A door panel felt bonding and curing online process closed-loop control system and method

Abstract

The application discloses a door board hemp felt bonding and curing online process closed-loop control system and method, relates to the field of intelligent control technology, and comprises the following steps: acquiring environment state data and thermal field data of a hemp felt bonding interface by using a multi-sensor array to perform asynchronous scanning and mapping to a plurality of initial hemp felt partitions of a mold preset; respectively calculating interface micro-impedance entropy and thermal escape nonlinear indexes in the partitions, inputting a mixing adaptive neural network model, generating a core feature tensor containing a future curing state prediction value through a dynamic weight adjustment mechanism; introducing a contact thermal resistance estimation model, judging a current dominant defect type according to the core feature tensor, and cooperatively adjusting heating power and local compression force of each partition through a partition thermal-pressure coupling execution mechanism until a final curing quality index generated satisfies a preset standard. The application solves the problem that a traditional method cannot dynamically balance external heating and internal heat release under variable interface conditions.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, specifically to a closed-loop control system and method for online bonding and curing of door panel hemp felt. Background Technology

[0002] The bonding and curing of door panel felt is a critical process in the manufacturing of automotive door assemblies, and its quality directly affects the structural integrity, NVH (noise, vibration, and harshness) performance, and appearance durability of the door. In actual production, the curing process is affected by a combination of factors, including adhesive formulation, adhesive thickness, mold geometry, clamping force distribution, and ambient temperature and humidity. Many defects are not visible on the surface but are hidden, often only becoming apparent after the vehicle has been in use or after high-temperature aging, leading to increased risks of rework and recalls. Therefore, building a process control system capable of real-time, multi-dimensional perception and closed-loop compensation of the curing process on the production line has become an urgent need to improve yield and reliability.

[0003] Currently used technologies have alleviated the above problems to some extent. Typical methods include: implementing PID control for key process parameters and using multi-source online sensing for process monitoring; reducing initial variation at the process front end through surface pretreatment, controlling the sizing mechanism, and accurate quantitative sizing; and establishing quality labels for high-risk batches to support subsequent data-driven models. Recent research has also introduced multimodal data fusion and machine learning methods for anomaly detection and quality prediction, thereby enabling more refined adjustments to the process.

[0004] However, in door panel manufacturing, the production workshop not only experiences diurnal fluctuations in ambient temperature and humidity (affecting the moisture content of the felt and the initial tack of the adhesive), but also suffers from random settling of airborne particles (contaminating the bonding interface) and drift in contact thermal resistance caused by mold thermal cycling. Furthermore, the high cost of cleanliness control means that hidden peeling defects are still highly likely to occur and be exposed in batches during subsequent operations. Existing PID or data-driven models cannot cope with the strong coupling between the randomness of the interface microstate and the nonlinearity of adhesive curing exothermics. When the presence of micro-dust at the interface increases the contact thermal resistance, conventional heating can lead to localized under-curing; blindly increasing power can cause localized ablation or carbonization of the felt due to the cumulative self-heating of the adhesive. It is impossible to dynamically balance external heating and internal exothermics under varying interface conditions.

[0005] Therefore, the present invention provides a closed-loop control system and method for online bonding and curing of door panel hemp felt. Summary of the Invention

[0006] The purpose of this invention is to provide a closed-loop control system and method for the online process of bonding and curing door panel hemp felt, so as to solve the existing problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a closed-loop control system for online bonding and curing of door panel hemp felt, comprising: The multi-dimensional field partitioning mapping module is configured to perform asynchronous scanning using a multi-sensor array during the mold closing stage to acquire environmental state data and thermal field data of the felt bonding interface, and then map the data to several preset independent control partitions of the mold. In the middle, it is defined as the initial hemp felt partition; The heterogeneous feature decoupling calculation module is configured to calculate the micro-impedance entropy of the interface within each of the initial felt partitions. and thermal escape nonlinear exponent ; The solidification field co-prediction tensor generation module is configured to generate the interface micro-impedance entropy. and the thermal escape nonlinear index The input mixture is used to create an adaptive neural network model, which generates a core feature tensor containing predictions of future solidification states through a dynamic weight adjustment mechanism. ; The contact thermal resistance adaptive compensation control module is configured to introduce a contact thermal resistance estimation model based on the core feature tensor. The current dominant defect type is determined, and the heating power of each zone is adjusted in a coordinated manner through a zoned thermo-pressure coupling actuator. With local clamping force Until the final cured quality index is generated. Meets the preset standards.

