Model-free adaptive iterative learning injection molding speed control method based on filter

Abstract

The invention provides a model-free adaptive iterative learning injection molding speed control method based on a filter. The method comprises the following steps: modeling an injection process as an actual output model containing measurement disturbance; constructing a control law with a filter, calculating control input, sending the control input to the injection molding system, and collecting measurement output after operation; if the sampling time is less than the maximum sampling time, updating the measurement output, and recalculating the control input; if the measurement error is greater than the error threshold value, entering the next iteration to recalculate the control input; and otherwise, taking the current control input as the control input of learning completion. Measurement disturbance is incorporated into the model, so that the model can reflect the injection molding system more truly; besides, high-frequency disturbance in measurement errors is filtered out through a filter, so that the high-frequency disturbance is closer to the real deviation of the injection speed, and the tracking precision of the injection speed is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of injection molding, in particular to a filter-based model-free adaptive iterative learning injection speed control method, system, device and storage medium. BACKGROUND

[0002] Injection molding, as an important plastic processing and molding process, melts solid plastic into a molten flow state, then injects it into a mold, and cools to form a plastic product with a specific shape and meet specific performance requirements. As a typical intermittent production process, the injection molding process consists of multiple stages, each of which is completed within a limited time and produces plastic products in batches periodically. During the entire injection molding process, precise control of multiple parameters is involved, such as barrel temperature, injection speed, nozzle pressure, and hydraulic pressure, among which injection speed is particularly critical as it directly affects the quality and production efficiency of plastic products. Therefore, precise control of injection speed is crucial to improving product quality and production efficiency.

[0003] In actual injection molding production, traditional PID (proportional-integral-derivative) control has been widely used in the control of injection speed due to its simple structure and ease of engineering implementation. However, this traditional PID method has weak anti-interference ability when facing high-frequency disturbances in the injection molding process: for example, high-frequency disturbances such as electronic thermal noise of sensors, pressure pulsation caused by high-frequency reciprocation of hydraulic pump plungers, and high-frequency valve core tremors of flow valves; since the parameters of traditional PID control are mainly based on static working condition setting, the response to such millisecond-level high-frequency disturbances is lagging, and it is difficult to correct their impact in real time, resulting in frequent deviation of the system output from the desired injection speed trajectory, and further affecting the consistency and qualification rate of injection molded products.

[0004] Therefore, how to design a filter-based model-free adaptive iterative learning injection speed control method, system, device and storage medium that can effectively resist high-frequency disturbances in the injection molding process is a technical problem that has not been solved in the prior art. SUMMARY

[0005] Therefore, the technical problem to be solved by the present application is to overcome the technical defect that the traditional PID control has weak anti-interference ability when facing high-frequency disturbances in the injection molding process, and to provide a filter-based model-free adaptive iterative learning injection speed control method, system, device and storage medium that can effectively resist high-frequency disturbances in the injection molding process.

[0006] The present application mainly includes the following aspects: In a first aspect, the present application provides a filter-based model-free adaptive iterative learning injection speed control method, which comprises: Step 1: obtaining the maximum sampling time of a single iteration , pseudo partial derivative initial value , initial control input of system operation , and initial system output data ; Wherein, is the control input, indicating the flow of the hydraulic system during the injection process at the th iteration, th sampling time; is the system output, indicating the injection speed during the injection process at the th iteration, th sampling time; is the sampling time, ; is the iteration number, ; Step 2: modeling the injection process as a nonlinear system model with iteration characteristics based on the input and output data; Step 3: linearizing the nonlinear system model with iteration characteristics in step 2 on the iteration axis by introducing a pseudo partial derivative , to obtain a compact format dynamic linearization model; Step 4: deriving the nonlinear system model with iteration characteristics in step 2 into a compact format dynamic linearization output model based on the compact format dynamic linearization model in step 3; Step 5: extending the compact format dynamic linearization output model in step 4 into an actual output model containing measurement disturbance by introducing a measurement disturbance : ; Wherein, is the measurement output containing measurement disturbance at the th iteration, th sampling time; the measurement disturbance is a random signal, and the statistical characteristics are , , wherein is a mathematical expectation factor, is a variance factor, the measurement disturbance is irrelevant to the sampling time , the iteration number , the measurement output , the system output , the control input , and the pseudo partial derivative , and the initial disturbance value of the measurement disturbance is 0; Step 6: Based on the desired output and the measured output , the measurement error is calculated Step 7: A low-pass filter is constructed and the measurement error is processed into an error signal by the low-pass filter ; wherein, is the measurement error of Step 6; is the low-pass filter transfer function used to filter out high-frequency disturbances in the measurement error ; wherein, the low-pass filter transfer function is: ; wherein, is the lead factor; is the filter parameter; is the filter order; filter for satisfies ; Step 8: A pseudo-derivative estimation algorithm is constructed and the pseudo-derivative estimate is calculated Step 9: Based on the error signal of Step 7 and the pseudo-derivative estimate of Step 8, a control law with filter is constructed and the control input is calculated and sent to the injection molding system, and the measured output after running is collected wherein, the control law with filter is: ; wherein, is the step factor, for making the control law with filter more general; is the weight factor; Step 10: The sampling time is compared with the maximum sampling time of the single iteration , if , then return to Step 6 and update the measured output of Step 6 to the measured output after running collected in Step 9; otherwise, Step 11 is executed Step 11: Collect all the measurement errors at all time instants in the current iteration, compare the absolute value of the maximum measurement error with the error threshold , if , go back to step 6, otherwise send the final control input to the injection molding system and end the iterative learning process.

