A composite film initial adhesion force intelligent detection method, system, device and medium

Abstract

The application discloses a kind of composite film initial adhesion force intelligent detection method, system, equipment and medium, it is related to flexible electronic packaging test technical field, the steps of this method include: first, through sensor acquisition micropore contact and probe peeling process in tension and compression, negative pressure and local microclimate data;Second, construct deformation decoupling strategy algorithm history background resistance bias, and establish transient initial adhesion model to infer the adhesion force prediction value of current interface.Subsequently, using multi-agent architecture to deduce peeling damage, identify the peeling mode of adhesive layer.Finally, based on the prediction value, the equipment control instruction sheet covering pressure maintenance, pump valve ratio and pose reset is generated by optimization algorithm, and the response delay is tracked.The application improves the precision and automation control level of composite film test.

Description

Technical Field

[0001] This invention relates to the field of flexible electronic packaging testing technology, specifically to a method, system, equipment, and medium for intelligent detection of initial tack of composite films. Background Technology

[0002] With the widespread application of polymer composite membrane materials in numerous fields, the initial tack strength (ITS) is a core indicator for evaluating the quality of interlayer bonding, and its accuracy is crucial. Conventional ITS testing often employs rigid clamps to hold the film and performs tensile and compressive tests with a fixed stroke. However, due to the extremely thin, easily rolled, and highly flexible physical characteristics of composite membranes, the transient mechanical process of probe contact and separation from the film is easily interfered with. This manifests not only as hysteresis and distortion in mechanical signal capture, but more importantly, the high coupling between the flexible film's own bending deformation resistance and the pure ITS at the interface leads to extremely poor repeatability of test data, making it difficult to accurately reflect the interfacial adhesion performance of the material.

[0003] To overcome these shortcomings, existing technologies typically employ solutions such as improving hardware sensing accuracy or introducing basic digital signal processing. Specifically, conventional solutions often attempt to capture millisecond-level instantaneous peak forces by replacing the sensor with a high-frequency dynamic force sensor, or by embedding low-pass filtering algorithms in the test control software to eliminate mechanical vibration noise. Meanwhile, some testing devices add a rigid backplate to the bottom of the film or use double-sided adhesive to force the film surface flattening, attempting to improve the consistency of the contact area when the probe is pressed down.

[0004] While existing solutions have mitigated basic noise and improved flatness to some extent, the following technical problems remain: First, traditional low-pass filtering algorithms typically employ a one-size-fits-all noise reduction logic, easily filtering out genuine high-frequency transient initial adhesion characteristic peaks as noise, resulting in incomplete extraction of key mechanical features; second, the forced fixing method relying on rigid backplates or physical adhesive bonding is not only cumbersome to operate but also introduces uncontrollable prestress into the flexible membrane, making it impossible to accurately peel off the resistance generated by the film's own deformation and the initial adhesion force at the interface at the moment of contact; finally, the fixed downward pressure stroke set by the system lacks adaptive micro-force closed-loop feedback, and the actual effective contact area in each test still exhibits uncontrollable fluctuations in the face of micro-undulations and local warping on the composite membrane surface.

[0005] Therefore, this invention proposes an intelligent detection method, system, device, and medium for the initial tack of composite films to address these problems. Summary of the Invention

[0006] The purpose of this invention is to provide a method, system, device and medium for intelligent detection of initial tack of composite films, 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 smart detection method for the initial tack of a composite film, comprising the following steps: S1. Collect time-series tension and compression data, micropore adsorption negative pressure data, and local microclimate interference data at each micropore contact and probe peeling node during the test rotation of the composite membrane using a flexible material mechanical sensing module. S2. Construct a thin film interface deformation decoupling strategy and calculate the background resistance bias in each historical test cycle. S3. Establish a transient initial adhesion inference model, using the time-series tensile and compressive data, microporous adsorption negative pressure data, and local microclimate disturbance data as input entities to infer the predicted value of interfacial adhesion force during the current test cycle. S4. Input the output items extracted in steps S3 and S2 into the multi-agent matching framework to perform peeling damage inference and output the composite film adhesive layer peeling mode identification results. S5. Configure the test parameter optimization algorithm, and generate a device control instruction sheet based on the predicted value of interface adhesion force during the current test cycle. The instruction sheet covers tasks such as pre-tightening and pressure holding, negative pressure pump valve ratio, and probe universal ball joint posture reset, and tracks the response delay of the instruction sheet nodes.

