Composite tile raw material dynamic proportioning control system

Abstract

The composite tile raw material dynamic proportioning control system of this invention belongs to the field of intelligent manufacturing technology of polymer composite materials. It includes an acoustic fingerprint acquisition module for acquiring composite acoustic signals generated by the flow of polymer melt in a mixing extruder; and a state decoupling AI model module for receiving the composite acoustic signals. This invention realizes the direct quantification of two previously unobservable core microstates, enabling the control system to have clear and real-time perception capabilities for the first time. It overcomes the problems of unknowable microstates and signal coupling confusion in the background technology, and can proactively predict and respond to state deviations caused by changes in raw materials, adaptively adjust process parameters, and thus maintain a high degree of stability in the production process even when raw materials fluctuate significantly. It broadens the process window and ultimately achieves the technical effect of maximizing the effectiveness of self-healing microcapsules in the final product while ensuring production stability.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology for polymer composite materials, specifically to a dynamic proportioning control system for composite tile raw materials. Background Technology

[0002] Self-healing composite tiles represent a cutting-edge technology for coping with extreme climates and extending the service life of building materials. The core of this technology involves uniformly dispersing functional microcapsules carrying a repair agent within a polymer matrix. When the tile is subjected to external force and develops microcracks, the stress field at the crack tip triggers the microcapsules to rupture and release the repair agent, thus achieving self-repair of the crack.

[0003] Current production processes generally employ static formulations and fixed process parameters, such as screw speed, temperature in each zone, and feeding rate, for open-loop control, or rely on offline sampling and testing of the final product followed by delayed closed-loop feedback adjustment. This traditional approach has the following inherent drawbacks that are difficult to overcome:

[0004] Significant differences exist in key rheological properties of polymer matrices within the same batch, such as melt index and molecular weight distribution, leading to substantial fluctuations in the macroscopic processability of the melt during processing. Fixed process parameters cannot accommodate these fluctuations, easily causing problems such as equipment torque overload, unstable extrusion, out-of-tolerance product dimensions, and even shutdowns. To ensure uniform dispersion of microcapsules in the matrix, sufficient shear force is required; however, excessive shear force or localized overheating of the melt can cause premature breakage of functional microcapsules during processing, resulting in the loss of their repair function in the final product, i.e., a significant reduction in microscopic functional effectiveness. This contradiction results in an extremely narrow production window; traditional feedback control is an ex-post adjustment, and by the time substandard finished product quality is detected, a large number of defective products have already been generated. More importantly, the control system cannot distinguish online whether the change in the current processing state is caused by normal fluctuations in melt viscosity or by a large number of microcapsule breakages, thus making it unable to make correct adjustment decisions.

[0005] While existing technologies attempt to alleviate the aforementioned problems through more sophisticated temperature control systems or more complex screw configurations, this not only significantly increases equipment costs but also fails to fundamentally solve the problems of melt rheology and microcapsule integrity, which are key parameters that are coupled and conflict in real time under process disturbances. Therefore, there is an urgent need in the field for an intelligent control system that can collaboratively optimize processes and proportions online, in real time, and in a feedforward manner to proactively adapt to raw material fluctuations and maximize product functionality while ensuring production stability.

[0006] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide a dynamic proportioning control system for composite tile raw materials to solve the problems mentioned in the background art.

[0008] The technical solution of the present invention includes: an acoustic fingerprint acquisition module, used to acquire composite acoustic signals generated by the flow of polymer melt in a mixing extruder;

[0009] The state decoupling AI model module is used to receive the composite acoustic signal and calculate the real-time melt viscosity estimate and the real-time microcapsule breakage rate estimate based on the composite acoustic signal.

[0010] The collaborative feedforward control module is used to generate a preliminary control command vector based on the real-time melt viscosity estimate and the real-time microcapsule breakage rate estimate, and by comparing them with the preset process target value.

[0011] The collaborative constraint and command arbitration module is used to receive the preliminary control command vector and the current equipment state vector obtained in real time from the hybrid extruder, and to arbitrate the preliminary control command vector according to the preset constraint parameters to output the final safety command vector.

[0012] Preferably, the acoustic fingerprint acquisition module includes a broadband acoustic sensor non-invasively installed at the end of the homogenization section of the mixing extruder barrel.

[0013] Preferably, the state decoupling AI model module includes:

[0014] The low-frequency feature extraction unit is used to perform low-pass filtering and short-time Fourier transform on the composite acoustic signal to extract the energy attenuation coefficient and phase delay as low-frequency feature vectors.

[0015] The viscosity calculation unit is used to output the real-time melt viscosity estimate based on the low-frequency feature vector and through a viscosity regression model.

