Digital visual control method for food production line control system

Abstract

The invention relates to the technical field of industrial process control and digital visualization, in particular to a digital visualization control method for a food production line control system. The method comprises the following steps that the surface hardness parameter of food to be grabbed is obtained, and a contact sensitivity triggering threshold value is set according to the surface hardness parameter; the flexible air bag gripper is driven to descend to make contact with food, and signals collected by the flexible touch skin are converted into visual contact dynamic pressure distribution data; virtual package fitting degree simulation verification is carried out on the contact dynamic pressure distribution data and a contact sensitivity triggering threshold value in a digital visualization background, and a self-adaptive differential pressure adjusting signal is generated; inflation and deflation linkage of the independent miniature air chamber is triggered based on the self-adaptive differential pressure adjusting signal. According to the invention, the independent air chamber is driven by flexible tactile feedback to carry out self-adaptive differential pressure regulation and full-flow digital visual monitoring, so that lossless accurate grabbing of vulnerable food with irregular shapes and transparent quality management in the production process are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of industrial process control and digital visualization, and particularly relates to a digital visualization control method for a food production line control system. BACKGROUND

[0002] Traditional food production line control systems mostly adopt "open loop" design or closed loop control based on simple unit feedback, resulting in the grabbing process becoming a "black box" of the system. The control system cannot obtain the microscopic force state of the contact surface between the gripper and the food, and neither the operator nor the system can quantitatively evaluate the grabbing quality in real time. Only passive inspection of the damage rate of the final product can be carried out, and process observability is lacking. At the same time, the existing control system generally adopts a control strategy with preset fixed parameters or simple threshold triggering. Such rigid control logic cannot adapt to the inherent high non-standard characteristics of food (such as different shapes and uneven hardness). Fixed control parameters result in poor system robustness, and the convex region is easily damaged due to overload extrusion, and the concave region is adsorbed due to control dead zone. More importantly, under dynamic disturbance conditions such as high-speed motion of the mechanical arm or vibration of the production line, the existing system lacks real-time online correction ability. It cannot identify the trend of small sliding or high-frequency oscillation, and usually triggers an alarm when the material is about to fall or has already fallen, lacking active inhibition and dynamic compensation ability in the process.

[0003] In summary, the existing technology has the problems of opaque control process, poor parameter adaptability and lack of dynamic compensation mechanism, which need to be solved. SUMMARY

[0004] Therefore, it is necessary to provide a digital visualization control method for a food production line control system to solve at least one of the above technical problems.

[0005] To achieve the above purpose, a digital visualization control method for a food production line control system comprises the following steps: Step S1: obtaining the surface hardness parameter of the food to be grabbed, and setting the contact sensitivity trigger threshold according to the surface hardness parameter; Step S2: driving the flexible air bag gripper to descend and contact the food, and converting the signal collected by the flexible tactile skin into visual contact dynamic pressure distribution data; Step S3: performing virtual package fit degree simulation verification on the contact dynamic pressure distribution data and the contact sensitivity trigger threshold in the digital visualization background, and generating an adaptive differential pressure adjustment signal; triggering the independent micro air chamber inside the flexible air bag gripper to charge and discharge gas based on the adaptive differential pressure adjustment signal; Step S4: when the digital visualization background determines that the contact dynamic pressure distribution data reaches the preset steady-state grabbing threshold, the grabbing lifting action is performed, and a three-dimensional pressure map is generated synchronously; when the digital visualization background determines that the contact dynamic pressure distribution data does not reach the preset steady-state grabbing threshold, the height of the flexible air bag gripper is maintained, and the differential pressure regulation is continuously performed, and a real-time pressure distribution state is output; and grabbing visualization management is completed according to the three-dimensional pressure map and the real-time pressure distribution state.

[0006] The application sets the contact sensitivity by obtaining the hardness parameter of the food surface, realizes zero-impact contact in combination with laser ranging, effectively avoids the initial damage risk of the fragile food, constructs the digital topographic features by using the flexible tactile skin, drives the independent micro air chamber to perform the partition differential pressure regulation based on the "peak filling and valley filling" strategy, realizes the adaptive and accurate fitting of the non-standard food surface topology, solves the problems of local overpressure and adsorption dead zone, further realizes the active inhibition and sliding prevention of the dynamic disturbance through the spectrum analysis under the non-steady state and the reverse pneumatic damping compensation, and the torque deviation grading pressure increasing correction in the lifting process, finally converts the physical interaction into the real-time three-dimensional pressure cloud map and the confidence score, realizes the change from "blind state" operation to full transparency and quantitative quality monitoring, and significantly improves the yield rate and process controllability of the production line. BRIEF DESCRIPTION OF DRAWINGS

[0007] Fig. 1 It is a step flowchart of a digital visualization control method for a food production line control system; Fig. 2 It is a self-adaptive differential pressure regulation control logic flowchart in the application. Fig. 3 It is a structure schematic diagram of the food production line control system in the application. DETAILED DESCRIPTION

[0008] The technical method of the application will be described clearly and completely in combination with the drawings. Obviously, the described embodiments are part of the embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor belong to the protection scope of the application.

[0009] In addition, the drawings are only schematic illustrations of the application, and are not necessarily drawn to scale. The same reference signs in the drawings represent the same or similar parts, and thus repeated descriptions thereof will be omitted. Some block diagrams shown in the drawings are functional entities, which do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in the form of software, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0010] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0011] To achieve the above objectives, please refer to Figs. 1 to 3 This invention provides a digital visualization control method for a food production line control system, comprising the following steps: Step S1: Obtain the surface hardness parameters of the food to be grasped, and set the contact sensitivity trigger threshold according to the surface hardness parameters; In this embodiment of the invention, the surface hardness parameters of the food to be grasped are obtained by database retrieval or online hardness detection. The system sets the contact sensitivity trigger threshold based on the linear mapping of the hardness value. For foods with low hardness, a lower threshold is set to improve the sensing sensitivity.

