Cable extrusion control method based on adaptive control

Abstract

The invention discloses a cable extrusion control method based on self-adaptive control, and the method comprises the steps: collecting technological parameter data after an electric connection extruder is started, taking the preprocessed parameter data as a standardized vector, introducing a least square identification model of a co-integration residual error compensation item and a nonlinear input disturbance suppression item, and carrying out the optimization of the standardized vector; carrying out state identification processing to obtain a model identification result; then, a trend sensing type sliding mode controller is introduced for adjustment, a control signal is determined by constructing a sliding mode surface function and designing a composite control structure fusing multiple nonlinear transformation terms, parameters in the controller are optimized based on a particle swarm optimization segmented structure parameter optimization strategy, an updated control signal is obtained and mapped into a control instruction, and the control instruction is sent to the controller. Cable extrusion control is realized; according to the invention, cable extrusion control with high control precision and self-adaptive control can be realized, the method is suitable for the field of cable extrusion control with self-adaptive control, and the control accuracy and self-adaptability are improved.

Description

Technical Field

[0001] This invention relates to the field of adaptive control technology, and in particular to a cable extrusion control method based on adaptive control. Background Technology

[0002] In the cable manufacturing industry, the extrusion process, as a crucial step in the formation of the insulation and sheath layers, directly impacts the geometric consistency, electrical performance stability, and overall production line efficiency of the cable products. Traditional extrusion control systems primarily rely on linear control methods such as PID control to independently control process parameters like temperature, pressure, and traction speed. While this control method is simple in structure and low in implementation cost, it often exhibits problems such as slow response, untimely adjustments, and severe overshoot when faced with complex operating conditions in modern production, including frequent batch variations in raw materials, changes in environmental temperature and humidity, and equipment aging and wear. This is especially problematic during non-steady-state transition phases, which can easily lead to deviations in cable outer diameter, uneven insulation thickness, and even cable breakage.

[0003] Furthermore, traditional control systems typically employ static, fixed parameters, making their control performance highly dependent on manual experience. When faced with the different melting characteristics of polymers such as polyethylene, polyvinyl chloride, and cross-linked polyolefins, they struggle to achieve effective adaptive adjustment, resulting in systems that can only maintain relatively stable control within a narrow range. Simultaneously, in actual production, nonlinear coupling relationships are prevalent between process variables (such as temperature and pressure, and pressure and velocity). Traditional linear controllers struggle to accurately characterize this dynamic coupling mechanism, often leading to undesirable behavior under disturbed conditions.

[0004] In recent years, although some high-end equipment has introduced intelligent methods such as fuzzy control and neural network control, their complex system structures, lack of interpretability, and high computational resource consumption make them difficult to deploy in resource-constrained industrial edge devices, resulting in the lack of large-scale application of these methods. In particular, current mainstream control systems generally lack a mechanism for sustainable learning and dynamic adjustment, making it impossible to achieve online modeling and real-time feedback correction of the system state. There is a serious disconnect between the model and the controller, making it difficult for the controller to understand the current system structure, thus losing its predictive and pre-tuning capabilities.

[0005] The existing technologies mentioned above also have technical problems such as low control accuracy, slow system response to fluctuations in raw material performance and external environmental disturbances, lack of adaptive capability in fixed parameter control methods, insufficient adjustment capability of the controller in the small error range, and separation between the model and the controller. Summary of the Invention

[0006] This invention provides a cable extrusion control method based on adaptive control to solve the technical problems of low control accuracy, slow system response to raw material performance fluctuations and external environmental disturbances, lack of adaptive capability in fixed parameter control, insufficient adjustment capability of the controller in the small error range, and separation between the model and the controller.

[0007] The present invention provides a cable extrusion control method based on adaptive control, which specifically includes the following technical solutions:

[0008] A cable extrusion control method based on adaptive control includes the following steps:

[0009] S1. After the electric extruder is started, process parameter data is collected. After preprocessing, the preprocessed parameter data is obtained and used as a standardized vector. Then, the state identification process is carried out by introducing a least squares identification model with cointegration residual compensation term and nonlinear input disturbance suppression term to obtain the model identification result.

