Process control method for multiple recycling of regenerated polyethylene for tailings imperviousness

Abstract

This invention relates to the field of process control technology and discloses a process control method for the multiple recycling of recycled polyethylene for tailings seepage prevention. The method includes: synchronously acquiring melt pressure, screw drive motor current, and control input sequence; calculating the cross-correlation function between current and pressure and identifying physical transmission lag time; real-time monitoring of the non-command residual component of the current; extrapolating the amplitude of impending pressure fluctuations based on the cross-correlation function and generating an inertial feedforward compensation quantity; using a dual-timescale model to update parameters online and constructing a disturbance observer to generate a feedback compensation control quantity; and superimposing the feedforward and feedback compensation onto the basic output quantity to generate the final drive command. This invention utilizes a heterogeneous signal timing correlation mechanism to implement reverse adjustment before pressure fluctuations are transmitted to the die head via the screw, avoiding control blind spots caused by physical transmission lag.

Description

Technical Field

[0001] This invention relates to a process control method for the multiple recycling of recycled polyethylene for tailings seepage prevention, belonging to the field of process control technology. Background Technology

[0002] In the current tailings dam seepage prevention engineering field, to reduce material costs and meet environmental protection and recycling requirements, recycled polyethylene, which has been recycled multiple times, has become the main raw material for manufacturing geomembranes. For the extrusion production process of such raw materials, the dynamic stability of melt pressure is crucial to ensuring the consistency of the physical properties of the geomembrane. Industrial sites generally adopt PID control strategies based on fixed parameters or conventional model predictive control algorithms to adjust the screw speed or feeding speed to maintain the melt pressure setpoint. This type of control method relies on the controlled object having a relatively certain static gain and time constant, and can achieve good steady-state control effect when processing raw materials with uniform properties. However, the inherent limitations of the above control methods are revealed when applied to the multiple recycling process of recycled polyethylene. The recycled material undergoes different degrees of photo-oxidative aging, shear degradation, and multiple thermal histories, and is mixed with impurities from complex sources, resulting in random time-varying molecular weight distribution, crosslinking degree, and rheological properties. The drastic fluctuations in raw material properties cause the process model parameters of the controlled object in the extruder to no longer maintain constant values, and to drift significantly with the batch of raw materials or even the feeding time within the same batch. Under this condition, traditional controllers based on linear time-invariant assumptions cannot detect changes in object parameters, and often experience response hysteresis or overshoot oscillation due to parameter mismatch, resulting in uneven product thickness or even local overheating and degradation.

[0003] Existing technologies often focus on maintaining production by optimizing screw combinations or setting static process windows, neglecting dynamic intervention in unsteady processes. For example, Chinese invention patent CN112390996B discloses a modified recycled polyethylene plastic for plastic pipes and its preparation method. Although it ensures the mixing effect of the substrate by limiting the hardware configuration of a single-screw counter-conical twin-screw two-stage extrusion and specific temperature and speed ranges, the control logic is essentially still in the category of open-loop or simple set-value regulation. Faced with the drastic rheological drift between batches of recycled materials, this kind of static control mode based on fixed formula and fixed parameters lacks real-time perception and feedforward response mechanisms for the process state. It cannot predict and compensate before melt pressure fluctuations occur, and it is difficult to eliminate transient impacts caused by material heterogeneity, resulting in the extrusion stability of the final product being difficult to fully control.

[0004] Therefore, how to construct a method for real-time identification of time-varying parameters of the controlled object, avoid physical transmission lag, and decouple mechanical noise control, so as to achieve high-precision and stable control of the recycled polyethylene extrusion process under conditions of strong fluctuations in raw material characteristics, has become the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A method for controlling the multiple recycling process of recycled polyethylene for tailings seepage prevention, operating in an electronic control unit connected to an extrusion production line, the extrusion production line including a screw drive motor and a melt pressure sensor installed at the end of the die head, the method comprising the following steps:

[0006] The multidimensional heterogeneous data synchronous acquisition step synchronously acquires the melt pressure sequence output by the melt pressure sensor, the drive current sequence of the screw drive motor, and the current control input sequence at a preset sampling frequency, and establishes a timestamp alignment reference between the melt pressure sequence, the drive current sequence, and the control input sequence.

[0007] The physical transmission lag feature extraction step involves calculating the cross-correlation function between the driving current sequence and the melt pressure sequence, and based on the time offset of the peak value of the cross-correlation function, identifying the physical transmission lag time of the load mutation feature in the driving current sequence relative to the pressure fluctuation feature in the melt pressure sequence.

[0008] The advanced disturbance targeted simulation step monitors the drive current sequence in real time and filters out the fundamental component caused by speed command changes to extract the non-command residual component. When the amplitude of the non-command residual component exceeds the preset noise shielding threshold, the amplitude of the melt pressure fluctuation that will reach the end of the machine head after the physical transmission lag time is simulated by using the physical transmission lag time and cross-correlation function.