[0008] A further improvement of this invention is that the time-series tension and compression data includes the stress gradient under the universal ball joint probe and the transient peak tension during probe pullback; the micropore adsorption negative pressure data includes the air pressure at the end of the high-density array base and the air flow rate to maintain surface flatness; and the local microclimate interference data includes the relative humidity of the test micro-area, the near-field temperature of the probe, and the electrostatic field strength interference.

[0009] A further improvement of this invention is that the construction process of the transient initial adhesion inference model includes: based on a one-dimensional deep residual network architecture, combining the number of historical test cycles, time-series tensile and compressive data, microporous adsorption negative pressure data, and local microclimate disturbance data into feature vectors, and using the set of all feature vectors as the input of the inference model; the transient initial adhesion inference model uses the interface adhesion force prediction value calculated by each set of feature vectors as the output, and uses the real interface initial adhesion force value collected in the dual-channel differential comparison test as the inference guide; the training guide is to minimize the sum of reconstruction errors of all inferred base interface adhesion force prediction values, and the training iteration stops when the sum of reconstruction errors drops to the convergence interval, and the reconstruction error is obtained by the square operation of the difference between the calculated value of the inference model and the real value.

[0010] A further improvement of this invention is that the transient initial adhesion inference model further includes: setting the average value of the background resistance bias during the historical test cycle of the composite membrane as the deformation tolerance threshold; comparing the deformation tolerance threshold with the predicted value of the interfacial adhesion during the current test cycle; when the predicted value of the interfacial adhesion during the current test cycle is greater than the deformation tolerance threshold, releasing a microporous negative pressure array side leakage warning waveform; when the predicted value of the interfacial adhesion during the current test cycle is less than or equal to the deformation tolerance threshold, calculating the expected value of deformation recovery during the current test cycle, wherein the expected value of recovery is obtained by calculating the frequency of the background resistance bias being greater than the deformation tolerance threshold during the historical test cycle of the composite membrane and the proportion of the total frequency during the historical test cycle.

[0011] A further improvement of this invention is that the transient initial adhesion inference model, when executing step S4, includes: taking the predicted interfacial adhesion force value of the current test cycle and the damage morphology sequence of the composite film adhesive layer during the historical test cycle as the bidder and the auction item, respectively, and solving the composite film adhesive layer peeling mode identification result based on the task bidding game algorithm; firstly, the historical adhesive layer damage morphology sequence is used as the auction target, and the damage morphology to be inferred is introduced into the bidding one by one from the first round; in each round of bidding, the current interfacial adhesion force prediction value puts up a bid for the historical damage morphology sequence, and the bid result is presented as the difference between the correlation gain of the current adhesion force value matching the historical damage morphology sequence and the sum of the bids that the current adhesion force value has won; the item with the highest bid result among the current adhesion force values ​​is selected to obtain the matching right of the item participating in the auction in the sequence; when there are still unsold items of the current interfacial adhesion force prediction value, the auction order is extended to start the next round of bidding, until all sequences are won and the auction ends.

[0012] A further improvement of this invention is that the thin film interface deformation decoupling strategy is obtained by normalizing and scaling the time-series tensile and compressive data, micropore adsorption negative pressure data, and local microclimate interference data at each node during the historical test cycle, assigning them anti-interference weight parameters, and adding them together.

[0013] A further improvement of this invention is that the specific steps of the test parameter optimization algorithm include: S51. Arrange all composite film test batches in order of their predicted interfacial adhesion values ​​from highest to lowest according to the current test cycle period. Set primary and secondary fluctuation thresholds. Classify test batches with interfacial adhesion values ​​greater than the predicted values ​​as high-frequency interference batches, classify test batches in the range of primary and secondary fluctuation thresholds as medium-frequency interference batches, and classify test batches with interfacial adhesion values ​​less than the predicted values ​​as low-frequency interference batches. S52. Create different probe poses and negative pressure pump valve ratio task libraries for different interference levels. S53. Configure the initial annealing parameters, including the initial thermodynamic constant, cooling rate of decrease, and termination constant; S54. Randomly generate an initial action arrangement scheme, including the servo action assignment, intervention order and dwell time slot of all devices; S55. Construct the objective function formula, which aims to minimize the system commissioning energy consumption and film peeling deviation. The calculation formula is expressed as follows: Among them, the first term in the summation operation represents the non-negative modulus of the difference between the actual and expected completion times of the servo action; the second term represents the ratio of negative pressure pumping power to the rated peak power of the pump body; the third term represents the summation of deformation resistance penalty values; and the fourth term represents the summation of electrostatic elimination energy consumption of the ion fan. S56. If the objective function is less than the set tolerance threshold, then a neighborhood search is performed on the current arrangement scheme to generate a new candidate scheme, and the objective function of the new scheme is calculated again. If the objective function of the new scheme is not less than the objective function of the old scheme, then the new scheme is accepted. S57. Iterate the current thermodynamic constant based on the cooling rate of decrease. The current thermodynamic constant is the product of the cooling rate of decrease and the previous constant. S58. Repeat steps S56 and S57 until the thermodynamic constant is less than the termination constant, and output the currently accepted optimal action arrangement scheme.