[0016] Preferably, the state decoupling AI model module further includes:

[0017] The high-frequency feature extraction unit is used to perform high-pass filtering and time-domain analysis on the composite acoustic signal to identify and count the acoustic emission pulses generated by the microcapsule rupture, and extract the pulse amplitude and occurrence frequency as high-frequency feature vectors.

[0018] The breakage rate calculation unit is used to output the real-time microcapsule breakage rate estimate based on the high-frequency feature vector and through a breakage rate regression model.

[0019] Preferably, the collaborative feedforward control module includes:

[0020] An error vector generation unit is used to construct an error vector by taking the difference between the process target value and the estimated real-time melt viscosity as a first error term and the difference between the process target value and the estimated real-time microcapsule breakage rate as a second error term.

[0021] The control command calculation unit is used to multiply the error vector with a preset coupling gain matrix to generate the preliminary control command vector.

[0022] Preferably, the functional component ratio constraint arbitration logic of the collaborative constraint and instruction arbitration module includes:

[0023] Based on the current microcapsule feeding rate in the current device state vector and the microcapsule feeding rate adjustment amount in the preliminary control command vector, the estimated microcapsule mass fraction is calculated.

[0024] The estimated microcapsule mass fraction is compared with the preset maximum microcapsule mass fraction;

[0025] When the estimated microcapsule mass fraction is greater than the maximum microcapsule mass fraction, the microcapsule feeding rate adjustment amount in the preliminary control command vector is reduced;

[0026] When the estimated microcapsule mass fraction is not greater than the maximum microcapsule mass fraction, the microcapsule feeding rate adjustment in the preliminary control command vector is adopted.

[0027] Preferably, the shear rate constraint arbitration logic of the collaborative constraint and instruction arbitration module includes:

[0028] Based on the current screw speed in the current equipment state vector and the screw speed adjustment amount in the preliminary control command vector, and in combination with the preset extruder shear geometry constant, the estimated shear rate is calculated;

[0029] The estimated shear rate is compared with the preset maximum allowable shear rate;

[0030] When the estimated shear rate is greater than the maximum permissible shear rate, the screw speed adjustment amount in the initial control command vector is trimmed to obtain the corrected screw speed adjustment amount;

[0031] When the estimated shear rate is not greater than the maximum permissible shear rate, the screw speed adjustment amount in the initial control command vector is adopted as the corrected screw speed adjustment amount.

[0032] Preferably, the collaborative constraint and instruction arbitration module further includes temperature compensation logic:

[0033] Based on the difference between the screw speed adjustment amount in the initial control command vector and the corrected screw speed adjustment amount, and combined with the preset temperature-viscosity compensation gain factor, a compensatory temperature adjustment amount is calculated and generated.

[0034] Preferably, the final constraint arbitration logic for process temperature in the collaborative constraint and instruction arbitration module includes:

[0035] The temperature adjustment amount in the initial control command vector is added to the compensatory temperature adjustment amount to obtain the final proposed temperature adjustment amount;

[0036] The final proposed temperature adjustment is added to the current region temperature in the current device state vector to calculate the final proposed temperature;

[0037] The final proposed temperature is compared with the upper and lower limits of the preset process temperature window;

[0038] When the final proposed temperature is within the process temperature window, the final proposed temperature adjustment is adopted as part of the final safety instruction vector;

[0039] When the final proposed temperature exceeds the process temperature window, the final proposed temperature adjustment is trimmed to keep it within the process temperature window, and the trimmed adjustment is used as part of the final safety command vector.

[0040] This invention provides an improved dynamic proportioning control system for composite tile raw materials, which, compared with the prior art, has the following improvements and advantages:

[0041] 1. By using acoustic fingerprinting and AI models, real-time and quantitative monitoring of the two core microscopic states, melt viscosity and microcapsule breakage rate, has been achieved. This overcomes the blind control problem caused by the inability to observe the internal state in traditional technologies and gives the production process a clear perception capability.

[0042] 2. The system can proactively anticipate and respond to state fluctuations caused by batch differences in raw materials. Through feedforward control and adaptive adjustment of process parameters, it can maintain a high degree of stability in the production process even when the characteristics of raw materials change significantly, thus avoiding downtime or defective products caused by adhering to static parameters.

[0043] 3. The system’s unique command arbitration and compensation logic intelligently coordinates the contradiction between the shear force required for enhanced mixing and the protection of microcapsule integrity. By setting an upper limit for the shear rate and performing intelligent temperature compensation, the process window is broadened, maximizing the effectiveness of the self-healing function in the finished product while ensuring production stability.