[0008] A further improvement of the present invention is that the multi-dimensional field partitioning mapping module includes a thermal field acquisition unit, an environmental field acquisition unit, and a data preprocessing unit; The thermal field acquisition unit is configured to acquire the infrared temperature distribution of the solidified area in real time at a first frequency using an infrared thermal imager. Based on the geometric topology of the mold surface, the temperature data is divided into the corresponding control partitions as the initial felt partitions; The environmental field acquisition unit is configured to acquire the surface equivalent resistance of the felt surface using a surface impedance probe array. Simultaneously, the spatial particle density of the current workstation is obtained using a vision or laser particle counter. ; The data preprocessing unit is configured to process the infrared temperature distribution. Surface equivalent resistance and spatial particle density are spatiotemporally aligned and Kalman filtered for noise reduction to construct initial feature vectors. .

[0009] A further improvement of this invention is that the mixing adaptive neural network model includes: Dual-channel input architecture, including independent processing of interface micro impedance entropy Characterized interface channels and thermal escape nonlinear exponent The input layer of the thermal channel is characterized; By setting an adaptive signal-to-noise ratio weight adjustment strategy, the micro-impedance entropy of the interface is monitored in real time. The amplitude, through the interface micro-impedance entropy The amplitude is adjusted in real time to control the interface channel weights; Output core feature tensor ,in To predict future curing uniformity, This is the recommended control increment.

[0010] A further improvement of this invention is that the signal-to-noise ratio adaptive weight adjustment strategy includes adjusting the interface micro impedance entropy when... When the weight is less than the preset threshold, the weights of the interface channel and the hot channel are kept balanced. When the interface micro-impedance entropy When the threshold is greater than or equal to the preset threshold, the dynamic weight adjustment mechanism formed by combining the effective confidence level and the model's historical predictive power with the original weight of the current input interface channel automatically reduces the weight coefficient of the interface channel and correspondingly increases the weight coefficient of the hot channel, so that the control strategy tilts towards a conservative mode based on thermal history.

[0011] A further improvement of the present invention is that the contact thermal resistance adaptive compensation control module includes a contact thermal resistance estimation model construction unit and a decoupling compensation strategy execution unit; The contact thermal resistance estimation model building unit is configured to use the interface micro-impedance entropy. Define contact thermal resistance and local clamping force The relation is expressed as ; The decoupling compensation strategy execution unit is configured as follows: When the micro-impedance entropy of a certain partition interface is monitored The increase leads to an increase in contact thermal resistance and a nonlinear thermal escape exponent. When the system is in a stable range, it enters a pressure-priority compensation mode, where the control system maintains a constant heating power while simultaneously controlling the current interface micro-impedance entropy. With thermal escape nonlinear exponent weighted sum The target reduction ratio is calculated, and then the target contact thermal resistance is mapped to the pressure compensation value. This drives the pressure actuator in that zone to increase the local clamping force. ; When the nonlinear exponent of thermal escape in a certain region is detected Greater than the corresponding threshold and continuously greater than When a new local clamping force is obtained from the pressure priority compensation mode within a sampling period, the decoupling compensation strategy execution unit executes the heat-pressure coordinated suppression mode, and the control system simultaneously reduces the heating power. And through the thermal escape nonlinear exponent Adjusting the local clamping force by varying the amount of change Heat absorption and suppression are achieved by utilizing the heat capacity of the mold.

[0012] A further improvement of this invention lies in the fact that the interface microscopic impedance entropy By setting a standard reference value for the resistance of clean hemp felt and particle sensitivity coefficient ; Obtain real-time surface equivalent resistance And calculate its baseline deviation from the standard clean hemp felt resistance reference value; using spatial particle density and its sensitivity coefficient An exponential weighting term is constructed to characterize the nonlinear contribution of particles to the disorder of the impedance distribution; the basic deviation is coupled with the exponential weighting term, and the result is logarithmically smoothed to finally obtain the dimensionless interfacial microscopic impedance entropy. .

[0013] A further improvement of this invention is that the process of obtaining the thermal escape nonlinear index includes: firstly, monitoring the real-time temperature of the partition. time derivative The heating rate characterization value is obtained; secondly, the mold clamping force currently applied to this partition is obtained. Based on pressure transmission coefficient and target temperature difference Calculate the expected rate of temperature rise under the theoretical linear model; The thermal escape nonlinearity index is obtained by calculating the difference between the temperature rise rate and the expected temperature rise rate: When the thermal escape nonlinear exponent A value greater than 0 indicates that a self-accelerating exothermic reaction occurs inside the adhesive, i.e., the risk of heat escape.

