch of the three fillers, the theoretical filling amount of each area, and the actual measurement amount. After the first inspection, the record file records rho_i, H_i, R_i, L_i, S_i, and E_i of each area. If compensation occurs, the record file further records the reason for compensation, the compensation action, the compensation amount, the retest result after compensation, and whether a verification instruction was triggered. For areas that did not trigger compensation, the time and corresponding parameters of entering the sealing candidate state are also saved. This recording method can link the sealing determination process of a single product with the actual inspection data, facilitating subsequent tracing of the area status by product number.

[0098] Under continuous production conditions, the controller can also statistically analyze the overall regional deviation of the same batch of products. When a certain batch of materials causes the same directional deviation in the same area of ​​multiple products, the controller can prompt adjustments to the initial mixing ratio or target parameter correction coefficient corresponding to that batch of materials. For example, if a batch of granules has a low bulk density, and multiple products show low rho_i in the sitting and lying buffer zones after initial filling, the controller can increase the theoretical filling amount in that area during the initial filling of the next product, without having to wait for each product to be refilled before reaching the target range. If a batch of sponge fragments has a low rebound recovery rate, and multiple products show low R_i and high L_i in the back support area, the controller can increase the fiber ratio or replace it with a high-resilience fragment batch. This statistical adjustment does not replace closed-loop detection for individual products, but rather reduces the number of repeated compensations based on single-product closed-loop detection.

[0099] For pet mat applications, products typically have a central lying area and surrounding areas, and the weight of the users varies considerably. The controller can map the central lying area to a sitting / lying buffer zone, the surrounding areas to edge stabilization zones, and the partial backrest area to a head and neck support zone. The target sitting comfort curve is set according to the pet's weight class; lower weight classes can reduce the baseline load F_0 and target hardness H_i. A larger weight class can improve the target density in the central lying area and the target shape retention ability in the edge stability area. Since pet mat edges usually require shape stability, the outward expansion after negative pressure release and the height difference between adjacent areas can be used as higher weighting indicators. Through the same weighing, visual contour, negative pressure adsorption, and pressure head rebound detection, the controller can determine whether the edge is too loose, the central area is too soft, and the backrest area has insufficient rebound before sealing.

[0100] For beanbag sofa applications, the product may lack a distinct rigid frame, and the filling material exhibits high fluidity during use. The controller can use temporary clamping fixtures to create repeatable partition boundaries during the filling stage and perform detection according to the partition target parameters before sealing. For models requiring a strong sense of enclosure, the sitting / lying buffer zone can be set with a lower target hardness and moderate rebound hysteresis; for models requiring strong back support, the lumbar support zone can be set with a higher target hardness and higher rebound recovery rate. The edge stabilization zone limits excessive migration of the filling material to the periphery. If the image shows insufficient edge height and excessive central area height, the controller can first redistribute the filling material through negative pressure shaping and material guide path adjustment before deciding whether to refill or remove material. This processing method is suitable for products with significant variations in the shape of their flexible internal cavities.

[0101] In terms of quality evaluation, this invention does not require all areas to have the same hardness or density, but rather requires that the actual parameters of each area match their corresponding target parameters. Therefore, the head and neck support area can be relatively stable, the sitting / lying buffer zone can be relatively soft, and the edge stability zone can focus more on maintaining its shape. Compared to methods that only control the total weight, zoned closed-loop control can identify local differences when the total weight is similar; compared to methods that only rely on material formulation, zoned closed-loop control can identify actual deviations within a single product caused by material guide paths, outer garment wrinkles, local accumulation, or material batch fluctuations; compared to post-sealing sampling inspection, pre-sealing inspection and compensation reduce the scope of rework. In this embodiment... Figure 8 The sources of the above control effect are explained by three sub-graphs: comprehensive deviation change, regional rebound recovery rate, and regional state matrix.

