Multi-parameter cooperative control method for cable insulation extrusion process

Abstract

The present application belongs to the technical field of cable manufacturing, and particularly relates to a multi-parameter collaborative control method for cable insulation extrusion process, which comprises the following steps: collecting the melt pressure of a machine head and the real-time rotating speed of a screw rod, mapping the online estimated value of the effective viscosity of a current batch, and obtaining the smoothed effective viscosity through smoothing processing; obtaining a reference effective viscosity and an initial cross-gain matrix, proportionally compensating the initial cross-gain matrix based on the ratio of the smoothed effective viscosity to the reference effective viscosity to obtain a modified cross-gain matrix; calculating a feedforward compensation amount based on the modified cross-gain matrix, combining the feedback adjustment amount calculated based on the measured value of the insulation thickness to obtain an output; extracting the periodic intensity index of the thickness deviation sequence, determining the excitation amplitude when the periodic intensity index exceeds a dynamic triggering threshold, and triggering forced excitation to update the matrix. The present application realizes the continuous guarantee of the control accuracy under the working condition of the batch characteristic drift of raw materials.

Description

Technical Field

[0001] This invention belongs to the field of cable manufacturing technology, specifically relating to a multi-parameter collaborative control method for the cable insulation extrusion process. Background Technology

[0002] Extrusion molding of cable insulation is a core process in wire and cable manufacturing that determines the electrical performance of the product. On the extrusion production line, insulating materials (such as thermoplastic polymers like XLPE and PVC) are heated and plasticized by an extruder, then extruded around the conductor surface through a die, and finally cooled and shaped to form an insulation layer with a specified thickness and uniformity. This process has stringent requirements for the uniformity of insulation thickness, eccentricity, and surface quality; these indicators directly affect the cable's breakdown voltage, partial discharge characteristics, and service life.

[0003] Patent application CN120353121A discloses a cross-linked cable production control method based on an improved PID control algorithm. This method involves establishing a collaborative control architecture for temperature, pressure, and traction speed, with vulcanizing pipe temperature as the primary controlled variable and extruder pressure and traction speed as secondary controlled variables, and constructing an optimal reference model. The primary controller tracks the temperature setpoint in real time and dynamically tunes the PID parameters based on process parameters such as insulation thickness deviation and material cross-linking degree. Precise control is achieved through a three-loop synergistic coupling improved PID controller: the temperature control loop converges the steady-state error between the actual temperature and the setpoint to zero; the pressure and speed control loops converge the fluctuation amplitude to the allowable process range. This patent can improve the temperature stability, pressure uniformity, and traction synchronization in cross-linked cable production, reduce process fluctuations and defect rates, and improve product quality and production efficiency.

[0004] To ensure the quality of the insulation layer molding, existing control schemes generally adopt a multi-parameter independent closed-loop control architecture, configuring independent PID controllers for each temperature zone of the extruder, screw speed, and cable traction speed. To address the issue of parameter cross-interference between these controllers, some advanced production lines have introduced a multivariable decoupling control method. This involves conducting step response experiments on the extruder under standard materials and operating conditions beforehand to identify the transfer functions between each control channel, thereby establishing a fixed cross-gain matrix. It is used for feedforward decoupling compensation throughout the production process, which can theoretically eliminate the interference between channels such as screw speed adjustment on temperature field and traction speed adjustment on extrusion pressure.

[0005] However, the aforementioned decoupling control scheme based on a fixed cross-gain matrix will experience systemic failure when the characteristics of the raw material batches drift. The melt flow index of the polymer raw material exhibits objective fluctuations between different production batches, and the shift in the melt flow index directly affects the effective viscosity of the melt within the die. Deviation from the reference value under calibration conditions Since the elements of the cross-gain matrix are physically proportional to the rheological transfer properties of the melt, when and After the deviation occurs, the original matrix A mismatch occurs between the gain value in the control module and the actual transmission relationship. In this case, decoupling compensation not only fails to eliminate cross-interference between channels, but also injects the mismatched gain into each control channel in the form of negative feedback. This prevents the insulation thickness deviation from self-healing through normal closed-loop adjustment, ultimately resulting in continuous thickness fluctuations with a fixed periodicity in the time series. These fluctuations are highly concealed and can accumulate continuously without triggering single-parameter monitoring alarms, until the insulation layer thinning point exceeds the factory inspection threshold, causing batch-wide downgrading or scrapping of the product. Summary of the Invention

[0006] This invention provides a multi-parameter collaborative control method for the cable insulation extrusion process to solve the technical problem of periodic thickness deviation caused by the mismatch of the fixed cross gain matrix when the characteristics of raw material batches drift.