[0014] A further improvement of the present invention is that the system further includes a core feature tensor. The quality prediction component in the curve is integrated over time to generate the final curing quality index. And construct a multidimensional secure hypersurface, only if all partitions When all surfaces converge to the multidimensional safe hypersurface, the system issues an opening command.

[0015] A further improvement of this invention is that the solidified field collaborative prediction tensor generation module also includes a dynamic partitioning reconstruction mechanism for feature clustering, including defining the initial felt partition as the smallest execution unit; in generating the core feature tensor Previously, the microscopic impedance entropy of the interface between adjacent smallest execution units was calculated. With thermal escape nonlinear exponent The similarity is represented by the feature difference coefficient; when the feature difference coefficient of an adjacent set of minimum execution units is less than a preset homogenization threshold, the control system logically merges the set of minimum execution units into a temporary logical partition in real time. The system issues unified heating power and clamping force commands to all execution units within the temporary logical partition to eliminate the control step effect at the partition boundary and ensure the consistency of curing in large-area defect areas.

[0016] On the other hand, the present invention provides a closed-loop control method for the online process of bonding and curing door panel hemp felt, comprising the following steps: S1. During the mold closing stage, an asynchronous scan is performed using a multi-sensor array to acquire environmental state data and thermal field data of the hemp felt bonding interface, and the data is mapped to several preset independent control zones of the mold. In the middle, it is defined as the initial hemp felt partition; S2. Based on each of the initial felt partitions, calculate the microscopic impedance entropy of the interface within each partition. and thermal escape nonlinear exponent ; S3, the interface micro-impedance entropy and the thermal escape nonlinear index The input mixture is used to create an adaptive neural network model, which generates a core feature tensor containing predictions of future solidification states through a dynamic weight adjustment mechanism. ; S4. Introduce a contact thermal resistance estimation model based on the core feature tensor. The current dominant defect type is determined, and the heating power of each zone is adjusted in a coordinated manner through a zoned thermo-pressure coupling actuator. With local clamping force Until the final cured quality index is generated. Meets the preset standards.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention first achieves early quantitative identification and zoning judgment of local interface contamination and self-heating exothermics by multi-dimensional field partitioning mapping and heterogeneous feature decoupling. This solves the identification difficulties of short production cycle, sudden heat / interface abnormalities and difficulty in real-time partitioning and positioning, and reduces the risk of hidden peeling and local over-curing from the source.

[0018] 2. Secondly, by using a contact thermal resistance estimation model and a decoupled heat-pressure collaborative compensation strategy, online adaptive adjustment of heating power and local clamping force for each zone is achieved. This solves the problem of compensation for the inability to directly measure the spatiotemporal changes in contact thermal resistance, which leads to local uncured areas, warping, or thermal stress concentration. Thus, curing uniformity and structural integrity are maintained without reducing production capacity.

[0019] 3. By constructing the final solidified quality index through the time integral of the quality prediction component in the core feature tensor output by the material mixing adaptive neural network, and combining dynamic partition reconstruction and multi-dimensional safety hypersurface judgment logic, part-level mold opening safety decision and partition linkage control are realized. This solves the stress concentration and misjudgment mold opening risks caused by partition boundary control step and hot spot drift, and ensures maximum yield under global consistency and traceability. Attached Figure Description

[0020] Figure 1 This is a framework diagram of a closed-loop control system for the online process of bonding and curing hemp felt for door panels according to the present invention.

[0021] Figure 2 This is a flowchart of an online closed-loop control method for the bonding and curing of door panel hemp felt, according to the present invention. Detailed Implementation

[0022] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0023] The term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three cases: A exists alone, A and B exist simultaneously, and B exists alone.

[0024] Example 1 Figure 1 This embodiment illustrates a framework diagram of a closed-loop control system for the online bonding and curing process of door panel hemp felt, as disclosed in this embodiment, including: The multi-dimensional field partitioning mapping module is configured to perform asynchronous scanning using a multi-sensor array during the mold closing stage to acquire environmental state data and thermal field data of the felt bonding interface, and then map the data to several preset independent control partitions of the mold. In the middle, it is defined as the initial hemp felt partition; The multi-dimensional field partitioning mapping module includes a thermal field acquisition unit, an environmental field acquisition unit, and a data preprocessing unit. The thermal field acquisition unit is configured to acquire the infrared temperature distribution of the solidified area in real time at a first frequency using an infrared thermal imager. Based on the geometric topology of the mold surface, the temperature data is divided into the corresponding control partitions as the initial felt partitions; The environmental field acquisition unit is configured to acquire the surface equivalent resistance of the felt surface using a surface impedance probe array. Simultaneously, the spatial particle density of the current workstation is obtained using a vision or laser particle counter. ; The data preprocessing unit is configured to process the infrared temperature distribution. Surface equivalent resistance and spatial particle density are spatiotemporally aligned and Kalman filtered for noise reduction to construct initial feature vectors. .