[0102] Regarding production safety and anomaly handling, the controller sets validity conditions for each type of sensor data. When weighing data is abnormal, the controller checks whether the tooling is in contact with the weighing platform or whether there is residual material in the feed tube; when visual data is abnormal, the controller checks the camera calibration point, lighting conditions, and any obstructions from the outer casing; when negative pressure data is abnormal, the controller checks the sealing status of the suction head and the pressure sensor reading; when the pressure head curve is abnormal, the controller checks the zero point of the pressure head, the force sensor range, and the product clamping status. The closed-loop control module only outputs compensation actions when the data is valid. If data from a certain area is invalid multiple times consecutively, the system does not directly determine that the product is qualified, nor does it continue to perform refilling or material extraction; instead, it pauses and outputs a verification command. This anomaly handling method reduces the risk of false compensation caused by sensor malfunctions or clamping abnormalities.

[0103] Furthermore, the controller can write the target comfort curve version, zoning scheme version, and sensor calibration version into the quality record. If the product model changes, the outer material batch changes, or the filler supply batch changes, the controller can require a zoning parameter review during the first piece production. After the first piece review is passed, the sealing judgment module allows entry into continuous production mode; in continuous production mode, if several consecutive products show a deviation trend close to the threshold in the same area, the controller can prompt recalibration of the target comfort curve or adjustment of the initial filling parameters. Through version records and trend reminders, a clear correspondence is maintained between target parameters, actual detection data, and compensation actions, enabling the closed-loop control process of this invention to be repeatedly implemented in long-term production.

[0104] The above description is merely a preferred embodiment of the present invention and does not limit the scope of patent protection of the present invention. Any equivalent structural substitutions, equivalent process changes, or combinations of related technical features made based on the technical content disclosed in the specification and drawings of this invention, as long as they do not depart from the technical concept of the present invention, should be covered within the scope of protection of this invention.

Claims

1. A method for adaptive control of filling density rebound in a beanbag chair, characterized in that, include: Obtain the product model and cavity partitioning scheme of the soft filling product to be filled, divide the cavity of the soft filling product to be filled into a head and neck support area, a waist and back support area, a sitting and lying buffer area and an edge stabilization area, and call the corresponding target sitting comfort curve for each filling area; The filling ports, weighing detection positions, negative pressure adsorption detection positions, visual detection positions, and pressure head loading detection positions of each filling zone are located and matched. According to the target seating comfort curve, fill each filling zone with a mixture of granules, sponge fragments and fibers; Weighing data, negative pressure adsorption data, visual expansion profile data, and pressure head loading rebound curves were collected for each filling zone. The filling density, local hardness, rebound recovery rate, rebound hysteresis and shape retention ability of each filling zone are calculated based on the collected data. The calculated actual parameters are compared with the target parameters corresponding to the target seating comfort curve to obtain the regional comprehensive deviation. Adjust the filling volume and material mixing ratio of the corresponding filling area according to the comprehensive deviation of the area, and repeatedly perform short-cycle compression and rebound tests according to the filling area number before sealing; when there is abnormal collapse, bulging or abnormal rebound in a certain filling area, perform supplementary filling, material extraction or material ratio correction for that filling area until each filling area meets the sealing judgment conditions, then seal and generate quality records.

2. The method according to claim 1, characterized in that, The internal cavity partitioning scheme is determined based on the product's flat outline, the bearing direction of the user posture, and the outer garment seam boundary. Each filling zone forms a relatively independent filling control space through breathable partitions, flexible limiting seams, local material guiding channels, or controllable clamping areas. The head and neck support zone corresponds to the upper part of the product and is used to support the head and neck. The waist and back support zone corresponds to the backrest or upper middle bearing area and is used to provide continuous back support. The sitting and lying buffer zone corresponds to the main contact bearing areas of the human body or pet and is used to provide compressible buffering. The edge stabilization zone is set along the outer periphery of the product and is used to limit the disorderly migration of the filler to the outer periphery. The volume reference, filling port position, sensor detection position, pressure head loading position, and allowable contour offset of each filling zone are written into the process data table of the controller according to the product model.