[0044] 4. The system establishes a multi-level constraint arbitration mechanism encompassing component ratios, shear rates, and process temperatures as the ultimate safety guarantee. This mechanism rigorously reviews and adjudicates the initial instructions generated by AI, ensuring the absolute safety of the instructions ultimately issued to the equipment and guaranteeing the safe operation of the equipment and the reliability of product quality. Attached Figure Description

[0045] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0046] Figure 1 This is a flowchart of the system of the present invention.

[0047] Figure 2 This is a flowchart of the state-decoupled AI model of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0049] Example 1

[0050] Please see Figure 1 The present invention provides a technical solution for a dynamic proportioning control system for composite tile raw materials, including: an acoustic fingerprint acquisition module, used to acquire composite acoustic signals generated by the flow of polymer melt in a mixing extruder;

[0051] The state decoupling AI model module is used to receive composite acoustic signals and calculate the real-time melt viscosity estimate and the real-time microcapsule breakage rate estimate based on the composite acoustic signals.

[0052] The collaborative feedforward control module is used to generate an initial control command vector based on the real-time melt viscosity estimate and the real-time microcapsule breakage rate estimate, and by comparing them with the preset process target value.

[0053] The collaborative constraint and command arbitration module is used to receive the preliminary control command vector and the current equipment state vector obtained in real time from the hybrid extruder, and to arbitrate the preliminary control command vector according to the preset constraint parameters to output the final safety command vector.

[0054] This embodiment provides a dynamic proportioning control system for composite tile raw materials. The system is applied to a production line that uses a co-rotating twin-screw extruder to produce self-healing polymer composite tiles. During operation, the acoustic fingerprint acquisition module captures the physical dynamics of the polymer melt in real time during processing. This dynamic information carries the coupling state between macroscopic flowability and the integrity of microscopic functional units. The state decoupling AI model module performs deep analysis on the acquired composite acoustic signals, thereby enabling the simultaneous acquisition of real-time melt viscosity estimates characterizing macroscopic processability and real-time microcapsule breakage rate estimates characterizing microscopic functional effectiveness. Based on the deviation between these two decoupled key states and the process objectives, the system works collaboratively... The feedforward control module proactively calculates the initial control command vector for multi-variable collaborative adjustment. As the final decision gateway, the collaborative constraint and command arbitration module reviews, compensates, and adjudicates the initial commands based on strict physical and process boundaries, outputting a set of final safety command vectors that satisfy both control intent and absolute safety, thereby achieving closed-loop intelligent control of the entire production process. The inherent logic of this architecture aims to solve the core technical bottlenecks in traditional processes, such as poor adaptability to raw material fluctuations, inherent conflicts between functions and processes, and blindness and lag in control. This ensures that the production stability and self-healing effectiveness of composite tile products can reach the optimal state when raw material batches fluctuate significantly.

[0055] Example 2

[0056] The acoustic fingerprint acquisition module includes a broadband acoustic sensor that is non-invasively installed at the end of the homogenization section of the mixing extruder barrel.

[0057] In this embodiment, the acoustic fingerprint acquisition module is implemented by non-invasively installing two VallenVS375-M broadband acoustic sensors along the outside of the barrel at the end of the homogenization section of the twin-screw extruder, adjacent to the die. This strategic deployment of broadband acoustic sensors at the end of the homogenization section forms the physical basis for accurate state perception. At this location, the polymer matrix has undergone sufficient melting and preliminary mixing, and its flow state best represents the final melt state before entering the die forming stage. Furthermore, this region is also the final and most critical stage of shearing, where the final breakage behavior of microcapsules is most intense and typical. The acoustic signal-to-noise ratio is highest at this location, containing the richest key information that can be used to decouple melt viscosity and microcapsule breakage state, thus providing the highest quality data input for the subsequent accurate analysis of the state decoupling AI model.

[0058] Example 3

[0059] Please see Figure 2 The state decoupling AI model module includes:

[0060] The low-frequency feature extraction unit is used to perform low-pass filtering and short-time Fourier transform on the composite acoustic signal to extract the energy attenuation coefficient and phase delay as low-frequency feature vectors.

[0061] The viscosity calculation unit is used to output real-time melt viscosity estimates based on low-frequency feature vectors and a viscosity regression model.

[0062] The state decoupling AI model module also includes:

[0063] The high-frequency feature extraction unit is used to perform high-pass filtering and time-domain analysis on the composite acoustic signal to identify and count the acoustic emission pulses generated by the rupture of microcapsules, and extract the pulse amplitude and occurrence frequency as high-frequency feature vectors.

[0064] The breakage rate calculation unit is used to output real-time microcapsule breakage rate estimates based on high-frequency feature vectors and a breakage rate regression model.