[0025] The heterogeneous feature decoupling calculation module is configured to calculate the interfacial micro-impedance entropy, which characterizes the degree of interfacial contamination and contact disorder, based on each of the initial felt partitions. And the thermal escape nonlinear index characterizing the deviation of the curing exothermic reaction of the adhesive. ; The interface micro impedance entropy By setting a standard reference value for the resistance of clean hemp felt and particle sensitivity coefficient ; Obtain real-time surface equivalent resistance And calculate its baseline deviation from the standard clean hemp felt resistance reference value; using spatial particle density and its sensitivity coefficient An exponential weighting term is constructed to characterize the nonlinear contribution of particles to the disorder of the impedance distribution; the basic deviation is coupled with the exponential weighting term, and the result is logarithmically smoothed to finally obtain the dimensionless interfacial microscopic impedance entropy. The calculation formula is expressed as: in, The larger the value, the more chaotic the physicochemical state of the interface, indicating the presence of micro-dust contamination, high moisture content, or micro-air gaps, resulting in a reduction in the effective bonding area.

[0026] Pollutants / particulate matter can simultaneously alter electrical resistance (conductivity / hygroscopicity) and particle number, through... This allows the coupling of the two to be placed under the same dimension. Sensor readings often span orders of magnitude; using ln can compress extreme values, mitigate the influence of outliers, and make small relative changes linearly additive, facilitating their use in subsequent models (e.g., linear approximation, RLS). When the particle number increases or the resistance deviates from the baseline... An increase indicates a more chaotic interface state, potentially higher contact thermal resistance, and a decrease in the effective bonding area.

[0027] Tiny contaminants or oil stains will first change the surface resistivity or particle count. These minute signals can be amplified and alerted at an early stage, allowing for localized cleaning or compaction before mold opening to avoid batch defects.

[0028] Because contact thermal resistance often increases with interface disorder (effective contact area decreases), (Or linear / power-law relationship) is a natural engineering approximation, which facilitates closed-form calculation and compensation.

[0029] It is more robust to sensor drift or single-point failures; for example, if vision is impaired by lighting conditions, the resistance / particle quantity can compensate for it, and vice versa.

[0030] The process of obtaining the thermal escape nonlinear index includes: firstly, by monitoring the real-time temperature of the partition. time derivative The actual temperature rise rate is used to obtain the temperature rise rate characterization value; secondly, the mold clamping force currently applied to this zone is obtained. Based on pressure transmission coefficient and target temperature difference Calculate the expected rate of temperature rise under the theoretical linear model; The thermal escape nonlinearity index is obtained by calculating the difference between the temperature rise rate and the expected temperature rise rate: When the thermal escape nonlinear exponent A value greater than 0 indicates that a self-accelerating exothermic reaction occurs inside the adhesive, i.e., the risk of heat escape.

[0031] This allows for the observation of the actual rate of temperature rise. It is based on the theoretically controllable heating contribution of the current clamping force and temperature difference. The residual obtained by subtracting the two can make the control layer more concerned with the heat release that exceeds the expectation, and provide direct evidence of the occurrence of nonlinear chemical reactions (curing) in a localized manner.

[0032] In this embodiment, sensitivity to the time derivative allows for the detection of exothermic acceleration earlier than absolute temperature; if High temperatures indicate a predominantly heat-related issue, necessitating priority for heat reduction and cooling; if... Low but High indicates that the interface is the primary issue, and compaction / cleaning should be prioritized.