[0007] This invention provides a multi-parameter coordinated control method for the cable insulation extrusion process, including: S1. Collect the die head melt pressure and screw real-time speed at the die head outlet section of the extruder. Based on the preset rheological model, map the die head melt pressure and screw real-time speed to the online estimated value of the effective viscosity of the current batch, and smooth the online estimated value to obtain the smoothed effective viscosity. S2, obtain the reference effective viscosity and initial cross-gain matrix of the calibration batch, calculate the ratio of the smoothed effective viscosity to the reference effective viscosity, and perform proportional compensation on the initial cross-gain matrix based on the ratio to obtain the corrected cross-gain matrix; S3: Collect the measured value of insulation thickness, calculate the feedforward compensation amount of cross-channel interference based on the corrected cross-gain matrix, and calculate the feedback adjustment amount in combination with the measured value of insulation thickness. After superimposing the feedforward compensation amount and the feedback adjustment amount, output them to the actuator for coordinated adjustment. During the excitation period, the feedforward compensation output is paused. S4. Calculate the autocorrelation function of the thickness deviation sequence corresponding to the measured insulation thickness to extract the periodic strength index. When the periodic strength index exceeds the dynamic trigger threshold, determine the excitation amplitude based on the standard deviation of the thickness deviation sequence and the main diagonal elements of the corrected cross-gain matrix, and trigger forced excitation according to the excitation amplitude to re-identify and update the corrected cross-gain matrix.

[0008] Its effect is that by mapping the directly collectable head pressure and screw speed to the effective viscosity through a rheological model, it achieves real-time non-destructive perception of the drift of raw material batch characteristics without the need to add additional online rheometer hardware, thus reducing the production line modification cost and maintenance complexity.

[0009] Furthermore, the online estimate is smoothed to obtain the smoothed effective viscosity, including: During the baseline acquisition period after material change, the autocorrelation function of the original sequence of online estimates is calculated; Extract the autocorrelation function decaying to its initial value. The number of delayed samples corresponding to the time interval is used as the relevant time constant; The window length of the moving average filter is set to twice the relevant time constant, and the online estimate is filtered using the window length to obtain the smoothed effective viscosity.

[0010] Its effects are as follows: by adaptively determining the filtering window through the autocorrelation function of the viscosity sequence, the viscosity estimate is smoothed, high-frequency interference generated by the screw and hydraulic system is filtered out, the influence of instantaneous sampling glitch on the gain matrix correction is eliminated, the viscosity data is kept in line with the steady-state rheological characteristics of the melt, the matrix is ​​prevented from being incorrectly corrected due to viscosity jumps, and the stability of decoupling compensation is improved.

[0011] Furthermore, based on a preset rheological model, the melt pressure at the die head and the real-time screw speed are mapped to an online estimate of the effective viscosity of the current batch, including: Obtain the equivalent flow channel radius and effective flow channel length of the die determined during the initial calibration of the production line, and calculate the geometric coefficients. Extract the flow index identified in the baseline batch calibration experiment; Divide the melt pressure at the die head by the product of the geometric coefficient and the flow exponent power of the real-time screw speed to obtain the online estimate of the effective viscosity of the current batch.

[0012] Its effect is that by merging the consistency coefficient and the geometric scaling coefficient into the geometric coefficient, an engineering mapping relationship that can be calculated online is formed, which reduces the number of on-site calibration parameters and simplifies the commissioning workload before production.