[0012] Step S2: Drive the flexible airbag gripper to descend and contact the food, and convert the signals collected by the flexible tactile skin into visualized dynamic contact pressure distribution data; In this embodiment of the invention, the driving robotic arm performs graded deceleration descent based on laser ranging data until the contact speed drops to zero impact speed. Simultaneously, voltage signals are collected through flexible tactile skin and baseline calibration is performed in combination with background cavity pressure. If overpressure is detected, the pressure is released instantaneously, and finally, the calibrated contact dynamic pressure distribution data is output.

[0013] Step S3: In the digital visualization background, the contact dynamic pressure distribution data and the contact sensitivity trigger threshold are used to perform virtual wrapping fit simulation verification, and an adaptive differential pressure adjustment signal is generated; based on the adaptive differential pressure adjustment signal, the independent micro air chambers inside the flexible airbag gripper are triggered to inflate and deflate in linkage. In this embodiment of the invention, the flexible airbag gripper has independent micro-air chambers corresponding to the tactile skin area. The pressure data is mapped to a local grid and the pressure gradient is calculated. The raised and recessed areas are identified through digital topography reconstruction. Based on this, targeted exhaust or suction commands are generated to drive the independent micro-air chambers to perform zoned pressure reshaping. The pressure distribution uniformity is balanced using a standard deviation iterative algorithm.

[0014] Step S4: When the digital visualization backend determines that the contact dynamic pressure distribution data has reached the preset steady-state grasping threshold, the grasping and lifting action is executed, and a three-dimensional pressure map is generated simultaneously; when the digital visualization backend determines that the contact dynamic pressure distribution data has not reached the preset steady-state grasping threshold, the height of the flexible airbag gripper is maintained and differential pressure adjustment is continuously executed, and the real-time pressure distribution status is output; the grasping visualization management is completed based on the three-dimensional pressure map and the real-time pressure distribution status. In this embodiment of the invention, the position of the robotic arm is locked when the pressure distribution data has not reached a steady state. Based on spectrum analysis, a reverse pneumatic pulse is generated to actively cancel the oscillation. After the data reaches the steady-state grasping threshold, the lifting is performed. Simultaneously, a color-mapped three-dimensional pressure map is generated, and the slip torque is detected and graded pressurization is corrected. Finally, the three-dimensional pressure map and confidence score are superimposed and displayed on the visualization terminal to complete the full process monitoring.

[0015] Of particular importance is that, in step S2, the process of driving the flexible airbag gripper to descend and contact the food specifically involves: The robotic arm uses a built-in laser ranging device to obtain real-time vertical distance data between the bottom of the gripper and the food surface. The preset velocity decay mapping table is invoked, and the real-time vertical distance data is substituted into the velocity decay mapping table for calculation to generate a dynamic descent speed command. The robotic arm's Z-axis motor is driven by a dynamic descent speed command, causing the descent rate to decrease in stages as the real-time vertical distance data decreases, until the speed drops to a preset zero-impact speed at the moment of contact.

[0016] In this embodiment, an industrial-grade laser displacement sensor is installed at the end of the robotic arm. This sensor emits a laser beam vertically downwards at a frequency of 1000Hz. When the laser beam illuminates the food surface on the conveyor belt, the reflected light is received by the sensor, and the vertical distance between the center point of the gripper's bottom surface and the highest point of the food surface is calculated in real time using triangulation. The system's built-in velocity decay mapping table stores the non-linear relationship between vertical distance and target descent velocity. This mapping table is pre-set according to the fragility level of the food, dividing the vertical distance into three segments: a rapid approach zone, a deceleration buffer zone, and a micro-contact zone.

[0017] In one implementation of this embodiment, when When the distance is greater than 100mm, it is in the rapid approach zone, and the mapping table outputs a constant maximum descent speed. At 500mm / s, the robotic arm is driven to quickly shorten the distance; when When the speed is between 10mm and 100mm, it is in the deceleration buffer zone, and the speed output by the mapping table is... It follows a nonlinear decay function, the formula of which is: ; in The attenuation coefficient is set to 50 to ensure a smooth decrease in speed as the distance decreases; when When the distance is less than 10mm, it is within the micro-touch contact zone, and the mapping table outputs a constant micro-touch speed. It is 5 mm / s.

[0018] In another implementation of this embodiment, the control system reads at a period of 1ms. and look up the table to obtain The system generates a dynamic descent speed command and sends it to the Z-axis servo drive. The servo drive employs a position-speed dual closed-loop control mode, adjusting the Z-axis motor speed in real time. This is achieved the instant the gripper contacts the food surface. When the descent rate drops to 0 mm, the current descent rate has precisely converged to... This achieves a zero-impact soft landing, avoiding mechanical damage to the food surface caused by inertial impact. Simultaneously, the laser sensor continuously monitors distance changes; once it detects that the distance is no longer decreasing, it determines that contact has occurred and immediately triggers the subsequent pressure acquisition process.

[0019] Preferably, in step S2, converting the signals collected by the flexible tactile skin into visualized dynamic contact pressure distribution data specifically involves: The contact point voltage signal is collected by a miniature air pressure sensor array, and the background cavity pressure data is collected simultaneously by the cavity air pressure monitoring component inside the flexible airbag gripper. The contact point voltage signal is digitally baseline-calibrated using background cavity pressure data to generate calibrated contact pressure values. Obtain the preset pressure safety range, monitor the contact pressure value in the visual data stream, and when the monitored contact pressure exceeds the pressure safety range, generate an overpressure visual alarm and perform instantaneous pressure relief step operation through the independent micro air chamber in the corresponding area. After each pressure relief step operation is completed, the contact point voltage signal is re-acquired and calibrated until the contact pressure value falls into the safe pressure range, and the adjusted contact dynamic pressure distribution data is output.