[0007] According to an embodiment of the present application, the nonlinear system model with iterative characteristics modeled in step 2 comprises: ; wherein, is the order of the output variable, is the order of the input variable.

[0008] According to an embodiment of the present application, the compact dynamic linearization model established in step 3 comprises: ; ; ; wherein, and respectively represent the control input variation and the system output variation under the operation of the adjacent batch of the system; for the arbitrary sampling time and the iteration number , both satisfy , wherein, is a constant; all the pseudo partial derivatives have the same sign, i.e. , wherein, is a constant.

[0009] According to an embodiment of the present application, the compact dynamic linearization output model derived in step 4 comprises: .

[0010] According to an embodiment of the present application, the method for calculating the measurement error in step 6 comprises: ; wherein, is the expected output, representing the known optimal injection speed; is the measurement output at the th iteration, time instant.

[0011] ​According to one embodiment of the present application, the step 8 of constructing a pseudo partial derivative estimation algorithm and calculating a pseudo partial derivative estimation value comprises: The step of constructing a pseudo partial derivative estimation function and determining a quantitative index of estimation error comprises: The pseudo partial derivative estimation function comprises: ; wherein, is a target function; is a weight coefficient; is an estimation value of the pseudo partial derivative; is a measurement output change, and the calculation formula is: ; is a control input change, and the calculation formula is: ; Based on the pseudo partial derivative estimation function and the optimal condition , the pseudo partial derivative estimation algorithm is obtained, and the pseudo partial derivative estimation value is calculated; the pseudo partial derivative estimation algorithm comprises: ; wherein, is a step factor, which is used for conveniently adjusting the pseudo partial derivative and enhancing the flexibility of the algorithm; is a weight coefficient; A reset algorithm is constructed, which is used for ensuring that the pseudo partial derivative estimation algorithm has good tracking performance; the reset algorithm comprises: ; If or or ; ; wherein, is the initial value of the pseudo partial derivative obtained in the step 1 ; is a reset threshold; is a sign function.

[0012] According to one embodiment of the present application, the step 9 of constructing a control law with a filter based on the error signal of the step 7 and the pseudo partial derivative estimation value of the step 8 comprises: A control input target function is constructed: ; ; wherein, is a target function; is a weight factor; Based on the control input target function and the optimal condition , a control law is obtained: ; wherein, is a step factor, for making the control law more general; is a weight factor; based on the control law, the error signal of step 7 and the pseudo-derivative estimate of step 8 establishes the control law with filter for calculating the control input .

[0013] In a second aspect, the embodiments of the present application further provide a filter-based model-free adaptive iterative learning injection speed control system, which comprises: an acquisition module configured to acquire the maximum sampling time of a single iteration , a pseudo-derivative initial value , an initial control input , initial system output data , an expected output and a measured output ; a storage module configured to store relevant data; a calculation module configured to calculate a measured error , a pseudo-derivative estimate , a control input ; the calculation module comprises a filter processing unit configured to process the measured error into an error signal by a low-pass filter; a comparison module configured to compare the sampling time with the maximum sampling time of a single iteration , and compare the absolute value of the measured error maximum value with an error threshold ; an output module configured to send the final control input to an injection system.

[0014] In a third aspect, the embodiments of the present application further provide a computer device, which comprises at least one processor, and a memory connected with the at least one processor in communication; wherein the memory stores instructions executable by the one processor, and the instructions are executed by the at least one processor, so that the at least one processor executes the instructions to implement the steps of the above-mentioned method.