[0013] Furthermore, the initial cross-gain matrix is ​​proportionally compensated based on the ratio to obtain the corrected cross-gain matrix, including: Calculate the absolute value of the difference between the smoothed effective viscosity and the reference effective viscosity, and use the ratio of the absolute value of the difference to the reference effective viscosity as the viscosity offset. Obtain the mean absolute value of the residuals of the linearized approximation of the rheological model in the calibration experiment, take the ratio of the mean absolute value of the residuals to the amplitude of the linear term as the linearization relative error, and extract the critical viscosity offset that causes the linearization relative error to exceed the insulation thickness control accuracy requirement as the boundary threshold. When the viscosity offset does not exceed the boundary threshold, the initial cross-gain matrix is ​​multiplied by the ratio of the smoothed effective viscosity to the reference effective viscosity to obtain the corrected cross-gain matrix.

[0014] Its effect is that by objectively calculating the boundary threshold, it avoids the directional error caused by driving feedforward compensation with linear correction results in the strong nonlinear region, thus ensuring the effectiveness and accuracy of matrix correction.

[0015] Furthermore, the feedback adjustment amount is calculated based on the measured insulation thickness, including: Extract the initial main diagonal elements from the initial cross-gain matrix, and the current main diagonal elements from the corrected cross-gain matrix; Multiply the preset initial proportional gain by the ratio of the initial main diagonal element to the current main diagonal element to obtain the dynamically corrected proportional gain. The measured value of insulation thickness is input to the closed-loop controller, and the dynamically corrected proportional gain is used for calculation to finally obtain the feedback adjustment amount.

[0016] Furthermore, the autocorrelation function of the thickness deviation sequence corresponding to the measured insulation thickness is calculated to extract the periodic strength index, including: The thickness deviation sequence corresponding to the measured insulation thickness is extracted using a sliding window of preset length; Calculate the autocorrelation function of the thickness deviation sequence within the sliding window; Within a preset delay interval, the maximum peak value of the autocorrelation function is extracted to obtain the periodic intensity index.

[0017] Furthermore, when the periodic intensity index exceeds the dynamic trigger threshold, including: During the healthy baseline data collection period of steady-state production, the mean and standard deviation of the statistical periodic intensity index series were collected. The health baseline collection period ends when the relative change in standard deviation is less than a preset percentage of the mean after multiple consecutive updates. By locking the sum of the mean and twice the standard deviation, the dynamic trigger threshold is finally obtained.

[0018] Its effect is that by dynamically determining the trigger threshold, it is strongly bound to the actual fluctuation baseline of the current batch of raw materials and operating conditions, avoiding the problem of false triggering or missed reporting under different operating conditions due to fixed thresholds, and improving the robustness of detection.

[0019] Furthermore, the excitation amplitude is determined based on the standard deviation of the thickness deviation sequence and the main diagonal elements of the corrected cross-gain matrix, including: During the healthy baseline data collection period, the standard deviation of the thickness deviation sequence was simultaneously calculated. Extract the main diagonal elements of the corrected cross-gain matrix; The standard deviation of twice the thickness deviation sequence is divided by the main diagonal element, and the calculation result is rounded to the minimum speed resolution level that the frequency converter can execute, thus obtaining the excitation amplitude.

[0020] Furthermore, the feedforward compensation for cross-channel interference is calculated based on the corrected cross-gain matrix, including: Monitor the step change of the set value; The expected disturbance magnitude caused by the step change in the remaining channels is calculated using the modified cross-gain matrix. Based on the expected disturbance amplitude, an equal and opposite compensation signal is generated, and finally the feedforward compensation amount for cross-channel interference is obtained.

[0021] Furthermore, triggering forced excitation according to the excitation magnitude to re-identify and update the modified cross-gain matrix includes: A speed step signal is sent to the actuator according to the excitation amplitude to trigger forced excitation and freeze the feedforward compensation amount; The change in process variables of each response channel is detected in real time, and the forced excitation ends when the change in process variables continuously exceeds the preset multiple of the baseline noise standard deviation. The measured response of each channel during the forced excitation period is recorded to re-identify the mismatch elements, and the identification results are used to replace the original calibration values, finally obtaining the re-identified and updated corrected cross-gain matrix.