[0010] S2. Based on the model identification results, a trend-aware sliding mode controller is introduced for adjustment. By constructing the sliding mode surface function and designing a composite control structure that integrates multiple nonlinear transformation terms, the control signal is determined. The parameters in the controller are optimized based on the piecewise structure parameter optimization strategy of the particle swarm algorithm to obtain the updated control signal, which is then mapped into control commands to realize cable extrusion control.

[0011] Preferably, S1 specifically includes:

[0012] After the cable extruder starts, high-frequency response sensors deployed at key process stages collect process parameter data. These key process stages include the middle of the screw barrel, the die exit, the screw tail end, and the traction wheel inlet. The process parameter data includes screw zone temperature, die exit pressure, screw back pressure, traction speed, and actual cable outer diameter. Specifically: a thermocouple temperature sensor is installed in the middle of the screw barrel to monitor the extrusion section temperature; a piezoresistive pressure sensor is installed at the die exit to monitor the melt exit pressure; a pressure feedback device is installed at the screw tail end to obtain back pressure; a front encoder of the traction motor is used to obtain the traction speed in real time; and a laser diameter gauge is installed at the cooling section outlet to obtain the cable outer diameter.

[0013] Preferably, S1 specifically includes:

[0014] To avoid differences in noise characteristics and physical dimensions of process parameter data, preprocessing is performed on the process parameter data, including noise reduction, cleaning, outlier detection, synchronization, standardization, and normalization, to obtain preprocessed parameter data. The preprocessed screw zone temperature, die outlet pressure, screw back pressure, and traction speed are then represented as standardized vectors.

[0015] Preferably, S1 specifically includes:

[0016] Based on standardized vectors, a least-squares identification model is introduced by introducing cointegration residual compensation terms and nonlinear input disturbance suppression terms. State identification processing is performed to obtain the modeling parameters of the input variables on the output variables in the current state, i.e., the model identification result.

[0017] Preferably, S1 specifically includes:

[0018] The least squares identification model that introduces cointegration residual compensation terms and nonlinear input disturbance suppression terms introduces dynamic residual compensation terms into the modeling structure to absorb structural shifts caused by historical errors, and introduces regularization constraints on input changes to control model sensitivity.

[0019] Preferably, S2 specifically includes:

[0020] Based on the model identification results, a trend-aware sliding mode controller is introduced for adjustment. Its core purpose is to minimize high-frequency oscillations and improve the steady-state accuracy and dynamic performance of the controller in non-stationary environments while ensuring response sensitivity.

[0021] Preferably, S2 specifically includes:

[0022] The specific adjustment process of the aforementioned trend-aware sliding mode controller is as follows:

[0023] First, a sliding mode surface function is constructed to determine the composite error relative to the target state at the current moment. Then, a composite control structure integrating multiple nonlinear transformation terms is designed to improve the system's flexible response capability and suppress controller chattering, thereby obtaining the control output signal.

[0024] Preferably, S2 specifically includes:

[0025] To maintain the robustness of the controller response, an existing piecewise structural parameter optimization strategy based on particle swarm optimization (PSO) is introduced to periodically optimize the parameter set in the controller online, obtain the updated control output signal, and then perform mapping processing through the existing command solver to obtain control commands, which are further sent to various control actuators, such as temperature controllers, screw frequency converters and traction motors, to realize cable extrusion control.

[0026] The beneficial effects of the technical solution of the present invention are:

[0027] 1. An improved least squares model is introduced, combining a cointegration residual compensation mechanism with an input disturbance regularization term, enabling real-time identification of dynamic changes in the system. During cable extrusion, common disturbances such as raw material batch fluctuations, equipment wear, and changes in ambient temperature and humidity can alter the system's structural characteristics. Using a static model can easily lead to accumulated control deviations or system oscillations. By using dynamic modeling and introducing residual correction feedback, the system becomes more sensitive to disturbances and provides more timely feedback, significantly enhancing the overall robustness and stability of the system.