[0015] In a fourth aspect, the embodiments of the present application further provide a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the steps of the method described above.

[0016] The technical scheme provided by the present application has the following advantages: 1. The present application incorporates the measurement disturbance into the model, so that the model can more truly reflect the injection molding system, and ensure that the control law designed based on the model can respond to the disturbance influence in a targeted manner, thereby guaranteeing the effectiveness of the control strategy in actual working conditions and avoiding the problem of error expansion or system instability caused by not considering the disturbance.

[0017] 2. The present application processes the measurement error into an error signal through a low-pass filter, filters out the high-frequency disturbance in the measurement error, so that the error signal is closer to the real deviation of the injection speed, and the control law constructed based on this can more accurately adjust the flow of the hydraulic system to correct the speed deviation, thereby significantly improving the tracking accuracy of the injection speed.

[0018] 3. The filtered error signal of the present application reduces the control input fluctuation caused by high-frequency interference, so that the action of the hydraulic system, the injection screw and other actuators is more stable, and the stability of the system in the key stage of the injection molding process is enhanced, thereby adapting to the nonlinear and repetitive characteristics of the injection molding process, and guaranteeing the consistency of the quality of plastic products.

[0019] 4. The filter of the present application can not only filter out high-frequency measurement disturbances caused by external environmental factors or the sensor itself, but also eliminate high-frequency signals caused by changes in system structure and parameters, thereby effectively suppressing the adverse effects of high-frequency disturbances and achieving better tracking control effect.

[0020] 5. The model-free adaptive iterative learning control strategy adopted in the embodiments of the present application does not need to rely on the accurate mathematical model of the injection molding process, but only relies on real-time system input and output data and an iterative learning mechanism, so that the control strategy can be dynamically optimized to adapt to the complex nonlinear characteristics in the injection molding process, thereby accurately controlling the injection speed.

[0021] 6. The model-free adaptive iterative learning control strategy adopted in the present application does not rely on the accurate mathematical model of a specific system, so it is not only suitable for injection speed control in the injection molding process, but also can be extended to other dynamic systems with nonlinear and repetitive characteristics, and has good universality.

[0022] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the following preferred embodiments are described in detail below, and the accompanying drawings are described as follows. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. It should be understood that the following drawings only show some of the embodiments of the present application, and therefore should not be regarded as a limitation on the scope. For those of ordinary skill in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0024] Figure 1 is the overall flowchart of the filter-based model-free adaptive iterative learning injection speed control method of the present application.

[0025] Figure 2 is the iterative running framework diagram of the present application.

[0026] Figure 3 is the system architecture diagram of the present application.

[0027] Figure 4 is the basic structure schematic diagram of the existing hydraulic screw injection molding machine.

[0028] Figure 5 is the schematic diagram of the existing optimal injection speed curve.

[0029] Figure 6 is the comparison schematic diagram of the maximum tracking error curve in the iterative domain of the present application and the general model-free adaptive iterative learning control scheme.

[0030] Figure 7 is the comparison schematic diagram of the maximum tracking error curve in the iterative domain of the present application under different measurement noise variance conditions.

[0031] Figure 8 is the injection speed tracking curve schematic diagram under different control schemes.

[0032] Figure 9 is the injection speed tracking error situation schematic diagram of the present application and the PID control scheme. DETAILED DESCRIPTION

[0033] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of description and illustration, and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn according to the actual proportions. The flowcharts used in the present application show the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowcharts can not be implemented in sequence, and the steps without logical context relationship can be reversed in sequence or implemented simultaneously. In addition, one or more other operations can be added to the flowcharts or one or more operations can be removed from the flowcharts under the guidance of the content of the present application.

[0034] In addition, the described embodiments are only some of the embodiments of the present application, not all the embodiments. The components of the embodiments of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0035] The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0036] Embodiments

