Intelligent control method and system for producing and processing a gel patch

Abstract

The application relates to the technical field of production process control, in particular to an intelligent control method and system for gel plaster production and processing. The method comprises the following steps: acquiring a visible light image of an extrusion coating interface of gel plaster production and processing, determining discrete thickness values of a plurality of discrete nodes on a gel surface layer by using the visible light image; acquiring an infrared image of the extrusion coating interface, and determining local temperature gradient values of the discrete nodes by using the infrared image; determining a dynamic rheological index by using the discrete thickness values and the local temperature gradient values of all the discrete nodes; determining a viscoelastic deviation degree by using a dynamic rheological index sequence at a current moment; and determining a target output speed by using the viscoelastic deviation degree and running state parameters of a gel extrusion device, so as to control the gel extrusion device to extrude a gel matrix according to the target output speed. Through the technical scheme, intelligent control of gel plaster production and processing can be realized.

Description

Technical Field

[0001] This application relates to the field of production process control technology, and in particular to an intelligent control method and system for the production and processing of gel patches. Background Technology

[0002] Gel patches, as a type of topical drug delivery product, are typically composed of a backing layer, a paste layer, and an anti-adhesive layer. In the actual production and processing of gel patches, the extrusion coating process is a key basic step that determines the final quality of the product.

[0003] A conventional gel plaster extrusion coating system typically includes an extrusion die, a feed pump assembly, and a conveyor belt. The feed pump assembly continuously pumps a gel matrix with a specific viscoelasticity to the extrusion die, and then a continuous plaster layer is formed on the backing material of the conveyor belt running at a constant speed.

[0004] Current methods for controlling the extrusion coating of gel patches have insufficient control precision when faced with dynamic fluctuations in the rheological properties of the gel matrix. Due to the non-Newtonian fluid properties of the gel matrix, it exhibits shear thinning effects and temperature sensitivity within the pipeline and at the extrusion die, resulting in a non-linear and non-stationary mapping relationship between the actual extrusion flow rate and the set output speed of the equipment.

[0005] When producing gel patches, related technologies may fail to maintain the spatial uniformity and temporal stability of the gel patch layer thickness, potentially leading to serious quality defects such as localized thickness deviations or even coating breaks. This could also result in the waste of polymer hydrogel matrix materials, increasing the overall manufacturing costs for pharmaceutical companies and limiting the development of gel patch production lines towards high-throughput and high-yield continuous automated intelligent manufacturing. Therefore, there is an urgent need for an intelligent control scheme that can track the rheological state of the gel matrix in real time and implement closed-loop feedforward regulation. Summary of the Invention

[0006] To achieve intelligent control of the production and processing of gel patches, this application provides an intelligent control method and system for the production and processing of gel patches.

[0007] According to a first aspect of the present application, an intelligent control method for the production and processing of gel patches is provided, comprising: acquiring a visible light image of the extrusion coating interface of the gel patch production and processing; performing edge detection and discretization processing on the thickness information of the gel surface layer in the visible light image to determine the discrete thickness values ​​of multiple discrete nodes distributed along the width direction of the extrusion die of the gel extrusion equipment; acquiring an infrared image of the extrusion coating interface and determining the surface temperature corresponding to the discrete nodes from the infrared image, and using the difference between the surface temperature and the ambient temperature as the local temperature gradient value of the discrete nodes; using the discrete thickness values ​​of all discrete nodes and the local temperature gradient values ​​to determine a dynamic rheological index for characterizing the viscoelastic state of the gel matrix at the coating interface; extracting a dynamic rheological index sequence with the current time as the end time, performing continuous differential and integral operations on the dynamic rheological index sequence to determine the viscoelastic deviation at the current time; constructing a thickness adjustment term and a damping adjustment term using the viscoelastic deviation and the operating state parameters of the gel extrusion equipment, and determining a target output speed using the thickness adjustment term and the damping adjustment term to control the gel extrusion equipment to extrude the gel matrix according to the target output speed.

[0008] In this way, intelligent control of the output speed of the gel extrusion equipment for gel matrix output can be achieved during the production and processing of gel patches.

[0009] Optionally, the dynamic rheological index is determined in the following way: ,in, The dynamic rheological index at the current moment. The total number of discrete nodes. For the current moment, the first The actual thickness of each discrete node. Standard coating thickness, For the current moment The local temperature gradient values ​​of discrete nodes. is the maximum temperature gradient value, and norm is the normalization function.

[0010] Optionally, the viscoelastic deviation is determined in the following way: ,in, The viscoelastic deviation at the current moment. as well as These are the preset first weight and second weight, respectively. It is the natural logarithm function. A preset positive number greater than or equal to 1. The dynamic rheological index at the current moment. Based on the basic rheological index, The preset time window length value. For integration time variable, This is the sign for the partial derivative.

[0011] In this way, by introducing a weighted natural logarithm transformation term and a time integral partial derivative term to form an evaluation model for viscoelastic deviation, the transient high-amplitude noise caused by random external disturbances such as bubble rupture in a short period of time is effectively smoothed out.

[0012] Optionally, the thickness adjustment term is constructed in the following way: the relative thickness error of the discrete thickness value of the discrete node relative to the standard coating thickness is determined, the average of the relative thickness errors of all discrete nodes is taken as the global average thickness error, and the product of the global average thickness error and the first preset adjustment coefficient is taken as the thickness adjustment term; the first preset adjustment coefficient is used to adjust the dimension of the thickness adjustment term to be consistent with the speed.

[0013] Optionally, the first preset adjustment coefficient is determined by: obtaining multiple sets of step output speed changes set for the gel extrusion equipment, and controlling the gel extrusion equipment to increase its output speed according to the multiple sets of step output speed changes; the speed difference between different adjacent step output speeds is equal; obtaining the thickness change of the gel patch obtained when the gel extrusion equipment is in a stable operating state after a single increase, relative to the thickness change before the single increase, to obtain the mapping ratio coefficient between different step output speed changes and the normalized thickness change; and using the average value of the mapping ratio coefficients corresponding to different step output speeds as the first preset adjustment coefficient.

[0014] In this way, by applying an arithmetic step velocity excitation signal and statistically analyzing the static gain mapping relationship under steady state, the uncertainty caused by the rheological differences of different non-Newtonian fluids can be avoided, and the first preset adjustment coefficient for conversion gain can be determined.

[0015] Optionally, the damping adjustment term is constructed as follows: determine the absolute difference between the current viscoelastic deviation and the historical viscoelastic deviation of the previous control cycle, use the ratio of the absolute difference to the preset deviation as the base term of the natural exponential function, use the reciprocal of the result of the natural exponential function as the damping attenuation coefficient at the current moment, and use the product of the current viscoelastic deviation, the damping attenuation coefficient, and the second preset adjustment coefficient as the damping adjustment term at the current moment.