[0014] On the other hand, the present invention provides an intelligent detection system for the initial tack of a composite film, comprising: The temporal alignment module is configured to collect temporal tension and compression data, micropore adsorption negative pressure data, and local microclimate interference data at various micropore contact and probe peeling nodes during the test cycle of the composite membrane through the flexible material mechanical sensing module. The interface deformation calculation module is configured to construct a thin film interface deformation decoupling strategy and calculate the background resistance bias in each historical test cycle. The transient initial adhesion inference model building module is configured to take the time-series tensile and compressive data, microporous adsorption negative pressure data and local microclimate disturbance data as input entities to infer the predicted value of interfacial adhesion force during the current test cycle. The peeling mode bidding game analysis module is configured to input the output terms extracted from the interface deformation calculation module and the interface deformation calculation module into the multi-agent matching framework to perform peeling damage inference and output the composite film adhesive layer peeling mode identification results. The test parameter optimization algorithm configuration module is configured to generate equipment control instruction sheets based on the predicted value of interface adhesion force during the current test cycle. These instruction sheets cover tasks such as pre-tightening and pressure holding, negative pressure pump valve ratio, and probe universal ball joint pose reset, and track the response delay of the instruction sheet nodes.

[0015] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described intelligent method for detecting the initial tack of a composite film.

[0016] A computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described intelligent method for detecting the initial tack of a composite film.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention first integrates multi-dimensional sensing features such as microporous adsorption negative pressure data, local microclimate interference data, and time-series tensile and compressive data of composite membranes, and uses dual-channel differential logic calculation to extract the background resistance bias that characterizes the base of flexible resistance. This solves the problem of stress decoupling difficulties and contact area fluctuations caused by the high coupling of the composite membrane's own elastic deformation force, bending stress, and contact surface adhesion force in transient pull-out tests. It achieves accurate quantification and elimination of nonlinear mechanical impedance interference, and significantly improves the purity of the mechanical test data of the flexible film substrate and the repeatability consistency of cross-batch testing.

[0018] 2. By constructing a transient initial adhesion inference model based on a one-dimensional deep residual network architecture, a nonlinear implicit mapping is performed using multi-dimensional feature vectors that combine environmental and mechanical aspects as input. The model is then iteratively reduced in dimension and reconstructed using the actual initial adhesion force calibration value as a guide. This solves the problem that traditional low-pass filtering algorithms easily filter out the real high-frequency transient initial adhesion feature peaks as mechanical noise when processing the contact separation process, leading to severe measurement distortion. This model accurately infers the predicted value of the interface adhesion force after removing deformation artifacts from the mixed tensile signal while retaining the characteristics of small stress change nodes, ensuring high fidelity in the evaluation of the microscopic interface mechanics of composite films. Attached Figure Description

[0019] Figure 1 This is a flowchart of an intelligent detection method for the initial tack of a composite film according to the present invention; Figure 2 This is a framework diagram of an intelligent detection system for the initial tack of a composite film according to the present invention. Detailed Implementation

[0020] 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.

[0021] 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.

[0022] Example 1 Figure 1 The flowchart of the intelligent detection method for initial tack of composite films disclosed in this embodiment is shown, and the steps are as follows: This embodiment applies to a high-frequency automated quality inspection scenario for interfacial adhesion during the processing and manufacturing of composite materials such as polymer flexible packaging materials and optical transparent films. The underlying equipment includes an intelligent peel strength testing machine equipped with a high-density microporous array adsorption base, a universal ball joint probe, and an Ethernet real-time communication bus. In conventional testing procedures, testing mechanisms often use rigid clamps to directly hold the film and control the probe to press down with a fixed stroke. This conventional method ignores the inherent flexibility and curlability of the composite film, resulting in significant fluctuations in the actual effective contact area between different test batches when the probe contacts the film with microscopic undulations on its surface.