[0022] The beneficial effects are: adaptive proportional correction of the cross-gain matrix based on viscosity ratio, combined with proportional gain synchronous self-tuning based on gain scheduling principle, effectively suppresses the risk of decoupling accuracy decrease and closed-loop oscillation caused by raw material batch switching, and significantly improves the convergence speed and stability of insulation thickness deviation.

[0023] The periodic deviation detection module based on autocorrelation analysis, together with the dynamically determined excitation amplitude, constitutes the verification and fallback layer for matrix correction quality. It automatically triggers local recorrection under strong nonlinear drift conditions that cannot be covered by proportional correction, enabling the entire control method to have graded self-healing capabilities for different degrees of viscosity drift. This significantly broadens the applicable batch range of the scheme and reduces the risk of batch non-compliance of the insulation layer due to raw material fluctuations. Attached Figure Description

[0024] Figure 1 This is a flowchart of the multi-parameter collaborative control method for the cable insulation extrusion process in this invention.

[0025] Figure 2 This is a comparison chart showing the combined effect of the standard deviation of thickness deviation and the periodic strength index under different drift intensity batches in this invention. Detailed Implementation

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

[0027] An embodiment of the multi-parameter coordinated control method for the cable insulation extrusion process provided by this invention: like Figure 1 As shown, the multi-parameter coordinated control method for the cable insulation extrusion process includes the following steps: S1 collects the melt pressure and screw speed at the die head in real time to estimate the effective viscosity of the smoothing process online.

[0028] In an optional embodiment, a pressure sensor is arranged at the extruder head outlet section to collect the melt pressure at the die head in real time. The unit is MPa, and the real-time screw speed is read synchronously. The unit is rpm. Both signals are low-pass filtered before entering the viscosity inversion module to remove high-frequency glitches caused by screw meshing vibration and hydraulic pulse superposition, ensuring that the data used in subsequent derivation truly reflects the steady-state rheological state of the current batch of melt.

[0029] The viscosity inversion module is based on a pre-defined rheological model (specifically, a power-law fluid model). The original form of the power-law model is as follows: ,in This is the consistency coefficient. Shear rate, This is the liquidity index. Because... In actual production, these quantities cannot be directly measured and must be mapped to collectable engineering quantities. For a single-screw extruder, the apparent shear rate at the die is related to the screw speed. Proportional, that is ,in This is a scaling factor related to the die geometry. Furthermore, according to the Hagen-Poiseuille circular pipe flow equation... The pressure at the die head can be obtained in relation to viscosity and flow rate, which satisfies the following conditions. The geometric coefficients are:

[0030] Obtain the equivalent flow channel radius of the die determined during the initial calibration of the production line. With effective flow channel length The geometric coefficients were calculated. It originates from objective physical measurements. And due to the extrusion volume... and Directly proportional to, Substituting the original power-law expression and rearranging it according to the pressure relationship, we can obtain the complete expression for the head pressure as follows:

[0031] in The flow index of the baseline batch is extracted from the flow index identified in the calibration experiment of the baseline batch. And store it as a fixed constant. Inverse the above equation, divide the melt pressure at the die head by the product of the geometric coefficient and the flow exponent power of the real-time screw speed, and finally obtain the online estimate of the effective viscosity of the current batch. Defined as:

[0032] The improvement over the original power-law formula lies in its ability to be directly collected. and Replaced the one that could not be tested. and the consistency coefficient Combined with geometric proportionality coefficient In this process, an engineering mapping relationship that can be calculated online is formed.

[0033] Then, every other sampling period Update once Exemplary The value can be 5 seconds, which is aligned with the refresh cycle of the temperature control PLC.

[0034] During the baseline data collection period after material change, The autocorrelation function is calculated from the original sequence. The autocorrelation function decays to its initial value. The corresponding delay at this point is often used in signal processing as a criterion for determining the substantial decay of sequence correlation. Therefore, the number of samples at this decay point is taken as the correlation time constant. The window length of the moving average filter is set to twice the relevant time constant. , that is to say Using window length to Filtering is performed, followed by windowing. The smoothed output is denoted as the smoothed effective viscosity. This is used as input for the subsequent matrix correction module.