[0065] The low-frequency feature extraction unit first performs a low-pass filter (below 100kHz) on the raw acoustic signal to remove high-frequency noise and transient interference from microcapsule rupture, focusing on the signal frequency band reflecting the bulk flow of the melt. The data from this unit is fed into the viscosity calculation unit, which uses a viscosity regression model. This formula is used to calculate real-time melt viscosity, aiming to solve the problem that melt viscosity cannot be directly measured during production. By establishing a mapping relationship between acoustic characteristics and viscosity, it achieves online, non-destructive, and real-time estimation of the viscosity inside a closed extruder; among which, This is the real-time melt viscosity estimate output by the model, in Pascal-seconds (Pa·s). The raw high-fidelity acoustic signal time series data input to the acoustic fingerprint acquisition module, where the subscript acou indicates acoustic and t indicates time; It is a low-frequency feature extraction function that performs low-pass filtering and short-time Fourier transform on the acoustic signal to extract a low-frequency feature vector composed of energy attenuation coefficient, phase delay, etc., which are strongly correlated with macroscopic fluid viscous damping. The subscript η indicates that it is related to viscosity; It is a standalone, lightweight viscosity regression model, such as support vector regression or a small neural network; It is a viscosity regression model The parameter set;

[0066] To further clarify, the viscosity regression model f η (·) If a neural network is used, it can be set as a fully connected feedforward neural network with two hidden layers; the first hidden layer can contain 64 neurons and the second hidden layer can contain 32 neurons, both of which use modified linear units as activation functions. The output layer is a linear neuron that directly outputs the real-time melt viscosity estimate.

[0067] Low-frequency feature vector The calculation method is as follows: signals collected by two acoustic sensors installed along the extruder barrel. and Perform a short-time Fourier transform (STFT) to obtain the values ​​at specific frequencies. Complex spectrum and ,in For time window indexing; the energy decay coefficient can be defined as the ratio of the energy spectra of two sensors within the same time window, i.e. Phase delay can be defined as the phase difference between two signals at the same frequency, i.e. Combining these two features and their statistics forms the final low-frequency feature vector. And input the model;

[0068] Model parameter set It was obtained through supervised learning offline training, the method of which was to run the extruder under various different process conditions and simultaneously collect acoustic signals. The corresponding real melt viscosity label values ​​were obtained using online rheometers or rapid sampling analysis, and the collected signal-label dataset was used to train a regression model. The system can then learn and solidify the quantitative relationship between acoustic characteristics and melt viscosity;

[0069] The high-frequency feature extraction unit performs high-pass filtering above 200kHz on the same raw acoustic signal to specifically capture high-frequency, transient acoustic emission pulses generated when microcapsules fracture under shear stress. This unit then uses a breakage rate calculation unit to apply a breakage rate regression model. This formula aims to estimate the real-time breakage rate of microcapsules; it addresses the fundamental problem of the inability to detect the breakage rate of functional microcapsules online during processing, and achieves quantitative monitoring of the effectiveness of microscopic functions through acoustic means; among which, This is the real-time microcapsule breakage rate estimate output by the model, expressed as a dimensionless percentage, with the subscript "break" indicating breakage. It is the original acoustic signal shared with viscosity calculation; It is a high-frequency feature extraction function that performs high-pass filtering and time-domain analysis on acoustic signals to identify and count acoustic emission pulses generated by microcapsule rupture, and extracts the pulse amplitude and occurrence frequency as high-frequency feature vectors. The subscript R indicates that it is related to the breakage rate; It is another independent damage rate regression model; This is the parameter set of the regression model;

[0070] To further clarify, the breakage rate regression model The network structure can be used with viscosity regression models Maintain consistency to facilitate deployment;

[0071] High-frequency feature vectors The calculation method is as follows: set an energy threshold for the high-pass filtered acoustic signal. This energy threshold The threshold can be determined by collecting background noise signals during the flow of a pure matrix melt without microcapsules. The threshold should be set slightly higher than the maximum instantaneous energy of the background noise signal to ensure a good signal-to-noise ratio and effectively distinguish between the actual rupture signal and equipment background noise. When the threshold is exceeded, it is recorded as an acoustic emission pulse event; the pulse amplitude is the maximum signal amplitude during the event; the occurrence frequency is the total number of AE pulse events recorded within a unit of time. The mean, maximum, and occurrence frequency of the pulse amplitudes are used to construct a high-frequency feature vector. Input model;

[0072] Parameter set Similarly, offline supervised learning was used to determine the process: samples were produced under different shear conditions, acoustic signals were collected simultaneously, and samples of the finished composite tiles were taken. Using microscopic analysis techniques such as scanning electron microscopy, the actual breakage rate of microcapsules in the samples was statistically analyzed as a label. These signal-label data were then used to... Once the model is trained, a precise correlation can be established between high-frequency acoustic features and microcapsule breakage rate;

[0073] This dual-channel decoupling mechanism enables the system to simultaneously and independently extract core state information from both macroscopic processing performance and microscopic functional attributes from a single, mixed acoustic signal source. This constitutes the fundamental premise for achieving subsequent collaborative control and optimization.