[0023] To address the challenges of flexible deformation coupling interference and inconsistent contact area, this embodiment introduces a micropore array forced flattening mechanism and a force-position dual-loop acquisition strategy in step S1. The flexible material mechanical sensing module collects temporal tension and compression data, micropore adsorption negative pressure data, and local microclimate interference data at various micropore contact and probe peeling nodes during the composite membrane testing cycle. The temporal tension and compression data includes the stress gradient under the universal ball joint probe and the transient peak tension during probe pullback. The micropore adsorption negative pressure data includes the air pressure at the high-density array base end and the airflow maintaining surface flatness. The local microclimate interference data includes the relative humidity of the test micro-area, the probe near-field temperature, and the electrostatic field strength interference. The relative humidity of the micro-area is expressed as a percentage, the probe near-field temperature as degrees Celsius, and the electrostatic field strength interference as kilovolts per meter. After obtaining these parameters, the complex three-dimensional warped flexible membrane can be reduced to a two-dimensional rigid plane, providing a reliable hard contact reference surface for subsequent mechanical calculations.

[0024] However, forced flattening only eliminates the fluctuations of the initial contact surface. After step S1 is completed, when the probe performs a pull-back peeling action, the elastic strain and bending resistance of the composite film itself will still be coupled with the adhesion force of the contact surface, forming a macroscopic mixed impedance. If the peak tensile force at this time is directly used as the test result, the data will deviate from the true interface adhesion force. Therefore, in this embodiment, a thin film interface deformation decoupling strategy is constructed in step S2 to calculate the background resistance bias in each historical test cycle. This calculation process relies on dual-channel differential comparison logic, and a blank reference group without adhesive is set up in the same test process. The thin film interface deformation decoupling strategy obtains the results by normalizing and scaling the temporal tensile and compressive data, micropore adsorption negative pressure data, and local microclimate interference data at each node in the historical test cycle, assigning anti-interference weight parameters, and adding them together. Its algebraic expression is: ,in, Characterizes the background resistance bias. The total number of sampling nodes representing the historical test rotation period, Characterizing the average elastic deformation tensile force collected on a glue-free reference surface, Representation Nodes The negative pressure value of microporous adsorption at the location, The environmental distortion factor, which characterizes temperature, humidity, and electrostatic field strength, is set by those skilled in the art. The elastic compensation weight is characterized by a preferred value of 0.75, which can retain the true bending resistance damping of the membrane while eliminating the mechanical resonance of the filtration system. The negative pressure coupling coefficient is characterized, with a preferred value of 0.12, which is used to compensate for the surge in internal stress of the thin film caused by excessive negative pressure. The environmental interference weight is characterized, with an optimal value of 0.08. To prevent signal drops or extreme values ​​during certain rotation periods from causing the denominator to become zero or data overflow, a small positive number is uniformly added to the denominator term during normalized division operations. Preferred .

[0025] After obtaining the baseline resistance bias, the problem of static baseline stripping can be solved. However, the formation and destruction of initial adhesion is a transient mechanical process at the millisecond level, and simple linear subtraction is insufficient to decouple the nonlinear dynamic high-frequency deformation coupling at the moment of contact. To solve this deep stress separation barrier, this embodiment establishes a transient initial adhesion inference model in step S3, using time-series tensile and compressive data, microporous adsorption negative pressure data, and local microclimate disturbance data as input entities to infer the predicted value of interfacial adhesion force during the current test cycle. The creation process of this inference model includes a one-dimensional deep residual network architecture, combining the number of historical test cycles and multi-source acquisition parameters into a reduced-dimensional tensor, and using the set of all reduced-dimensional tensors as the input of the inference model. The inference model outputs the predicted value of interfacial adhesion force calculated by each set of tensors, and uses the real interfacial initial adhesion force value collected in the dual-channel differential comparison test as the inference guide. Its training guide aims to minimize the sum of reconstruction errors of all inferred interfacial adhesion force prediction values ​​and stop iteration when it reaches the convergence interval. The reconstruction error calculation formula is: in, Characterizes the total reconstruction error. Characterizing batch sample size Characterizing the first The predicted value of interfacial adhesion force from the second calculation. Characterizes the corresponding actual interface initial tack value. The weight matrix representing the one-dimensional convolution kernel. The regularization penalty constant is preferably $0.001$, which effectively suppresses overfitting of the network to high-frequency transient noise and enhances the model's generalization resolution for different batches of membrane materials. Simultaneously, the inference model sets the mean of the background resistance bias during the historical test cycle of the composite membrane as the deformation tolerance threshold. Then, the threshold is compared with the current predicted value.