[0020] In this embodiment, the flexible tactile skin consists of an array of 64×64 miniature piezoresistive sensor units, which output analog voltage signals at each point in parallel at a sampling rate of 500Hz. An analog-to-digital converter quantizes the analog voltage into a 12-bit digital signal. ,in , These represent the row and column numbers of the sensor in the array, respectively. Simultaneously, a high-precision pressure transmitter located in the main air path of the gripper collects the baseline inflation pressure inside the airbag in real time and outputs the background chamber pressure data. The unit is kPa.

[0021] In one implementation of this embodiment, the digital baseline calibration process eliminates the interference of airbag pre-inflation on tactile measurements using a differential method. The calculation formula is as follows: ; in This is the actual contact pressure value after calibration. The sensitivity coefficient of the sensor is set to 0.5 kPa / mV. To the current background cavity pressure The zero-point voltage reference value when the sensor is not in contact with an object is obtained by consulting a pre-calibrated reference voltage lookup table.

[0022] In another implementation of this embodiment, the system's preset pressure safety range is [0.5 kPa, 5.0 kPa]. Control traverses all... Once the pressure value at any point exceeds 5.0 kPa, a bright red flashing mark is immediately superimposed on the corresponding visual pixel, forming an overpressure visual alarm. Simultaneously, the system locates the individual micro-chamber number of the overpressure point and drives the corresponding high-speed solenoid valve to open for 2 ms to perform a small amount of venting. Venting causes the local air bladder to collapse, thereby reducing the pressure on the food surface. 10 ms after the venting action ends, the system reads the voltage signal at that point again and calculates the pressure value. If it is still higher than 5.0 kPa, the venting step operation is repeated until the pressure value drops below 5.0 kPa, ensuring that the pressure distribution data output is within a safe range each time.

[0023] Preferably, before performing virtual wrapping fit simulation verification on the contact dynamic pressure distribution data and contact sensitivity trigger threshold in the digital visualization background in step S3, the following steps are also included: The contact dynamic pressure distribution data is mapped into several local force grids according to the preset arrangement rules in the flexible tactile skin; Calculate the pressure gradient value of the pressure values ​​in each local force grid in the virtual mapping layer, and perform an independent verification operation between the pressure gradient value and the contact sensitivity trigger threshold. The number of locally stressed grids that pass the statistical verification is counted, and a preset effective contact area threshold is obtained. Only when the number of consecutive adjacent grids that pass the verification exceeds the effective contact area threshold is the adaptive differential pressure adjustment signal generation command triggered to generate the adaptive differential pressure adjustment signal.

[0024] In this embodiment, the flexible tactile skin physically contains 4096 sensor points. To improve computational efficiency, the system aggregates data according to a 4×4 sensor specification, and the physical surface is divided into 256 local stress grids. Pressure values ​​within each grid cell The arithmetic mean of the pressure values ​​at the 16 sensor points covered by the grid is taken. This mapping process is completed in the preprocessing module of the digital visualization backend, constructing a low-resolution but high signal-to-noise ratio virtual mapping layer.

[0025] In one implementation of this embodiment, the system calculates the pressure difference between each local stress grid and its eight neighboring grids, and takes the absolute value of the maximum difference as the pressure gradient value of that grid. Contact sensitivity trigger threshold The threshold is set to 0.2 kPa, representing the minimum rate of pressure change required for effective contact between the gripper and the food. The logic comparison unit scans all... ,like Greater than If the condition is met, the grid is determined to be in an effective stress state, and its status bit is marked as 1; otherwise, it is marked as 0.

[0026] In another implementation of this embodiment, the system uses a connected component labeling method to scan the state bit matrix and counts all spatially adjacent grid groups marked as 1. A preset effective contact area threshold is also included. The grid is set to 12 grid units, corresponding to a physical contact area of ​​approximately 3 cm². The maximum number of grid units contained in the largest connected grid group is then determined. Greater than At this point, it is determined that the gripper has formed an effective covering contact with the food, providing a basis for adjustment. At this time, the central processing unit sends a high-level enable signal, allowing the adaptive differential pressure regulation method to intervene in the control; conversely, if... Less than If the contact area is insufficient or it is merely a false alarm, the system will maintain the current air pressure and will not trigger any adjustment action to prevent air path oscillation caused by false adjustment.

[0027] Preferably, step S3 involves triggering the independent micro-cell inflation / deflation linkage based on the adaptive differential pressure adjustment signal, including identifying morphological features and switching the adjustment mode: Based on the adaptive differential pressure regulation signal, digital mesh reconstruction is performed on each local stress mesh to form a digital topography feature reconstruction result, including the reconstructed local stress mesh and pressure gradient value distribution data structure; In the reconstructed local stress grid and pressure gradient value distribution data structure, when the pressure at the center of the grid is greater than the pressure at the edge and the pressure gradient value is positive, it is marked as a digital convex feature region; when the pressure at the center of the grid is less than the pressure at the edge and is close to zero, it is marked as a digital concave feature region. The exhaust control parameters are applied to the independent micro-chambers mapped to the digital raised feature regions, and the suction control parameters are applied to the independent micro-chambers mapped to the digital recessed feature regions. The system automatically switches the on / off state of the gas path based on the reconstruction results of digital topography features, and performs zone pressure reshaping operations.