[0074] Example 4

[0075] The collaborative feedforward control module includes:

[0076] The error vector generation unit is used to construct an error vector by taking the difference between the process target value and the real-time melt viscosity estimate as the first error term and the difference between the process target value and the real-time microcapsule breakage rate estimate as the second error term.

[0077] The control command calculation unit is used to multiply the error vector with a preset coupling gain matrix to generate a preliminary control command vector.

[0078] In this embodiment, after obtaining a clear state understanding decoupled from the AI ​​model, the core task of the collaborative feedforward control module is to quickly and collaboratively calculate the adjustment command; this function is achieved through a simplified control law based on the idea of ​​a linear quadratic regulator. To achieve this, the control law aims to establish a multi-input multi-output feedforward control framework, transforming independent error signals into a set of coordinated control commands, thereby anticipating and overcoming control conflicts and achieving synchronous optimization of multiple objectives; among which, The initial control command vector output by the control law can be constructed as follows: \left [ {\Delta N,\Delta {T}_{zone3},\Delta {m}_{c}} \right ]^{T} Each component represents the proposed adjustment amount for screw speed (in rpm), the proposed temperature adjustment amount for the third temperature zone (in °C), and the proposed adjustment amount for microcapsule feeding rate (in kg / h). The error vector is constructed by the error vector generation unit, and its structure is: E(t)=\left [ {{\eta}_{target}-{\eta}_{melt},{R}_{target}-{R}_{break}} \right ]^{T} , and These are the process target values ​​preset by the process engineer. and This is the real-time output of the state-decoupled AI model module; It is a preset. The coupling gain matrix, its elements The weight and direction of the influence of the j-th error term on the i-th control output are defined; the coupling gain matrix is ​​also defined. This is the core of the control law, and its numerical value is determined offline through system identification or model-based optimization algorithms. By applying a series of small step inputs to a simulation model or actual equipment and observing the changes in the state output, the dynamic relationship between the system input and output can be identified. Based on this dynamic model, and using optimal control theory algorithms to minimize control error and control energy consumption, the optimal gain matrix can be calculated. ;

[0079] Specifically, the system identification process can be carried out as follows: Operate the system near its steady state, and for each control input and screw speed... The third temperature zone Microcapsule feeding rate The applied amplitude is The step disturbance is continuously monitored, and the output quantities of the two states are continuously monitored, including the melt viscosity. Microcapsule breakage rate The dynamic response curve;

[0080] Each input-output response curve can be fitted with a first-order or second-order transfer function model with time delay. For example, the effect of screw speed on melt viscosity can be modeled as follows: In this way, a complete [data / information] can be obtained. Transfer function matrix It describes the input vector from the control input vector To the error vector This refers to the dynamic relationship of state changes, where: For transfer functions; For Laplace variables; Change in melt viscosity The Laplace transform form; The change in screw speed The Laplace transform form; For process gain, For time delay, It is a time constant; The base of the natural logarithm is a mathematical constant.

[0081] To obtain this transfer function matrix Then, mature multivariable control design methods, such as internal model control or decoupling control strategies, can be used to directly calculate the controller gain matrix. For example, in decoupled control, the controller It can be calculated pseudo-reversal And combined with the desired closed-loop response filter To design, for example ,in A diagonal filter matrix is ​​used to achieve the desired closed-loop response speed and robustness. The matrix represents the steady-state gain of the controller at zero frequency.

[0082] Example 5

[0083] The functional components of the collaborative constraint and instruction arbitration module, namely the proportional constraint arbitration logic, include:

[0084] The estimated microcapsule mass fraction is calculated based on the current microcapsule feeding rate in the current equipment state vector and the microcapsule feeding rate adjustment amount in the preliminary control command vector.

[0085] The estimated microcapsule mass fraction is compared with the preset maximum microcapsule mass fraction;

[0086] When the estimated microcapsule mass fraction is greater than the maximum microcapsule mass fraction, the microcapsule feeding rate adjustment amount in the initial control command vector is adjusted.

[0087] When the estimated microcapsule mass fraction is not greater than the maximum microcapsule mass fraction, the microcapsule feeding rate adjustment in the initial control command vector is adopted.