[0033] This embodiment introduces the micro-impedance entropy of the partition interface. With thermal escape nonlinear exponent Two indicators. The former, derived from the logarithmic transformation of the product of surface equivalent resistance and particle count, sensitively reflects contact mismatch caused by interfacial microparticles, oil contamination, or micro-air gaps. The latter, expressed as the difference between the rate of temperature rise and the linear thermal response based on clamping force, enables early detection of localized chemical exothermics or heat dissipation. The two are used together to determine the priority compensation strategy (clamping priority or thermal suppression priority), thereby minimizing the risk of defects such as localized over-curing, peeling, and warpage while ensuring production capacity. The solidification field co-prediction tensor generation module is configured to generate the interface micro-impedance entropy. and the thermal escape nonlinear index The input mixture is used to create an adaptive neural network model, which generates a core feature tensor containing predictions of future solidification states through a dynamic weight adjustment mechanism. ; The mixing adaptive neural network model includes: Dual-channel input architecture, including independent processing of interface micro impedance entropy Characterized interface channels and thermal escape nonlinear exponent The input layer of the thermal channel is characterized; By setting an adaptive signal-to-noise ratio weight adjustment strategy, the micro-impedance entropy of the interface is monitored in real time. The amplitude, through the interface micro-impedance entropy The amplitude is adjusted in real time to control the interface channel weights; Output core feature tensor ,in To predict future curing uniformity, Recommended control increment; The signal-to-noise ratio adaptive weight adjustment strategy includes adjusting the interface micro impedance entropy when... When the weight is less than the preset threshold, the weights of the interface channel and the hot channel are kept balanced. When the interface micro-impedance entropy When the threshold is greater than or equal to the preset threshold (indicating extreme interface disorder / sensor signal distortion), the weight coefficient of the interface channel is automatically reduced, and the weight coefficient of the thermal channel is increased accordingly, so that the control strategy is tilted towards a conservative mode based on thermal history. Historical interface micro-impedance entropy was examined using the VAR method. and curing quality index The data was used to calculate the interfacial micro-impedance entropy within a set time window using the impulse response (IRF). The magnitude of the impact, after normalization, is denoted as the historical predictive power within partition i. ; Calculate instantaneous variance using EMA This yields the effective confidence level (a higher confidence level indicates greater credibility). ,in Historical influence amplification factor (example) Then, the effective confidence score is normalized and multiplied by the original weight of the current input interface channel to obtain the weight coefficient of the interface channel; like "Big" indicates historical It has strong predictive power for quality, even Even at a medium level, it will improve. ; like Small, indicating historical If the predictive power for quality is weak, then even Small, It will not be overestimated.

[0034] The contact thermal resistance adaptive compensation control module is configured to introduce a contact thermal resistance estimation model based on the core feature tensor. The current dominant defect type is determined, and the heating power of each zone is adjusted in a coordinated manner through a zoned thermo-pressure coupling actuator. With local clamping force Until the final cured quality index is generated. Meets the preset standards.

[0035] Example 2 Based on the inventive concept of Embodiment 1, this embodiment provides a specific construction process for the contact thermal resistance adaptive compensation control module, including a contact thermal resistance estimation model construction unit and a decoupling compensation strategy execution unit. The contact thermal resistance estimation model building unit is configured to use the interface micro-impedance entropy. Define contact thermal resistance and local clamping force (Define contact thermal resistance) With interface micro-impedance entropy Proportional to local compressive force (Non-linear inverse proportion), the relationship is expressed as: ; The decoupling compensation strategy execution unit is configured as follows: (1) Pressure-priority compensation mode (for interface contamination): When the microscopic impedance entropy of a certain partition interface is detected... The increase leads to an increase in contact thermal resistance and a nonlinear thermal escape exponent. When the system is in a stable range, it enters a pressure-priority compensation mode. The control system maintains a constant heating power (to prevent the adhesive from overheating) while calculating a pressure compensation value that can offset the increase in thermal resistance. This drives the pressure actuator in that zone to increase the local clamping force. Furthermore, the contact thermal resistance is reduced by physically crushing interfacial dust or squeezing out air gaps. The target of the pressure-priority compensation mode is set as: the increase in contact thermal resistance due to interface contamination, expressed as: By increasing local pressure Bundle Return to target At the same time, the heating power should not be changed or should be changed as little as possible. .

[0036] Current value Through the target Get the new pressure you need : ; Therefore, the pressure compensation amount is expressed as... Target contact thermal resistance is defined as ,in This indicates the target reduction percentage, with a lower bound constraint added to prevent overvoltage. Regarding the target reduction percentage: When interface contamination is high ( When the pressure is large, stronger compression compensation is required. (The value is large), but the physical upper limit of the pressure must also be considered; High heat dissipation / self-heating ( When (large), increase This is not the only strategy; often, reducing heating and combining it with a small increase in pressure (or even prohibiting large increases in pressure to avoid extruding the adhesive) is preferred. right The response should be more conservative or designed in conjunction with thermal suppression action.

[0037] It must be subject to upper and lower bounds and rate limits, and smoothing / hysteresis is usually used to avoid jitter.

[0038] Therefore, in this embodiment, the target reduction ratio is achieved by normalizing the interface micro-impedance entropy. With thermal escape nonlinear exponent weighted sum get: parameter Controlling the turning points and steepness; and using historical predictive power to amplify or reduce them. response ,in , Control the gain (e.g., 0.3–1). If the history indicates interfacial micro-impedance entropy... If the product is highly sensitive to quality, increase or decrease the percentage; if it is not sensitive to historical factors, decrease the percentage. .