[0028] In this embodiment, the digital mesh reconstruction process smooths the pressure data of the original 256 meshes using a bicubic interpolation method to generate a high-resolution pressure surface model. This model simultaneously includes the pressure scalar value P and the pressure gradient vector G for each mesh point. The data structure is stored in video memory as a three-dimensional array, containing coordinates, pressure values, and gradient directions.

[0029] In one implementation of this embodiment, the feature recognition method traverses the reconstructed data structure. For any grid point... The system compares its pressure values. The average pressure value of the edge points within a 3×3 neighborhood centered on it. .like And gradient vector The direction of the gradient diverges outward from the center (defined as a positive gradient). This determines if the point is located at a peak or edge on the food surface, and its attribute is marked as a "protruding feature." If... ,and If the value is less than 0.1 kPa (the threshold close to zero), the point is determined to be located in a surface pit or non-contact area, and its attribute bit is marked as "pit feature".

[0030] In another implementation of this embodiment, the control system reads the grid coordinates of all grids marked as "protruding features," retrieves their corresponding independent micro-air chamber IDs, and calls the exhaust control parameters: opening the exhaust valve and setting the exhaust time to 50ms to quickly reduce the pressure inside the airbag in that area. Simultaneously, the system reads the grid coordinates of all grids marked as "depressed features" and their corresponding air chamber IDs, and calls the suction control parameters: opening the vacuum valve, setting the suction negative pressure to -20kPa for a duration of 100ms, forcing the flexible skin in that area to indent.

[0031] It should be noted that, based on the digital topography feature reconstruction results, the air circuit controller outputs multiple switching signals in parallel through the FPGA to drive the solenoid valve groups of each independent air chamber. The air chambers in the raised areas perform pressure relief to eliminate excessive compression on the food tip; the air chambers in the recessed areas perform adsorption to establish a local negative pressure connection. This zoned pressure reshaping operation is completed within 200ms, enabling precise complementary engagement between the physical deformation of the gripper's inner surface and the topological features of the food's outer surface.

[0032] Preferably, performing the partitioned pressure reshaping operation further includes iterative balancing: After executing the exhaust control parameters or suction control parameters, pressure data is collected and fed back through the flexible tactile skin; Calculate the average pressure value of each grid pressure value in the feedback pressure data, and calculate the pressure difference between each grid pressure value and the average pressure value; Generate a dispersion distribution map of the feedback pressure data, calculate the standard deviation of the dispersion distribution map, and determine whether the standard deviation is less than the normalization threshold by combining it with the preset normalization threshold. If the standard deviation is greater than the homogenization threshold, the target adjustment amount is obtained by multiplying the absolute value of the pressure difference with the step adjustment coefficient using the preset step adjustment coefficient. For grid regions with positive pressure differentials, the corresponding independent micro-chambers are controlled to perform pressure reduction operations according to the target adjustment amount; For grid regions with negative pressure differentials, the corresponding independent micro-chambers are controlled to perform negative pressure enhancement operations according to the target adjustment amount; Refresh the dispersion distribution map and calculate the standard deviation until the pressure distribution meets the homogenization condition.

[0033] In this embodiment, 10ms after the initial partition pressure reshaping is completed, the system scans the pressure values ​​of all grids again using a flexible tactile skin. The central processing unit quickly calculates the arithmetic mean of the pressures of all effective contact meshes. Then, the pressure deviation for each grid was calculated. The system build includes all The dispersion distribution of the values ​​is shown in the histogram stored in memory. The horizontal axis represents the pressure deviation range, and the vertical axis represents the frequency.

[0034] In one implementation of this embodiment, the standard deviation of the distribution plot is calculated. Preset homogenization threshold The threshold is set to 0.5 kPa, defining the maximum allowable non-uniformity of the gripping force field. (Logic judgment unit comparison) and ,like If the pressure distribution has reached equilibrium, stop adjusting; Initiate the incremental correction process. Step adjustment coefficient. Set to 0.8ms / kPa, this coefficient defines the proportional relationship between pressure deviation and valve action time.

[0035] In another implementation of this embodiment, for The positive deviation grid (i.e., the region of excessive stress) is used to calculate the target exhaust time. Drive the corresponding air chamber exhaust valve to open Duration; for The negative deviation grid (i.e., the region of insufficient force) is used to calculate the target pumping time. This drives the corresponding chamber vacuum valve to open. Duration.

[0036] It should be noted that after the above adjustments are completed, the system immediately refreshes the data collection. And recalculate The process then proceeds to the next iteration. This iteration cycle runs at a frequency of 50Hz, and typically after 3-5 iterations, the standard deviation is... When the pressure converges to below 0.5 kPa, the dispersion distribution plot shows a narrow peak normal distribution centered at 0, indicating that an isotropic uniform contact force field has been established between the gripper and the food.

[0037] Preferably, in step S4, when the digital visualization backend determines that the contact dynamic pressure distribution data has not reached the preset steady-state grasping threshold, maintaining the height of the flexible airbag gripper and continuously performing differential pressure adjustment specifically involves: When the dynamic pressure distribution data does not reach the preset steady-state capture threshold, the non-steady-state area is marked on the visualization interface and a position servo lock command is generated. Using position servo locking commands, the three-dimensional spatial coordinates of the robotic arm are fixed, and digital vibration pressure waveforms in the locked state are simultaneously acquired through flexible tactile skin in a high-frequency sampling mode. Spectral features were extracted from the digital jitter pressure waveform to identify the jitter frequency component with the highest energy in the digital jitter pressure waveform. Calculate the real-time phase angle and amplitude information of the jitter frequency components, and generate a cancellation control signal with opposite phase and matching amplitude based on the real-time phase angle; The independent micro-air chambers are driven by the cancellation control signal to generate reverse pneumatic pulses, which actively cancel the oscillations on the airbag surface.