[0016] Optionally, the target output speed is determined using the thickness adjustment item and the damping adjustment item, including: determining a first sum between the thickness adjustment item and the preset reference output speed, determining a first difference between the first sum and the damping adjustment item; obtaining the lower limit of the dynamic speed and the upper limit of the dynamic speed of the gel extrusion equipment in the next control cycle, taking the minimum of the first difference and the upper limit of the dynamic speed as the initial adjustment speed, and taking the maximum of the initial adjustment speed and the lower limit of the dynamic speed as the target output speed.

[0017] Optionally, the upper and lower limits of the dynamic speed are determined by: obtaining the maximum output speed, minimum output speed, maximum allowable rate of change of speed, and historical output speed in the previous control cycle of the gel extrusion equipment; determining a second sum of the historical output speed and the maximum rate of change of speed, and using the minimum of the second sum and the maximum output speed as the upper limit of the dynamic speed; determining a second difference between the historical output speed and the maximum rate of change of speed, and using the maximum of the second difference and the minimum output speed as the lower limit of the dynamic speed.

[0018] Optionally, the method further includes: emitting ultrasonic probe waves into the extrusion die using a piezoelectric ceramic transducer arranged outside the extrusion die of the gel extrusion equipment; acquiring ultrasonic echo information reflected by the ultrasonic probe waves upon encountering the gel matrix; performing inversion processing using the time delay information and amplitude attenuation information of the ultrasonic echo information to obtain an acoustic impedance distribution mapping map of the channel cross-section inside the extrusion die; obtaining the area ratio of high impedance characteristic regions based on the acoustic impedance distribution mapping map; and outputting a prompt message when the area ratio is greater than a preset residual warning threshold; the prompt message is used to prompt the treatment of solidified residues inside the extrusion die.

[0019] In this way, by using an externally arranged piezoelectric ceramic transducer to emit ultrasonic detection waves and analyze the echo information to perform acoustic impedance inversion, it is helpful to deal with extrusion dies containing solidified residues in a timely manner.

[0020] Optionally, the time delay information and amplitude attenuation information are determined in the following ways: obtain the transmission timestamp of the ultrasonic probe wave and the reception timestamp of the ultrasonic echo information, and use the difference between the reception timestamp and the transmission timestamp as the time delay information; obtain the initial transmission amplitude of the ultrasonic probe wave and the reception amplitude of the ultrasonic echo information, divide the initial transmission amplitude and the reception amplitude, and take the logarithm of the division result, and use the logarithmic result as the amplitude attenuation information.

[0021] Optionally, the method further includes: obtaining the historical dynamic rheological index, historical viscoelastic deviation, and historical output velocity of the gel patch production process within a historical production cycle; combining the historical dynamic rheological index, historical viscoelastic deviation, and historical output velocity into a multivariate physical feature sequence; using the thickness standard deviation of the gel patch in the corresponding batch of the multivariate physical feature sequence as a supervision label; training a pre-constructed initial network model using the multivariate physical feature sequence and the corresponding supervision label; using the network model obtained after training as a quality prediction model; and using the quality prediction model to output the predicted thickness standard deviation of the target production batch.

[0022] In this way, by combining the dynamic rheological characteristics and control variables within the historical cycle into a multivariate physical feature sequence, and using it as input tensor features to train the quality prediction model, it is easy to use the obtained quality prediction model to predict the quality of the subsequently obtained gel patches.

[0023] Optionally, the method further includes: obtaining a multivariate physical feature sequence within the latest production cycle, inputting the multivariate physical feature sequence into a quality prediction model, obtaining the predicted thickness standard deviation for a future time period output by the quality prediction model; outputting an early warning intervention signal when the predicted thickness standard deviation is greater than or equal to a preset quality safety threshold; the early warning intervention signal is used to prompt a reduction in the reference output speed of the extrusion equipment.

[0024] Optionally, the discrete thickness value of the discrete node is determined as follows: Feature recognition is performed on the visible light image using an edge detection algorithm to extract the first contour boundary representing the gel layer surface and the second contour boundary representing the nonwoven fabric backing; the vertical pixel spacing between the first and second contour boundaries is determined, and the vertical pixel spacing is converted into a thickness distribution curve; the effective coating width of the gel extrusion device is obtained, and the effective coating width is proportionally divided into multiple sampling intervals; the center physical coordinates of the sampling intervals in the width direction are determined; and the position is mapped on the thickness distribution curve based on the center physical coordinates to determine the discrete thickness value of the discrete node.

[0025] Optionally, determining the surface temperature corresponding to discrete nodes from the infrared image includes: aligning the visible light image and the infrared image in a coordinate system based on a preset spatial alignment parameter, obtaining the corresponding coordinates of multiple discrete nodes in the infrared image, and extracting the surface temperature corresponding to the discrete nodes from the infrared image based on the corresponding coordinates.

[0026] According to a second aspect of the present application, an intelligent control system for the production and processing of gel patches is provided, comprising: a processor and a memory, wherein the memory stores computer program instructions, and the computer program instructions, when executed by the processor, implement the steps of the intelligent control method for the production and processing of gel patches provided in the first aspect of the present application.

[0027] The technical solutions provided by the embodiments of this application may include the following beneficial effects: By simultaneously extracting the spatial thickness distribution characteristics and thermodynamic temperature gradient characteristics of the coating interface, since the coupling relationship between the temperature field and the thickness field contains the true rheological response of the gel matrix at the current shear rate and the ambient temperature, by fusing the dual-field characteristics to construct a dynamic rheological index characterizing the viscoelastic state, and by performing calculus calculations on the dynamic evolution trajectory of the dynamic rheological index in the time dimension to capture the deviation trend, a thickness compensation and damping attenuation mechanism can be constructed to achieve dynamic closed-loop control of the output speed of the gel extrusion equipment, which can avoid uneven coating thickness or discontinuity defects caused by fluctuations in gel rheological properties, and improve the adaptive adjustment capability of the gel patch production and processing process and the stability of the finished product quality.

[0028] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0029] Figure 1 This is a flowchart illustrating an intelligent control method for the production and processing of gel patches according to an exemplary embodiment; Figure 2 This is a schematic diagram illustrating the change in the output speed of the gel matrix by the gel extrusion equipment; Figure 3 This is a schematic diagram of the structure of an intelligent control system for the production and processing of gel patches, according to an exemplary embodiment. Detailed Implementation

[0030] To achieve intelligent control over the production and processing of gel patches, embodiments of this application provide an intelligent control method and system for the production and processing of gel patches. Figure 1 This is a flowchart illustrating an intelligent control method for the production and processing of gel patches according to an exemplary embodiment, such as... Figure 1 As shown, the method includes the following steps.