[0026] when When this occurs, it indicates that the local area may not be properly adsorbed and adhered, and the system immediately releases a microporous negative pressure array side leakage warning waveform; when At that time, the expected value of deformation recovery during the current test cycle is calculated. The formula for calculating this expected value is: ,in, For the expected value of deformation recovery, For history The frequency of occurrence, Total frequency Replace the smallest positive number with the invalid point, and take constant values. To prevent division by zero errors during cold starts; The material fatigue relaxation coefficient is preferably 0.95, which closely matches the viscoelastic recovery decay process of conventional polymer materials.

[0027] After obtaining the highly accurate pure initial adhesion force through step S3, the test system faces a new information gap: a single mechanical scalar cannot reverse-map the microscopic tearing state of the test interface. Conventional methods rely on manual observation of the cross-section after pull-out, which is easily influenced by subjective experience and cannot achieve a closed-loop mechanism. Therefore, in this embodiment, the pure interface adhesion force value is input into a multi-agent matching framework in step S4 for peel damage estimation, outputting the composite film adhesive layer peeling mode identification result. The pure interface adhesion force value of the current test cycle is then used... Damage morphology sequence of composite film adhesive layer during historical test rotation period Each participant acts as both a bidder and an auction item, and the problem is solved using a task-based bidding game algorithm. In each round of bidding, the current adhesion force value is used to submit a bid against the historical damage sequence, with the bid gain formula being: in, Characterizes the degree of bidding correlation gain. The adhesion force sensitivity scaling factor, preferably 10, is used to amplify the weighting of small force value differences in the bidding process. Characterizing historical sequence Typical destructive power values ​​recorded in the data. Characterizing error-proofing micro-values, constant values ​​are taken. , This represents the cumulative suppression of currently winning bids, used to prevent a single type from excessively absorbing bidding share. The top bid is selected to receive matching rights, and the order is extended until the auction ends, outputting coded identification results such as cohesive fracture or interfacial debonding.

[0028] After identifying the peeling mode and force value in step S4, the next challenge is how to adaptively adjust the electromechanical components in subsequent cycle periods based on the detection results of the current batch to smooth out the oscillations caused by continuous testing. Conventional fixed-formula testing machines struggle to handle this dynamic drift. Therefore, in step S5 of this embodiment, a test parameter optimization algorithm is configured to generate equipment control instruction sheets based on the inferred force value. All test batches are arranged in descending order of force value, and primary and secondary fluctuation thresholds are set to classify the interference batches. Initial annealing parameters and initial thermodynamic constants are configured. Set to 1000, cooling rate decrease Set to 0.98, termination constant. Set to 0.1. This annealing combination ensures that the algorithm has global escape capability in the initial stage and converges quickly to the steady-state command at the end. Construct the objective function to minimize system integration energy consumption and thin film peeling deviation: in, Characterizes the overall operating cost of a single rotation period. Characterizes the actual completion time of the servo action. Characterize the expected settlement time. Characterizing negative pressure pumping power, Characterizes the rated peak power of the pump body. The numerical value characterizing the deformation resistance penalty caused by non-coplanar contact of the probe. Characterizing local static electricity removal energy consumption, weighted parameter combination Preferred value is assigned This distribution relies on the priority settings of the measurement and control system, focusing on suppressing timing delays and mechanical force interference. If the algorithm finds that the objective function is less than the tolerance, it uses a Markov chain to implement neighborhood state transitions; if the new scheme is inferior to the old scheme, it is accepted based on the decay criterion; based on... Iterative cooling is performed until the output includes the optimal action arrangement scheme covering pre-tightening and negative pressure ratio, and the response delay is tracked to achieve closed-loop correction of electromechanical equipment.

[0029] The threshold and weight settings involved in this embodiment can be set by default according to the present invention, or can be set by those skilled in the art.

