Chip implantation intelligent control method based on multi-stage data feedback

Abstract

The invention relates to the technical field of data sensing, in particular to a chip implantation intelligent control method based on multi-stage data feedback, which comprises the following steps: executing loading-holding-unloading circulation at a target implantation position of a rubber green tire by adopting a pre-pressing head, acquiring a stress relaxation curve, and extracting a relaxation time constant and peak stress as material characteristic parameters; and matching an impedance control model from a database based on the parameters, and determining a dynamic balance depth threshold and a pressing speed curve. Impedance self-adaptive subsection control is adopted in the implanting process, impedance change is calculated in real time, the dynamic balance depth is judged, and meanwhile it is judged that stress relaxation is completed and release is triggered according to the reactive force attenuation rate. Then, the three-dimensional contours of the chip and the rubber extrusion ring are obtained through structured light, the height uniformity, continuity and symmetry of the chip and the rubber extrusion ring are analyzed, the combination quality grade is generated, self-adaptive correction is conducted on implantation parameters of the next period when the quality is insufficient, and therefore stable and high-consistency implantation control is achieved.

Description

Technical Field

[0001] This invention relates to the field of data sensing technology, specifically to a chip implantation intelligent control method based on multi-stage data feedback. Background Technology

[0002] With the application of IoT technology in tire lifecycle management, embedding electronic chips such as RFID before tire vulcanization has become a critical process. However, this process currently relies mainly on manual operation, which is not only inefficient but also cannot guarantee the consistency and accuracy of chip placement. Even when conventional automated mounting equipment is introduced, it mostly adopts an open-loop control mode based on preset coordinates, failing to fully consider the technical characteristics of green rubber tires, such as viscoelasticity, easy surface deformation, and low flatness.

[0003] The main drawback of existing technologies lies in the lack of an intelligent feedback mechanism for the unique mating relationship between the "rigid chip" and the "flexible rubber substrate." Specifically, existing equipment cannot effectively overcome visual positioning interference against a black rubber background, and it lacks closed-loop monitoring of the chip's stress state and functional integrity throughout the entire process of grasping, transporting, and rolling. Due to the lack of real-time correction capabilities based on multi-stage data, when dealing with rubber rheology or equipment micro-vibrations, it is highly susceptible to chip implantation position displacement, uncontrollable implantation depth, or microcracks caused by excessive rigid-flexible contact stress. Ultimately, this results in tires containing faulty chips flowing into the vulcanization process, severely impacting product yield and safety. Therefore, an intelligent control method based on multi-stage data feedback is urgently needed to address these issues.

[0004] To address this, a chip implantation intelligent control method based on multi-stage data feedback is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent control method for chip implantation based on multi-stage data feedback. This method obtains the characteristic parameters of the rubber material through pre-testing and matches them with an impedance control model. It then employs impedance adaptive segmented control to achieve precise implantation depth positioning, determines the chip release timing based on stress relaxation characteristics, and uses structured light 3D detection to evaluate implantation quality and provide feedback for parameter correction. This method achieves automated and intelligent control of chip implantation in rubber embryos, solving technical challenges such as inaccurate depth control, difficulty in determining the release timing, and lack of quantitative basis for quality assessment during viscoelastic material implantation.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A chip implantation intelligent control method based on multi-stage data feedback includes: A pre-compression head is used to perform a loading-holding-unloading cycle at the target implantation position of the rubber green tire. The stress relaxation curve of the pressure change over time is collected. The time required for the stress to decay to the preset proportion of the peak value is calculated as the relaxation time constant, and the initial peak stress is used as the hardness characterization parameter. Based on the relaxation time constant and hardness characterization parameters, the corresponding impedance control model parameters are matched from the preset database, including at least the target impedance dynamic equilibrium depth threshold and the pressing speed curve. The impedance adaptive segmented control chip enters the rubber through the compression speed curve and calculates the impedance value in real time. When the slope of the impedance value change meets the preset conditions, it is determined that the impedance dynamic equilibrium depth has been reached. The decay rate of the rubber reaction force is calculated in real time. When the decay rate is lower than the release threshold set based on the relaxation time constant, it is determined that the stress relaxation is completed and the chip release action is executed. Structured light is used to acquire three-dimensional contour data of the chip and the surrounding rubber extrusion ring. The height uniformity, continuity and symmetry characteristics of the extrusion ring are analyzed to generate the bonding quality level. If the bonding quality level is lower than the preset standard, the parameters for the next implantation cycle are corrected.

[0007] Preferably, the process of obtaining the relaxation time constant is as follows: control the pre-pressure head to press into the rubber surface at a constant rate to a preset depth and then stop and keep the position fixed; take the stopping moment as the starting point, continuously collect the real-time pressure data fed back by the pressure sensor, and construct the pressure decay curve as a function of time; extract the time span corresponding to when the pressure value in the decay curve drops to a preset percentage of the initial peak stress; and mark this time span as the relaxation time constant.

[0008] Preferably, the impedance-adaptive segmental implantation process specifically includes: The rapid approach phase is used to control the robotic arm to move at high acceleration to a preset safe height from the rubber surface. During this phase, the position loop gain is at its maximum, and impedance calculation is not performed. The contact sensing section is used to decelerate the robotic arm and monitor changes in the readings of the end effector force sensor. When the reading exceeds the contact threshold, the contact zero point is marked, and the system switches to impedance control mode. The impedance working section is used to adjust the end effector stiffness of the robotic arm from the zero contact point, based on the impedance control model and the deviation between the real-time contact force and the reference speed, until the impedance dynamic equilibrium depth is determined to be reached.

[0009] Preferably, the process for determining the depth of impedance dynamic equilibrium is as follows: During the impedance operation segment, the mechanical impedance value at the current moment is calculated in real time, and the derivative of the mechanical impedance value with respect to the implantation depth is obtained to obtain the impedance change rate. The numerical characteristics of the impedance change rate are monitored in real time. When the impedance change rate changes from a linear and stable state to an exponentially increasing state, and the increase exceeds the preset matrix stiffness threshold, it is determined that the chip has passed through the surface rheological layer and contacted the dense rubber matrix, and the current Z-axis coordinate is immediately locked.

[0010] Preferably, the execution process of the chip release action is as follows: Once stress relaxation is complete, a release command is sent to the end effector. During the initial micro-displacement phase of the robotic arm's retraction action, the vertical tension data is continuously monitored. If the tension data exceeds the preset adhesion safety threshold, the retraction is paused and high-frequency micro-vibration is performed to break the interface vacuum adhesion force through vibration. Once the tension data falls back to the safe range, the full-speed retraction action is performed.