[0028] 2. By introducing multiple nonlinear transformation structures—including a hyperbolic tangent compression function, an arctangent trend-following function of the error derivative, and a hybrid control term of exponential square root function—differential control for different error ranges is achieved. Especially in the small error range, the exponential-square root hybrid function structure can compress the output adjustment amplitude to a continuous, differentiable region, significantly reducing the oscillation of the control signal and the risk of system overshoot, thus achieving the technical goal of smooth control. Attached Figure Description

[0029] Figure 1 This is a flowchart of a cable extrusion control method based on adaptive control according to the present invention. Detailed Implementation

[0030] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0032] The following description, in conjunction with the accompanying drawings, details a specific scheme for an adaptive control-based cable extrusion control method provided by the present invention.

[0033] See attached document Figure 1 The diagram illustrates a flowchart of an adaptive control-based cable extrusion control method according to an embodiment of the present invention, which includes the following steps:

[0034] S1. After the electric extruder is started, process parameter data is collected. After preprocessing, the preprocessed parameter data is obtained and used as a standardized vector. Then, the state identification process is carried out by introducing a least squares identification model with cointegration residual compensation term and nonlinear input disturbance suppression term to obtain the model identification result.

[0035] After the cable extruder starts, high-frequency response sensors deployed at key process stages collect process parameter data. These key process stages include the middle of the screw barrel, the die exit, the screw tail end, and the traction wheel inlet. The process parameter data includes screw zone temperature, die exit pressure, screw back pressure, traction speed, and actual cable outer diameter. Specifically: a thermocouple temperature sensor is installed in the middle of the screw barrel to monitor the extrusion section temperature; a piezoresistive pressure sensor is installed at the die exit to monitor the melt exit pressure; a pressure feedback device is installed at the screw tail end to obtain back pressure; a front encoder of the traction motor is used to obtain the traction speed in real time; and a laser diameter gauge is installed at the cooling section outlet to obtain the cable outer diameter.

[0036] To avoid differences in noise characteristics and physical dimensions among process parameter data, preprocessing is performed on the process parameter data, including noise reduction, cleaning, outlier detection, synchronization, standardization, and normalization. This results in preprocessed parameter data, with the preprocessed screw zone temperature, die outlet pressure, screw back pressure, and traction speed represented as standardized vectors. ,in These represent the pre-treated screw zone temperature, mold outlet pressure, screw back pressure, and traction speed, respectively. The pre-treatment process employs techniques well-known to those skilled in the art, which will not be elaborated upon here.

[0037] Based on standardized vectors, a least-squares identification model is introduced by incorporating cointegration residual compensation and nonlinear input disturbance suppression terms. State identification processing is then performed to obtain the modeling parameters of the input variables on the output variables in the current state, i.e., the model identification result. The least-squares identification model incorporating cointegration residual compensation and nonlinear input disturbance suppression terms introduces a dynamic residual compensation term into the modeling structure to absorb structural shifts caused by historical errors, and introduces regularization constraints on input changes to control model sensitivity. Its iterative formula is as follows:

[0038] ;

[0039] in, It is a moment Control input (i.e., normalized vector) and system output ( When modeling the mapping relationship between ), the parameter vector estimated in real time is the model identification result; It is a moment The parameter vector; It is a moment The system output, i.e., time... The outer diameter of the pre-treated cable; It is a moment The weighting factor is used to dynamically adjust the impact of the current cycle on the overall model update. ,in, It is a moment The system output, i.e., time... The outer diameter of the pre-treated cable; This is the disturbance sensitivity coefficient, used to dynamically adjust the model's confidence level in response to disturbances, thereby ensuring identification stability under abrupt changes. It is determined based on an adaptive adjustment method using statistical volatility estimation, with a reference value range of [value missing]. ; The modeling residuals from the previous period are used as cointegration residuals; Indicates transpose; For a moment Parameter vector The covariance matrix is ​​used to measure the uncertainty of parameter estimation and is updated using the recursive formula of the covariance matrix in standard recursive least squares theory. This is a forgetting factor, used to regulate the rate at which the weights of historical observation data decay. It is based on an adaptive adjustment method using residuals, with a reference value range. The historical observation data refers to all control inputs and system outputs that have been used for model parameter iteration from the start of control modeling to the current moment; Input the change terms; To mitigate unreasonable model jumps caused by abrupt input changes, the perturbation penalty weight coefficients are determined based on a Bayesian optimization algorithm, with a reference value range of [value range missing]. ; This is the cointegration residual feedback gain, used to mitigate slow model drift. It is determined based on expert experience, with a reference value range of [value missing]. ; It is a cointegration residual compensation term that provides short-term feedback correction and suppresses error accumulation; It is a nonlinear input perturbation suppression term, which describes the nonlinear penalty suppression mechanism for large updates of model parameters when the input changes drastically.