[0039] And set each control cycle to be rate-limited. : (e.g., 0.1 MPa / s), execute in stages to prevent sudden changes from damaging the material or extruding excessive amounts of adhesive. If A combined heat-pressure strategy is adopted.

[0040] Meanwhile, this embodiment also considers It will also decrease as the pressure decreases (crushing particles and expelling air gaps), that is The pressure will gradually decrease, thus alleviating the problem—therefore, iterative / verification-based execution should be adopted, rather than a one-time brute-force application of pressure.

[0041] (2) Thermal-pressure co-suppression mode (for thermal escape): When the nonlinear exponent of thermal escape in a certain region is detected... Greater than the corresponding threshold When this occurs, cooperative inhibition is triggered; to avoid false triggering, a timing condition is usually added: continuous Second sampling When the average pressure exceeds the limit, or when a new local clamping force is obtained from the pressure priority compensation mode, the decoupling compensation strategy execution unit executes the heat-pressure coordinated suppression mode, and the control system simultaneously reduces the heating power. And fine-tune the local clamping force Heat absorption and suppression are achieved by utilizing the heat capacity of the mold.

[0042] First, set a conservative heat reduction range. ( (Assuming a preset reduction ratio), then solve using a closed-form expression to obtain: In the thermo-pressure co-suppression mode, the controller employs the following linearized energy approximation: And define the linearized sensitivity coefficient. ; To make the current over-limit quantity To descend to the target, the required control variable should satisfy a linear equation: ; and on Execute after applying a lower bound of 0, an upper bound, and a rate limit; The upper and lower bounds and rate constraints include, but are not limited to: (1) Changes in heating power satisfy and ; (2) Changes in clamping force satisfy and ; (3) When any constraint block achieves the desired control effect (i.e., it still cannot make...) When the value drops to a safe level, alternative suppression measures (including local cooling, vacuum assistance, part rejection, or production line shutdown) are triggered and an anomaly log is recorded.

[0043] The decoupling compensation strategy execution unit adopts an iterative-verification execution process: in each execution... or After the increment, short intervals Remeasurement And recalculate With sensitivity coefficient If the expected convergence is not achieved, the next increment is calculated and executed again until the exit condition is met (e.g., ...). , Or reach the constraint limit).

[0044] To ensure model applicability and control accuracy, the system also includes a parameter identification and calibration unit. This unit is used to identify and periodically update parameters through offline design experiments (different pressures, different heating power pulses, and artificial injection of interface contamination) or online recursive least squares (RLS) methods. The identification results are then written back to the contact thermal resistance estimation model construction unit and the decoupling compensation strategy execution unit for subsequent control use.

[0045] For core feature tensors The quality prediction component in the curve is integrated over time to generate the final curing quality index. ; Constructing a multidimensional secure hypersurface only if all partitions When all surfaces converge to the multidimensional safe hypersurface, the system issues an opening command.

[0046] The solidification field collaborative prediction tensor generation module also includes a dynamic partitioning reconstruction mechanism based on feature clustering, including... If a contaminated area happens to cross the boundary of two pre-defined zones, the fixed zones may cause the two zones to adopt different countermeasures, resulting in stress concentration at the boundary.

[0047] If several adjacent zones have completely identical states, calculating and controlling them separately would be a waste of resources. It would be better to merge them into a single "large zone" with unified instructions to ensure consistency.

[0048] Therefore, the initial felt partition is defined as the smallest execution unit; In generating core feature tensors Previously, the microscopic impedance entropy of the interface between adjacent smallest execution units was calculated. With thermal escape nonlinear exponent The similarity is represented by the feature difference coefficient; when the feature difference coefficient of an adjacent set of minimum execution units is less than a preset homogenization threshold, the control system logically merges the set of minimum execution units into a temporary logical partition in real time. The system issues unified heating power and clamping force commands to all execution units within the temporary logical partition to eliminate the control step effect at the partition boundary and ensure the consistency of curing in large-area defect areas.

[0049] For example, during the solidification process of hemp felt, hotspot drift (adjustment) may occur, affecting the central (hotspot) region of the exothermic reaction. Slowly drifted to and At the boundary. At this point, the system detected an abnormal boundary temperature gradient and temporarily... and The locking mechanism works together to perform cooling and pressure relief operations, preventing overheating at the boundary.

[0050] The threshold and weight settings involved in the above embodiments can be set by default according to the present invention, or can be set by those skilled in the art.