[0088] In this embodiment, the first and highest priority arbitration executed by the collaborative constraint and instruction arbitration module is the functional component ratio constraint arbitration; this logic first bases on the current microcapsule feeding rate obtained from the device PLC. and current polymer matrix feeding rate Based on the microcapsule feeding rate proposed from upstream, the amount of adjustment is... Through the formula: To calculate the estimated microcapsule mass fraction Where, the subscript current indicates the current value obtained from the PLC in real time, cap indicates the microcapsule, and est indicates the estimated value; this estimated value is then compared with a preset maximum microcapsule mass fraction. Compare; the threshold This was not arbitrarily set, but determined through a series of rigorous offline experiments: small batches of blend samples with different microcapsule mass fractions were prepared, and their microstructure was observed using scanning electron microscopy. The critical addition ratio at which the microcapsules began to show significant aggregation and could no longer maintain a good dispersion in the matrix was determined; this is the critical addition ratio. ;if Exceeded The arbitration logic will initiate a pruning process, rejecting the original argument. Instead, it recalculates a safe adjustment amount. Ensure that the final feeding rate is exactly equal to the microcapsule mass fraction. This approach ensures the physical effectiveness of the product formula, effectively solves the agglomeration problem caused by excessive addition of functional components, and ensures that each microcapsule can be uniformly dispersed in the composite matrix, laying the most basic microstructural foundation for exerting its self-repair function.

[0089] Example 6

[0090] The shear rate constraint arbitration logic of the collaborative constraint and instruction arbitration module includes:

[0091] Based on the current screw speed in the current equipment state vector and the screw speed adjustment amount in the preliminary control command vector, and combined with the preset extruder shear geometry constant, the estimated shear rate is calculated;

[0092] Compare the estimated shear rate with the preset maximum allowable shear rate;

[0093] When the estimated shear rate is greater than the maximum allowable shear rate, the screw speed adjustment amount in the initial control command vector is reduced to obtain the corrected screw speed adjustment amount;

[0094] When the estimated shear rate is not greater than the maximum permissible shear rate, the screw speed adjustment amount in the initial control command vector is adopted as the corrected screw speed adjustment amount.

[0095] The collaborative constraint and instruction arbitration module also includes temperature compensation logic:

[0096] Based on the difference between the screw speed adjustment amount in the initial control command vector and the corrected screw speed adjustment amount, and combined with the preset temperature-viscosity compensation gain factor, a compensatory temperature adjustment amount is calculated and generated.

[0097] In this embodiment, after completing the component ratio verification, the collaborative constraint and instruction arbitration module immediately enters the second core priority arbitration: shear rate constraint arbitration and intelligent temperature compensation.

[0098] Shear rate constraint arbitration logic is based on the current screw speed. and the proposed adjustment amount Combined with an extruder shear geometry constant provided by the equipment manufacturer or calculated based on the screw configuration. Through formula To calculate the estimated shear rate ;in, Represents the shear rate; compare this estimate with the maximum permissible shear rate. The threshold is compared. The design is also based on rigorous experiments: samples are processed at different shear rates, and then the morphology and integrity of the microcapsules are observed using scanning electron microscopy to determine the critical shear rate at which the microcapsules begin to break down on a large scale; if Exceed The system will forcibly reduce the proposed speed adjustment. Calculate the corrected safe speed adjustment amount. This makes the final rotational speed exactly correspond to This step is a direct barrier to protect the functionality of the microcapsules. By applying an absolute upper limit to the mechanical stress during processing, it physically ensures that the vast majority of microcapsules can remain intact in the final composite tile product.

[0099] As a direct extension of this logic, a deeper level of intelligent control logic, namely temperature compensation logic, is activated; the system recognizes that the adjustment amount of the screw speed has been reduced, that is... < This means that the upstream module's attempt to adjust the melt state by increasing shear was not fully achieved. To compensate for this control deviation, this module activates temperature as an auxiliary control measure. To ensure the rigor of the physical meaning and the correctness of the dimensions, a compensatory temperature adjustment is made. The following was calculated using a transformation model: This formula aims to address the problem of how to intelligently compensate for unmet control intentions after control commands have been pruned under physical constraints; it embodies the logical advancement of the system from passive limitation to active compensation; among which, This is a temperature adjustment amount generated internally by this module and used for compensation. It is the original screw speed adjustment proposed from upstream. It is the safe speed adjustment amount obtained after shear rate constraint; It is a preset temperature-viscosity gain factor, with units of (Pa·s) / °C, which characterizes the degree of influence of temperature change on melt viscosity. The value is usually negative and is identified by a small amount of offline process experimental data or estimated based on the Arrhenius model of the material. It is a newly added screw speed-viscosity gain factor, which is necessary to ensure the feasibility of the model. The unit is (Pa·s) / rpm. It characterizes the degree of influence of screw speed change on melt viscosity. It is also obtained through system identification experiments.