[0038] In this embodiment, the system calculates the coefficient of variation of the pressure values ​​at all current contact points in real time. The preset steady-state capture threshold is set to... If in real time If the value is greater than 0.1, the grasping is determined to be unstable, and the visualization backend immediately marks the area with pressure fluctuation exceeding ±10% as a flashing yellow block. At the same time, the motion controller sends a position holding command to the servo drive of the six-axis robotic arm, so that the end effector of the robotic arm is completely stationary in the XYZ coordinate system and enters the hovering mode.

[0039] In one implementation of this embodiment, the flexible tactile skin switches to a 2kHz high-frequency sampling mode to continuously acquire 200ms of pressure timing data, forming a digital vibration pressure waveform. Digital signal processors (DSPs) Perform a Fast Fourier Transform (FFT) to obtain the spectrum. The system scans the spectrum, searching for the frequency point with the highest power spectral density in the range of 1Hz to 100Hz, and identifies this as the main jitter frequency. This frequency typically corresponds to the natural frequency of the robotic arm's end effector or the resonant frequency of the environment.

[0040] In another implementation of this embodiment, the system uses the Hilbert transform to solve the problem. Instantaneous phase of the component and amplitude The control method generates a cancellation control signal. This signal, via a pneumatic proportional valve controller, drives the corresponding independent micro-chamber. The chamber performs a high-frequency inflation / deflation operation, generating a frequency of... However, a reverse aerodynamic pulse with a phase lag of 180 degrees is generated. The tiny aerodynamic force generated by the reverse pulse is superimposed on the physical vibration force at the end of the robotic arm. Utilizing the principle of wave interference, the amplitude of the physical vibration on the gripper surface is attenuated to less than 5% of the original amplitude, thus achieving active vibration suppression.

[0041] Preferably, step S4, which involves performing the grasping and lifting action, further includes slippage detection and graded pressure increase correction: Real-time load torque data during the lifting process is collected by torque sensors at the joints of the robotic arm; Obtain the pre-stored food standard gravity torque, compare the real-time load torque data with the food standard gravity torque to generate a load holding deviation value; When the load holding deviation is greater than the preset slip compensation threshold and less than the preset detachment alarm threshold, it is determined to be a compensable slip state, where the detachment alarm threshold is greater than the slip compensation threshold. When the state is determined to be compensable slip, the secondary pressurization threshold of the independent micro-chamber is activated, and a staged air replenishment operation is performed. After each stage of air replenishment, the real-time load torque data is re-acquired for comparison until the load deviation value returns to zero.

[0042] In this embodiment, the fourth joint (wrist pitch axis) of the robotic arm integrates a high-sensitivity strain gauge torque sensor to monitor torque changes in the vertical direction at a frequency of 500Hz. The system database pre-stores the standard weight W_std of various food products and their corresponding standard gravitational torques. ,in This represents the current boom span. During the lifting process, the sensor outputs the load torque in real time. The system calculates the load holding deviation value. .

[0043] In one implementation of this embodiment, a preset slip compensation threshold is used. Set as The 5% threshold, meaning when a torque attenuation exceeding 5% is detected, indicates that the food has experienced slight displacement or loosening; this is the detachment alarm threshold. Set as When the torque decreases by more than 20%, it indicates that the food is about to or has already detached from the gripper. The logic comparison unit determines the interval of ΔT in real time: if... If so, the compensable slip state flag will be triggered.

[0044] In another implementation of this embodiment, once the flag is triggered, the gas path control system immediately invokes the secondary pressurization strategy. At this time, the base pressure setpoint of all independent micro-chambers is increased from the primary threshold. Raise to Level 2 threshold The pressurization is performed via a stepped pulse method, with each pulse width being 50ms. After each pressurization, the pressure in the gas chamber increases by 0.5kPa. During the 10ms stabilization period following the end of each pulse stage, the system reads the pressure again. And calculate .like Still greater than Then execute the next level of boost pulse; if Down to The following indicates that the slippage has been corrected, pressurization has stopped and the current air pressure has been maintained, completing the closed-loop correction.

[0045] Preferably, the synchronous generation of the three-dimensional pressure map in step S4 specifically involves: Start the texture mapping process of the visualization rendering unit; Read the final pressure values ​​of each sensor on the flexible tactile skin at the steady-state grasping moment; Based on the preset color lookup table, the values ​​of the negative pressure adsorption region are mapped to cool color RGB values, and the values ​​of the positive pressure support region are mapped to warm color RGB values, thus forming the mapped RGB values; The mapped RGB values ​​are applied in real time to the digital geometric surface of the flexible airbag gripper to generate a three-dimensional pressure map that dynamically changes with the gripping force.

[0046] In this embodiment, the visualization rendering unit is developed based on the OpenGL graphics library and runs on the GPU of the industrial control computer. The system loads a high-precision triangular mesh model of the flexible airbag gripper, which contains vertex coordinate data corresponding one-to-one with the physical sensor array. Once the steady-state gripping determination is successful, the system reads the final pressure value stream from 4096 sensor points from the shared memory area.

[0047] In one implementation of this embodiment, the chromatographic lookup table (LUT) predefines the pressure values. A linear mapping relationship with RGB color values. For In the negative pressure adsorption region <0, the LUT maps the pressure range [-20kPa, 0kPa] to a cool color scheme that gradually changes from deep blue (0, 0, 139) to cyan (0, 255, 255); for For the positive pressure support area >0, the LUT maps the pressure range [0kPa, 10kPa] to a warm color scheme that gradually changes from yellow (255, 255, 0) to dark red (139, 0, 0); for the uncontacted area P=0, it is mapped to a translucent gray (128, 128, 128, 0.5).