[0011] Preferably, the process of acquiring three-dimensional contour data is as follows: Multiple sets of sinusoidal fringe gratings with different phase shifts are projected onto the chip and its surrounding area; a sequence of deformed fringe images modulated by the height of the object surface is acquired simultaneously; phase demodulation is performed on the image sequence to obtain a folded phase map; the folded phase map is unfolded using the multi-frequency heterodyne principle to obtain a continuous phase distribution; the continuous phase distribution is converted into point cloud data containing spatial coordinates according to the phase-height mapping relationship; and the annular region data located outside the chip edge is segmented from the point cloud data as three-dimensional contour data.

[0012] Preferably, the process for determining the quality level specifically includes continuity analysis, uniformity analysis, and symmetry analysis; specifically: The gradient change of height data is calculated along the circumferential path of the extrusion ring, the number of gradient breakpoints is counted, and a continuity score is generated; the standard deviation of the height values ​​of all point cloud values ​​in the extrusion ring region is calculated, and a uniformity score is generated; the vector deviation between the geometric center of the extrusion ring and the physical center of the chip is calculated, and a symmetry score is generated; the continuity score, uniformity score, and symmetry score are input into a preset weighted evaluation model, and the corresponding combination quality level is output.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention introduces a pre-compression test process of loading-holding-unloading before chip implantation to acquire the stress relaxation curve of the rubber green tire at the target implantation site in real time. The relaxation time constant and initial peak stress are extracted as material state characterization parameters, thereby accurately characterizing the differences in rheological properties and equivalent hardness between different batches and regions of rubber. This invention can quantitatively identify the mechanical properties of the material before implantation and match the corresponding impedance control model parameters accordingly, enabling targeted settings for the compression speed and target equilibrium depth. This significantly reduces overpressure, underpressure, and implantation instability caused by uneven rubber hardness or time-varying properties, improving the chip implantation process's adaptability and consistency to complex material states.

[0014] 2. This invention employs an impedance-adaptive segmented control mechanism during chip implantation. By calculating the mechanical impedance and its rate of change in real time, it identifies key points where the impedance transitions from a steady change to an exponential increase, thereby accurately determining the position where the chip has penetrated the surface rheological layer and entered the dense rubber matrix, and locking the impedance dynamic equilibrium depth. Furthermore, by combining the attenuation rate of the rubber reaction force with a release threshold set based on the relaxation time constant, a "release after mechanical stabilization" control strategy is achieved. This invention effectively avoids springback, displacement, or interface tearing caused by insufficient stress relaxation, significantly reducing the risk of damage to the internal rubber structure while ensuring implantation depth accuracy, thus improving chip positioning stability and long-term reliability.

[0015] 3. After chip release, this invention introduces structured light 3D imaging technology to reconstruct the 3D contour of the rubber extrusion ring surrounding the chip with high precision. It then quantitatively analyzes the morphology of the extrusion ring from three dimensions: continuity, uniformity, and symmetry, generating an objective evaluation result for the bonding quality level. This invention can reflect the true bonding state of the chip-rubber interface using point cloud-level data, avoiding subjectivity and the risk of missed detections. More importantly, the bonding quality level is directly used to correct the control parameters for the next implantation cycle, forming a self-learning closed-loop optimization mechanism based on result feedback, thereby continuously improving implantation consistency, finished product qualification rate, and the overall stability of the automated production line. Attached Figure Description

[0016] Figure 1 A flowchart of a chip implantation intelligent control method based on multi-stage data feedback provided by the present invention; Figure 2 A schematic diagram of the impedance adaptive segmented structure provided by the present invention; Figure 3 The flowchart for obtaining the chip bonding quality level provided by this invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the invention.

[0018] This invention provides a chip implantation intelligent control method based on multi-stage data feedback, which is executed by a dedicated automated implantation device.

[0019] Example 1: Please see Figure 1 This invention provides a chip implantation intelligent control method based on multi-stage data feedback. The technical solution is as follows: A pre-pressure head is used to perform a loading-holding-unloading cycle at the target implantation position of the rubber tire, and the stress relaxation curve of the pressure changing with time is collected. The time required for the stress to decay to a preset proportion of the peak value is calculated as the relaxation time constant, and the initial peak stress is used as the hardness characterization parameter. Based on the relaxation time constant and the hardness characterization parameter, the corresponding impedance control model parameters are matched from a preset database, including at least the target impedance dynamic equilibrium depth threshold and the pressing speed curve. The chip is controlled by impedance adaptive segmentation to enter the rubber according to the pressing speed curve, and the impedance value is calculated in real time. When the slope of the impedance value change meets the preset condition, it is determined that the impedance dynamic equilibrium depth has been reached, and the decay rate of the rubber reaction force is calculated in real time. When the decay rate is lower than the release threshold set based on the relaxation time constant, it is determined that the stress relaxation is completed, and the chip release action is executed. The three-dimensional contour data of the chip and the surrounding rubber extrusion ring are acquired using structured light, and the height uniformity, continuity and symmetry characteristics of the extrusion ring are analyzed to generate a bonding quality level. If the bonding quality level is lower than the preset standard, the parameters of the next implantation cycle are corrected.

[0020] The pre-built database primarily stores material property and control parameter mapping data based on experiments or simulations. Specifically, it includes: impedance control model parameter sets indexed by relaxation time constants and hardness characterization parameters. Each parameter set contains at least the target impedance dynamic equilibrium depth threshold, indentation velocity curve, and release threshold. Furthermore, the database includes standard test temperatures corresponding to each parameter set, as well as temperature-viscoelastic correlation data for rubber materials used for temperature correction. For multi-point implantation scenarios, the database also pre-stores stress attenuation lookup tables, recording residual stress amplitudes at different depths, chip sizes, and distances.

[0021] Preferably, the process of obtaining the relaxation time constant is as follows: control the pre-pressure head to press into the rubber surface at a constant rate to a preset depth and then stop and keep the position fixed; take the stopping moment as the starting point, continuously collect the real-time pressure data fed back by the pressure sensor, and construct the pressure decay curve as a function of time; extract the time span corresponding to when the pressure value in the decay curve drops to a preset percentage of the initial peak stress; and mark this time span as the relaxation time constant.

[0022] Specifically, before the chip implantation process begins, a pre-pressure head equipped with a spherical probe is first moved to the target implantation position on the rubber green tire. The pre-pressure head is then pressed vertically into the rubber surface at a constant low speed until a preset detection depth (e.g., 0.5 mm) is reached. At this point, the servo motor stops feeding and maintains a locked position. The moment of stopping is taken as the zero point of timing. The instantaneous pressure value collected by the force sensor at this time is the initial peak stress, which is used as a parameter characterizing the local hardness of the rubber. Subsequently, real-time pressure data fed back from the pressure sensor is continuously collected at a high frequency (e.g., 1000 Hz), constructing an exponential decay curve of pressure changing over time. The central processing unit monitors this curve in real time and calculates the ratio of the current pressure value to the initial peak stress. When the pressure value drops to a preset percentage of the initial peak stress (e.g., 36.8%), the time point at this moment is recorded. The time span between this time point and the zero point of timing is marked as the relaxation time constant. This parameter reflects the rate of stress release of the rubber material after being compressed and is directly used to guide the waiting time in the subsequent chip release stage.