[0040] The model identification result is the mapping function from control variables to output variables under the current operating conditions, which is used for predictive control of the controller in the next stage.

[0041] S2. Based on the model identification results, a trend-aware sliding mode controller is introduced for adjustment. By constructing the sliding mode surface function and designing a composite control structure that integrates multiple nonlinear transformation terms, the control signal is determined. The parameters in the controller are optimized based on the piecewise structure parameter optimization strategy of the particle swarm algorithm to obtain the updated control signal, which is then mapped into control commands to realize cable extrusion control.

[0042] Based on the model identification results, a trend-aware sliding mode controller is introduced for adjustment. Its core objective is to minimize high-frequency oscillations while ensuring response sensitivity, thereby improving the controller's steady-state accuracy and dynamic performance in non-stationary environments. The specific adjustment process of the trend-aware sliding mode controller is as follows:

[0043] First, construct the sliding surface function, the specific formula of which is as follows:

[0044] ;

[0045] in, It is a sliding surface function, representing the composite error relative to the target state at the current moment; For instantaneous error, It involves setting a target, specifically the target cable outer diameter, which is determined based on customer requirements. The integral gain factor is used to control the long-term error elimination rate. It is determined using a Bayesian optimization algorithm, with a reference value range of [value missing]. ; It is an integral term.

[0046] Furthermore, a composite control structure integrating multiple nonlinear transformation terms is designed to improve the system's flexible response capability and suppress controller chattering. The expression for the control output signal is:

[0047] ;

[0048] in, It is the control output signal, that is, the controller at a certain time. The output general regulation signal will be projected to each control actuator: temperature controller, screw frequency converter and traction motor; This is the sliding mode nonlinear feedback gain coefficient, which controls the direct influence of the sliding mode surface error on the control output. The reference value range is... ; This is the trend integral term gain coefficient, representing the strength of the influence of the control error change trend on the controller response. The reference value range is... ; It is the sliding mode error scaling factor, which controls the amount of composite error. exist The degree of stretching in the function controls the saturation rate; the reference value range is... ; It is the first derivative of the instantaneous error, representing the rate of change of the error, that is, the growth or convergence speed of the current instantaneous error; This refers to the length of the trend review time window, which determines how long the controller reviews error changes over a specific period. The reference range is [range to be specified]. ; These are the coefficients of the exponential-square root mixed term, representing the nonlinear control gain in the micro-error range, used for small-range adjustment. The reference value range is... ; It is the exponential amplification factor, which controls the strength of the effect of error growth on the exponential term and determines the output amplification rate. The reference range is [range missing]. ; It features a dual nonlinear adjustment term, providing fine adjustment in the small error range, resulting in smooth control behavior and avoiding large jumps. This is the prediction correction coefficient, representing the degree to which the control modeling error corrects the controller output. The reference value range is... ;

[0049] Furthermore, to maintain the robustness of the controller response, an existing piecewise structural parameter optimization strategy based on particle swarm optimization (PSO) is introduced to optimize the parameter set in the controller. Periodic online optimization is performed to obtain updated control output signals, which are then mapped and processed by existing command solvers to obtain control commands. These commands are then sent to various control actuators, such as temperature controllers, screw frequency converters, and traction motors, to achieve cable extrusion control.

[0050] In summary, a cable extrusion control method based on adaptive control has been developed.

[0051] The order of the embodiments is for illustrative purposes only and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0052] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0053] 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 the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A cable extrusion control method based on adaptive control, characterized in that, Including the following steps: S1. After the electric extruder is started, process parameter data is collected. After preprocessing, the preprocessed parameter data is obtained and used as a standardized vector. Then, the state identification process is carried out by introducing a least squares identification model with cointegration residual compensation term and nonlinear input disturbance suppression term to obtain the model identification result. S2. Based on the model identification results, a trend-aware sliding mode controller is introduced for adjustment. By constructing the sliding mode surface function and designing a composite control structure that integrates multiple nonlinear transformation terms, the control signal is determined. The parameters in the controller are optimized based on the piecewise structure parameter optimization strategy of the particle swarm algorithm to obtain the updated control signal, which is then mapped into control commands to realize cable extrusion control.