[0048] In another implementation of this embodiment, the rendering engine writes the calculated RGB color values ​​into the texture buffer of the gripper model. Using UV mapping technology, the color values ​​are precisely overlaid onto the corresponding geometric surfaces of the model as texels. For the blank areas between sensor points, bilinear interpolation is used for smooth color transitions. As the GPU refreshes the video memory at a frequency of 60 frames per second, a colorful and real-time pulsating 3D pressure cloud map is displayed on the screen, intuitively showing the stress state at every point on the gripper surface.

[0049] Preferably, step S4, which involves completing the capture visualization management based on the three-dimensional pressure map and real-time pressure distribution status, includes: When the robotic arm is in a moving and rotating state, the rendering data frames of the 3D pressure map are continuously refreshed; Obtain a preset standard grasp fingerprint map, calculate the image similarity between the 3D stress map and the standard grasp fingerprint map, and generate a grasp quality confidence score. The grasping quality confidence score is transmitted to the visualization control terminal. In the production line layout diagram of the visualization control terminal, the 3D pressure map and confidence score are superimposed on the corresponding robotic arm position as floating labels, and the label border color changes according to the score.

[0050] In this embodiment, the system database stores a standard gripping fingerprint map for each food SKU. This map is a baseline pressure distribution image matrix M_std generated under ideal laboratory conditions when a standard-shaped food is perfectly gripped. As the robotic arm carries the food between different workstations on the production line, the background calculation module extracts the two-dimensional projection matrix of the current three-dimensional pressure map at a frequency of 10Hz. .

[0051] In one implementation of this embodiment, the image similarity can be calculated using the Normalized Cross-Correlation (NCC) algorithm to obtain the capture quality confidence score. The value range is [0,1]. The crawl quality confidence score quantifies the degree of agreement between the current actual crawl state and the ideal state.

[0052] In another implementation of this embodiment, the visualization control terminal loads a WebGL-based digital twin factory scene. The front-end program receives the data via the WebSocket protocol. The data stream includes the values ​​and compressed 3D pressure map data. Above the corresponding robotic arm coordinates on the virtual production line, a moving HTML5 floating label component is rendered. This component internally draws and displays a real-time thumbnail of the 3D pressure map. Percentage value. Logical judgment script monitoring. Value: If The label border is rendered in green, indicating excellent crawling performance; if The border is rendered in yellow, indicating that the capture is acceptable but has some deviation; if The border is rendered in red and accompanied by a flashing animation to indicate a risk of falling, allowing inspection personnel to make immediate decisions.

[0053] Preferably, the method further includes a visual monitoring operation for residual pressure elimination during the release phase: When the robotic arm reaches the target placement position, it performs a full-cavity positive pressure injection operation through an independent micro air chamber; Real-time pressure attenuation visualization curves are generated using flexible tactile skin; When a separation abrupt change point is detected in the pressure decay curve, it is determined that the food has detached, the positive pressure injection is stopped, and the robotic arm is reset.

[0054] In this embodiment, after the robotic arm motion controller sends a positioning signal, the pneumatic system switches to release mode. The control valves of all independent micro-air chambers are synchronously connected to a positive pressure air source, with the air source pressure set to 5 kPa. Compressed air is injected into the airbag, rapidly disrupting the previous negative pressure adsorption state and generating an outward thrust.

[0055] In one implementation of this embodiment, a separate monitoring window pops up in the visualization interface, with the horizontal axis of the coordinate system representing time. (Unit: ms), the vertical axis represents the average contact pressure of flexible tactile skin. (Unit: kPa). The system samples and plots at a frequency of 1 kHz. The trajectory changing over time forms a real-time pressure decay visualization curve. In the initial stage of positive pressure injection, It shows a linear downward trend.

[0056] In another implementation of this embodiment, the release logic method calculates the second derivative of the curve in real time. The instant the food physically separates from the gripper surface, the contact reaction force suddenly disappears. A step drop will occur, causing the second derivative to have an amplitude exceeding a threshold. A negative spike (set to 10 kPa / ms²) was observed. The system marked this moment as the point of separation abruptly. Once detected The control system immediately shuts off the positive pressure air source solenoid valve, stops the air injection, and sends a reset command to the robotic arm. The robotic arm then quickly raises its height and returns to its initial position, completing a single work cycle.

[0057] Of particular importance is that step S4, during the grasping and lifting action and subsequent movement, also includes an active inertial compensation operation: Real-time acceleration vector of the robotic arm end effector is acquired. Based on the direction of the real-time acceleration vector, the independent micro-air chambers of the flexible airbag gripper are divided into an inertial release side air chamber located in front of the direction of motion and an inertial compression side air chamber located behind the direction of motion. The preset inertial compensation gain table is called, and the magnitude of the real-time acceleration vector is substituted into the inertial compensation gain table for retrieval to obtain the corresponding dynamic compensation air pressure value. The inertial release side air chamber is pressurized by adding a dynamic compensation pressure value to the current air pressure, while the inertial compression side air chamber is depressurized by subtracting the dynamic compensation pressure value from the current air pressure, in order to maintain the force balance of the food during the variable speed movement.

[0058] In this embodiment, a six-axis IMU (Inertial Measurement Unit) is rigidly connected to the flange at the end of the robotic arm, outputting the three-axis acceleration vector of the end in the world coordinate system at a frequency of 1000Hz. The control method first filters out the gravitational component and extracts the pure motion acceleration. The system calculates the projection vector of 'a' onto the horizontal plane in real time. and its direction angle .