[0100] The system identification experiment is performed as follows:

[0101] To determine the screw speed-viscosity gain factor While keeping all feed rates and temperature settings constant, adjust the screw speed centered on the current operating point. Make multiple small, step adjustments, for example, rpm; After each adjustment, wait for the system to reach a new steady state and record the melt viscosity estimate output by the state-decoupled AI model. ; Fitting multiple pairs using linear regression The slope of the data points can be obtained. The estimated value is given in (Pa·s) / rpm. Mathematically, ;

[0102] To determine the temperature-viscosity gain factor Using a similar method, the screw speed and feed rate are kept constant, and the set temperature of the third temperature zone is adjusted. Make multiple small, step adjustments, for example, °C; similarly, record the change in viscosity when the new steady state is reached. ; through multiple groups By performing linear regression on the data, we can obtain... The estimated value is given in units of (Pa·s) / °C; mathematically, This value is usually negative, which aligns with physical intuition;

[0103] The underlying logic of this formula is: through The speed adjustment that failed to be executed Converted to an equivalent viscosity control deviation, and then through... The system calculates the required temperature change to produce a compensation effect equivalent to the viscosity deviation. This mechanism transforms the originally independent constraint checking and control execution into an intelligent arbitration closed loop with compensation logic, enabling the system to exhibit higher control flexibility and goal achievement capability when facing physical boundaries.

[0104] Example 7

[0105] The final constraint arbitration logic for process temperature in the collaborative constraint and instruction arbitration module includes:

[0106] The temperature adjustment amount in the initial control command vector is added to the compensatory temperature adjustment amount to obtain the final proposed temperature adjustment amount;

[0107] The final proposed temperature adjustment is added to the current region temperature in the current device state vector to calculate the final proposed temperature;

[0108] The final proposed temperature is compared with the upper and lower limits of the preset process temperature window;

[0109] When the final proposed temperature is within the process temperature window, the final proposed temperature adjustment amount is adopted as part of the final safety command vector;

[0110] When the final proposed temperature exceeds the process temperature window, the final proposed temperature adjustment is trimmed to keep it within the process temperature window, and the trimmed adjustment is used as part of the final safety command vector.

[0111] In this embodiment, the final line of defense of the collaborative constraint and instruction arbitration module is the final constraint arbitration of process temperature; this logic will determine the temperature adjustment amount originally proposed by the upstream collaborative feedforward control module. Compensatory temperature adjustment amount arising from the higher-level arbitration Algebraic summation yields a final proposed temperature that integrates both proactive and compensatory adjustment intentions. The calculation formula is: In this formula, The current actual temperature of the third temperature zone is obtained in real time from the equipment's PLC; the next step for the system is to... The safe process temperature window is set based on the thermal properties of the polymer matrix and the thermal stability of the microcapsules. [{{T}_{min},{T}_{max}}] Compare; if If the temperature adjustment command falls within this window, it is considered safe and will be ultimately adopted; otherwise, if... Exceeding the upper boundary or below the lower boundary The instructions will be forcibly calibrated to ensure that the final instructions issued to the heater maintain the temperature precisely at the boundary value. This final safety baseline arbitration ensures that regardless of the decisions made by the front-end AI model and control algorithm, all instructions ultimately executed on the physical device remain within an absolutely safe operating range. This guarantees the normal processing of polymer materials while preventing functional microcapsules from failing due to overheating, forming the cornerstone of the stable and reliable operation of the entire intelligent control system. Finally, the safety instruction vector after all three levels of arbitration is defined as follows: The calibrated adjustment amount is included as part of the final safety instruction vector, denoted as... ;The value is determined by: {U}_{safe}(t)=\left [ {\Delta {N}_{safe},\Delta {T}_{zone3,safe},\Delta {m}_{c,safe}} \right ]^{T} The output is sent to the PLC actuator, where the subscript "safe" indicates the final instruction that is absolutely safe after arbitration.

[0112] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A dynamic proportioning control system for composite tile raw materials, characterized in that, include: The acoustic fingerprint acquisition module is used to acquire composite acoustic signals generated by the flow of polymer melt in the mixing extruder; The state decoupling AI model module is used to receive the composite acoustic signal and calculate the real-time melt viscosity estimate and the real-time microcapsule breakage rate estimate based on the composite acoustic signal. The collaborative feedforward control module is used to generate a preliminary control command vector based on the real-time melt viscosity estimate and the real-time microcapsule breakage rate estimate, and by comparing them with the preset process target value. The collaborative constraint and command arbitration module is used to receive the preliminary control command vector and the current equipment state vector obtained in real time from the hybrid extruder, and to arbitrate the preliminary control command vector according to the preset constraint parameters to output the final safety command vector.