[0031] In step S101, a visible light image of the extrusion coating interface of the gel patch production process is obtained. Edge detection and discretization processing are performed on the thickness information of the gel surface layer in the visible light image to determine the discrete thickness values ​​of multiple discrete nodes distributed along the width direction of the extrusion die of the gel extrusion equipment.

[0032] The intelligent control method for gel patch production and processing provided in this application embodiment can be applied to control systems, terminal equipment, or cloud platforms for intelligent control of industrial production.

[0033] Industrial camera arrays can continuously capture images of a moving extrusion coating interface to obtain a visible light image matrix with high spatial resolution. By deploying a visual extraction algorithm that includes Sobel operator edge detection logic and adaptive threshold segmentation strategy, the pixel gradient change pattern in the two-dimensional visible light image is captured. The captured reflective contour line of the gel surface is discretized and sliced ​​along the vertical pixel position. The real-time discrete thickness values ​​of, for example, 100 discrete monitoring nodes uniformly distributed along the physical width direction of the extrusion die can be determined.

[0034] When the gel matrix passes through the slit extrusion die, the nonlinear distribution of fluid shear stress at different locations within the die's internal flow channel may cause the actual ejection volume at each extrusion lip to exhibit a parabolic distribution characteristic, with a larger central region and a smaller peripheral region.

[0035] High-density discrete sampling of the gel surface morphology based on high-frequency visual images can reduce the morphology of the three-dimensional paste coated on the non-woven fabric backing to a one-dimensional spatial node thickness array. This provides a basic geometric feature data source for subsequent quantitative analysis of coating sagging and uneven thickness caused by the instability of polymer cross-linking networks, which helps to ensure the non-destructive nature of online monitoring of the production process and the real-time nature of data acquisition.

[0036] In one embodiment, the discrete thickness value of a discrete node is determined as follows: A feature recognition algorithm is used to analyze a visible light image to extract a first contour boundary representing the surface of the gel layer and a second contour boundary representing the nonwoven fabric backing; the vertical pixel spacing between the first and second contour boundaries is determined, and the vertical pixel spacing is converted into a thickness distribution curve; the effective coating width of the gel extrusion device is obtained, and the effective coating width is proportionally divided into multiple sampling intervals; the center physical coordinates of the sampling intervals in the width direction are determined; and the position is mapped onto the thickness distribution curve based on the center physical coordinates to determine the discrete thickness value of the discrete node.

[0037] The image processing unit can run an edge detection algorithm that includes Gaussian blur preprocessing and non-maximum suppression logic, perform feature recognition on the high-frequency components of the pixel matrix of the visible light image input to the video memory, and extract the first contour boundary curve for characterizing the light reflection abrupt interface on the surface of the gel layer and the second contour boundary curve for characterizing the diffuse reflection texture of the nonwoven backing material using the gray-scale abrupt change characteristics.

[0038] For example, the vertical pixel spacing values ​​of two contour boundaries at the same horizontal image coordinate scale can be scanned column by column. Using the physical scale conversion coefficient provided by the camera calibration matrix, the vertical pixel spacing at each column vector position can be transformed into a continuous thickness distribution curve that characterizes the physical thickness of the full coating area.

[0039] The system can retrieve the pre-set effective coating width specifications of the gel extrusion equipment from the memory of the industrial control system, divide the effective coating width into, for example, 100 equal physical sampling intervals according to the preset equal division principle, and sequentially determine the central physical coordinates of the geometric center position of each independent sampling interval in the width direction.

[0040] Based on the determined center physical coordinates, a reverse interpolation mapping operation of the geometric position can be performed on the thickness distribution curve obtained by converting the pixel spacing to determine the precise discrete thickness value corresponding to the center position of each sampling interval.

[0041] Due to its translucent and highly viscous physical properties, the gel layer exhibits a microscopic permeation effect when bonded to the rough nonwoven fabric surface in the early stages of molding, which may lead to blurred visual features of the lower boundary. By independently extracting the first contour boundary characterizing the upper surface and the second contour boundary of the backing with stable optical features, and calculating the relative difference between the first and second contour boundaries, the influence of environmental stray light interference on the extraction of reflective features of a single surface can be reduced.

[0042] By using central physical coordinates for mapping, the disordered continuous image thickness distribution curve can be reduced in dimensionality and transformed into a low-dimensional regular set of state variables aligned with the equipment control node. This reduces the computational overhead of subsequent complex rheological index calculations and improves the generalization and adaptability of the thickness measurement model during the switching process of different width production specifications.

[0043] In step S102, an infrared image of the extrusion coating interface is acquired, and the surface temperature corresponding to the discrete nodes is determined from the infrared image. The difference between the surface temperature and the ambient temperature is used as the local temperature gradient value of the discrete nodes. Using the discrete thickness values ​​of all discrete nodes and the local temperature gradient values, the dynamic rheological index used to characterize the viscoelastic state of the gel matrix at the coating interface is determined.

[0044] Infrared images of the extrusion coating interface can be acquired using an infrared thermal imaging sensor. The array of infrared thermal imaging sensors performs long-wave infrared thermal radiation acquisition on the extrusion coating interface at an optical axis angle that coincides with the field of view of a visible light camera, thereby acquiring a high signal-to-noise ratio infrared image for characterizing the thermodynamic energy distribution of the gel matrix.

[0045] The absolute surface temperature values ​​of pixel regions corresponding to discrete nodes can be extracted from infrared images using a spatial coordinate mapping matrix. For example, the current ambient temperature data of the cleanroom can be retrieved from the real-time broadcast of the temperature and humidity sensor network deployed in the cleanroom. The computer's subtraction arithmetic logic unit is then used to subtract the extracted absolute surface temperature values ​​of each node from the current ambient temperature data.

[0046] The difference obtained by subtraction is used as the real-time local temperature gradient value of each corresponding discrete node. When the polymer network structure in the gel patch undergoes strong mechanical shearing inside the screw extruder and slit die, it will convert a large amount of mechanical energy into heat energy, resulting in significant thermal stress accumulation inside the freshly extruded gel and exhibiting a thermal distribution state higher than room temperature.

[0047] The uneven spatial distribution of residual heat energy may directly change the relaxation time and viscosity of the polymer inside the gel, thereby affecting the leveling characteristics of the gel during curing. By extracting the gradient difference variable between the surface temperature of the coating interface and the ambient reference temperature, the thermodynamic potential energy dissipation rate of the gel matrix during the heat exchange process with the surrounding environment after being freed from the physical constraints of the mold can be quantified, providing support for the subsequent analysis of the coupling characteristics of the temperature field and the thickness field.