[0023] The relaxation time constant extraction method of the present invention is based on the linear viscoelastic theory of viscoelastic materials and holds true under the following conditions: The pre-compression strain amplitude is ≤3% (corresponding to a preset depth of 0.3-0.7mm constraint); the ambient temperature is stable within the range of 15-35°C, with temperature fluctuation ≤5°C; the test time is ≤60 seconds; the rubber used is a standard tire raw rubber formula (filler volume fraction <35%).

[0024] If, in addition to the above conditions, the material exhibits multi-stage decay characteristics, the material property parameters should be updated.

[0025] By performing standardized stress relaxation tests on the rubber green tire before chip implantation, the relaxation time constant and local hardness parameters are quantitatively obtained, enabling precise characterization of the rubber's rheological properties. Through constant-rate preloading, position locking, and high-frequency pressure sampling, a stable stress decay curve can be constructed, and the relaxation time constant is determined using a uniform proportional threshold, reducing interference from environmental and material differences. This parameter directly guides the chip release timing, matching the release process with the rubber's stress release behavior, effectively reducing rebound, adhesion, and implantation instability, and improving overall implantation consistency and reliability.

[0026] Preferably, impedance adaptive segmented reference Figure 2 The implantation process is as follows: The rapid approach phase is used to control the robotic arm to move at high acceleration to a preset safe height from the rubber surface. During this phase, the position loop gain is at its maximum, and impedance calculation is not performed. The contact sensing section is used to decelerate the robotic arm and monitor changes in the readings of the end effector force sensor. When the reading exceeds the contact threshold, the contact zero point is marked, and the system switches to impedance control mode. The impedance working section is used to adjust the end effector stiffness of the robotic arm from the zero contact point, based on the impedance control model and the deviation between the real-time contact force and the reference speed, until the impedance dynamic equilibrium depth is determined to be reached.

[0027] Specifically, after obtaining the above parameters, the robotic arm grips the chip and performs segmented implantation. First, it enters the rapid approach phase, where the robotic arm moves downward with high acceleration and a maximum speed limit (e.g., 500 mm / s) until it reaches a preset safe height of about 2 mm to 5 mm from the rubber surface. During this phase, a position control mode is adopted, and the position loop gain is set to the maximum value to pursue the highest efficiency.

[0028] Then, the robot arm enters the contact sensing phase, decelerates to a medium speed (e.g., 10 mm / s to 50 mm / s), and monitors the changes in the readings of the end effector force sensor in real time. A small contact threshold (e.g., 0.5 N) is set. When the sensor reading exceeds the threshold, the position is marked as the contact zero point, and the control mode is immediately switched to impedance control mode.

[0029] Next, the impedance operation phase begins. The central control unit invokes the impedance control model based on the aforementioned relaxation time constant matching. In this mode, the robotic arm is no longer forced to execute rigid position commands. Instead, the deviation between the "real-time contact force" and the "reference speed" is used as input to adjust the end effector stiffness (or compliance) of the robotic arm through an impedance algorithm. Specifically, if the rubber is locally harder, the feedback force increases, and the downward pressure speed is automatically reduced, exhibiting high damping characteristics. If the rubber is softer, the speed is maintained or slightly increased. This process continues until the deep locking condition is met.

[0030] The impedance control model is the core algorithm unit for achieving "rigid-flexible coupling" adaptive implantation. It consists of a basic matching layer and a higher-order prediction layer: the basic layer uses database matching, while the higher-order layer employs regression-based machine learning algorithms (such as random forest). The model uses the relaxation time constant, hardness characterization parameters, and ambient temperature obtained from the previous stage of pre-testing as input features to establish a nonlinear mapping relationship between the material's rheological properties and the optimal robotic arm control parameters. Its purpose is to dynamically generate the optimal target impedance dynamic equilibrium depth threshold and compression velocity curve when the rubber material has uneven hardness or the ambient temperature fluctuates, ensuring accurate implantation depth without overload. This embodiment uses a random forest regression algorithm to build the prediction model. The training dataset comes from a historical production database. Each sample contains an input feature vector (including relaxation time constant, initial peak stress, and rubber surface temperature) and an output label vector (including the target impedance dynamic equilibrium depth threshold, compression velocity curve parameters, and release threshold). After standardizing the features, the dataset is divided into a training set and a validation set in an 8:2 ratio. Five-fold cross-validation is used to train the model, with mean squared error as the loss function, until the validation error converges to a preset threshold (e.g., depth prediction error < 0.1 mm). When high-confidence samples (i.e., samples with excellent combination quality) are accumulated, they are given higher weights for model retraining.

[0031] In the "impedance operation segment" of the implantation process, the model first receives real-time relaxation time constant and initial peak stress data obtained from the pre-compression test. The model outputs a recommended parameter set, which the controller uses to set the end effector stiffness coefficient and reference trajectory of the robotic arm. During execution, the model receives contact force and position data from force sensors in real time, and calculates the mechanical impedance and its derivative with respect to depth (resistance change rate). When the impedance change rate is detected to change from linear and stable to an exponential increase, and the magnitude exceeds the matrix stiffness threshold set by the model, it is determined that a dynamic equilibrium point has been reached.

[0032] By dividing the chip implantation process into three stages—rapid approach, contact sensing, and impedance operation—efficiency and accuracy are balanced. The rapid approach stage employs high-gain position control to shorten idle travel time; the contact sensing stage precisely calibrates the contact zero point using a force threshold to avoid accidental contact or overpressure; and the impedance operation stage adaptively adjusts the end-effector stiffness and pressing speed based on material properties, allowing the implantation process to automatically adjust to changes in rubber hardness. This segmented impedance control strategy effectively suppresses impact and overload, improves the stability and consistency of implantation depth determination, and ensures chip implantation quality.

[0033] Preferably, the process for determining the depth of impedance dynamic equilibrium is as follows: During the impedance operation segment, the mechanical impedance value at the current moment is calculated in real time, and the derivative of the mechanical impedance value with respect to the implantation depth is obtained to obtain the impedance change rate. The numerical characteristics of the impedance change rate are monitored in real time. When the impedance change rate changes from a linear and stable state to an exponentially increasing state, and the increase exceeds the preset matrix stiffness threshold, it is determined that the chip has passed through the surface rheological layer and contacted the dense rubber matrix, and the current Z-axis coordinate is immediately locked.