3. The method according to claim 1, characterized in that, The target seating comfort curve includes a target parameter set T_i corresponding to each filling zone, T_i=[rho_i , H_i , R_i , L_i , S_i ], where i is the filling zone number, rho_i For the target fill density, H_i For the target local hardness, R_i For the target rebound recovery rate, Li For the target rebound hysteresis, S_i To maintain the target shape, the controller calls up the target parameter set based on the product model, the weight class of the target user, and the target sitting or lying posture, and corrects the target parameter set based on the elongation of the outer fabric, the loose packing density of the filling batch, and the ambient temperature and humidity, so that the same product model has a traceable source of target parameters under different material batches; the target parameter set is established from qualified sample test data or historical qualified batch statistical data, and is written into the quality record along with the version of the target sitting comfort curve.

4. The method according to claim 1, characterized in that, The visual expansion contour data is obtained through visual acquisition devices positioned above and to the side of the product. The controller estimates the effective volume V_i of the filling area based on the calibrated outer contour height, projected area, and local curvature, where V_i = V_i0 × gamma_i, and gamma_i = 1 + k_a × (A_i - A_i) ) / A_i +k_h×(h_i-h_i ) / h_i Where V_i0 is the reference volume of the i-th filling region, gamma_i is the profile correction coefficient, and A_i is the measured projected area. h_i is the reference projected area, and h_i is the measured profile height. The baseline contour height is defined by k_a and k_h, which are the area correction weight and height correction weight, respectively. The filling density of the i-th filling area is rho_i = M_i / V_i, where M_i is the area filling quality after deducting the influence of the fixture and the outer jacket. When V_i is zero or lower than the preset effective volume lower limit, the controller re-acquires the visual expansion contour data or outputs a verification command.

5. The method according to claim 1, characterized in that, The negative pressure adsorption data includes adsorption pressure P_i, holding time t_i, and contour rebound displacement d_i within the holding time. The controller determines the fit and migration trend of the filler in the local area based on the adsorption pressure, holding time, and contour rebound displacement. When a filling area shows insufficient contour height recovery, outward expansion of the edge area, or abrupt height change in adjacent areas after the negative pressure is released, the controller marks the filling area as an area with abnormal shape retention. In subsequent compensation control, the proportion of edge stabilizing material in the area is increased, the proportion of large-diameter particles is reduced, or the material guiding path is adjusted to reduce the shape instability caused by local material migration after sealing.

6. The method according to claim 1, characterized in that, The indenter loading rebound curve is obtained by pressing the indenter into the corresponding filling area at a preset loading speed to the target displacement or target pressure and then unloading. The controller calculates the local hardness H_i, rebound recovery rate R_i, and rebound hysteresis L_i based on the maximum indentation displacement x_i,max, the residual displacement x_i,res after unloading, the loading stage energy E_load, and the unloading stage energy E_unload, where H_i=F_0 / x_i,max, R_i=1-x_i,res / x_i,max, and L_i=(E_load-E_unload) / E_load, and F_0 is the reference load during the loading process. When E_load is zero or lower than the effective energy lower limit, the controller determines that the current curve is invalid and reloads for detection. When L_i is higher than the target range and R_i is lower than the target range, the controller prioritizes increasing the proportion of sponge fragments or high-resilience fibers. When H_i is lower than the target range and rho_i is lower than the target range, the controller prioritizes performing supplementary filling.

7. The method according to claim 1, characterized in that, The regional comprehensive deviation E_i is calculated as E_i = w_rho × |rho_i - rho_i | / rho_i +w_H×|H_i-H_i | / H_i +w_R×|R_i-R_i | / R_i +w_L×|L_i-L_i | / L_i +w_S×|S_i-S_i | / S_i The calculation is performed, where rho_i, H_i, R_i, L_i, and S_i represent the actual filling density, actual local hardness, actual rebound recovery rate, actual rebound hysteresis, and actual shape retention capability of the i-th filling zone, respectively, and w_rho, w_H, w_R, w_L, and w_S represent the weights of the corresponding indicators. When a target parameter is zero or lower than the preset normalization lower limit, the preset safety lower limit is used to replace the denominator. When E_i exceeds the regional comprehensive deviation threshold, the controller selects to replenish, extract, change the mixing ratio, change the filling sequence, or re-execute negative pressure shaping based on the contribution of the deviation item. The controller identifies the deviation item with the largest contribution as the main source of anomaly and uses the height difference or edge expansion of adjacent filling zones as auxiliary judgment conditions.