[0030] Example 2 Figure 2 This invention presents a framework diagram of an intelligent detection system for the initial tack of a composite film. Based on the same inventive concept as Embodiment 1, this invention provides an intelligent detection system for the initial tack of a composite film, comprising: The temporal alignment module is configured to collect temporal tension and compression data, micropore adsorption negative pressure data, and local microclimate interference data at various micropore contact and probe peeling nodes during the test cycle of the composite membrane through the flexible material mechanical sensing module. The interface deformation calculation module is configured to construct a thin film interface deformation decoupling strategy and calculate the background resistance bias in each historical test cycle. The transient initial adhesion inference model building module is configured to take the time-series tensile and compressive data, microporous adsorption negative pressure data and local microclimate disturbance data as input entities to infer the predicted value of interfacial adhesion force during the current test cycle. The peeling mode bidding game analysis module is configured to input the output terms extracted from the interface deformation calculation module and the interface deformation calculation module into the multi-agent matching framework to perform peeling damage inference and output the composite film adhesive layer peeling mode identification results. The test parameter optimization algorithm configuration module is configured to generate equipment control instruction sheets based on the predicted value of interface adhesion force during the current test cycle. These instruction sheets cover tasks such as pre-tightening and pressure holding, negative pressure pump valve ratio, and probe universal ball joint pose reset, and track the response delay of the instruction sheet nodes.

[0031] The time-domain alignment module includes a multi-source signal acquisition unit and a timestamp synchronization unit.

[0032] The multi-source signal acquisition unit communicates directly with the underlying hardware, including the flexible material mechanical sensing module, microporous negative pressure gauge, temperature and humidity sensors, and electrostatic sensors, to acquire raw tension / compression, air pressure, and microclimate waveform data. The timestamp synchronization unit eliminates sampling frequency differences between different sensors, using a unified clock source to precisely align the aforementioned multidimensional heterogeneous data on the time axis, and outputs a structured time-series data stream.

[0033] The interface deformation calculation module includes a historical feature extraction unit and a background bias calculation unit.

[0034] The historical feature extraction unit is used to retrieve and clean the control group data from the historical test cycle of photovoltaic / composite film; the background bias calculation unit is used to execute the interface deformation decoupling strategy, remove the elasticity and bending interference of the film itself, and calculate the background resistance bias.

[0035] The transient initial viscosity inference model establishment module includes a multi-dimensional input mapping unit and an initial viscosity force prediction output unit.

[0036] The multidimensional input mapping unit receives aligned time-series tensile and compressive data, microclimate disturbances, and other variables, and converts them into a feature vector matrix that the inference model can recognize. The initial adhesion prediction output unit runs the core deep network (such as a one-dimensional residual network), calculates and outputs the predicted value of the interface adhesion force after removing disturbances during the current rotation period based on the input feature matrix.

[0037] The stripping mode bidding game analysis module includes a bidding parameter configuration unit and a damage mode deduction unit.

[0038] The bidding parameter configuration unit receives the inferred adhesion force prediction value and the historical baseline resistance bias, and converts these values ​​into bidders' quotations and auction participants in a multi-agent game architecture. The damage mode inference unit executes specific bidding matching logic, calculates matching gain, and finally maps and outputs the specific physical peeling state of the composite film adhesive layer, such as the identification results of cohesive failure and interface debonding.

[0039] The test parameter optimization algorithm configuration module includes a device status optimization unit, an action command generation unit, and an execution delay tracking unit.

[0040] The equipment state optimization unit calculates the optimal hardware matching parameters for the next round by running optimization algorithms such as simulated annealing based on the currently identified initial adhesion force prediction value and damage mode. The action command generation unit is used to convert the optimization results into specific low-level control code, generating control command sheets covering actions such as pre-tightening and pressure holding, negative pressure pump valve ratio, and probe pose reset. The execution delay tracking unit is used to monitor the response time of the low-level servo motor and air valve after receiving the command, record the delay, and provide closed-loop feedback for the next optimization iteration.

[0041] Example 3 This embodiment provides an electronic device, including: a processor and a memory, wherein the memory stores a computer program that can be called by the processor; The processor executes the aforementioned intelligent detection method for the initial tack of composite films by calling computer programs stored in memory.