[0034] Specifically, during the impedance operation segment, a depth locking algorithm is executed. The processor calculates the mechanical impedance value at the current moment (i.e., the ratio of the real-time contact force change to the displacement change, representing dynamic stiffness) in real time. To more sensitively identify the rubber substrate, the derivative of this mechanical impedance value with respect to the implantation depth is obtained to obtain the impedance change rate. In the early stage of implantation, the chip mainly penetrates the surface rheological layer, and the impedance change rate is relatively stable or shows a linear increase. When the bottom surface of the chip contacts the dense rubber substrate or the top of the cord layer, the impedance value will change abruptly. The numerical characteristics of the impedance change rate are monitored in real time. When the impedance change rate is detected to change from a linear stable state to an exponential increase state, and the increase exceeds the preset substrate stiffness threshold, it is determined that the chip is in the optimal bonding position. At this time, a command is immediately sent to lock the current Z-axis coordinate, which is defined as the impedance dynamic equilibrium depth. The robotic arm then stops pressing down and enters the pressure holding state.

[0035] By analyzing mechanical impedance and its variation with implantation depth in real time, this method enables adaptive determination of the optimal implantation depth for the chip. Introducing the rate of impedance change as a criterion distinguishes the differences in mechanical response between the rheological region of the rubber surface and the dense substrate, accurately identifying when the chip has reached a stable bonding position when impedance abrupt changes occur. This method avoids misjudgments caused by relying solely on fixed depths or force thresholds, effectively preventing over- or under-pressure implantation, improving the reliability and repeatability of depth locking, and providing a stable benchmark for subsequent pressure holding and release control.

[0036] Preferably, the execution process of the chip release action is as follows: Once stress relaxation is complete, a release command is sent to the end effector. During the initial micro-displacement phase of the robotic arm's retraction action, the vertical tension data is continuously monitored. If the tension data exceeds the preset adhesion safety threshold, the retraction is paused and high-frequency micro-vibration is performed to break the interface vacuum adhesion force through vibration. Once the tension data falls back to the safe range, the full-speed retraction action is performed.

[0037] Specifically: When stress relaxation monitoring determines that the chip can be released, a release command is sent to the end effector (e.g., shutting off the vacuum generator solenoid valve). To prevent the chip from being pulled out due to negative pressure adhesion on the rubber surface, during the initial micro-displacement phase of the robotic arm's upward retraction (e.g., during an upward lift of 0.1mm), a force sensor continuously monitors the vertical tension data. If the tension data exceeds a preset adhesion safety threshold, it indicates that there is residual vacuum or adhesion between the chip and the nozzle. The robotic arm immediately pauses the retraction action and activates the piezoelectric ceramic vibration unit inside the end effector to perform high-frequency micro-amplitude vibration (e.g., frequency 20kHz, amplitude 5 micrometers). The air layer introduced by the high-frequency vibration breaks the interface vacuum adhesion force. Only after the tension data is detected to have fallen back to the safe range (close to zero) does the robotic arm perform a full-speed retraction action, ensuring that the chip remains inside the tire.

[0038] By combining chip release timing with stress relaxation state, precise control of the release action is achieved. Force monitoring and adhesion assessment are introduced during the initial withdrawal phase after release to identify residual vacuum or adhesion risks between the chip and the adsorption end in real time. If an anomaly is detected, high-frequency micro-vibration is used to disrupt the interfacial adsorption force, effectively preventing the chip from being carried out or its position from being disturbed. This release and withdrawal strategy improves the reliability and success rate of chip placement, reduces the risk of implantation failure and rework, and enhances overall process stability.

[0039] Preferably, the process of acquiring three-dimensional contour data is as follows: Multiple sets of sinusoidal fringe gratings with different phase shifts are projected onto the chip and its surrounding area; a sequence of deformed fringe images modulated by the height of the object surface is acquired simultaneously; phase demodulation is performed on the image sequence to obtain a folded phase map; the folded phase map is unfolded using the multi-frequency heterodyne principle to obtain a continuous phase distribution; the continuous phase distribution is converted into point cloud data containing spatial coordinates according to the phase-height mapping relationship; and the annular region data located outside the chip edge is segmented from the point cloud data as three-dimensional contour data.

[0040] Preferably, the process for determining the quality level specifically includes continuity analysis, uniformity analysis, and symmetry analysis; specifically: The gradient change of height data is calculated along the circumferential path of the extrusion ring, the number of gradient breakpoints is counted, and a continuity score is generated; the standard deviation of the height values ​​of all point cloud values ​​in the extrusion ring region is calculated, and a uniformity score is generated; the vector deviation between the geometric center of the extrusion ring and the physical center of the chip is calculated, and a symmetry score is generated; the continuity score, uniformity score, and symmetry score are input into a preset weighted evaluation model, and the corresponding combination quality level is output.

[0041] Specifically, after chip implantation, the vision inspection module intervenes. The projection unit projects multiple sets of sinusoidal fringe gratings with different phase shifts (e.g., 0, π / 2, π, 3π / 2) onto the chip and surrounding area. Simultaneously, the CMOS camera acquires a sequence of deformed fringe images modulated by the object's surface height. The processor uses a four-step phase-shift algorithm to demodulate the image sequence, obtaining a folded phase map. Phase unwrapping is then performed using the multi-frequency heterodyne principle to obtain a continuous absolute phase distribution. Based on the pre-calibrated phase-height mapping relationship, point cloud data containing spatial coordinates is generated.

[0042] Automatically identify the chip edge and extract the point cloud of a ring-shaped region with a preset width (e.g., 2mm) outside the chip edge as 3D contour data (i.e., rubber extrusion ring), and perform multidimensional analysis on this data: Continuity analysis: Calculate the break point of the height gradient along the circumference of the extrusion ring. If a break exists, it indicates that the rubber compound is not fully coated. Uniformity analysis: Calculate the standard deviation of the extrusion ring height value. The smaller the standard deviation, the more uniform the extrusion. Symmetry analysis: Calculate the deviation between the geometric center of the extrusion ring and the chip center; finally, input the above three scores into the weighted evaluation model and output a combined quality level. If the level does not meet the standard, the deviation will be calculated and fed back to the main control unit to correct the impedance parameters when implanting the next chip (e.g., increase the target dynamic stiffness threshold).

[0043] The weighted evaluation model is primarily used for the quantitative assessment of the bonding quality after chip implantation. Its core logic is based on 3D point cloud data acquired using structured light to deconstruct the rubber extrusion ring from a geometrical perspective. The model receives continuity, uniformity, and symmetry scores as input vectors, and performs linear weighted fusion using a pre-defined weighting coefficient matrix, ultimately outputting a discrete bonding quality level. This model aims to transform unstructured 3D morphology data into a decision-making basis for closed-loop control, achieving objective and automated judgment of implantation quality. The model is constructed based on supervised learning or expert system methods. First, a sample set is established, collecting a large amount of historical 3D point cloud data of implanted rubber extrusion rings. Senior process engineers manually label each data set according to appearance defect standards (such as bubbles, fractures, and eccentricity), classifying them into four levels: "Excellent," "Good," "Medium," and "Poor." Second, for the three feature dimensions of continuity, uniformity, and symmetry, the weighting coefficients for each dimension are determined using the analytic hierarchy process (AHP) or principal component analysis (PCA).