[0048] In one embodiment, determining the surface temperature corresponding to a discrete node from an infrared image includes: aligning the visible light image and the infrared image in a coordinate system based on a preset spatial alignment parameter, obtaining the corresponding coordinates of multiple discrete nodes in the infrared image, and extracting the surface temperature corresponding to the discrete node from the infrared image based on the corresponding coordinates.

[0049] During the pre-calibration stage of the image acquisition equipment, a standard black and white checkerboard grid and a heating target can be used to jointly calibrate the spatial alignment parameters, which include rotation matrices and translation vectors. An affine transformation algorithm can be applied to perform coordinate system alignment processing on the visible light image pixel matrix and the infrared image thermal matrix, which have physical differences in resolution and field of view. Edge detection and proportional division can be used to determine the visual spatial position coordinates of multiple discrete monitoring nodes.

[0050] The two-dimensional coordinates of multiple discrete monitoring nodes in the infrared image grid system after affine transformation are obtained, and the average surface temperature of each discrete node in a small neighborhood range is extracted from the thermal data matrix of the infrared image based on the two-dimensional coordinates.

[0051] Heterogeneous binocular vision systems have inherent baseline distance deviations and lens distortion differences in physical installation space, which may cause the target object to appear displaced on sensors of different wavelengths. If visible light coordinates are used directly to read infrared temperature data, feature mismatch may occur and introduce fatal temperature measurement bias errors.

[0052] By applying an affine transformation matrix based on high-precision calibration parameters and implementing a coordinate system alignment operation mechanism, the correctness of the correlation between the spatial nodes used to describe macroscopic geometric thickness characteristics and their absolute positions in the microscopic thermodynamic state can be ensured, thus avoiding misjudgment of rheological characteristics due to coordinate offset.

[0053] In one embodiment, the dynamic rheological index is determined in the following manner: ,in, The dynamic rheological index at the current moment. The total number of discrete nodes. For the current moment, the first The actual thickness of each discrete node. Standard coating thickness, For the current moment, the first The local temperature gradient values ​​of discrete nodes. is the maximum temperature gradient value, and norm is the normalization function.

[0054] For each discrete node containing, for example, 100 node elements, the actual thickness at the current moment can be obtained. Array sequence and corresponding local temperature gradient values An array sequence can be used to obtain a pre-set standard coating thickness. And the maximum temperature gradient value obtained in advance based on thermodynamic experience. These two reference parameters.

[0055] For a specific discrete node, the absolute value of the deviation between the actual thickness and the standard coating thickness can be calculated sequentially. The absolute value of the deviation can be divided by the standard coating thickness to obtain the relative thickness deviation ratio. The relative thickness deviation ratio can then be subjected to a squared exponentiation operation to amplify the abnormal fluctuation amplitude of the thickness and construct a geometric distortion term. .

[0056] Dividing the local temperature gradient value of the corresponding node by the maximum temperature gradient value allows us to construct a thermodynamic compensation term that incorporates the nonlinear dissipation characteristics of thermodynamic activation energy. Multiplying the geometric distortion term by the thermodynamic compensation term yields the deviation eigenvalues ​​of the nodes. We then iterate through and calculate the deviation eigenvalues ​​of all discrete nodes and apply the summation operator. After summing, the summation result is divided by the total number of discrete nodes to obtain the global rheological deviation mean.

[0057] Apply a normalization function to the global rheological deviation from the mean. By restricting the numerical scale to a preset non-negative range, the dynamic rheological index at the current moment can be obtained.

[0058] Normalization processing function For example, a maximum-minimum normalization algorithm can be used. By extracting the maximum and minimum statistical limits of the historical global rheological deviation from the mean, the minimum statistical limit is subtracted from the currently calculated global rheological deviation from the mean, and then divided by the difference between the maximum and minimum statistical limits. This linearly maps the broadly distributed deviation from the mean to a strict standard interval of 0 to 1.

[0059] As a soft fluid with relaxation memory properties, the ability of the gel matrix to recover transient geometric deformation during actual industrial extrusion is affected by the current internal temperature field conditions. When the local temperature gradient is too high, the polymer molecular chain movement is violent, resulting in a sharp drop in apparent viscosity and making it more prone to irreversible flow deformation, thus amplifying thickness error defects.

[0060] In the calculation formula of dynamic rheological index, by introducing the square term of relative thickness, a nonlinear penalty weight of power-two level can be applied to the thickness abrupt change that exceeds the tolerance boundary. Through this multi-dimensional physical parameter fusion and global averaging, the complex spatiotemporal coupling state of the coating interface is quantified, so that the output dynamic rheological index can evaluate the rheological steady-state characteristics of the extrusion system.

[0061] In step S103, the dynamic rheological index sequence ending at the current time is extracted, and continuous differentiation and integration operations are performed on the dynamic rheological index sequence to determine the viscoelastic deviation at the current time.

[0062] In one embodiment, the viscoelastic deviation is determined in the following manner: ,in, The viscoelastic deviation at the current moment. as well as These are the preset first weight and second weight, respectively. It is the natural logarithm function. A preset positive number greater than or equal to 1. The dynamic rheological index at the current moment. Based on the basic rheological index, The preset time window length value. For integration time variable, This is the sign for the partial derivative.

[0063] It can extract a series of dynamic rheological indexes that are continuously generated over a period of time and arranged in chronological order of sampling timestamps. It can also call the numerical calculus package to perform numerical operations on values ​​within a preset time window. Dynamic rheological index sequences within the range are subjected to integral time variables. The combination of differentiation and integration operations is used to track the time evolution trajectory of the rheological state.

[0064] Dividing the dynamic rheological index obtained at the current moment by the basic rheological index, which reflects the system under ideal and stable coating conditions, allows us to construct a real-time rheological ratio parameter. , ratio parameter After superimposing a preset positive number, the natural logarithm function is used to perform nonlinear compression mapping. The result of the nonlinear compression mapping is multiplied by a preset first weight, which can form a steady-state offset feedback term to suppress exponential transient spikes.

[0065] Applying partial derivative symbols By performing first-order partial derivative calculations on the dynamic rheological sequence within a preset time window, a gradient curve reflecting the instantaneous rate of change can be obtained. After squaring the gradient curve, a continuous integral operator is applied within the integration time variable interval. Perform surface accumulation and multiply by the length value Perform a square root operation on the product to extract the dynamic fluctuation energy value with the meaning of the physical root mean square rate of change. .

[0066] Multiplying the dynamic fluctuation energy value by a preset second weight forms a transient fluctuation penalty term. The steady-state offset feedback term and the transient fluctuation penalty term can be added together to determine the viscoelastic deviation at the current moment. In addition to being affected by the current working conditions, the forced shear conditions over a period of time will also generate residual stress accumulation with long-term memory effect in the gel rheological process.