[0009] The time-domain reverse compensation step generates an inverse inertial feedforward compensation amount based on the derived melt pressure fluctuation amplitude, and superimposes the inertial feedforward compensation amount into the basic control amount generated by the electronic control unit to generate the final drive command. The final drive command drives the screw drive motor to perform reverse adjustment action before the load change characteristics are physically transmitted to the end of the die head and cause pressure fluctuations.

[0010] Preferably, the method further includes a dual-timescale model evolution step. While performing the time-domain inverse compensation step, the recursive least squares algorithm is used to process the melt pressure sequence and control input sequence to update the coefficients of the discretized parameter model describing the dynamic characteristics of the extrusion process online. The updated discretized parameter model is used to construct a disturbance observer to calculate the deviation between the theoretical model output and the measured melt pressure. The deviation of the disturbance observer output is processed by a low-pass filter to generate a feedback compensation control quantity, and the feedback compensation control quantity and the inertial feedforward compensation quantity are superimposed on the basic control quantity.

[0011] Preferably, in the dual-timescale model evolution step, the coefficient update of the discretized parameter model follows the following recursive operation rule that includes a forgetting factor: ,in, For a moment The model parameter vector, For a moment The measured value of the melt pressure. A regression vector containing historical input and output data. The gain matrix is ​​calculated using a pre-defined forgetting factor. This forgetting factor reduces the weight of historical data on the current model parameter updates to track process gain drift caused by component wear in the extrusion production line.

[0012] Preferably, the deviation output by the disturbance observer needs to undergo periodic noise decoupling processing before entering the low-pass filter. This processing includes: acquiring the rotational speed command of the screw drive motor in real time, calculating the current inherent pulsation frequency of the equipment based on the screw's geometric parameters, configuring a dynamic notch filter whose center frequency is locked to the inherent pulsation frequency of the equipment in real time, using the dynamic notch filter to process the deviation, eliminating the periodic component synchronized with the screw rotation, and retaining the random disturbance component caused by fluctuations in raw material characteristics.

[0013] Preferably, the cross-correlation function is calculated based on the data segment within the sliding time window; the physical transmission lag feature extraction step further includes: when the maximum correlation coefficient of the cross-correlation function is lower than the preset confidence threshold, it is determined that no raw material fluctuation transmission feature that meets the preset correlation requirements has appeared at present, and the physical transmission lag time identified at the previous moment remains unchanged until the new maximum correlation coefficient exceeds the confidence threshold.

[0014] Preferably, the extraction of non-command residual components is achieved by: constructing a command response reference model of the screw drive motor, inputting the control input sequence into the command response reference model to generate a theoretical current response sequence; calculating the difference between the drive current sequence and the theoretical current response sequence as a non-command residual component, which characterizes the load fluctuation caused by changes in the rheological properties of the raw materials.

[0015] Preferably, the generation of the inertial feedforward compensation amount follows a single-value limit logic; a maximum allowable compensation gradient is set, and when the rate of change of the compensation amount generated according to the deduction exceeds the maximum allowable compensation gradient, the slope of the inertial feedforward compensation amount is limited to prevent the rate of change of screw speed caused by the compensation action from exceeding the mechanical bearing limit of the screw drive motor.

[0016] Preferably, the final drive commands include screw speed commands and feeding speed commands; the time-domain reverse compensation step uses a preset decoupling allocation matrix to proportionally allocate the inertial feedforward compensation amount to the screw speed commands and feeding speed commands, thereby maintaining material mass conservation within the extrusion production line while maintaining stable melt pressure.

[0017] Preferably, the method further includes a melt temperature safety constraint step, which involves real-time monitoring of the melt temperature and rate of change at the end of the die head; when the rate of change of the melt temperature exceeds a preset shear heat accumulation threshold, the gain coefficient of the inertial feedforward compensation is reduced, and the control strategy is switched to a safety mode that prioritizes suppressing temperature rise, so as to prevent thermal degradation of recycled polyethylene due to reverse adjustment action.

[0018] Preferably, the control input sequence includes a screw speed setpoint, and the physical transmission lag time is defined as a variable that is dynamically adjusted as the screw speed setpoint changes. The method includes establishing an inverse mapping table between the physical transmission lag time and the screw speed setpoint. When an effective cross-correlation function cannot be calculated, the estimated physical transmission lag time is obtained by consulting the inverse mapping table based on the current screw speed setpoint.