[0044] By introducing structured light 3D reconstruction technology, the spatial morphology of the rubber extrusion ring after chip implantation is acquired and quantitatively analyzed with high precision, enabling an objective assessment of the bonding quality level. Point cloud data is generated through phase-shifted fringe projection and phase unfolding, accurately reconstructing the true height distribution of the rubber surrounding the chip. Furthermore, the extrusion ring is comprehensively evaluated from three dimensions: continuity, uniformity, and symmetry, effectively reflecting the integrity of the rubber coating and the consistency of stress. The weighted evaluation model outputs the bonding quality level, and the evaluation results are fed back for adaptive parameter correction in the next cycle, forming a closed-loop optimization mechanism. This improves the consistency, reliability, and overall process stability of chip implantation.

[0045] This invention achieves high-precision, adaptive, and closed-loop optimized control of the chip implantation process in rubber embryos by constructing a multi-stage data feedback control mechanism spanning pre-implantation, implantation, and post-implantation. Before implantation, stress relaxation testing is introduced to quantitatively obtain the local hardness and relaxation time constant of the rubber, providing a reliable feedforward basis for impedance control model parameter matching. During implantation, a segmented impedance control and dynamic impedance balance depth determination strategy is adopted, enabling the robotic arm to automatically adjust its pressing speed and compliance according to changes in the rubber's rheological properties, accurately locking the optimal implantation depth and avoiding over- or under-pressure. In the chip release stage, the stress relaxation completion state is combined with force feedback, supplemented by high-frequency micro-amplitude vibration, effectively eliminating the risk of adhesion and ensuring stable chip placement. After implantation, structured light 3D reconstruction is used to analyze the continuity, uniformity, and symmetry of the rubber extrusion ring, forming an objective bonding quality grade, as detailed in the following reference. Figure 3 The evaluation results are then fed back to subsequent parameter adjustments, forming a complete closed-loop self-learning optimization system. The overall solution significantly improves the consistency, reliability, and process stability of chip implantation, making it suitable for automated chip implantation scenarios with high precision and quality requirements.

[0046] Example 2: Based on Embodiment 1, this embodiment further discloses the collaborative control method, temperature compensation mechanism, and specific implementation of the predictive optimization module based on historical data when multiple implantation needs to be performed on the same rubber green tire.

[0047] When multiple chips need to be implanted at different locations on the same green tire during tire production—for example, temperature monitoring chips symmetrically implanted on the sidewall or multiple RFID tag chips implanted in the bead—stress field interference occurs between adjacent implantation points. Specifically, after the first chip is implanted, a residual stress field is formed within the rubber matrix. This stress field spreads outward from the implantation point, altering the local stress state of the surrounding material. When a second implantation is performed within this affected area, the relaxation time constant and hardness characterization parameters measured during the pre-test phase no longer accurately reflect the true material properties at that location, causing the matched impedance control model parameters to fail. Simultaneously, the dynamic stress waves generated by the new implantation operation propagate to the already implanted chip locations, potentially causing minute displacements of the fixed chips. To address the aforementioned issues, this embodiment provides the following technical solution: An implantation sequence planning module is established to calculate the minimum safe distance between adjacent implantation points based on their spatial locations. Before performing pre-testing and implantation operations at the Nth implantation point, the position of the nearest implanted chip is measured, and the stress influence attenuation coefficient is calculated. The impedance control model parameters obtained from a preset database are corrected based on the stress influence attenuation coefficient to compensate for local stress field changes caused by adjacent chips. After completing the implantation of the Nth chip and determining stress relaxation, structured light is used to simultaneously detect the positional offset of the Nth implantation point and all implanted chips within its influence range. If the offset exceeds the position tolerance threshold, the area is marked as a high-risk area, and subsequent implantation sequences are adjusted.

[0048] Specifically, before commencing multi-point implantation, the three-dimensional spatial coordinates of all implantation points are first acquired. These coordinates can be pre-calibrated using visual positioning or directly imported from the chip layout scheme in the tire design drawings. A built-in stress influence model, based on the elastic modulus and Poisson's ratio of the rubber material, calculates the spatial distribution of the stress field generated at a single implantation point. The stress field intensity decreases exponentially with increasing distance. All pairs of implantation points are traversed, and the Euclidean distance between each pair is calculated. When the distance between a pair of implantation points is less than the critical interference distance calculated based on the stress influence model, these two points are marked as an interference pair. The critical interference distance is 30 to 50 millimeters, determined based on the rubber hardness and chip size. For interference pairs, a greedy algorithm or genetic algorithm is used to generate the optimal implantation sequence, minimizing the impact of the existing stress field during each implantation. The priority order is determined by prioritizing implantation of chips with dispersed locations, maximizing the time interval between implantations of adjacent chips, and utilizing the stress relaxation characteristics of rubber to allow the existing stress field to decay naturally.

[0049] The stress influence model is specifically designed to address stress field interference issues in multi-point implantation scenarios. Based on the principles of elasticity, the model treats the rubber green tire as a viscoelastic half-space, describing the attenuation of the residual stress field generated by a single-point implantation with spatial distance. The model calculates the Euclidean distance between the implantation point and the implanted chips, and, using the superposition principle, quantifies the interference of the known stress field on the current implantation point. The model's function is to output correction coefficients to compensate for the preset parameters of the impedance control model, preventing stress concentration or misjudgment caused by adjacent chips, and assisting in planning the optimal implantation sequence. A simulation model of rubber-chip interaction is established through finite element analysis, setting different implantation depths (2-8mm), chip sizes (small / medium / large), and spacing (10-100mm) as variables to calculate the residual stress distribution under various conditions. The simulation results are extracted and stored in a three-dimensional array, constructing a lookup table. Simultaneously, combined with actual testing, the critical interference distance of the rubber material (e.g., 30-50mm) is determined to define the effective domain of the model. Before implanting the Nth chip, the system obtains the coordinates of the current target and all implanted chips within an 80mm radius. The model iterates through and calculates the distances between the current point and each implanted point. If the distance is less than the critical interference distance, a lookup table query is triggered. The model indexes the stress components generated by each implanted chip at the current position based on the distance, size, and depth, and performs vector superposition to calculate the total stress bias value.