[0067] A single static rheological index feedback may cause the control system to fall into the oscillation error of frequent adjustment and fail to smoothly track long-period trends. The formula for calculating viscoelastic deviation uses the natural logarithm function to smooth the transient high-amplitude noise spectrum caused by bubble rupture in a short period of time. The root mean square term can reflect the second-order dynamic evolution kinetic energy information of the gel matrix produced by the gel extrusion equipment in the high-frequency oscillation intensity. Through these two physical and mathematical characteristics that respectively characterize the macroscopic slow trend deviation and the microscopic rapid kinetic energy dissipation, the mechanism can be extracted and assigned different weight ratios for summation and reconstruction, thereby improving the ability to evaluate the multi-scale rheological behavior of the non-Newtonian fluid gel matrix coated on the coating interface.

[0068] The first weight and the second weight can be adjusted according to the contribution of these two different physical characteristics. The first weight and the second weight can both be between 0 and 1, and the sum of the first weight and the second weight is equal to 1. For example, the first weight and the second weight can both be equal to 0.5. The specific values ​​of the first weight and the second weight are not limited in the embodiments of this application.

[0069] In step S104, thickness adjustment terms and damping adjustment terms are constructed using viscoelastic deviation and operating status parameters of the gel extrusion equipment. The target output speed is determined using the thickness adjustment terms and damping adjustment terms to control the gel extrusion equipment to extrude the gel matrix according to the target output speed.

[0070] By establishing a thickness adjustment term to directly correct the steady-state thickness deviation of the system, and a nonlinear damping adjustment term to suppress the inertial overshoot behavior during high-speed dynamic adjustment, the gel matrix extrusion of the gel extrusion equipment can be controlled using these two adjustment terms.

[0071] Using the obtained target output speed, the digital-to-analog conversion module can be driven to send analog electrical signals to the servo driver that controls the variable frequency motor of the feed pump group, so as to control the gel extrusion equipment to perform the extrusion action of the gel matrix according to the target output speed, thereby realizing the adaptive adjustment of the gel extrusion speed.

[0072] In one embodiment, the thickness adjustment term is constructed by: determining the relative thickness error of the discrete thickness value of the discrete node relative to the standard coating thickness, taking the average of the relative thickness errors of all discrete nodes as the global average thickness error, and taking the product of the global average thickness error and a first preset adjustment coefficient as the thickness adjustment term; the first preset adjustment coefficient is used to adjust the dimensions of the thickness adjustment term to be consistent with the speed.

[0073] The latest discrete thickness values ​​corresponding to all monitored coordinate positions can be extracted sequentially. The absolute difference between the values ​​corresponding to these real-time spatial monitoring points and the standard coating thickness required by the industrial formulation process document can be calculated respectively. Then, the difference is further divided by the standard coating thickness to obtain the dimensionless relative thickness error rate. After summing the relative thickness error rates of all nodes and dividing by the total number of monitoring points, the global average thickness error parameter that characterizes the degree of macroscopic deformation of the entire wide-width coating surface can be obtained.

[0074] By calculating the global average state containing the spatial distribution characteristics of relative error and performing dimensional transformation processing using the first preset adjustment coefficient, a proportional feedback adjustment term that maps the dimensionless geometric distortion ratio to the speed change required by the motor actuator can be constructed, enabling the servo drive to achieve closed-loop control of the front-end raw material pumping power.

[0075] In one embodiment, the first preset adjustment coefficient is determined by: acquiring multiple sets of step output speed changes set for the gel extrusion equipment, and controlling the gel extrusion equipment to increase its output speed according to the multiple sets of step output speed changes; the speed difference between different adjacent step output speeds is equal; acquiring the thickness change of the gel patch obtained when the gel extrusion equipment is in a stable operating state after a single increase, relative to the thickness change before the single increase, to obtain the mapping ratio coefficient between different step output speed changes and the normalized thickness change; and using the average of the mapping ratio coefficients corresponding to different step output speeds as the first preset adjustment coefficient.

[0076] During the calibration test in the equipment's factory commissioning phase, five sets of pre-designed step output speed change instructions covering the entire working range from low speed to high speed can be obtained. The servo motor of the gel extrusion equipment is controlled through the communication bus to strictly follow the above instruction sequence to increase the basic output speed, ensuring that the speed difference between different adjacent step output speed control signals set in each test phase remains equal.

[0077] The absolute thickness data of the coating is acquired synchronously when the gel extrusion equipment is in a stable operating state again after experiencing a speed increase command and mechanical response time. The relative absolute thickness change parameter is calculated relative to the previous stable state before the single speed increase control command is issued. The absolute thickness change parameter is normalized to obtain multiple sets of mapping ratio coefficients between the driving speed increment and the coating deformation increment under different step speed excitation conditions.

[0078] By averaging the collected mapping ratio coefficients, a first preset adjustment coefficient can be obtained. The different internal molecular chain entanglement densities of different components of polymer hydrogels may lead to a significant formulation-specific characteristic in the hydrodynamic hysteresis response sensitivity of hydrogels to changes in external pumping speed.

[0079] The system identification operation was performed by applying an arithmetic step velocity excitation signal, and the static gain mapping relationship of the system under steady state was statistically analyzed. The influence of different non-Newtonian fluid rheological differences was eliminated, and a matching conversion gain parameter was found for the proportional control loop.

[0080] In one embodiment, the damping adjustment term is constructed as follows: the absolute difference between the current viscoelastic deviation and the historical viscoelastic deviation of the previous control cycle is determined; the ratio of the absolute difference to a preset deviation is used as the base term of the natural exponential function; the reciprocal of the result of the natural exponential function is used as the damping attenuation coefficient at the current moment; and the product of the current viscoelastic deviation, the damping attenuation coefficient, and the second preset adjustment coefficient is used as the damping adjustment term at the current moment.

[0081] The system can obtain the viscoelastic deviation determined at the current moment and the historical viscoelastic deviation stored in the previous digital control cycle, perform a difference subtraction operation to extract the absolute difference, divide the absolute difference by a preset deviation constant to form the relative deviation change rate, and use the relative deviation change rate as the base term of the natural exponential function. The system then performs a reciprocal operation on the result of the natural exponential function to generate the damping attenuation coefficient at the current moment, which is numerically in the non-negative dynamic range.

[0082] The original viscoelastic deviation parameter at the current moment, the damping attenuation coefficient calculated together, and the second preset adjustment coefficient set manually are continuously multiplied. The product result is used as the damping adjustment term at the current moment to implement the overshoot suppression characteristic of the control loop.

[0083] When a polymeric gel with strong viscoelastic memory properties experiences a sudden change in pumping pressure, its internal stress field will not be released immediately. Instead, it will slowly relax in subsequent cycles and is prone to resonant coupling with the next round of speed intervention commands from the controller, thereby inducing a pressure pulsation disaster in the entire extrusion system. The absolute difference in viscoelastic deviation between adjacent control cycles can be used as the base of exponential decay to characterize the severity of state changes in the gel patch production system.