2. The cable extrusion control method based on adaptive control according to claim 1, characterized in that, In S1, in order to fully describe the dynamic change capability, based on the standardized vector, a least squares identification model is introduced by introducing a cointegration residual compensation term and a nonlinear input disturbance suppression term to perform state identification processing, and obtain the modeling parameters of the input variable on the output variable in the current state, that is, the model identification result.

3. The cable extrusion control method based on adaptive control according to claim 1, characterized in that, The least squares identification model in S1 introduces a cointegration residual compensation term and a nonlinear input disturbance suppression term. A dynamic residual compensation term is introduced into the modeling structure to absorb the structural shift of historical errors, and a regularization constraint on input changes is introduced to control the model's sensitivity. Its iterative formula is as follows: ； in, It is a moment Control input (i.e., normalized vector) and system output ( When modeling the mapping relationship between ), the parameter vector estimated in real time is the model identification result; It is a moment The parameter vector; It is a moment The system output, i.e., time... The outer diameter of the pre-treated cable; It is a moment The weighting factor is used to dynamically adjust the impact of the current cycle on the overall model update. ,in, It is a moment The system output, i.e., time... The outer diameter of the pre-treated cable; It is the perturbation sensitivity coefficient, which is used to dynamically adjust the model's confidence in perturbations, thereby ensuring identification stability under abrupt changes. It is a standardized vector; The modeling residuals from the previous period are used as cointegration residuals; Indicates transpose; For a moment Parameter vector The covariance matrix is ​​used to measure the uncertainty of parameter estimation and is updated using the recursive formula of the covariance matrix in standard recursive least squares theory. Forgetting factor, a modulatory factor used to regulate the rate at which the weights of historical observation data decay; Input the change terms; To mitigate the impact of perturbation on the weight coefficients and prevent unreasonable model jumps caused by abrupt input changes; It is the cointegration residual feedback gain, used to mitigate slow model drift. The reference value range is... .

4. The cable extrusion control method based on adaptive control according to claim 1, characterized in that, In step S2, a trend-aware sliding mode controller is introduced for adjustment based on the model identification results to ensure response sensitivity.

5. The cable extrusion control method based on adaptive control according to claim 1, characterized in that, The specific process of adjusting the trend-aware sliding mode controller in S2 is as follows: The first step is to construct the sliding surface function and determine the composite error relative to the target state at the current moment; The second step is to design a composite control structure that integrates multiple nonlinear transformation terms based on the composite error quantity, and determine the output control signal.

6. The cable extrusion control method based on adaptive control according to claim 1, characterized in that, The composite control structure in S2, which integrates multiple nonlinear transformation terms, has the following control output signal expression: ； in, It is the control output signal, that is, the controller at a certain time. The output general regulation signal will be projected to each control actuator: temperature controller, screw frequency converter and traction motor; It is the sliding mode nonlinear feedback gain coefficient, which controls the intensity of the direct influence of the sliding mode surface error on the control output; It is the trend integral term gain coefficient, representing the intensity of the influence of the control error change trend on the controller response; It is the sliding mode error scaling factor, which controls the amount of composite error. exist The degree of stretching in the function controls the saturation rate; It is the first derivative of the instantaneous error, representing the rate of change of the error, that is, the growth or convergence speed of the current instantaneous error; It is the length of the trend review time window, which indicates how long the controller reviews the error changes over the past period; It is the coefficient of the exponential-square root mixed term, the nonlinear control gain in the micro-error range, used for small-range adjustment; It is the exponential amplification factor, which controls the strength of the effect of error growth on the exponential term and determines the output amplification rate.

7. The cable extrusion control method based on adaptive control according to claim 1, characterized in that, In step S2, in order to maintain the robustness of the controller response, a segmented structure parameter optimization strategy based on particle swarm optimization (PSO) is introduced to periodically optimize the parameter set in the controller online, thereby obtaining the updated control output signal.

Images