[0050] When performing the implantation of the Nth chip, the structured light 3D scanning module is first invoked to scan the area within an 80mm radius around the current implantation point, identifying the actual coordinates of all implanted chips. The coordinates of the current implantation point are compared with the coordinates of all implanted chips to identify the closest one or more chips. For each implanted chip, based on its implantation depth, chip size, and spatial distance from the current point, a pre-established stress attenuation lookup table is consulted to obtain the residual stress amplitude generated at the current location. The stress attenuation lookup table is established through finite element simulation or experimental testing. The table structure is a 3D array with three index dimensions: implantation depth of the implanted chip (range 2 to 8 mm, step size 0.5 mm); chip size class (small, medium, and large, corresponding to diameter ranges of 5 to 8 mm, 8 to 12 mm, and 12 to 20 mm, respectively); and spatial distance from the current implantation point (range 10 to 100 mm, step size 5 mm). Each cell in the table stores the residual stress amplitude generated at the current location under those conditions, in kilopascals (kPa). The lookup table supports continuous value lookups via linear interpolation. When multiple implanted chips exist, the stress components generated by each chip are vector-superimposed to obtain the total stress bias at the current location. Based on the calculated stress bias value, the impedance control model parameters are corrected. Specifically, due to the local prestress, the apparent hardness of the rubber increases, and the target implantation depth threshold obtained from the database is reduced by 5% to 15% to prevent the chip from being pressed too deeply. Simultaneously, the initial contact speed in the pressing speed curve is reduced to decrease the impact load and prevent stress concentration from causing rubber tearing. The release threshold is correspondingly reduced because the stress relaxation rate is faster in areas with prestress.

[0051] After the Nth chip is implanted and stress relaxation is determined, the scanner does not immediately withdraw but restarts a structured light 3D scan. The scan area covers the current implantation point and all implanted chips within the influence radius determined by the stress influence model. The feature contours of each chip are extracted using an image recognition algorithm, and its geometric center coordinates are calculated. These coordinates are then compared with the initial position coordinates of the chip recorded before implantation to obtain the position offset vector.

[0052] The position offset vector comprises three components: X, Y, and Z. The magnitude of the offset vector, i.e., the total offset distance, is calculated. If the total offset distance exceeds a preset position tolerance threshold, such as 0.2 mm, the chip is considered to have undergone unacceptable displacement. The chip location and its surrounding 30 mm radius are marked as a high-risk area, and this area is dynamically adjusted in subsequent implantation sequences to avoid implanting new chips near this area, or to increase the implantation interval of chips within this area to allow for further stress release. If the position offset of an implanted chip exceeds a failure threshold, such as 0.5 mm, an alarm signal is immediately issued, subsequent implantation operations are suspended, and the operator is notified for manual inspection. The operator can choose to reimplant the chip or adjust the entire implantation plan.

[0053] The viscoelastic parameters of rubber materials are highly sensitive to temperature. As temperature increases, the speed of movement of rubber molecular chain segments increases, the relaxation time constant shortens, and the material exhibits greater fluidity. Conversely, as temperature decreases, rubber tends to harden, and the relaxation process slows down. The ambient temperature in production environments typically fluctuates between 15 and 35 degrees Celsius; such temperature variations significantly affect the validity of material property parameters obtained during the preliminary testing phase.

[0054] The temperature compensation mechanism provided in this embodiment is as follows: While executing the pre-compression head loading-holding-unloading cycle, a non-contact infrared temperature sensor is used to measure the rubber surface temperature at the target implantation location in real time; the measured real-time temperature is compared with the standard test temperature corresponding to each group of impedance control model parameters in the preset database, and the temperature deviation value is calculated; based on the time-temperature equivalence principle, the time scale translation factor corresponding to the temperature deviation is calculated using preset WLF equation parameters or Arrhenius activation energy, and the relaxation time constant is corrected; at the same time, the hardness characterization parameter is compensated according to the temperature dependence of rubber modulus; during the continuous implantation process of the same production batch, the change of ambient temperature is periodically monitored, and when the cumulative temperature deviation exceeds the preset temperature drift threshold, the pre-test process is re-executed before the start of the next implantation cycle to update the material characteristic parameters and impedance control model.

[0055] Throughout the loading-hold-unloading cycle performed by the preload head, a non-contact infrared temperature sensor mounted on the side of the preload head remains continuously aligned with the rubber surface at the target implantation site, acquiring temperature data at a sampling frequency of 5 Hz. The sensor probe is kept perpendicular to the rubber surface at a distance of 80 to 120 mm to ensure measurement accuracy. The sensor's measurement wavelength range covers 8 to 14 micrometers, suitable for the emissivity characteristics of rubber materials.

[0056] Record the initial temperature at the start of loading, the average temperature during the holding phase, and the final temperature at the unloading point. Since the contact time between the preload head and the rubber is short, typically 10 to 30 seconds, and the preload head itself does not generate significant heat, the rubber temperature changes very little during the test. The average temperature during the holding phase can be used as the representative temperature at that location.

[0057] Each set of impedance control model parameters stored in the pre-set database corresponds to a standard test temperature, typically set at 23 degrees Celsius, which meets the standard laboratory conditions for material testing. The real-time measured rubber surface temperature is compared with 23 degrees Celsius to calculate the temperature deviation.

[0058] A positive temperature deviation value indicates that the actual temperature is higher than the standard temperature; a negative temperature deviation value indicates that the actual temperature is lower than the standard temperature. The temperature-viscoelasticity correlation data for the built-in rubber material is derived from technical documents provided by the material supplier or laboratory test results. This correlation data describes the variation of the relaxation time constant and hardness characterization parameters with temperature.

[0059] For temperature correction of the relaxation time constant, a calculation method based on the time-temperature equivalence principle is adopted. The time-scale shift factor corresponding to the temperature deviation is calculated based on the glass transition temperature of the rubber material and two empirical constants. The specific calculation steps are as follows: First, calculate the difference between the reference temperature and the actual temperature, denoted as the temperature difference value; then, calculate the product of the first constant and the temperature difference value, divided by the sum of the second constant and the temperature difference value, to obtain the logarithmic shift factor; finally, take the negative exponent of the logarithmic shift factor to obtain the time-scale shift factor. When the actual temperature is 5 degrees Celsius higher than the reference temperature, the time-scale shift factor is approximately 0.6 to 0.8; when the actual temperature is 5 degrees Celsius lower than the reference temperature, the time-scale shift factor is approximately 1.2 to 1.5. When the actual temperature is higher than the standard temperature, the shift factor is less than 1, indicating that the relaxation process is accelerated, and the measured relaxation time constant is multiplied by this shift factor for correction. Conversely, when the actual temperature is lower than the standard temperature, the shift factor is greater than 1, and the corrected relaxation time constant increases.