[0019] Compared with the prior art, the beneficial effects of the present invention are:

[0020] 1. This invention constructs a dual-timescale parameter identification and disturbance observation architecture to resolve the technical contradiction that fixed parameters of the control model cannot adapt to the time-varying characteristics of the controlled object. The control system uses a recursive algorithm to correct the process gain and time constant in real time on a slow timescale, locking in the static characteristic drift caused by equipment wear or raw material batch switching. On a fast timescale, a disturbance observer is used to estimate and compensate for instantaneous pressure fluctuations caused by raw material impurities. Time-frequency domain decoupling enables the control system to maintain the basic accuracy of the model while having the ability to quickly suppress high-frequency nonlinear disturbances. This ensures that the melt pressure control system maintains dynamic balance under the condition of random changes in the rheological properties of the raw materials, eliminating the risk of system oscillation caused by model mismatch.

[0021] 2. By utilizing the time-series cross-correlation characteristics between the screw drive motor current and the melt pressure, the physical transmission lag problem caused by relying solely on pressure feedback is avoided. The system calculates the time window in which current fluctuations lead pressure fluctuations, establishes a dynamic correlation model between heterogeneous signals, and generates an inertial feedforward compensation quantity before the pressure sensor senses the fluctuation. The actuator load current signal is multiplexed into an advanced sensing signal reflecting the sudden change in the rheology of the raw material, filling the control blind spot caused by fluid transmission on the time axis, and achieving zero or even negative lag response to sudden disturbances in the raw material.

[0022] 3. By introducing dynamic frequency locking filtering based on screw speed command, the inherent pulsation noise of the equipment and random disturbances of raw materials are separated in the frequency domain. The control system calculates the fundamental frequency characteristic frequency based on the real-time screw speed and locks the center frequency of the dynamic notch filter to the characteristic frequency in real time. This accurately filters out the pressure pulsation component caused by the mechanical cycle of screw rotation, ensuring that the disturbance observer only generates compensation commands for random deviations caused by changes in raw material characteristics. This avoids the actuator from making meaningless high-frequency adjustments to eliminate normal mechanical pulsation, improves the signal-to-noise ratio of the control signal, and reduces the energy consumption and mechanical wear of the drive mechanism. Attached Figure Description

[0023] Figure 1 This is a flowchart of the collaborative control logic for dual-timescale model evolution and physical lag compensation in this invention.

[0024] Figure 2 This is a use case diagram illustrating the interaction between the multi-dimensional functional modules and hardware of the process control system of this invention.

[0025] Figure 3 This is a diagram showing the hardware topology and signal transmission architecture of the control system integrating heterogeneous sensing nodes in this invention. Detailed Implementation

[0026] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0027] This invention discloses a process control method for the multiple recycling of recycled polyethylene for tailings seepage prevention. The method operates on an electronic control unit connected to an extrusion production line, which includes a screw drive motor and a melt pressure sensor installed at the end of the die head. The electronic control unit performs a multi-dimensional heterogeneous data synchronous acquisition step, synchronously acquiring the melt pressure sequence output by the melt pressure sensor at a preset sampling frequency. Drive current sequence of screw drive motor and the current control input sequence The system establishes a melt pressure sequence. Drive current sequence With control input sequence The timestamp alignment benchmark between the two is used to eliminate communication transmission timing deviations; based on the extraction of physical transmission lag characteristics, the electronic control unit calculates the drive current sequence. With melt pressure sequence The drive current sequence is identified based on the time offset of the cross-correlation function within the sliding time window and the peak value of the cross-correlation function. Medium load mutation characteristics relative to melt pressure sequence Physical transmission lag time of medium pressure fluctuation characteristics When the maximum correlation coefficient of the cross-correlation function is lower than the preset confidence threshold, the system maintains the physical transmission lag time identified at the previous moment unchanged or consults the pre-established inverse mapping table based on the current screw speed setting to obtain the estimated physical transmission lag time.

[0028] In the advanced disturbance targeting simulation step, the electronic control unit monitors the drive current sequence in real time. To eliminate the fundamental frequency component caused by speed command changes, the system constructs a command response reference model for the screw drive motor, and incorporates the control input sequence... Input the model to generate the theoretical current response, and calculate the measured driving current sequence. The difference between the current response and the theoretical current response is used to extract the non-command residual component. When the amplitude of this non-command residual component exceeds a preset noise shielding threshold, the system utilizes the identified physical transmission lag time. Using the cross-correlation function, the amplitude of the melt pressure fluctuation that will reach the end of the head after the physical transmission lag time is calculated. Based on the calculated amplitude of the melt pressure fluctuation, an inverse inertial feedforward compensation is generated. The inertial feedforward compensation is then added to the basic control quantity. This step drives the screw drive motor to perform a reverse adjustment action before the load abrupt change characteristics are physically transmitted to the end of the die head and cause pressure fluctuations. The method performs a dual-timescale model evolution step. On the slow timescale, the electronic control unit uses a recursive least squares algorithm to process the melt pressure sequence. and control input sequence The coefficients of the discretized parameter model describing the dynamic characteristics of the extrusion process are updated online. This model update follows a recursive operation rule that includes a forgetting factor. The forgetting factor reduces the weight of historical data on the current model parameter update, thereby tracking the process gain drift caused by component wear or raw material batch switching in the extrusion production line. On a fast timescale, a disturbance observer is constructed using the updated discretized parameter model to calculate the deviation between the theoretical model output and the measured melt pressure. To achieve decoupling of periodic noise, this deviation needs to be processed by a dynamic notch filter before entering the low-pass filter. The electronic control unit obtains the screw drive motor speed command in real time, calculates the current inherent pulsation frequency of the equipment based on the screw geometry parameters, and locks the center frequency of the dynamic notch filter to the inherent pulsation frequency of the equipment in real time. The dynamic notch filter is used to remove the periodic component of the deviation that is synchronized with the screw rotation, and retains the random disturbance component caused by the fluctuation of raw material characteristics. The filtered deviation generates a feedback compensation control quantity through the low-pass filter. The feedback compensation control quantity and the inertial feedforward compensation quantity These are combined and superimposed on the basic control quantity to generate the final drive instruction.