[0051] Example 3 Figure 2 This invention presents a flowchart of an online closed-loop control method for the bonding and curing of door panel hemp felt, based on the same inventive concept as Embodiment 1. The invention provides an online closed-loop control method for the bonding and curing of door panel hemp felt, comprising: S1. During the mold closing stage, an asynchronous scan is performed using a multi-sensor array to acquire environmental state data and thermal field data of the hemp felt bonding interface, and the data is mapped to several preset independent control zones of the mold. In the middle, it is defined as the initial hemp felt partition; S2. Based on each of the initial felt partitions, calculate the microscopic impedance entropy of the interface within each partition. and thermal escape nonlinear exponent ; S3, the interface micro-impedance entropy and the thermal escape nonlinear index The input mixture is used to create an adaptive neural network model, which generates a core feature tensor containing predictions of future solidification states through a dynamic weight adjustment mechanism. ; S4. Introduce a contact thermal resistance estimation model based on the core feature tensor. The current dominant defect type is determined, and the heating power of each zone is adjusted in a coordinated manner through a zoned thermo-pressure coupling actuator. With local clamping force Until the final cured quality index is generated. Meets the preset standards.

[0052] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0053] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0054] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0055] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0056] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A closed-loop control system for online bonding and curing of door panel felt, characterized in that: The system includes: The multi-dimensional field partition mapping module is configured to perform asynchronous scanning using a multi-sensor array during the mold closing stage to acquire environmental state data and thermal field data of the hemp felt bonding interface, and map the data to several preset independent control partitions of the mold, which are defined as initial hemp felt partitions. The heterogeneous feature decoupling calculation module is configured to calculate the interface micro impedance entropy and thermal escape nonlinearity index within each of the initial felt partitions. The solidification field collaborative prediction tensor generation module is configured to input the interface micro impedance entropy and the thermal escape nonlinear exponent into the mixing adaptive neural network model, and generate a core feature tensor containing the prediction value of the future solidification state through a dynamic weight adjustment mechanism. The contact thermal resistance adaptive compensation control module is configured to introduce a contact thermal resistance estimation model, determine the current dominant defect type based on the core feature tensor, and coordinate the heating power and local clamping force of each zone through a zoned thermal-pressure coupling actuator until the final curing quality index meets the preset standard.

2. The closed-loop control system for online bonding and curing of door panel felt according to claim 1, characterized in that: The multi-dimensional field partitioning mapping module includes a thermal field acquisition unit, an environmental field acquisition unit, and a data preprocessing unit. The thermal field acquisition unit is configured to acquire the infrared temperature distribution of the solidified area in real time at a first frequency using an infrared thermal imager. Based on the geometric topology of the mold surface, the temperature data is divided into the corresponding control partitions as the initial felt partitions; The environmental field acquisition unit is configured to acquire the surface equivalent resistance of the felt surface using a surface impedance probe array. Simultaneously, the spatial particle density of the current workstation is obtained using a vision or laser particle counter. ; The data preprocessing unit is configured to process the infrared temperature distribution. Surface equivalent resistance and spatial particle density are spatiotemporally aligned and Kalman filtered for noise reduction to construct initial feature vectors. .

3. The closed-loop control system for online bonding and curing of door panel felt according to claim 1, characterized in that: The mixing adaptive neural network model includes: Dual-channel input architecture, including independent processing of interface micro impedance entropy Characterized interface channels and thermal escape nonlinear exponent The input layer of the thermal channel is characterized; By setting an adaptive signal-to-noise ratio weight adjustment strategy, the micro-impedance entropy of the interface is monitored in real time. The amplitude, through the interface micro-impedance entropy The amplitude is adjusted in real time to control the interface channel weights; Output core feature tensor ,in To predict future curing uniformity, This is the recommended control increment.

4. The closed-loop control system for online bonding and curing of door panel felt according to claim 3, characterized in that: The signal-to-noise ratio adaptive weight adjustment strategy includes adjusting the interface micro impedance entropy when... When the weight is less than the preset threshold, the weights of the interface channel and the hot channel are kept balanced. When the interface micro-impedance entropy When the threshold is greater than or equal to the preset threshold, the dynamic weight adjustment mechanism formed by combining the effective confidence level and the model's historical predictive power with the original weight of the current input interface channel automatically reduces the weight coefficient of the interface channel and correspondingly increases the weight coefficient of the hot channel, so that the control strategy tilts towards a conservative mode based on thermal history.