[0084] When a drastic change in the deviation state of an adjacent cycle is detected, the reciprocal characteristic of the operation will reduce the amplitude of the damping attenuation coefficient, thereby weakening the disturbance contribution rate of the current deviation to the overall control command, and can always maintain a stable extrusion compensation power output under complex impedance change conditions.

[0085] In one embodiment, determining the target output speed using a thickness adjustment term and a damping adjustment term includes: determining a first sum between the thickness adjustment term and a preset reference output speed; determining a first difference between the first sum and the damping adjustment term; obtaining the lower limit of the dynamic speed and the upper limit of the dynamic speed of the gel extrusion device in the next control cycle; using the minimum of the first difference and the upper limit of the dynamic speed as the initial adjustment speed; and using the maximum of the initial adjustment speed and the lower limit of the dynamic speed as the target output speed.

[0086] The thickness adjustment term, which has a proportional correction function, is algebraically added to the preset benchmark output speed, which represents the current batch's normal constant speed production requirements, to obtain a first sum. The aforementioned damping adjustment term, which has an overshoot suppression function, is then subtracted from the first sum to obtain a first difference after nonlinear damping disturbance rejection correction.

[0087] Obtain the absolute dynamic speed lower limit and dynamic speed upper limit threshold information of the gel extrusion equipment in the next future control cycle, apply the minimum value algorithm to compare the first difference with the dynamic speed upper limit, and extract the smaller value of the two as the initial adjustment speed of the intermediate transition state.

[0088] The maximum value algorithm is used to compare the obtained initial adjustment speed with the lower limit of the underlying dynamic speed, and the maximum of the two is extracted as the target output speed.

[0089] Complex electromechanical coupling systems, under the combined effect of multiple gain feedback mechanisms, are prone to internal calculation results exceeding the physical safety tolerance limit of mechanical transmission devices due to extreme external interference. By combining thickness compensation enhancement with nonlinear damping attenuation, and incorporating the multidimensional dynamic rheological characteristics of polymers, and on this basis, through multi-level boundary comparison and amplitude limiting operations, the target output speed of the calculation output can always be limited within the range that conforms to the fluid dynamic safety boundary.

[0090] In one embodiment, the dynamic speed upper limit and dynamic speed lower limit are determined by: obtaining the maximum output speed, minimum output speed, maximum allowable speed change rate, and historical output speed of the gel extrusion equipment in the previous control cycle; determining a second sum of the historical output speed and the maximum speed change rate, and using the minimum of the second sum and the maximum output speed as the dynamic speed upper limit; determining a second difference between the historical output speed and the maximum speed change rate, and using the maximum of the second difference and the minimum output speed as the dynamic speed lower limit.

[0091] The maximum and minimum output speed parameters that the current machine model's hardware can withstand can be read from the process parameter configuration library of the production execution system. The maximum speed change rate allowed to ensure that the motor gears are not torn by transient large torque, as well as the historical output speed of the drive motor in the previous control cycle, can be obtained simultaneously.

[0092] By defining the upper and lower limits of the dynamic speed, we can avoid exceeding the upper or lower limits and prevent equipment vibration or abnormal product output caused by excessive adjustment of the speed of the gel patch extrusion equipment.

[0093] For example, it can prevent the motor from falling into continuous overload but cannot prevent the water hammer effect caused by acceleration and deceleration in a short period of time from impacting the pipeline. By combining the physical constraints of the actual operating speed and the maximum safe acceleration to construct a dynamic window limiting protection strategy that slides in real time with the operating conditions, it can avoid the damage of the fine macromolecular network cross-linked structure of the hydrogel to sudden commands.

[0094] Figure 2 This is a schematic diagram illustrating the change in the output speed of the gel matrix by the gel extrusion equipment, as shown below. Figure 2 As shown, the dynamic response trajectory of a traditional tracking model that does not employ the damping attenuation mechanism of this application embodiment is compared with that of the embodiment of this application when facing abnormal operating conditions.

[0095] like Figure 2As shown, at the abnormal sudden anchor point, when the viscoelastic deviation increases sharply due to fluctuations in gel rheological properties or external disturbances, the traditional tracking model, lacking nonlinear damping disturbance correction, still maintains a constant high speed command, which can easily cause overload of the feeding motor and cause severe stress concentration inside the die head.

[0096] like Figure 2 As shown, the intelligent control method of this application, after sensing the sudden change in viscoelastic deviation state, suppresses the overshoot of the command through the damping attenuation coefficient, rapidly and smoothly reduces the command speed of the drive motor, and effectively releases the internal thermal stress of the polymer gel material; at the anomaly elimination anchor point, as the abnormal situation is gradually eliminated and the system deviation trend is alleviated, the command speed of the motor under the control of the cooperative damping model exhibits adaptive smooth recovery characteristics.

[0097] The commanded speed gradually increases at a gentle slope, eventually returning smoothly to the reference output speed. Figure 2 This application demonstrates the target output speed determined by the thickness adjustment term and the damping adjustment term, which can effectively respond to transient high-amplitude interference and avoid the control system from falling into frequent adjustment oscillations, thus ensuring the leveling characteristics and physical morphology stability of the gel matrix during the extrusion process.

[0098] In one embodiment, an ultrasonic probe wave can be emitted into the extrusion die using a piezoelectric ceramic transducer arranged outside the extrusion die of the gel extrusion equipment; the ultrasonic echo information reflected by the ultrasonic probe wave upon encountering the gel matrix is ​​acquired; the time delay information and amplitude attenuation information of the ultrasonic echo information are used for inversion processing to obtain an acoustic impedance distribution map of the cross-section of the channel inside the extrusion die; the area ratio of high impedance characteristic regions is obtained based on the acoustic impedance distribution map; when the area ratio is greater than a preset residual warning threshold, a prompt message is output; the prompt message is used to prompt the treatment of the solidified residue inside the extrusion die.

[0099] An ultrasonic pulse generator drives a piezoelectric ceramic transducer array that is attached to the outside of a metal extrusion die. This array can perform high-frequency oscillation excitation and emit a high-frequency ultrasonic probe wave group with a preset center frequency into the high-pressure flow channel space inside the die, which is in a closed state. The signal acquisition component can capture the multi-channel ultrasonic echo information formed by the reflection of the ultrasonic probe wave after it comes into contact with the viscous polymer gel matrix. Using an embedded reconstruction algorithm, the time delay information and the amplitude attenuation information caused by the accompanying energy dissipation contained in the ultrasonic echo information can be inversely processed and calculated.