2. The composite tile raw material dynamic proportioning control system according to claim 1, characterized in that, The acoustic fingerprint acquisition module includes a broadband acoustic sensor that is non-invasively installed at the end of the homogenization section of the mixing extruder barrel.

3. The composite tile raw material dynamic proportioning control system according to claim 1, characterized in that, The state decoupling AI model module includes: The low-frequency feature extraction unit is used to perform low-pass filtering and short-time Fourier transform on the composite acoustic signal to extract the energy attenuation coefficient and phase delay as low-frequency feature vectors. The viscosity calculation unit is used to output the real-time melt viscosity estimate based on the low-frequency feature vector and through a viscosity regression model.

4. The composite tile raw material dynamic proportioning control system according to claim 1, characterized in that, The state decoupling AI model module also includes: The high-frequency feature extraction unit is used to perform high-pass filtering and time-domain analysis on the composite acoustic signal to identify and count the acoustic emission pulses generated by the microcapsule rupture, and extract the pulse amplitude and occurrence frequency as high-frequency feature vectors. The breakage rate calculation unit is used to output the real-time microcapsule breakage rate estimate based on the high-frequency feature vector and through a breakage rate regression model.

5. The composite tile raw material dynamic proportioning control system according to claim 1, characterized in that, The collaborative feedforward control module includes: An error vector generation unit is used to construct an error vector by taking the difference between the process target value and the estimated real-time melt viscosity as a first error term and the difference between the process target value and the estimated real-time microcapsule breakage rate as a second error term. The control command calculation unit is used to multiply the error vector with a preset coupling gain matrix to generate the preliminary control command vector.

6. The composite tile raw material dynamic proportioning control system according to claim 1, characterized in that, The functional component proportional constraint arbitration logic of the collaborative constraint and instruction arbitration module includes: Based on the current microcapsule feeding rate in the current device state vector and the microcapsule feeding rate adjustment amount in the preliminary control command vector, the estimated microcapsule mass fraction is calculated. The estimated microcapsule mass fraction is compared with the preset maximum microcapsule mass fraction; When the estimated microcapsule mass fraction is greater than the maximum microcapsule mass fraction, the microcapsule feeding rate adjustment amount in the preliminary control command vector is reduced; When the estimated microcapsule mass fraction is not greater than the maximum microcapsule mass fraction, the microcapsule feeding rate adjustment in the preliminary control command vector is adopted.

7. The composite tile raw material dynamic proportioning control system according to claim 1, characterized in that, The shear rate constraint arbitration logic of the collaborative constraint and instruction arbitration module includes: Based on the current screw speed in the current equipment state vector and the screw speed adjustment amount in the preliminary control command vector, and in combination with the preset extruder shear geometry constant, the estimated shear rate is calculated; The estimated shear rate is compared with the preset maximum allowable shear rate; When the estimated shear rate is greater than the maximum permissible shear rate, the screw speed adjustment amount in the initial control command vector is trimmed to obtain the corrected screw speed adjustment amount; When the estimated shear rate is not greater than the maximum permissible shear rate, the screw speed adjustment amount in the initial control command vector is adopted as the corrected screw speed adjustment amount.

8. The composite tile raw material dynamic proportioning control system according to claim 7, characterized in that, The collaborative constraint and instruction arbitration module also includes temperature compensation logic: Based on the difference between the screw speed adjustment amount in the initial control command vector and the corrected screw speed adjustment amount, and combined with the preset temperature-viscosity compensation gain factor, a compensatory temperature adjustment amount is calculated and generated.

9. The composite tile raw material dynamic proportioning control system according to claim 8, characterized in that, The final constraint arbitration logic for process temperature in the collaborative constraint and instruction arbitration module includes: The temperature adjustment amount in the initial control command vector is added to the compensatory temperature adjustment amount to obtain the final proposed temperature adjustment amount; The final proposed temperature adjustment is added to the current region temperature in the current device state vector to calculate the final proposed temperature; The final proposed temperature is compared with the upper and lower limits of the preset process temperature window; When the final proposed temperature is within the process temperature window, the final proposed temperature adjustment is adopted as part of the final safety instruction vector; When the final proposed temperature exceeds the process temperature window, the final proposed temperature adjustment is trimmed to keep it within the process temperature window, and the trimmed adjustment is used as part of the final safety command vector.

Images