[0029] The system operates on melt temperature safety constraint logic, monitoring the melt temperature and rate of change at the end of the die head in real time. When the rate of change of the melt temperature exceeds the preset shear heat accumulation threshold, the electronic control unit reduces the inertial feedforward compensation. The gain coefficient is adjusted to switch the control strategy to a safety mode that prioritizes suppressing temperature rise, preventing thermal degradation of recycled polyethylene due to excessive screw speed adjustment. The final drive command is proportionally distributed to the screw speed command and feed rate command through a preset decoupling allocation matrix, maintaining material mass conservation within the extrusion production line. Before closed-loop control, non-command residual component reference noise calibration is initiated. Under constant screw speed and stable thermal steady-state feed rate, the screw drive motor current sequence is continuously acquired at 50Hz for 300 seconds and detrended, and the statistical standard deviation of the residual signal is calculated. Noise shielding threshold Set as ,coefficient Using values ​​of 3 to 5, the statistical boundaries of background electromagnetic noise and load mutation are defined. The decoupling allocation matrix is ​​constructed following the open-loop step response identification procedure. During extrusion steady state, independent step excitations of 5% of the rated value are applied to the screw speed command and the feed speed command respectively. The melt pressure response curve is recorded, and the change after reaching the new steady state is extracted. Based on this, the screw speed sensitivity coefficient to pressure is calculated. and the sensitivity coefficient of feeding speed to pressure A two-dimensional decoupling matrix is ​​constructed, and the scalar inertial feedforward compensation amount is used by the matrix inverse. Decomposed into complementary screw speed adjustment vector Feeding speed adjustment vector Pressure is adjusted under the constraint of material mass conservation; the melt temperature safety constraint is based on offline calibration data of material thermal stability boundary. The calibration is based on ASTM D3895-03 (Standard Test Method for Determining Oxidation Induction Time of Polyolefins by Differential Scanning Calorimetry) to determine the oxidation induction period (OIT) of recycled polyethylene raw materials, and the relationship between OIT decay rate and melt temperature change rate is established. The nonlinear mapping defines the rate of temperature change corresponding to the OIT value falling to 80% of its initial value as the critical degradation threshold. The runtime shear heat accumulation threshold is limited to When the rate of temperature change exceeds the safety boundary, the gain decay logic is forcibly triggered to suppress the generation of shear heat.

[0030] Example 1: This example is set in a specific scenario of a continuous extrusion production line for tailings dam geomembrane. In this scenario, the production line uses recycled polyethylene containing randomly distributed hard impurities and with drastic fluctuations in melt flow index as raw material. The rheological properties of the raw material drift nonlinearly between batches and within batches, and the unpredictable mixing of hard particles causes sudden changes in screw load, resulting in lag and overshoot oscillation in the conventional feedback control system, which threatens the physical performance stability of the final geomembrane material.

[0031] Under this operating condition, the electronic control unit synchronously acquires the melt pressure sequence in real time. With screw drive motor current sequence Calculate and lock the physical transmission lag time of current fluctuations relative to pressure fluctuations. When high-viscosity, hard impurities enter the plasticizing section, causing a sudden change in screw load, the system compares the measured current with the output of the command response reference model, extracts the non-command residual component, and based on this residual component and... The system generates an inverse inertial feedforward compensation quantity before the melt pressure sensor detects the pressure fluctuation. This is then superimposed onto the drive command. Simultaneously, the disturbance observer utilizes a dynamic notch filter with its center frequency locked to the inherent pulsation frequency of the equipment to eliminate periodic mechanical noise synchronized with the screw speed in the melt pressure signal. Feedback compensation control quantities are then generated to address random disturbances caused by the raw material characteristics. The FF-RLS algorithm updates the coefficients of the discretized parameter model online on a slow time scale, tracking the process gain changes caused by the rheological state of the raw materials. This control mechanism uses the time lead characteristic of the current signal to fill the physical lag blind zone of the pressure feedback, and completes the speed counter-adjustment before impurities reach the die head and cause pressure fluctuations. The dynamic notch filter separates the inherent pulsation of the equipment and the random disturbance of the raw materials, preventing the actuator from making ineffective adjustments for mechanical noise. Under the constraint of keeping the melt temperature change rate below the shear heat accumulation threshold, the system maintains the melt pressure fluctuations in the recycled polyethylene extrusion process within the set range, ensuring the consistency of geomembrane product quality.