[0100] A three-dimensional acoustic impedance distribution map is constructed by using an acoustic inverter algorithm to reconstruct the two-dimensional matrix of the microscopic acoustic properties of the hidden fluid channel cross-section inside the extrusion die. Based on threshold segmentation logic, the geometric pixels of the abnormal feature regions exhibiting significantly high acoustic impedance characteristics in the acoustic impedance distribution map are accumulated and statistically analyzed to obtain the macroscopic area ratio parameter relative to the entire cross-section.

[0101] The control system can compare the monitored area percentage with a preset residual warning threshold. When the area percentage is confirmed to be greater than the preset residual warning threshold, it can broadcast a high-priority prompt to the alarm center of the human-machine interface and the workshop engineer's handheld terminal. The prompt can remind the operation and maintenance personnel to clean and maintain the solidified residue accumulated inside the extrusion die flow channel during the downtime window.

[0102] After prolonged local shear heat accumulation and stress concentration, the polymer in gel plaster is prone to irreversible chemical cross-linking and phase separation in the slow-flow area inside the mold head, thus forming a hardened stone.

[0103] Because it is located inside a closed metal shell, conventional optical visual sensing methods cannot detect the internal state. By using ultrasonic acoustic impedance tomography inversion technology based on piezoelectric ceramic media, the physical and chemical evolution of the internal flow field can be effectively monitored. It can keenly capture solidified and deformed matrices with sudden changes in acoustic impedance and perform predictive alarm maintenance before blockage and shutdown are induced.

[0104] In one embodiment, the time delay information and amplitude attenuation information are determined as follows: the transmission timestamp of the ultrasonic probe wave and the reception timestamp of the ultrasonic echo information are obtained, and the difference between the reception timestamp and the transmission timestamp is used as the time delay information; the initial transmission amplitude of the ultrasonic probe wave and the reception amplitude of the ultrasonic echo information are obtained, the initial transmission amplitude and the reception amplitude are divided, and the logarithm of the division result is taken as the amplitude attenuation information.

[0105] It can record the precise transmission timestamp of the ultrasonic probe wave generated by the piezoelectric ceramic transducer at the moment of excitation by the electric pulse, and the accurate reception timestamp of the ultrasonic echo information of the obvious reflected energy front signal captured by the analog-to-digital conversion circuit. By subtracting the reception timestamp from the transmission timestamp using a subtractor, the flight time difference representing the total flight time of the sound wave in different media can be obtained. The flight time difference can be used as the time delay information required for the subsequent inversion model.

[0106] It can simultaneously acquire the initial transmission amplitude parameters carried by the ultrasonic probe wave when it is excited at the transmitting end, as well as the peak received amplitude parameters of the attenuated ultrasonic echo information sensed by the piezoelectric receiving module. The initial transmission amplitude parameters are divided by the received amplitude parameters after medium dissipation to perform a division ratio processing operation, and the logarithmic function component is used to perform a base-based logarithmic numerical transformation operation on the division result. The result after the logarithmic transformation is used as the key amplitude attenuation information required for acoustic model calculation.

[0107] When ultrasound propagates in media with different physical densities, the sound wave propagation rate and mechanical energy absorption efficiency exhibit physical differences. By introducing logarithmic operations to process the ratio of transmitted amplitude to received amplitude, we can match the exponential absorption and dissipation model of sound mechanical energy propagating in viscoelastic polymer media. By utilizing the time-of-flight difference and logarithmic energy attenuation information, we can effectively eliminate noise aliasing interference caused by multipath reflection of ultrasonic signals between multi-layer metal shells.

[0108] In one embodiment, the historical dynamic rheological index, historical viscoelastic deviation, and historical output velocity of the gel patch production process within a historical production cycle can also be obtained. The historical dynamic rheological index, historical viscoelastic deviation, and historical output velocity are combined into a multivariate physical feature sequence. The thickness standard deviation of the gel patch in the corresponding batch of the multivariate physical feature sequence is used as a supervision label. The pre-constructed initial network model is trained using the multivariate physical feature sequence and the corresponding supervision label. The network model obtained after training is used as a quality prediction model. The quality prediction model is used to output the predicted thickness standard deviation of the target production batch.

[0109] Through the data interface, we can obtain a large amount of historical dynamic rheological index arrays, historical viscoelastic deviation curves, and historical output speed records actually issued by the system from the server database, which are accumulated during the historical production cycle of the corresponding gel patch production and processing process.

[0110] By using time series to perform high-dimensional tensor splicing and combination of historical dynamic rheological index records, historical viscoelastic deviation records, and historical output velocity sequences belonging to the same production time window, a multivariate physical feature sequence sample for characterizing the operational evolution trajectory can be constructed. Simultaneously, the thickness standard deviation of the gel patch corresponding to the multivariate physical feature sequence confirmed by actual testing in the downstream quality traceability system can be retrieved, and the thickness standard deviation information can be used as a supervision label variable.

[0111] By using multivariate physical feature sequences as input tensors and corresponding supervised label data, iterative backpropagation weight update training can be performed on a pre-built initial network model. The final network model obtained after the loss function convergence process is then packaged and deployed as a quality prediction model.

[0112] The pre-built initial network model can be a convolutional neural network, a temporal fully connected network, or an attention mechanism-enhanced model, etc. The embodiments of this application do not impose any restrictions on the model structure of the pre-built initial network model.

[0113] While relying solely on feedforward control logic with real-time feedback can mitigate short-term fluid fluctuations, it may lack the ability to predict long-term chronic quality drift caused by uneven distribution of crosslinking agents in batches of raw materials. By introducing deep neural networks with time memory capabilities, we can delve into the mapping patterns between historical rheological fluctuation data and the quality indicators of the final macroscopic finished product, which can help achieve more accurate predictive control over the production of gel patches.

[0114] In one embodiment, after using the trained model as a quality prediction model, a multivariate physical feature sequence within the latest production cycle can be obtained. This multivariate physical feature sequence is then input into the quality prediction model to obtain the predicted thickness standard deviation for the future time period output by the quality prediction model. If the predicted thickness standard deviation is greater than or equal to a preset quality safety threshold, an early warning intervention signal is output. The early warning intervention signal is used to prompt a reduction in the baseline output speed of the extrusion equipment.

[0115] It can extract real-time data from the local data bus within a short production cycle to form a multivariate physical feature sequence. By inputting the multivariate physical feature sequence into a quality prediction model that has been trained, it can obtain the predicted thickness standard deviation of the finished product in the future time period, which is represented by the quality prediction model output.

[0116] If the predicted thickness standard deviation is greater than or equal to the preset quality and safety threshold boundary condition, an early warning intervention signal can be sent to the main controller through the communication interruption channel and highlighted on the dashboard. The early warning intervention signal is used to prompt the operation of reducing the preset reference output speed of the main feed pump servo drive equipment.