[0042] The electronic device can vary considerably depending on its configuration or performance. It may include one or more processors (Central Processing Units, CPUs) and one or more memories, wherein the memory stores at least one computer program, which is loaded and executed by the processor to implement the intelligent detection method for the initial tack of a composite film provided in the above-described embodiment. The electronic device may also include other components for implementing its functions; for example, it may have wired or wireless network interfaces and input / output interfaces for data input and output. Details will not be elaborated upon in this embodiment.

[0043] Example 4 This embodiment proposes a computer-readable storage medium on which an erasable and rewritable computer program is stored. When the computer program runs on the computer device, it causes the computer device to perform the aforementioned intelligent detection method for the initial tack of composite films.

[0044] For example, computer-readable storage media can be read-only memory (ROM), random access memory (RAM), compact disc read-only memory (CD-ROM), magnetic tape, floppy disk, and optical data storage devices.

[0045] 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.

[0046] 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.

[0047] 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.

[0048] 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.

[0049] 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 method for intelligent detection of initial tack of composite films, characterized in that: Includes the following steps: S1. Collect time-series tension and compression data, micropore adsorption negative pressure data, and local microclimate interference data at each micropore contact and probe peeling node during the test rotation of the composite membrane using a flexible material mechanical sensing module. S2. Construct a thin film interface deformation decoupling strategy and calculate the background resistance bias in each historical test cycle. S3. Establish a transient initial adhesion inference model, using the time-series tensile and compressive data, microporous adsorption negative pressure data, and local microclimate disturbance data as input entities to infer the predicted value of interfacial adhesion force during the current test cycle. S4. Input the output items extracted in steps S3 and S2 into the multi-agent matching framework to perform peeling damage inference and output the composite film adhesive layer peeling mode identification results. S5. Configure the test parameter optimization algorithm, and generate a device control instruction sheet based on the predicted value of interface adhesion force during the current test cycle. The instruction sheet covers tasks such as pre-tightening and pressure holding, negative pressure pump valve ratio, and probe universal ball joint posture reset, and tracks the response delay of the instruction sheet nodes.

2. The intelligent detection method for initial tack of a composite film according to claim 1, characterized in that: The time-series tension and compression data includes the stress gradient under the universal ball joint probe and the transient peak tension when the probe is pulled back; the micropore adsorption negative pressure data includes the air pressure at the end of the high-density array base and the air flow rate to maintain surface flatness; the local microclimate interference data includes the relative humidity of the test micro-area, the near-field temperature of the probe, and the electrostatic field strength interference.

3. The intelligent detection method for initial tack of a composite film according to claim 2, characterized in that: The construction process of the transient initial adhesion inference model includes: based on a one-dimensional deep residual network architecture, combining the number of historical test cycles, time-series tensile and compressive data, microporous adsorption negative pressure data, and local microclimate disturbance data into feature vectors, and using the set of all feature vectors as the input of the inference model; the transient initial adhesion inference model outputs the interface adhesion force prediction value calculated by each set of feature vectors, and uses the real interface initial adhesion force value collected in the dual-channel differential comparison test as the inference guide; the training guide is to minimize the sum of reconstruction errors of all inferred base interface adhesion force prediction values, and the training iteration stops when the sum of reconstruction errors drops to the convergence interval. The reconstruction error is obtained by squaring the difference between the calculated value of the inference model and the real value.

4. The intelligent detection method for initial tack of a composite film according to claim 3, characterized in that: The transient initial adhesion inference model further includes: setting the average value of the background resistance bias during the historical test cycle of the composite membrane as the deformation tolerance threshold; comparing the deformation tolerance threshold with the predicted value of the interfacial adhesion during the current test cycle; releasing a microporous negative pressure array side leakage warning waveform when the predicted value of the interfacial adhesion during the current test cycle is greater than the deformation tolerance threshold; calculating the expected value of deformation recovery during the current test cycle when the predicted value of the interfacial adhesion during the current test cycle is less than or equal to the deformation tolerance threshold, wherein the expected value of recovery is obtained by calculating the frequency of the background resistance bias being greater than the deformation tolerance threshold during the historical test cycle of the composite membrane and the proportion of the total frequency during the historical test cycle.