[0032] Example 2: This example aims to verify the effectiveness of the process control method of the present invention under high noise and strong nonlinear conditions through comparative experiments, and to empirically define the preferred numerical range of key control parameters. The experiment relies on an industrial-grade single-screw extrusion experimental platform, which is equipped with a sampling frequency of [missing information]. The high-frequency data acquisition system was used to construct a rigorous testing environment that closely approximates the actual tailings geomembrane production site. The experimental raw material was waste polyethylene shredded material that had undergone three thermal history cycles, with a melt flow index (MI) of [missing value]. to The values ​​exhibit drastic and random fluctuations, and the premixed mass fraction in the raw materials is... High-melting-point polypropylene particles were used to simulate unpredictable hard impurity disturbances. Simultaneously, to verify the system's ability to suppress signal noise, a signal-to-noise ratio of [value missing] was actively superimposed on the original signal channel of the melt pressure sensor. Gaussian white noise, coupled at a frequency of Power frequency interference signals.

[0033] To construct a legally valid chain of evidence, the experimental design incorporates a multi-dimensional control system with missing gradient parameters and partial features. Control group 1 uses a conventional fixed-parameter PID controller as the benchmark for existing technologies. Control group 2 employs a dual-timescale model evolution step but removes the inertial feedforward compensation based on current cross-correlation, aiming to verify the synergistic effect of heterogeneous signal fusion through partial missing features. Control groups 3 and 4 employ the complete control strategy, but respectively remove the forgetting factor from the recursive least squares algorithm. Set as and Located in the scope defined by this invention to Outside the scope, the aim is to verify the critical significance of the parameter range; finally, the sample group of the present invention is set to enable the complete technical solution, and... Set as Each group has its screw speed set to [value missing]. The target melt pressure is set to Continuous operation under the same working conditions During the test, the melt pressure and temperature sequences at the die head were recorded in real time, and the pressure standard deviation during steady-state operation was calculated. and maximum temperature deviation As a quantitative evaluation indicator.

[0034] Table 1: Comparison of Performance Indicators under Different Control Strategies

[0035]

[0036] Referring to Table 1, the data from control group 1 show that under conditions where the rheological properties of the raw material change over time and are accompanied by impacts from hard impurities, the fixed-parameter PID controller cannot maintain pressure stability. Gundam Furthermore, accompanied by temperature fluctuations, data from control group 2 showed that although adaptive model updates were introduced, the system could not overcome physical transmission lag due to the lack of current-based inertial feedforward compensation. Consequently, when impurities caused sudden load changes through the screw, pressure fluctuations still reached [a certain level]. The comparison results between the sample group of this invention and control group 2 confirm that the time-domain inverse compensation step has an irreplaceable synergistic effect in suppressing transient disturbances; further comparison of the data of control groups 3 and 4 with the sample group of this invention shows that the forgetting factor The value of directly determines the stability and agility of the model's recognition. At times, the model forgets historical data too quickly, causing parameter estimates to fluctuate wildly due to measurement noise. Deteriorated to ;when At that time, the model was not sensitive enough to new data and could not track the process gain drift caused by raw material batch switching in a timely manner, resulting in a lag in control response. Only the sample of this invention could achieve this. Realize for The optimal control accuracy is achieved. Furthermore, spectrum analysis shows that the control commands output by the sample group of this invention do not contain any elements that appear in the control commands. The peak energy corresponding to the power frequency and the fundamental frequency of the screw rotation speed.

[0037] Example 3: This example combines Figures 1 to 3 The process control method for multiple recycling of recycled polyethylene used for tailings seepage prevention is explained, such as... Figure 1 As shown, this step establishes a timestamp alignment benchmark for pressure, current, and control sequences. The processing flow is divided into two parallel logical branches. The left branch performs dual-timescale model evolution to update process gain online and track the drift of raw material rheological properties. Then, it locks the inherent pulsation frequency of the equipment through a dynamic notch filter to eliminate periodic mechanical noise. Subsequently, it uses a disturbance observer to generate feedback compensation control quantity based on denoising deviation. The right branch performs physical transmission lag feature extraction to identify the lead time window of current fluctuations relative to pressure fluctuations. It performs targeted inference of advance disturbances to extract non-command residual components and infer the amplitude of pressure fluctuations. It performs time-domain inverse compensation to generate anti-phase inertial feedforward quantity to eliminate physical lag blind zone. The calculation results of the two branches are finally converged to generate drive command. This command is superimposed with compensation quantity and proportionally distributed to screw speed and feed rate. At the same time, the system executes melt temperature safety constraint logic in parallel, monitors the shear heat accumulation threshold in real time, and prevents thermal degradation of recycled material through forced yield reduction path when necessary.