[0059] In one implementation of this embodiment, the flexible airbag gripper has a circular symmetrical structure, with independent micro-air chambers distributed circumferentially inside, each air chamber corresponding to an azimuth angle. The system is based on Dynamically divide all air chambers into two: satisfy The air chamber is defined as the inertial detachment side air chamber (i.e., the side where the food tends to detach), satisfying... The air chamber is defined as the inertial compression side air chamber (i.e., one side of the food compression air bladder).

[0060] In another implementation of this embodiment, the system calculates the magnitude |a| of the real-time acceleration vector. A preset inertial compensation gain table stores |a| and the dynamic compensation air pressure value. The nonlinear relationship. When When <1m / s², =0; when 1m / s²≤ When ≤5m / s², Among them, the inertial gain coefficient The target pressure is set to 0.5 kPa / (m / s²). The control unit sends a pressurization command to the inertial separation side chamber, with the target pressure set at [value missing]. To enhance the gripping force on that side and resist detachment; simultaneously, a depressurization command is sent to the inertial compression side air chamber, with the target pressure set at [value missing]. This softens the airbag on that side to absorb the impact force. The response time of this compensation action is less than 5ms, ensuring that the food does not shift relative to the center of the gripper.

[0061] Please see Fig. 2 This is a schematic diagram of the adaptive differential pressure regulation control logic in this invention. The process starts with "pressure acquisition" to obtain pressure data between the gripper and the food. After "shape recognition" analyzes the shape characteristics of the food, "venting" or "suction" operations are performed to adjust the pressure of the gripper's independent air chamber. Then, "standard deviation judgment" verifies whether the pressure distribution is uniform and meets the standard. If it does not meet the requirements, "pressure acquisition" is performed again. If it meets the requirements, it enters "locked state" to complete the pressure adaptation of the gripper to the current food, providing a stable contact pressure guarantee for gripping.

[0062] Please see Fig. 3 The diagram shows the structure of the food production line control system of the present invention. The food production line control system includes a visual control terminal 10, a robotic arm 20 on the food production line, and a flexible airbag gripper 30 connected to the end of the robotic arm 20. The flexible airbag gripper 30 has a flexible tactile skin 40 with a micro pressure sensor array integrated on its surface.

[0063] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0064] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A digital visualization control method for a food production line control system, characterized in that, The method, applied to a food production line control system, includes a visual control terminal and a robotic arm on the food production line. The robotic arm's end is connected to a flexible airbag gripper, the surface of which integrates a flexible tactile skin with a miniature air pressure sensor array. The method comprises the following steps: Step S1: Obtain the surface hardness parameters of the food to be grasped, and set the contact sensitivity trigger threshold according to the surface hardness parameters; Step S2: Drive the flexible airbag gripper to descend and contact the food, and convert the signals collected by the flexible tactile skin into visualized dynamic contact pressure distribution data; Step S3: In the digital visualization background, the contact dynamic pressure distribution data and the contact sensitivity trigger threshold are used to perform virtual wrapping fit simulation verification, and an adaptive differential pressure adjustment signal is generated; based on the adaptive differential pressure adjustment signal, the independent micro air chambers inside the flexible airbag gripper are triggered to inflate and deflate in linkage. Step S4: When the digital visualization backend determines that the contact dynamic pressure distribution data has reached the preset steady-state grasping threshold, the grasping and lifting action is executed, and a three-dimensional pressure map is generated simultaneously; when the digital visualization backend determines that the contact dynamic pressure distribution data has not reached the preset steady-state grasping threshold, the height of the flexible airbag gripper is maintained and differential pressure adjustment is continuously executed, and the real-time pressure distribution status is output; the grasping visualization management is completed based on the three-dimensional pressure map and the real-time pressure distribution status.

2. The digital visualization control method for a food production line control system according to claim 1, characterized in that, In step S2, the signals collected by the flexible tactile skin are converted into visualized dynamic contact pressure distribution data, specifically as follows: The contact point voltage signal is collected by a miniature air pressure sensor array, and the background cavity pressure data is collected simultaneously by the cavity air pressure monitoring component inside the flexible airbag gripper. The contact point voltage signal is digitally baseline-calibrated using background cavity pressure data to generate calibrated contact pressure values. Obtain the preset pressure safety range, monitor the contact pressure value in the visual data stream, and when the monitored contact pressure exceeds the pressure safety range, generate an overpressure visual alarm and perform instantaneous pressure relief step operation through the independent micro air chamber in the corresponding area. After each pressure relief step operation is completed, the contact point voltage signal is re-acquired and calibrated until the contact pressure value falls into the safe pressure range, and the adjusted contact dynamic pressure distribution data is output.

3. The digital visualization control method for a food production line control system according to claim 1, characterized in that, Before performing virtual wrapping fit simulation verification using the contact dynamic pressure distribution data and contact sensitivity trigger threshold in the digital visualization background in step S3, the following steps are also included: The contact dynamic pressure distribution data is mapped into several local force grids according to the preset arrangement rules in the flexible tactile skin; Calculate the pressure gradient value of the pressure values ​​in each local force grid in the virtual mapping layer, and perform an independent verification operation between the pressure gradient value and the contact sensitivity trigger threshold. The number of locally stressed grids that pass the statistical verification is counted, and a preset effective contact area threshold is obtained. Only when the number of consecutive adjacent grids that pass the verification exceeds the effective contact area threshold is the adaptive differential pressure adjustment signal generation command triggered to generate the adaptive differential pressure adjustment signal.