[0117] During long-cycle, high-throughput extrusion processing of polymer gels, the accumulation of thermal stress and mechanical aging within the material exhibit a hidden, gradual evolution pattern. By combining deep learning inference models to extract features of rheological dynamics and predict future quality, when the prediction results indicate a trend of thickness runaway, a deceleration command can be proactively issued in advance to alleviate the excessive mechanical shear rate burden inside the die head, giving the polymer gel material more time to release internal stress and ensuring the production quality of gel patches.

[0118] Figure 3 A schematic diagram of the structure of an intelligent control system 1000 for the production and processing of gel patches is shown according to an exemplary embodiment. (Refer to...) Figure 3The intelligent control system 1000 for the production and processing of gel patches includes a processor 1100 and a memory 1200. The memory 1200 stores computer program instructions. When the computer program instructions are executed by the processor 1100, they implement all or part of the steps of the intelligent control method for the production and processing of gel patches in this application.

[0119] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only.

[0120] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope.

Claims

1. A smart control method for the production and processing of gel patches, characterized in that, include: The visible light image of the extrusion coating interface of the gel patch production process is obtained. The thickness information of the gel surface layer in the visible light image is subjected to edge detection and discretization processing to determine the discrete thickness values ​​of multiple discrete nodes distributed along the width direction of the extrusion die of the gel extrusion equipment. Infrared images of the extrusion coating interface are acquired, and the surface temperatures corresponding to discrete nodes are determined from the infrared images. The difference between the surface temperature and the ambient temperature is used as the local temperature gradient value of the discrete nodes. Using the discrete thickness values ​​of all discrete nodes and the local temperature gradient values, a dynamic rheological index characterizing the viscoelastic state of the gel matrix at the coating interface is determined, including: , The dynamic rheological index at the current moment. The total number of discrete nodes. For the current moment The actual thickness of each discrete node. Standard coating thickness, For the current moment The local temperature gradient values ​​of discrete nodes. is the maximum temperature gradient value, and norm is the normalization function; Extract the dynamic rheological index sequence with the current time as the end time, perform continuous differentiation and integration operations on the dynamic rheological index sequence, and determine the viscoelastic deviation at the current time, including: , The viscoelastic deviation at the current moment. as well as These are the preset first weight and second weight, respectively. It is the natural logarithm function. A preset positive number greater than or equal to 1. Based on the rheological index, The preset time window length value. For integration time variable, The sign of the partial derivative; By utilizing the viscoelastic deviation and the operating parameters of the gel extrusion equipment, a thickness adjustment term and a damping adjustment term are constructed. The target output speed is determined using the thickness adjustment term and the damping adjustment term, so as to control the gel extrusion equipment to extrude the gel matrix according to the target output speed.

2. The intelligent control method for the production and processing of gel patches according to claim 1, characterized in that, The thickness adjustment item is constructed in the following way: The relative thickness error of the discrete thickness value of the discrete node relative to the standard coating thickness is determined. The average value of the relative thickness errors of all discrete nodes is taken as the global average thickness error. The product of the global average thickness error and the first preset adjustment coefficient is taken as the thickness adjustment term. The first preset adjustment coefficient is used to adjust the dimensions of the thickness adjustment item to be consistent with the speed.

3. The intelligent control method for the production and processing of gel patches according to claim 1, characterized in that, The first preset adjustment coefficient is determined in the following way: The system acquires multiple sets of step output speed changes set for the gel extrusion equipment, and controls the gel extrusion equipment to increase its output speed according to these multiple sets of step output speed changes; the speed difference between different adjacent step output speeds is equal. The thickness change of the gel patch obtained after the gel extrusion equipment reaches a stable operating state after a single increment is obtained relative to the thickness change before the single increment. This is used to obtain the mapping ratio coefficient between the change in output speed at different steps and the normalized thickness change. The average value of the mapping ratio coefficients corresponding to different output speeds is used as the first preset adjustment coefficient.

4. The intelligent control method for the production and processing of gel patches according to claim 1, characterized in that, The damping adjustment term is constructed in the following way: Determine the absolute difference between the current viscoelastic deviation and the historical viscoelastic deviation of the previous control cycle. Use the ratio of the absolute difference to the preset deviation as the base of the natural exponential function, and use the reciprocal of the result of the natural exponential function as the damping attenuation coefficient at the current moment. The product of the current viscoelastic deviation, damping attenuation coefficient, and the second preset adjustment coefficient is used as the damping adjustment term for the current moment.

5. The intelligent control method for the production and processing of gel patches according to claim 1, characterized in that, The target output speed is determined using thickness adjustment and damping adjustment parameters, including: Determine the first sum between the thickness adjustment item and the preset reference output speed, and determine the first difference between the first sum and the damping adjustment item; The lower limit and upper limit of the dynamic speed of the gel extrusion equipment in the next control cycle are obtained. The smaller of the first difference and the upper limit of the dynamic speed is taken as the initial adjustment speed, and the larger of the initial adjustment speed and the lower limit of the dynamic speed is taken as the target output speed.

6. The intelligent control method for the production and processing of gel patches according to claim 1, characterized in that, The method further includes: Ultrasonic detection waves are emitted into the extrusion die by a piezoelectric ceramic transducer arranged outside the extrusion die of the gel extrusion equipment. The ultrasonic echo information reflected by the ultrasonic probe wave upon encountering the gel matrix is ​​obtained. The time delay information and amplitude attenuation information of the ultrasonic echo information are used for inversion processing to obtain the acoustic impedance distribution mapping map of the internal channel cross section of the extrusion die. The area ratio of high impedance characteristic regions is obtained based on the acoustic impedance distribution map. When the area ratio is greater than the preset residual warning threshold, a prompt message is output. The prompt message is used to remind the user to deal with the solidified residue inside the extrusion die.

7. The intelligent control method for the production and processing of gel patches according to claim 1, characterized in that, The method further includes: The historical dynamic rheological index, historical viscoelastic deviation, and historical output velocity of the gel patch production process within a historical production cycle are obtained, and the historical dynamic rheological index, historical viscoelastic deviation, and historical output velocity are combined into a multivariate physical feature sequence. The thickness standard deviation of the gel patch corresponding to the multivariate physical feature sequence is used as the supervision label; the pre-constructed initial network model is trained using the multivariate physical feature sequence and the corresponding supervision label, and the network model obtained after training is used as the quality prediction model; the quality prediction model is used to output the predicted thickness standard deviation of the target production batch.

8. An intelligent control system for the production and processing of gel patches, characterized in that, include: A processor and a memory, the memory storing computer program instructions, which, when executed by the processor, implement the intelligent control method for the production and processing of gel patches according to any one of claims 1-7.

Images