[0035] By mapping the die head pressure and screw speed to effective viscosity using a power-law rheological model, real-time non-destructive sensing of batch-specific drift of raw materials is achieved. This eliminates the need for additional online rheometer hardware on the production line, significantly reducing modification costs and subsequent maintenance complexity.

[0036] S2 is the initial cross-gain matrix based on viscosity ratio proportional compensation.

[0037] In an optional embodiment, the system obtains the baseline effective viscosity of the calibration batch during a standard batch calibration experiment before production. With the initial cross gain matrix .in Steady state of the calibrated batch of raw materials under rated operating conditions and Substituting into the inversion formula, we can calculate the result. The rows and columns correspond to the control output channels and process response channels, respectively, and each element is identified from the step response curve in the calibration experiment. For example, the control output channels may include screw speed, traction speed, and main temperature zone heating power, while the process response channels may include die head pressure, linear velocity deviation, and extrusion quantity deviation. It is a 3×3 matrix.

[0038] The online matrix correction is based on the rheological conclusion that the flow rate and pressure transmission of power-law fluids are approximately linearly proportional, and viscosity is approximately linearly proportional. It calculates the ratio of the smoothed effective viscosity to the reference effective viscosity, and then proportionally compensates the initial cross-gain matrix based on this ratio to obtain the corrected cross-gain matrix. The specific correction process is as follows: Calculation... and The absolute value of the difference, and then the absolute value of the difference is compared with... The ratio is used as the viscosity offset; the mean absolute value of the residuals of the linearized approximation of the rheological model in the calibration experiment is obtained. ,Will The ratio of the linear term amplitude to the linearization relative error is used as the linearization relative error, and the critical viscosity offset that makes the linearization relative error just exceed the insulation thickness control accuracy requirement is extracted as the boundary threshold. .

[0039] viscosity offset Not exceeding the boundary threshold At this point, the initial cross-gain matrix is ​​multiplied by the ratio of the smoothed effective viscosity to the reference effective viscosity to obtain the corrected cross-gain matrix, with the following relationship:

[0040] The meaning of this ratio correction is: when the melt becomes thinner, that is... When the same rotational speed disturbance causes a decrease in the response amplitude in the pressure channel, the corresponding gain term should be proportionally reduced; when the melt thickens, it should be amplified in the opposite direction. The greater the deviation of this ratio from 1, the greater the difference in rheological properties between the current batch of raw materials and the calibration batch, and the more significant the matrix correction amplitude; the closer the ratio is to 1, the more consistent the rheological properties of the two, and the closer the correction amount is to 0. When the viscosity offset exceeds the boundary threshold, it is determined that the current viscosity offset has exceeded the effective range of linear correction, triggering a recalibration process instead of continuing online correction, thus avoiding the directional error caused by using linear correction results to drive feedforward compensation in the strongly nonlinear region.

[0041] update frequency and To maintain synchronization, immediately replace the gain matrix in the feedforward decoupling controller after the refresh, and all subsequent calculations of feedforward compensation amounts will be based on the current... To proceed, rather than following a fixed approach. .

[0042] By adaptively scaling the cross-gain matrix with viscosity ratios, the matrix gain always corresponds to the actual rheological properties of the current batch of raw materials, effectively eliminating the problem of additional interference injected into each control channel due to gain mismatch when switching batches by a fixed matrix.

[0043] S3 calculates the feedforward and feedback quantities and superimposes them to coordinate the output adjustment.

[0044] In an optional embodiment, real-time correction is obtained. Subsequently, the multi-parameter collaborative controller employs a combined feedforward and feedback structure to control the screw speed. traction speed Main temperature zone heating power The three channels implement coordinated regulation.

[0045] The feedforward section works as follows: it calculates the feedforward compensation for cross-channel interference based on the modified cross-gain matrix. Specifically, this includes: monitoring the step change in the setpoint; calculating the expected disturbance amplitude caused by the step change in other channels using the modified cross-gain matrix; generating an equal and inverse compensation signal based on the expected disturbance amplitude; and finally obtaining the feedforward compensation for cross-channel interference, thus offsetting the interference and blocking the transmission chain from speed regulation to pressure fluctuations and then to thickness deviations. The feedforward compensation is sent out within the first control cycle after the setpoint change, without waiting for the process variables to deviate, thereby eliminating the inherent hysteresis response of pure feedback control.