4. The digital visualization control method for a food production line control system according to claim 3, characterized in that, Step S3 involves triggering the independent micro-cell inflation / deflation linkage based on the adaptive differential pressure adjustment signal, including identifying morphological features and switching the adjustment mode: Based on the adaptive differential pressure regulation signal, digital mesh reconstruction is performed on each local stress mesh to form a digital topography feature reconstruction result, including the reconstructed local stress mesh and pressure gradient value distribution data structure; In the reconstructed local stress grid and pressure gradient value distribution data structure, when the pressure at the center of the grid is greater than the pressure at the edge and the pressure gradient value is positive, it is marked as a digital convex feature region; when the pressure at the center of the grid is less than the pressure at the edge and is close to zero, it is marked as a digital concave feature region. The exhaust control parameters are applied to the independent micro-chambers mapped to the digital raised feature regions, and the suction control parameters are applied to the independent micro-chambers mapped to the digital recessed feature regions. The system automatically switches the on / off state of the gas path based on the reconstruction results of digital topography features, and performs zone pressure reshaping operations.

5. The digital visualization control method for a food production line control system according to claim 4, characterized in that, Performing partitioned stress reshaping operations also includes iterative balancing: After executing the exhaust control parameters or suction control parameters, pressure data is collected and fed back through the flexible tactile skin; Calculate the average pressure value of each grid pressure value in the feedback pressure data, and calculate the pressure difference between each grid pressure value and the average pressure value; Generate a dispersion distribution map of the feedback pressure data, calculate the standard deviation of the dispersion distribution map, and determine whether the standard deviation is less than the normalization threshold by combining it with the preset normalization threshold. If the standard deviation is greater than the homogenization threshold, the target adjustment amount is obtained by multiplying the absolute value of the pressure difference with the step adjustment coefficient using the preset step adjustment coefficient. For grid regions with positive pressure differentials, the corresponding independent micro-chambers are controlled to perform pressure reduction operations according to the target adjustment amount; For grid regions with negative pressure differentials, the corresponding independent micro-chambers are controlled to perform negative pressure enhancement operations according to the target adjustment amount; Refresh the dispersion distribution map and calculate the standard deviation until the pressure distribution meets the homogenization condition.

6. The digital visualization control method for a food production line control system according to claim 1, characterized in that, In step S4, when the digital visualization backend determines that the contact dynamic pressure distribution data has not reached the preset steady-state grasping threshold, maintaining the height of the flexible airbag gripper and continuously performing differential pressure adjustment specifically involves: When the dynamic pressure distribution data does not reach the preset steady-state capture threshold, the non-steady-state area is marked on the visualization interface and a position servo lock command is generated. Using position servo locking commands, the three-dimensional spatial coordinates of the robotic arm are fixed, and digital vibration pressure waveforms in the locked state are simultaneously acquired through flexible tactile skin in a high-frequency sampling mode. Spectral features were extracted from the digital jitter pressure waveform to identify the jitter frequency component with the highest energy in the digital jitter pressure waveform. Calculate the real-time phase angle and amplitude information of the jitter frequency components, and generate a cancellation control signal with opposite phase and matching amplitude based on the real-time phase angle; The independent micro-air chambers are driven by the cancellation control signal to generate reverse pneumatic pulses, which actively cancel the oscillations on the airbag surface.

7. The digital visualization control method for a food production line control system according to claim 1, characterized in that, Step S4 involves performing a gripping and lifting action, which also includes slip detection and graded pressure correction. Real-time load torque data during the lifting process is collected by torque sensors at the joints of the robotic arm; Obtain the pre-stored food standard gravity torque, compare the real-time load torque data with the food standard gravity torque to generate a load holding deviation value; When the load holding deviation is greater than the preset slip compensation threshold and less than the preset detachment alarm threshold, it is determined to be a compensable slip state, where the detachment alarm threshold is greater than the slip compensation threshold. When the state is determined to be compensable slip, the secondary pressurization threshold of the independent micro-chamber is activated, and a staged air replenishment operation is performed. After each stage of air replenishment, the real-time load torque data is re-acquired for comparison until the load deviation value returns to zero.

8. The digital visualization control method for a food production line control system according to claim 1, characterized in that, The specific steps for generating the three-dimensional pressure map in step S4 are as follows: Start the texture mapping process of the visualization rendering unit; Read the final pressure values ​​of each sensor on the flexible tactile skin at the steady-state grasping moment; Based on the preset color lookup table, the values ​​of the negative pressure adsorption region are mapped to cool color RGB values, and the values ​​of the positive pressure support region are mapped to warm color RGB values, thus forming the mapped RGB values; The mapped RGB values ​​are applied in real time to the digital geometric surface of the flexible airbag gripper to generate a three-dimensional pressure map that dynamically changes with the gripping force.

9. The digital visualization control method for a food production line control system according to claim 8, characterized in that, Step S4 involves completing the capture and visualization management based on the 3D pressure map and real-time pressure distribution status, including: When the robotic arm is in a moving and rotating state, the rendering data frames of the 3D pressure map are continuously refreshed; Obtain a preset standard grasp fingerprint map, calculate the image similarity between the 3D stress map and the standard grasp fingerprint map, and generate a grasp quality confidence score. The grasping quality confidence score is transmitted to the visualization control terminal. In the production line layout diagram of the visualization control terminal, the 3D pressure map and confidence score are superimposed on the corresponding robotic arm position as floating labels, and the label border color changes according to the score.

10. The digital visualization control method for a food production line control system according to claim 1, characterized in that, The method also includes a visual monitoring operation for residual pressure elimination during the release phase: When the robotic arm reaches the target placement position, it performs a full-cavity positive pressure injection operation through an independent micro air chamber; Real-time pressure attenuation visualization curves are generated using flexible tactile skin; When a separation abrupt change point is detected in the pressure decay curve, it is determined that the food has detached, the positive pressure injection is stopped, and the robotic arm is reset.

Images