[0060] For temperature correction of hardness characterization parameters, linear compensation is used; the modulus of rubber decreases with increasing temperature. Based on the modulus-temperature correlation coefficient of the material, the hardness change corresponding to the temperature deviation is calculated. This change is subtracted from the measured initial peak stress to obtain the corrected hardness characterization parameters.

[0061] The corrected relaxation time constant and hardness characterization parameters were used in the subsequent database matching process to replace the original measurements, thereby obtaining more accurate impedance control model parameters.

[0062] In continuous implantation operations within the same production batch, a temperature monitoring log is established to record the rubber surface temperature at each implantation operation. Using the temperature at the time of the first implantation as the baseline temperature, the cumulative temperature deviation for each subsequent implantation is calculated. The cumulative temperature deviation is defined as the absolute value of the difference between the current temperature and the baseline temperature.

[0063] When the cumulative temperature deviation exceeds a preset temperature drift threshold, such as five degrees Celsius, it is determined that the ambient temperature has changed significantly, and the original material property parameters may no longer be applicable. Before the next implantation cycle begins, a recalibration process is automatically triggered. The recalibration process includes: re-executing the preload-hold-unload cycle of the preload head under the new temperature conditions, re-acquiring the stress relaxation curve, recalculating the relaxation time constant and hardness characterization parameters, and updating various parameters in the impedance control model, including the target implantation depth threshold, the indentation velocity curve, and the release threshold.

[0064] After recalibration, the current temperature is updated to the new reference temperature, the cumulative temperature deviation is reset to zero, and subsequent implantation operations continue. This batch-to-batch periodic recalibration mechanism can adapt to temperature fluctuations in the production environment, maintaining the accuracy and applicability of control parameters at all times.

[0065] Traditional parameter matching methods rely on pre-established discrete databases. When the measured material property parameters fall between sample points in the database, matching can only be achieved through interpolation or by selecting the closest sample point, resulting in limited accuracy. This embodiment introduces machine learning technology to establish a continuous mapping from material property parameters to optimal impedance control model parameters, achieving more accurate parameter prediction. Predictive optimization module based on historical data: Establish an implantation process database to record the relaxation time constant, stiffness characterization parameters, actual implantation depth, stress relaxation completion time, and integration quality grade for each implantation operation; A regression-based machine learning algorithm was used to train historical data to establish a predictive model from material property parameters to optimal control parameters, and the accuracy was evaluated through cross-validation. When the validation error of the prediction model is lower than the preset threshold, when performing a new implantation task, the relaxation time constant and stiffness characterization parameters obtained from the pretest are input into the trained prediction model to generate recommended impedance control model parameters as a supplement to the preset database matching results. When the quality level of multiple consecutive implantations reaches the excellent standard, the corresponding material property parameters and control parameters are marked as high-confidence samples and given higher weights during model retraining.

[0066] Specifically, it is equipped with a dedicated data storage module that automatically records all data from each chip implantation operation after completion. The recorded data includes: relaxation time constant, hardness characterization parameters, and rubber surface temperature measured during the pre-testing phase; impedance control model parameters actually used during the implantation phase, including the target implantation depth threshold and key node data of the compression velocity curve; the actual implantation depth recorded during implantation, the time it takes for the chip to reach the target depth, and the actual waiting time when stress relaxation is complete; extrusion ring feature data obtained through structured light measurement during the quality inspection phase; and the final assessed bonding quality level.

[0067] Each record constitutes a complete sample, with the sample format being a multidimensional feature vector. A unique identifier and timestamp are assigned to each sample for easy subsequent retrieval and tracing. The database employs a relational or time-series database architecture, supporting efficient data writing and query operations.

[0068] As production continues, the number of samples accumulated in the database grows. When the number of samples reaches a preset training threshold, such as 100 samples, the first training of machine learning is automatically triggered.

[0069] A predictive model is built using regression-based machine learning algorithms, including support vector regression, random forest regression, gradient boosting decision trees, or multilayer perceptron neural networks. This embodiment preferably uses the random forest regression algorithm, which performs well on small to medium-sized datasets and exhibits strong resistance to overfitting and robustness to missing values.

[0070] The input features of the prediction model are the relaxation time constant, hardness characterization parameters, and rubber surface temperature. These three features are standardized before training. Specifically, for each feature in the training set, its mean and standard deviation are calculated; then, each sample value of that feature is subtracted from the mean and divided by the standard deviation to obtain the standardized feature value. The standardized feature value has a mean of zero and a standard deviation of one. For the validation set and new samples during prediction, the same standardization transformation is performed using the mean and standard deviation of the training set to ensure consistent feature distribution. The model's output targets are the key parameters of the impedance control model, including the target implantation depth threshold and release threshold. Since the compression velocity curve is multi-point time-series data, it is represented parametrically, for example, using three parameters: initial velocity, acceleration, and steady-state velocity, and thus incorporated into the model output.

[0071] The samples in the database were randomly divided into training and validation sets in an 8:2 ratio. The training set was used for model parameter learning, and the validation set was used to evaluate the model's generalization performance. A cross-validation strategy was adopted, in which the training set was further divided into five folds. In each iteration, four folds were selected for training, and the remaining fold was used for validation. This process was repeated five times, and the average validation error was taken as the model performance metric.

[0072] During the training of the prediction model, the changing trends of training error and validation error are monitored. When the validation error no longer decreases for several consecutive rounds, the model is considered to have converged, and training is stopped. After training is completed, the prediction error of the model on the independent validation set is calculated, including mean squared error and mean absolute error. When the prediction error does not exceed a preset threshold, for example, the prediction error for the target implantation depth threshold does not exceed 0.1 mm, and the prediction error for the release threshold does not exceed 0.5 seconds, the model is marked as usable and officially put into use.

[0073] In subsequent chip implantation, after completing pre-testing and obtaining the relaxation time constant, hardness characterization parameters, and rubber surface temperature, the discrete database is no longer directly queried for matching. Instead, these three features are input into the trained prediction model. The model calculates and outputs recommended impedance control model parameters in real time.

[0074] Simultaneously, both traditional database matching and machine learning prediction processes are executed to obtain two sets of impedance control model parameters. The differences between the two sets of parameters are compared, and the percentage deviation for each parameter is calculated. When the percentage deviation does not exceed a safety threshold, such as 10%, the machine learning prediction result is prioritized due to its higher accuracy. When the percentage deviation exceeds the safety threshold, it is determined that the current material properties may be outside the coverage of the training data. In this case, the traditional database matching result is used, and the sample is marked as an outlier for subsequent model optimization.

[0075] The quality of each implantation procedure is evaluated. When the integration quality of three or more consecutive implantation procedures reaches the excellent standard, the samples corresponding to these procedures are marked as high-confidence samples. High-confidence samples indicate that, under the conditions of the material properties, the impedance control model parameters output by the prediction model can stably achieve high-quality implantation results and have high reliability.