[0038] like Figure 2 As shown, the process operator is responsible for setting the process target parameters. The system performs multi-dimensional heterogeneous data synchronous acquisition and timestamp benchmark alignment by connecting the melt temperature sensor and the melt pressure sensor. The core processing functions include generating inertial feedforward compensation by identifying physical transmission lag and extracting non-command current residuals; generating feedback compensation control by updating the dual time scale model and decoupling periodic noise; and executing melt temperature safety constraints and triggering a safety degradation mode when the shear heat accumulation exceeds the threshold. The above functional modules work together and finally converge into the output drive command function, directly controlling the screw drive motor to perform corresponding adjustment actions; such as Figure 3As shown, the system hardware topology is centered on the Electronic Control Unit (ECU). This unit integrates a dual-timescale evolution model, a cross-correlation hysteresis analysis algorithm, and a dynamic notch filter algorithm. It is connected to the human-machine interface terminal above via industrial Ethernet communication for parameter configuration and alarm recording. The system input is connected to a field sensing node consisting of a melt pressure sensor and a melt temperature sensor at the end of the die head, and a current transformer on the drive side. Signals are transmitted through a multi-dimensional heterogeneous data acquisition link. The system output is connected to a reverse adjustment node consisting of a screw drive motor servo driver and a feed speed regulating motor. The drive commands output by the ECU act on this adjustment node to implement closed-loop control of the tailings seepage-proof recycled polyethylene extrusion production line, which includes the screw, barrel, and physical transmission hysteresis characteristics.

[0039] Example 4: This example aims to systematically explain the calibration procedure for the noise shielding threshold and the construction logic of the decoupling allocation matrix involved in the process control method, in order to eliminate potential uncertainties in parameter setting and multivariable collaborative control, and ensure the reproducibility and stability of the control strategy under different equipment and operating conditions. After the extrusion production line completes preheating and enters thermal steady-state operation, the electronic control unit executes the reference noise calibration procedure for the non-command residual components. During this period, the system forcibly locks the screw speed command and the feed rate command, keeping them constant, and... The sampling frequency and continuous acquisition duration are The system performs detrending processing on the current data within the screw drive motor for a given time window to filter out extremely low-frequency components caused by temperature drift or mechanical break-in, and calculates the standard deviation of the remaining high-frequency random components. Based on statistical principles, the electronic control unit sets a preset noise shielding threshold. Set as this standard deviation Doubled times, that is (In this embodiment) Values This calibration procedure establishes a criterion based on measured noise levels to ensure that the system responds only to statistically significant load fluctuations.

[0040] Furthermore, the system executes the identification procedure of the decoupling allocation matrix to quantify the degree of coupling influence of the two input variables, screw speed and feed rate, on the output variable, melt pressure. In open-loop control mode, the system applies an amplitude of [missing value] to the screw speed command. The step signal was recorded while the feed rate remained constant. The melt pressure response curve was recorded and the change after reaching a new steady state was extracted. Based on this, the first sensitivity coefficient of screw speed to pressure was calculated. After the system returns to its initial equilibrium state, an amplitude of [value] is applied to the feeding speed command. While maintaining a constant screw speed, the step signal is recorded, the change in melt pressure is observed, and the second sensitivity coefficient of the feed rate to pressure is calculated. Based on these two measured sensitivity coefficients, the electronic control unit constructs a two-dimensional decoupling allocation matrix; during the execution of the time-domain reverse compensation step, when the system generates a scalar inertial feedforward compensation amount... Subsequently, this signal is input as a target pressure adjustment request to the decoupling allocation matrix, and the system accordingly... and The weighting ratio will The analysis decomposes the components into screw speed adjustment components. Adjust the amount of feed with the feeding speed This decomposition logic follows the material conservation constraint, meaning that while adjusting the pressure, the feed rate is adjusted in the opposite direction to offset the fluctuations in theoretical output caused by changes in screw speed, or the optimal solution is found under the constraint of maintaining a constant output. and Combining, the final generated contains and The vector drive commands are synchronously sent to their respective servo drives, thereby maintaining the dynamic balance of the extruder head discharge flow while suppressing melt pressure fluctuations.