5. The closed-loop control system for online bonding and curing of door panel felt according to claim 1, characterized in that: The contact thermal resistance adaptive compensation control module includes a contact thermal resistance estimation model construction unit and a decoupling compensation strategy execution unit. The contact thermal resistance estimation model building unit is configured to use the interface micro-impedance entropy. Define contact thermal resistance and local clamping force The relation is expressed as ; The decoupling compensation strategy execution unit is configured as follows: When the micro-impedance entropy of a certain partition interface is monitored The increase leads to an increase in contact thermal resistance and a nonlinear thermal escape exponent. When the system is in a stable range, it enters a pressure-priority compensation mode, where the control system maintains a constant heating power while simultaneously controlling the current interface micro-impedance entropy. With thermal escape nonlinear exponent weighted sum The target reduction ratio is calculated, and then the target contact thermal resistance is mapped to the pressure compensation value. This drives the pressure actuator in that zone to increase the local clamping force. ; When the nonlinear exponent of thermal escape in a certain region is detected Greater than the corresponding threshold and continuously greater than When a new local clamping force is obtained from the pressure priority compensation mode within a sampling period, the decoupling compensation strategy execution unit executes the heat-pressure coordinated suppression mode, and the control system simultaneously reduces the heating power. And through the thermal escape nonlinear exponent Adjusting the local clamping force by varying the amount of change Heat absorption and suppression are achieved by utilizing the heat capacity of the mold.

6. The closed-loop control system for online bonding and curing of door panel felt according to claim 1, characterized in that: The interface micro impedance entropy By setting a standard reference value for the resistance of clean hemp felt and particle sensitivity coefficient ; Obtain real-time surface equivalent resistance And calculate its baseline deviation from the standard clean hemp felt resistance reference value; using spatial particle density and its sensitivity coefficient An exponential weighting term is constructed to characterize the nonlinear contribution of particles to the disorder of the impedance distribution; The fundamental deviation is coupled with the exponential weighting term, and the result is logarithmically smoothed to obtain the dimensionless interface microscopic impedance entropy. .

7. The closed-loop control system for online bonding and curing of door panel felt according to claim 1, characterized in that: The process of obtaining the thermal escape nonlinear index includes: firstly, by monitoring the real-time temperature of the partition. time derivative The heating rate characterization value is obtained; secondly, the mold clamping force currently applied to this partition is obtained. Based on pressure transmission coefficient and target temperature difference Calculate the expected rate of temperature rise under the theoretical linear model; The thermal escape nonlinearity index is obtained by calculating the difference between the temperature rise rate and the expected temperature rise rate: When the thermal escape nonlinear exponent A value greater than 0 indicates that a self-accelerating exothermic reaction occurs inside the adhesive, i.e., the risk of heat escape.

8. The closed-loop control system for online bonding and curing of door panel felt according to claim 1, characterized in that: The system also includes core feature tensors. The quality prediction component in the curve is integrated over time to generate the final curing quality index. ; And construct a multidimensional secure hypersurface, only if all partitions When all surfaces converge to the multidimensional safe hypersurface, the system issues an opening command.

9. The closed-loop control system for online bonding and curing of door panel felt according to claim 3, characterized in that: The solidified field collaborative prediction tensor generation module also includes a dynamic partitioning reconstruction mechanism for feature clustering, which includes defining the initial felt partition as the smallest execution unit; and generating the core feature tensor. Previously, the microscopic impedance entropy of the interface between adjacent smallest execution units was calculated. With thermal escape nonlinear exponent The similarity is represented by the feature difference coefficient; when the feature difference coefficient of an adjacent set of minimum execution units is less than a preset homogenization threshold, the control system logically merges the set of minimum execution units into a temporary logical partition in real time. The system issues unified heating power and clamping force commands to all execution units within the temporary logical partition to eliminate the control step effect at the partition boundary and ensure the consistency of curing in large-area defect areas.

10. A closed-loop control method for online bonding and curing of door panel hemp felt, used to execute the closed-loop control method for online bonding and curing of door panel hemp felt as described in any one of claims 1-9, characterized in that: Includes the following steps: S1. During the mold closing stage, an asynchronous scan is performed using a multi-sensor array to acquire environmental state data and thermal field data of the hemp felt bonding interface, and the data is mapped to several preset independent control zones of the mold. In this context, the initial felt partition is defined. S2. Based on each of the initial felt partitions, calculate the microscopic impedance entropy of the interface within each partition. and thermal escape nonlinear exponent ; S3, the interface micro-impedance entropy and the thermal escape nonlinear index The input mixture is used to create an adaptive neural network model, which generates a core feature tensor containing predictions of future solidification states through a dynamic weight adjustment mechanism. ; S4. Introduce a contact thermal resistance estimation model based on the core feature tensor. The current dominant defect type is determined, and the heating power of each zone is adjusted in a coordinated manner through a zoned thermo-pressure coupling actuator. With local clamping force Until the final cured quality index is generated. Meets the preset standards.

Images