5. The intelligent detection method for initial tack of a composite film according to claim 4, characterized in that: The transient initial adhesion inference model, in executing step S4, includes: taking the predicted interfacial adhesion force value of the current test cycle and the damage morphology sequence of the composite film adhesive layer during the historical test cycle as the bidder and the auction item, respectively, and solving the composite film adhesive layer peeling mode identification result based on the task bidding game algorithm; firstly, the historical adhesive layer damage morphology sequence is used as the auction target, and the damage morphology to be inferred is introduced into the bidding one by one from the first round; in each round of bidding, the current interfacial adhesion force prediction value puts up a bid for the historical damage morphology sequence, and the bid result is presented as the difference between the correlation gain of the current adhesion force value matching the historical damage morphology sequence and the sum of the bids that have been won by the current adhesion force value; the item with the highest bid result among the current adhesion force values ​​is selected to obtain the matching right of the item participating in the auction in the sequence; when there are still unsold items of the current interfacial adhesion force prediction value, the auction order is extended to start the next round of bidding, until all sequences are won and the auction ends.

6. The intelligent detection method for initial tack of a composite film according to claim 5, characterized in that: The thin film interface deformation decoupling strategy obtains the results by normalizing and scaling the time-series tensile and compressive data, micropore adsorption negative pressure data, and local microclimate disturbance data at each node during the historical test cycle, assigning them anti-interference weight parameters, and then summing them up.

7. The intelligent detection method for initial tack of a composite film according to claim 6, characterized in that: The specific steps of the test parameter optimization algorithm include: S51. Arrange all composite film test batches in order of their predicted interfacial adhesion values ​​from highest to lowest according to the current test cycle period. Set primary and secondary fluctuation thresholds. Classify test batches with interfacial adhesion values ​​greater than the predicted values ​​as high-frequency interference batches, classify test batches in the range of primary and secondary fluctuation thresholds as medium-frequency interference batches, and classify test batches with interfacial adhesion values ​​less than the predicted values ​​as low-frequency interference batches. S52. Create different probe poses and negative pressure pump valve ratio task libraries for different interference levels. S53. Configure the initial annealing parameters, including the initial thermodynamic constant, cooling rate of decrease, and termination constant; S54. Randomly generate an initial action arrangement scheme, including the servo action assignment, intervention order and dwell time slot of all devices; S55. Construct the objective function formula, which aims to minimize the system commissioning energy consumption and film peeling deviation. The calculation formula is expressed as follows: Among them, the first term in the summation operation represents the non-negative modulus of the difference between the actual and expected completion times of the servo action; the second term represents the ratio of negative pressure pumping power to the rated peak power of the pump body; the third term represents the summation of deformation resistance penalty values; and the fourth term represents the summation of electrostatic elimination energy consumption of the ion fan. S56. If the objective function is less than the set tolerance threshold, then a neighborhood search is performed on the current arrangement scheme to generate a new candidate scheme, and the objective function of the new scheme is calculated again. If the objective function of the new scheme is not less than the objective function of the old scheme, then the new scheme is accepted. S57. Iterate the current thermodynamic constant based on the cooling rate of decrease. The current thermodynamic constant is the product of the cooling rate of decrease and the previous constant. S58. Repeat steps S56 and S57 until the thermodynamic constant is less than the termination constant, and output the currently accepted optimal action arrangement scheme.

8. A composite film initial tack intelligent detection system, used to perform the composite film initial tack intelligent detection method as described in any one of claims 1-7, characterized in that: include: The temporal alignment module is configured to collect temporal tension and compression data, micropore adsorption negative pressure data, and local microclimate interference data at various micropore contact and probe peeling nodes during the test cycle of the composite membrane through the flexible material mechanical sensing module. The interface deformation calculation module is configured to construct a thin film interface deformation decoupling strategy and calculate the background resistance bias in each historical test cycle. The transient initial adhesion inference model building module is configured to take the time-series tensile and compressive data, microporous adsorption negative pressure data and local microclimate disturbance data as input entities to infer the predicted value of interfacial adhesion force during the current test cycle. The peeling mode bidding game analysis module is configured to input the output terms extracted from the interface deformation calculation module and the interface deformation calculation module into the multi-agent matching framework to perform peeling damage inference and output the composite film adhesive layer peeling mode identification results. The test parameter optimization algorithm configuration module is configured to generate equipment control instruction sheets based on the predicted value of interface adhesion force during the current test cycle. These instruction sheets cover tasks such as pre-tightening and pressure holding, negative pressure pump valve ratio, and probe universal ball joint pose reset, and track the response delay of the instruction sheet nodes.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements a method for intelligent detection of the initial tack of a composite film according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for intelligent detection of the initial tack of a composite film according to any one of claims 1-7.

Images