8. The method according to claim 1, characterized in that, The material mixing ratio is represented by P_i=[p_i1,p_i2,p_i3], where p_i1 is the particle ratio, p_i2 is the sponge fragment ratio, p_i3 is the fiber ratio, and p_i1+p_i2+p_i3=1. The controller generates an initial mixing ratio based on the target hardness and target rebound hysteresis of each filling zone, and adjusts the mixing ratio according to the regional comprehensive deviation during the closed-loop correction process. When a bulge appears in the filling zone and the local hardness is higher than the target range, the particle ratio is reduced or material extraction is performed. When a collapse occurs in the filling zone and the rebound recovery rate is lower than the target range, the sponge fragment or fiber ratio is increased. When the edge stable zone shows an outward expansion trend, the ratio of materials with supporting and stabilizing effects is increased and the material guiding speed of adjacent areas is reduced. After the ratio adjustment, the visual expansion contour data and the contour offset after negative pressure release are re-acquired.

9. A beanbag chair filling density rebound adaptive control system, used to implement the method described in any one of claims 1-8, characterized in that, It includes a partition modeling module, a filling execution module, a weighing and detection module, a negative pressure adsorption detection module, a visual contour detection module, a pressure head rebound detection module, a parameter calculation module, a closed-loop control module, a sealing judgment module, and a quality recording module; The partitioning modeling module is used to obtain the product model and the cavity partitioning scheme, and to divide the cavity of the soft filler product to be filled into the head and neck support area, the waist and back support area, the sitting and lying buffer area and the edge stabilization area. The filling execution module is used to fill each filling zone with a mixture of granules, sponge fragments and fibers according to the target seating comfort curve; the weighing detection module is used to collect the weighing data of each filling zone; The negative pressure adsorption detection module is used to collect negative pressure adsorption data of each filling zone; the visual contour detection module is used to collect visual expansion contour data of each filling zone. The pressure head rebound detection module is used to collect the pressure head loading rebound curve of each filling zone; The parameter calculation module is used to calculate filler density, local hardness, rebound recovery rate, rebound hysteresis, and shape retention capability. The closed-loop control module is used to adjust the filling amount and material mixing ratio based on the comparison results between the actual parameters and the target sitting comfort curve. The sealing determination module is used to issue a sealing command after each filling zone meets the sealing determination conditions. The quality record module is used to save product model, partition parameters, test data, compensation actions, and sealing results.

10. The system according to claim 9, characterized in that, The filling execution module includes a granule feeding unit, a sponge fragment feeding unit, a fiber feeding unit, a mixing and metering unit, and a zoned material guiding unit. The zoned material guiding unit is connected to the filling ports of the head and neck support area, the back support area, the sitting and lying buffer area, and the edge stabilization area, respectively. The closed-loop control module is communicatively connected to the mixing and metering unit, the zoned material guiding unit, the negative pressure adsorption detection module, and the pressure head rebound detection module. It is used to control the replenishment amount, extraction amount, mixing ratio, and number of test retests for the corresponding filling area based on the comprehensive deviation of the area. The quality recording module associates and stores the target sitting comfort curve of the same product, the actual parameters of each filling area, the comprehensive deviation of the area, the anomaly type, the compensation action, the material batch, and the sealing time to form a traceable production quality record. The parameter calculation module outputs a retest mark when the weighing data, visual expansion contour data, negative pressure adsorption data, or pressure head loading rebound curve is invalid. When the closed-loop control module receives the retest mark, it pauses the compensation action and controls the corresponding detection module to re-collect data.

Images