[0076] Assigning higher weights to high-confidence samples allows them to have a greater impact on model parameters during periodic retraining, thus guiding the model towards optimization in higher-quality regions. Setting retraining triggers can also be implemented, such as automatically retraining after every fifty new samples, or triggering retraining when the model's prediction error exceeds a threshold a certain number of times.

[0077] During training, newly accumulated samples are merged with existing samples, the training and validation sets are redefined, and the model training and validation processes are repeated. As the amount of data increases and high-confidence samples accumulate, the model's prediction accuracy gradually improves, and the system's adaptability and intelligence level continuously increase.

[0078] When multi-point implantation, temperature compensation, and predictive optimization are used in combination, the accuracy of material property parameters is first ensured through the temperature compensation mechanism, then high-precision impedance control model parameters are obtained through the predictive optimization module, and finally stress interference between adjacent implantation points is eliminated through the multi-point collaboration mechanism, ultimately achieving high-quality and high-efficiency batch implantation of multiple chips.

[0079] The data flow relationship between the various schemes is as follows: the corrected material property parameters output by the temperature compensation module are used as input features for the prediction and optimization module; the recommended impedance control model parameters output by the prediction and optimization module are further corrected in the multi-point collaboration module based on the stress influence coefficient; after the multi-point collaboration module performs the implantation operation and completes the quality inspection, it feeds back the entire process data to the database for the prediction and optimization module to continuously learn.

[0080] Through this multi-level and multi-dimensional collaborative optimization, it is possible to adapt to complex and ever-changing production environments, handle implantation tasks with different material properties, different temperature conditions, and different spatial layouts, and significantly improve the intelligence level and process stability of chip implantation operations.

[0081] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A chip implantation intelligent control method based on multi-stage data feedback, characterized in that, include: A pre-compression head is used to perform a loading-holding-unloading cycle at the target implantation position of the rubber green tire. The stress relaxation curve of the pressure change over time is collected. The time required for the stress to decay to the preset proportion of the peak value is calculated as the relaxation time constant, and the initial peak stress is used as the hardness characterization parameter. Based on the relaxation time constant and hardness characterization parameters, the corresponding impedance control model parameters are matched from the preset database, including at least the target impedance dynamic equilibrium depth threshold and the pressing speed curve. The impedance adaptive segmented control chip enters the rubber through the compression speed curve and calculates the impedance value in real time. When the slope of the impedance value change meets the preset conditions, it is determined that the impedance dynamic equilibrium depth has been reached. The decay rate of the rubber reaction force is calculated in real time. When the decay rate is lower than the release threshold set based on the relaxation time constant, it is determined that the stress relaxation is completed and the chip release action is executed. Structured light is used to acquire three-dimensional contour data of the chip and the surrounding rubber extrusion ring. The height uniformity, continuity and symmetry characteristics of the extrusion ring are analyzed to generate the bonding quality level. If the bonding quality level is lower than the preset standard, the parameters for the next implantation cycle are corrected.

2. The chip implantation intelligent control method based on multi-stage data feedback according to claim 1, characterized in that: The process of obtaining the relaxation time constant is as follows: control the pre-pressure head to press into the rubber surface at a constant rate to a preset depth and then stop and keep the position fixed; take the stopping moment as the starting point, continuously collect the real-time pressure data fed back by the pressure sensor, and construct the pressure decay curve with time. Extract the time span corresponding to when the pressure value in the attenuation curve drops to a preset percentage of the initial peak stress; The time span is labeled as a relaxation time constant.

3. The chip implantation intelligent control method based on multi-stage data feedback according to claim 1, characterized in that: The impedance-adaptive segmental implantation process specifically includes: The rapid approach phase is used to control the robotic arm to move at high acceleration to a preset safe height from the rubber surface. During this phase, the position loop gain is at its maximum, and impedance calculation is not performed. The contact sensing section is used to decelerate the robotic arm and monitor changes in the readings of the end effector force sensor. When the reading exceeds the contact threshold, the contact zero point is marked, and the system switches to impedance control mode. The impedance working section is used to adjust the end effector stiffness of the robotic arm from the zero contact point, based on the impedance control model and the deviation between the real-time contact force and the reference speed, until the impedance dynamic equilibrium depth is determined to be reached.

4. The chip implantation intelligent control method based on multi-stage data feedback according to claim 1, characterized in that: The process for determining the depth of impedance dynamic equilibrium is as follows: During the impedance operation segment, the mechanical impedance value at the current moment is calculated in real time, and the derivative of the mechanical impedance value with respect to the implantation depth is obtained to obtain the impedance change rate. The numerical characteristics of the impedance change rate are monitored in real time. When the impedance change rate changes from a linear and stable state to an exponentially increasing state, and the increase exceeds the preset matrix stiffness threshold, it is determined that the chip has passed through the surface rheological layer and contacted the dense rubber matrix, and the current Z-axis coordinate is immediately locked.

5. The chip implantation intelligent control method based on multi-stage data feedback according to claim 1, characterized in that: The specific execution process of the chip release action is as follows: Once stress relaxation is complete, a release command is sent to the end effector. During the initial micro-displacement phase of the robotic arm's retraction action, the vertical tension data is continuously monitored. If the tension data exceeds the preset adhesion safety threshold, the retraction is paused and high-frequency micro-vibration is performed to break the interface vacuum adhesion force through vibration. Once the tension data falls back to the safe range, the full-speed retraction action is performed.

6. The chip implantation intelligent control method based on multi-stage data feedback according to claim 1, characterized in that: The specific process of acquiring 3D contour data is as follows: Multiple sets of sinusoidal fringe gratings with different phase shifts are projected onto the chip and its surrounding area; a sequence of deformed fringe images modulated by the height of the object surface is acquired simultaneously; phase demodulation is performed on the image sequence to obtain a folded phase map; the folded phase map is unfolded using the multi-frequency heterodyne principle to obtain a continuous phase distribution; the continuous phase distribution is converted into point cloud data containing spatial coordinates according to the phase-height mapping relationship; and the annular region data located outside the chip edge is segmented from the point cloud data as three-dimensional contour data.

7. The chip implantation intelligent control method based on multi-stage data feedback according to claim 1, characterized in that: The process for determining quality levels specifically includes continuity analysis, uniformity analysis, and symmetry analysis; specifically: The gradient change of height data is calculated along the circumferential path of the extrusion ring, the number of gradient breakpoints is counted, and a continuity score is generated; the standard deviation of the height values ​​of all point cloud values ​​in the extrusion ring region is calculated, and a uniformity score is generated; the vector deviation between the geometric center of the extrusion ring and the physical center of the chip is calculated, and a symmetry score is generated; the continuity score, uniformity score, and symmetry score are input into a preset weighted evaluation model, and the corresponding combination quality level is output.

Images