[0046] The feedback section collects the measured values ​​of insulation thickness. The measured insulation thickness is collected in real time by an online laser diameter gauge, and the feedback adjustment amount is calculated based on the measured insulation thickness. The specific calculation process is as follows: extract the initial principal diagonal element representing the self-gain of screw speed on extrusion volume. and the current main diagonal element in the corrected cross-gain matrix. ; Set the preset initial scaling gain Multiplying by the ratio of the initial main diagonal element to the current main diagonal element yields the dynamically adjusted scaling gain. The relation is:

[0047] The measured insulation thickness is input to the closed-loop controller, and the dynamically corrected proportional gain is used for calculation to obtain the feedback adjustment amount. Because... Scaling synchronously with the viscosity ratio, when the melt becomes thinner... When shrinking, The corresponding increase ensures that the overall open-loop gain of the feedback loop remains constant; when the melt thickens... When magnified, This corresponding reduction avoids the risk of oscillation caused by excessively high closed-loop gain after matrix update.

[0048] The feedforward compensation and feedback adjustment are superimposed and output to the actuator for coordinated adjustment. The final control quantity of the three channels is linearly superimposed by the feedforward compensation and feedback adjustment and output to each actuator. The superposition calculation of each channel is completed within the same PLC scan cycle, ensuring the timing synchronization of multi-channel actions and eliminating brief inter-channel disturbances caused by timing differences in control quantities.

[0049] By combining feedforward compensation and feedback adjustment, and synchronously self-tuning of the proportional gain with matrix updates, the system can intercept the viscosity drift caused by raw material batch switching within the first control cycle of the disturbance, and ensure that the stability of the feedback loop does not decrease with matrix correction.

[0050] S4 triggers an incentive update matrix when the extracted metric exceeds the threshold.

[0051] In an optional embodiment, when the nonlinear characteristics of the raw material rheological curve are prominent, linear proportional correction may not completely eliminate residual mismatch, and small fluctuations with periodic characteristics may still remain in the insulation thickness deviation sequence. Therefore, a periodic deviation detection module based on autocorrelation analysis is added above the multi-channel coordinated adjustment control execution layer as a verification layer for the matrix correction quality.

[0052] The detection module calculates the autocorrelation function of the thickness deviation sequence corresponding to the measured insulation thickness values ​​to extract periodic strength indicators. Specifically, this includes: extracting the thickness deviation sequence corresponding to the measured insulation thickness values ​​using a sliding window of a preset length. The window length is determined based on a combination of the production line speed and the settling time of a single step response; for example, the window length can be 240 sampling points; and calculating the autocorrelation function of the thickness deviation sequence within the sliding window. Extract the maximum peak value of the autocorrelation function within a preset delay interval. The periodic intensity index is finally obtained. The lower limit of the delay interval is determined by the number of sampling periods corresponding to a single rotation of the screw, and the upper limit is determined by the number of sampling periods corresponding to the power frequency. For example, the lower limit of the delay interval can be 20 sampling points, and the upper limit can be 120 sampling points to ensure that the detection window accurately covers the mid-frequency periodic deviation band induced by the drift of raw material batches.

[0053] When the periodic intensity index exceeds the dynamic trigger threshold, the subsequent recalibration process is triggered. Dynamic trigger threshold The data was obtained by: during the health baseline collection period of steady-state production, by statistically analyzing the mean of the periodic intensity index series. with standard deviation When the relative change in standard deviation is less than a preset percentage of the mean after multiple consecutive updates, the statistic is considered to have converged, and the healthy baseline collection period ends. The number of consecutive updates can be up to 3, and the preset percentage can be 5%. The sum of the mean and twice the standard deviation is locked to obtain the dynamic trigger threshold. This threshold is strongly tied to the actual fluctuation baseline of the current batch of raw materials and operating conditions, avoiding the problem of false triggering or missed reporting of fixed thresholds under different operating conditions.