[0041] Example 5: This example aims to construct a set of on-site pre-deployment calibration and commissioning procedures for different extrusion production lines and raw material characteristics. This procedure covers two core aspects: offline identification of the basic characteristics of the controlled object and adaptive initialization of control parameters. Before formally connecting the control system to the new extrusion production line, an offline identification program for basic characteristics is executed. With the extruder in an unloaded state, the system drives the screw motor at a preset stepped speed, covering the entire range from the lowest operating speed to the rated speed. The drive current value at each steady-state speed point is recorded, and a reference load current model is constructed accordingly. The model serves as a reference benchmark for subsequent extraction of non-command residual components. It can effectively eliminate the inherent nonlinear effects of mechanical friction and motor efficiency changes with speed. After the extruder is loaded with standard raw materials and enters thermal steady-state operation, an open-loop step response test is performed. The system applies small-amplitude step signals to the screw speed command and the feeding speed command respectively, and records the dynamic response curve of the melt pressure. By analyzing the time constant and steady-state gain of the response curve, the initial coefficients of the discretized parameter model are initially determined, and the initial value of the covariance matrix of the recursive least squares algorithm is set to accelerate the convergence speed of the online identification process.

[0042] To address the differences in rheological properties among different batches of recycled polyethylene raw materials, the system executes adaptive initialization logic for control parameters. In the initial stage of each batch of new raw materials being put into production, the system automatically runs a self-learning mode for a period of time. During this period, the control system only uses the dual-timescale model evolution step for parameter tracking and does not apply any active control actions. By monitoring the root mean square value of the model prediction error in real time, the system evaluates the degree of fit of the current model to the rheological properties of the new raw materials. When the prediction error converges to the preset allowable range, the system automatically calculates and sets the bandwidth of the disturbance observer and the gain coefficient of the feedforward compensation based on the identified model parameters. This ensures that the control system has completed the optimal parameter configuration for the current raw material characteristics before formally engaging in closed-loop control, avoiding initial control oscillations caused by parameter mismatch.

[0043] Example 6: This example provides a standardized destructive thermal stress calibration procedure and a quantitative execution logic for a safety degradation mode to address the thermal stability boundary of recycled polyethylene under extreme operating conditions. This quantitatively defines the melt temperature safety constraint characteristics, ensuring the physical safety of the control system during operation. Before formal production, a destructive thermal stress calibration procedure is executed to determine the preset shear heat accumulation threshold. With the extruder barrel temperature set to the standard process value and the feed rate constant, the system drives the screw to increase its speed in a stepped manner, with each speed increment being a fraction of the rated speed. The holding time is Minutes, rate of change of melt temperature in real-time computer head of electronic control unit Simultaneously, offline sampling of the extruded material at each speed step was performed, and its oxidation induction period (OIT) was tested according to ASTM D3895-03 (Standard Test Method for Determining Oxidation Induction Time of Polyolefins by Differential Scanning Calorimetry). When the OIT value of the sample dropped to the initial value of the raw material, the OIT was determined. When the rate of change of the melt temperature reaches the critical point of thermal degradation of the material, it is denoted as . Considering sensor noise and thermal inertia in industrial environments, the system sets a safety threshold during operation as follows: This calibration process maps the chemical degradation characteristics of materials into a rate of change of physical quantities that can be observed in real time, providing a clear quantitative basis for safety constraints.

[0044] Based on the above calibration The electronic control unit executes dynamic gain scheduling logic in real-time operation to achieve a safety degradation mode. The system defines the value range as follows: dynamic attenuation coefficient In each control cycle The system calculates the current rate of temperature change. ,like Then set The system performs full-amplitude inertial feedforward compensation if This indicates that the rate of accumulation of shear heat has approached the safety boundary, and the system immediately responds according to the formula. Calculate the attenuation coefficient, where The preset sensitivity gain (typical value is...) to The system will use the originally calculated inertial feedforward compensation amount. Revised to It outputs a smooth and uninterrupted switch from a pressure-priority to a temperature-priority control strategy by continuously and linearly reducing the control amplitude within the overheat risk zone.

[0045] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0046] Finally, 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.