[0054] When the system detects that the periodic intensity index exceeds the dynamic trigger threshold for multiple consecutive sliding windows, it determines the current... Linear correction is insufficient to eliminate residual mismatch. The excitation amplitude is determined based on the standard deviation of the thickness deviation sequence and the main diagonal elements of the corrected cross-gain matrix. Specifically, during the healthy baseline acquisition period, the standard deviation of the thickness deviation sequence is simultaneously calculated. Extract the main diagonal elements of the corrected cross-gain matrix. Divide twice the standard deviation of the thickness deviation sequence by the main diagonal element, and round the result to the minimum speed resolution level that the frequency converter can execute, to finally obtain the excitation amplitude. The relation is:

[0055] Subsequently, a forced excitation is triggered according to the excitation amplitude to re-identify and update the corrected cross-gain matrix. Specifically, this includes: sending a speed step signal to the actuator according to the excitation amplitude to trigger forced excitation and freezing the feedforward compensation; real-time monitoring of the process variable changes in each response channel, ending forced excitation when the process variable changes continuously exceed a preset multiple of the baseline noise standard deviation. The preset multiple is determined comprehensively based on the production line background noise level and the required signal-to-noise ratio; for example, a preset multiple of 2 can be used; recording the measured response of each channel during forced excitation to re-identify mismatched elements, and replacing the original calibration values ​​with the identification results, ultimately obtaining the re-identified and updated corrected cross-gain matrix. Feedforward compensation is immediately restored after recalibration.

[0056] Thus, by combining periodic deviation detection based on autocorrelation analysis with dynamically determined excitation amplitude, local recalibration is automatically triggered under strongly nonlinear drift conditions that cannot be covered by linear proportional correction. This enables the control method to have graded self-healing capabilities for different degrees of viscosity drift, significantly expanding the applicable batch range of the scheme.

[0057] like Figure 2 As shown, it can be observed that with the increase of drift intensity, the thickness deviation standard deviation and periodic intensity index of the existing method both show a significant upward trend, and the periodic intensity index continuously exceeds the dynamic trigger threshold; the two corresponding indices of the method of the present invention remain at a low level under each drift intensity batch, and the periodic intensity index is always near or below the dynamic trigger threshold, indicating that the method of the present invention has a stable graded self-healing control capability for different degrees of viscosity drift of raw material batches.

[0058] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A multi-parameter coordinated control method for cable insulation extrusion process, characterized in that, include: S1. Collect the die head melt pressure and screw real-time speed at the die head outlet section of the extruder. Based on the preset rheological model, map the die head melt pressure and screw real-time speed to the online estimated value of the effective viscosity of the current batch, and smooth the online estimated value to obtain the smoothed effective viscosity. S2, obtain the reference effective viscosity and initial cross-gain matrix of the calibration batch, calculate the ratio of the smoothed effective viscosity to the reference effective viscosity, and perform proportional compensation on the initial cross-gain matrix based on the ratio to obtain the corrected cross-gain matrix; S3: Collect the measured value of insulation thickness, calculate the feedforward compensation amount of cross-channel interference based on the corrected cross-gain matrix, and calculate the feedback adjustment amount in combination with the measured value of insulation thickness. After superimposing the feedforward compensation amount and the feedback adjustment amount, output them to the actuator for coordinated adjustment. During the excitation period, the feedforward compensation output is paused. S4. Calculate the autocorrelation function of the thickness deviation sequence corresponding to the measured insulation thickness to extract the periodic strength index. When the periodic strength index exceeds the dynamic trigger threshold, determine the excitation amplitude based on the standard deviation of the thickness deviation sequence and the main diagonal elements of the corrected cross-gain matrix, and trigger forced excitation according to the excitation amplitude to re-identify and update the corrected cross-gain matrix.

2. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, The online estimate is smoothed to obtain the smoothed effective viscosity, including: During the baseline acquisition period after material change, the autocorrelation function of the original sequence of online estimates is calculated; Extract the autocorrelation function decaying to its initial value. The number of delayed samples corresponding to the time interval is used as the relevant time constant; The window length of the moving average filter is set to twice the relevant time constant, and the online estimate is filtered using the window length to obtain the smoothed effective viscosity.

3. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, Based on a pre-defined rheological model, the melt pressure at the die head and the real-time screw speed are mapped to an online estimate of the effective viscosity of the current batch, including: Obtain the equivalent flow channel radius and effective flow channel length of the die determined during the initial calibration of the production line, and calculate the geometric coefficients. Extract the flow index identified in the baseline batch calibration experiment; Divide the melt pressure at the die head by the product of the geometric coefficient and the flow exponent power of the real-time screw speed to obtain the online estimate of the effective viscosity of the current batch.

4. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, The initial cross-gain matrix is ​​proportionally compensated based on the ratio to obtain the corrected cross-gain matrix, including: Calculate the absolute value of the difference between the smoothed effective viscosity and the reference effective viscosity, and use the ratio of the absolute value of the difference to the reference effective viscosity as the viscosity offset. Obtain the mean absolute value of the residuals of the linearized approximation of the rheological model in the calibration experiment, take the ratio of the mean absolute value of the residuals to the amplitude of the linear term as the linearization relative error, and extract the critical viscosity offset that causes the linearization relative error to exceed the insulation thickness control accuracy requirement as the boundary threshold. When the viscosity offset does not exceed the boundary threshold, the initial cross-gain matrix is ​​multiplied by the ratio of the smoothed effective viscosity to the reference effective viscosity to obtain the corrected cross-gain matrix.

5. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, The feedback adjustment amount is calculated based on the measured insulation thickness, including: Extract the initial main diagonal elements from the initial cross-gain matrix, and the current main diagonal elements from the corrected cross-gain matrix; Multiply the preset initial proportional gain by the ratio of the initial main diagonal element to the current main diagonal element to obtain the dynamically corrected proportional gain. The measured value of insulation thickness is input into the closed-loop controller, and the dynamically corrected proportional gain is used for calculation to finally obtain the feedback adjustment amount.

6. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, The autocorrelation function of the thickness deviation sequence corresponding to the measured insulation thickness is calculated to extract periodic strength indices, including: The thickness deviation sequence corresponding to the measured insulation thickness is extracted using a sliding window of preset length; Calculate the autocorrelation function of the thickness deviation sequence within the sliding window; Within a preset delay interval, the maximum peak value of the autocorrelation function is extracted to obtain the periodic intensity index.

7. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, When the periodic intensity index exceeds the dynamic trigger threshold, including: During the healthy baseline data collection period of steady-state production, the mean and standard deviation of the statistical periodic intensity index series were collected. The health baseline collection period ends when the relative change in standard deviation is less than a preset percentage of the mean after multiple consecutive updates. By locking the sum of the mean and twice the standard deviation, the dynamic trigger threshold is finally obtained.

8. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, The excitation amplitude is determined based on the standard deviation of the thickness deviation sequence and the main diagonal elements of the corrected cross-gain matrix, including: During the healthy baseline data collection period, the standard deviation of the thickness deviation sequence was simultaneously calculated. Extract the main diagonal elements of the corrected cross-gain matrix; The standard deviation of twice the thickness deviation sequence is divided by the main diagonal element, and the calculation result is rounded to the minimum speed resolution level that the frequency converter can execute, thus obtaining the excitation amplitude.

9. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, The feedforward compensation for cross-channel interference is calculated based on the corrected cross-gain matrix, including: Monitor the step change of the set value; The expected disturbance magnitude caused by the step change in the remaining channels is calculated using the modified cross-gain matrix. Based on the expected disturbance amplitude, an equal and opposite compensation signal is generated, and finally the feedforward compensation amount for cross-channel interference is obtained.

10. The multi-parameter coordinated control method for the cable insulation extrusion process according to claim 1, characterized in that, Forced excitation is triggered according to the excitation magnitude to re-identify and update the modified cross-gain matrix, including: A speed step signal is sent to the actuator according to the excitation amplitude to trigger forced excitation and freeze the feedforward compensation amount; The change in process variables of each response channel is detected in real time, and the forced excitation ends when the change in process variables continuously exceeds the preset multiple of the baseline noise standard deviation. The measured response of each channel during the forced excitation period is recorded to re-identify the mismatch elements, and the identification results are used to replace the original calibration values, finally obtaining the re-identified and updated corrected cross-gain matrix.

Images