Claims

1. A process control method for multiple recycling of tailings impervious regenerated polyethylene, characterized by, An electronic control unit connected to an extrusion production line, including a screw drive motor and a melt pressure sensor mounted at the end of the die head, is used. The method includes the following steps: The multidimensional heterogeneous data synchronous acquisition step synchronously acquires the melt pressure sequence output by the melt pressure sensor, the drive current sequence of the screw drive motor, and the current control input sequence at a preset sampling frequency, and establishes a timestamp alignment reference between the melt pressure sequence, the drive current sequence, and the control input sequence. The physical transmission lag feature extraction step involves calculating the cross-correlation function between the driving current sequence and the melt pressure sequence, and based on the time offset of the peak value of the cross-correlation function, identifying the physical transmission lag time of the load mutation feature in the driving current sequence relative to the pressure fluctuation feature in the melt pressure sequence. The advanced disturbance targeted simulation step monitors the drive current sequence in real time and filters out the fundamental component caused by speed command changes to extract the non-command residual component. When the amplitude of the non-command residual component exceeds the preset noise shielding threshold, the amplitude of the melt pressure fluctuation that will reach the end of the machine head after the physical transmission lag time is simulated by using the physical transmission lag time and cross-correlation function. The time-domain reverse compensation step generates an inverse inertial feedforward compensation amount based on the derived melt pressure fluctuation amplitude, and superimposes the inertial feedforward compensation amount onto the basic control amount generated by the electronic control unit to generate the final drive command. The final drive command drives the screw drive motor to perform reverse adjustment action before the load change characteristics are physically transmitted to the end of the die head and cause pressure fluctuations. Furthermore, the method also includes a dual-timescale model evolution step. While performing the time-domain inverse compensation step, a recursive least squares algorithm is used to process the melt pressure sequence and control input sequence to update the coefficients of the discretized parameter model describing the dynamic characteristics of the extrusion process online. A disturbance observer is constructed using the updated discretized parameter model to calculate the deviation between the theoretical model output and the measured melt pressure. A low-pass filter is used to process the deviation of the disturbance observer output to generate a feedback compensation control quantity, and the feedback compensation control quantity and the inertial feedforward compensation quantity are superimposed on the basic control quantity. In the dual-timescale model evolution steps, the coefficient update of the discretized parameter model follows the following recursive operation rules that include a forgetting factor: ,in, For a moment The model parameter vector, For a moment The measured value of the melt pressure, A regression vector containing historical input and output data. The gain matrix is ​​calculated by introducing a preset forgetting factor. The forgetting factor is used to reduce the weight of historical data on the current model parameter update in order to track the process gain drift caused by component wear in the extrusion production line. Before entering the low-pass filter, the deviation output by the disturbance observer needs to undergo periodic noise decoupling processing. This processing includes: acquiring the speed command of the screw drive motor in real time, calculating the current inherent pulsation frequency of the equipment based on the screw's geometric parameters, configuring a dynamic notch filter whose center frequency is locked to the inherent pulsation frequency of the equipment in real time, using the dynamic notch filter to process the deviation, eliminating the periodic component that is synchronized with the screw rotation, and retaining the random disturbance component caused by the fluctuation of raw material characteristics. The control input sequence includes the screw speed setpoint, and the physical transmission lag time is defined as a variable that is dynamically adjusted as the screw speed setpoint changes. The method includes establishing an inverse mapping table between the physical transmission lag time and the screw speed setpoint. When an effective cross-correlation function cannot be calculated, the estimated physical transmission lag time is obtained by consulting the inverse mapping table based on the current screw speed setpoint.

2. The process control method for multiple recycling of a regenerated polyethylene for tailings impoundment according to claim 1, characterized in that, The cross-correlation function is calculated based on the data segment within the sliding time window; The physical transmission lag feature extraction step further comprises: when the maximum correlation coefficient of the cross-correlation function is lower than a preset confidence threshold, determining that no material fluctuation transmission feature meeting the preset correlation requirement is present at present, and maintaining the physical transmission lag time identified at the last time unchanged until the new maximum correlation coefficient exceeds the confidence threshold.

3. The process control method for multiple recycling of a regenerated polyethylene for tailings impoundment according to claim 1, characterized in that, The non-instructional residual component is extracted by: constructing an instruction response reference model of the screw drive motor, inputting the control input sequence into the instruction response reference model to generate a theoretical current response sequence; and calculating the difference between the drive current sequence and the theoretical current response sequence as the non-instructional residual component, which represents the load fluctuation caused by the change in the material flow property.

4. The process control method for multiple recycling of a regenerated polyethylene for tailings impoundment according to claim 1, characterized in that, The generation of the inertial feedforward compensation amount follows an amplitude limiting logic; a maximum allowed compensation gradient is set, and when the compensation amount change rate generated according to the deduction exceeds the maximum allowed compensation gradient, the inertial feedforward compensation amount is subjected to slope limiting processing to prevent the screw speed change rate caused by the compensation action from exceeding the mechanical bearing limit of the screw drive motor.

5. The process control method for multiple recycling of a regenerated polyethylene for tailings impoundment according to claim 1, characterized in that, The final drive instruction includes a screw speed instruction and a feeding speed instruction; the time domain reverse compensation step proportionally distributes the inertial feedforward compensation amount to the screw speed instruction and the feeding speed instruction through a preset decoupling distribution matrix, thereby maintaining the material mass conservation in the extrusion production line while maintaining the melt pressure stability.

6. The process control method for multiple recycling of a regenerated polyethylene for tailings impoundment according to claim 1, characterized in that, The method further comprises a melt temperature safety constraint step of monitoring the melt temperature and its change rate at the end of the die head in real time; when the change rate of the melt temperature exceeds a preset shear heat accumulation threshold, the gain coefficient of the inertial feedforward compensation amount is reduced, and the control strategy is switched to a safety mode with temperature rise suppression as the priority.

Images