[0037] As shown in Figure 1 and Figure 2 , the present embodiment provides a filter-based model-free adaptive iterative learning injection speed control method, comprising the following steps: Step 1: obtaining the maximum sampling time of a single iteration , the initial control input of system operation and the initial system output data ; ; wherein, is the control input, indicating the flow of the hydraulic system in the injection process at the th iteration, time; is the system output, indicating the injection speed in the injection process at the th iteration, time; is the sampling time, ; is the number of iterations, ; Step 2: based on the input and output data, modeling the injection process as a nonlinear system model with iterative characteristics; Step 3: linearizing the nonlinear system model with iterative characteristics of Step 2 on the iteration axis by introducing pseudo partial derivatives to obtain a compact format dynamic linearization model; Step 4: based on the compact format dynamic linearization model of Step 3, deriving the nonlinear system model with iterative characteristics of Step 2 into a compact format dynamic linearization output model; Step 5: extending the compact format dynamic linearization output model of Step 4 into an actual output model with measurement disturbance by introducing measurement disturbance ; ; wherein, is the measurement output with measurement disturbance at the kth iteration, is the kth iteration, is the measurement output with measurement disturbance at the kth iteration, is a random signal with statistical characteristics , wherein is a mathematical expectation factor, is a variance factor, the measurement disturbance is irrelevant to the sampling time , the iteration number , the measurement output , the system output , the control input and the pseudo partial derivative , the initial disturbance value of the measurement disturbance is 0; Step 6: based on the expected output and the measurement output , calculating the measurement error ; Step 7: constructing a low-pass filter and processing the measurement error into an error signal through the low-pass filter: ; wherein, is the measurement error of Step 6; is the low-pass filter transfer function for filtering out high-frequency disturbance in the measurement error ; wherein, the low-pass filter transfer function is: ; wherein, is a forward shift factor; is a filter parameter; is a filter order; filter for satisfies ; Step 8: Construct a pseudo-derivative estimation algorithm and calculate a pseudo-derivative estimation value ; Step 9: Based on the error signal of step 7 and the pseudo-derivative estimation value of step 8 , construct a control law with filter, calculate the control input , send the control input to the injection molding system, collect the measured output after running ; wherein, the control law with filter is: ; wherein, is a step factor, , for making the control law with filter more general; is a weight factor; Step 10: Compare the sampling time with the maximum sampling time of the single iteration , if , return to step 6 and update the measured output of step 6 to the measured output after running collected in step 9 ; otherwise, perform step 11; Step 11: Collect the measurement error of all times within the current iteration, compare the absolute value of the maximum measurement error with the error threshold , if , return to step 6, otherwise send the final control input to the injection molding system and end the iterative learning process.

[0038] The following describes each step of the filter-based model-free adaptive iterative learning injection speed control method in detail: Step 1: Obtain the maximum sampling time of a single iteration , the pseudo-derivative initial value , the initial control input of system running and the initial system output data ; wherein, is the control input, representing the i-th the i-th iteration, the flow rate of the hydraulic system at the i-th iteration, is the system output, representing the i-th iteration, the i-th iteration, the injection speed at the i-th iteration, is the sampling time, ; is the iteration number, ; Here, the maximum sampling time of a single iteration The initial value of the pseudo partial derivative is 10, the initial control input at the first iteration is 0, the initial system output data is 0.

[0039] It should be noted that, as shown in Figure 4 , the basic structure of the hydraulic screw injection molding machine includes a clamping unit, an injection unit, a power unit, and a control unit; wherein the injection unit includes a plasticizing device, a screw driving device, an injection device, a measuring device, an injection seat, a rack, and a barrel feeding device; wherein the injection device includes a cylinder, a screw, and a nozzle, and during injection, hydraulic oil is injected into the cylinder to generate pressure to push the screw to press the melt in front of the cylinder, so that the melt enters the mold cavity through the nozzle.

[0040] Based on the structure of the above hydraulic screw injection molding machine, it can be obtained that the control input of the embodiment of the present application is the flow rate of the hydraulic system during the injection process, specifically the flow rate of the hydraulic oil injected into the cylinder; correspondingly, the system output of the embodiment of the present application is the injection speed during the injection process, specifically the speed of the melt injected into the mold cavity through the nozzle, which can be measured by a speed sensor.

[0041] Step 2: based on the input and output data, modeling the injection process as a nonlinear system model with iterative characteristics; In one possible implementation, the nonlinear system model with iterative characteristics modeled in step 2 includes: ; wherein, is the order of the output variable, is the order of the input variable.

[0042] It should be noted that the injection molding production process begins with the closure of the mold, the mold is closed, and the nozzle of the machine cylinder is in contact with the main bushing of the mold. When the mold is closed and the nozzle is pressed against the main bushing, the injection process begins: the hydraulic system provides hydraulic oil to the injection cylinder to push the screw forward, and the high polymer melt at the head of the screw is pushed by the screw to enter the mold cavity through the nozzle, the runner, the flow channel, and the gate. After the injection process is completed, the holding stage begins: to prevent the melt in the mold cavity from flowing back due to the action of the reverse pressure gradient, and to compensate for the shrinkage of the plastic material due to cooling and solidification, the injection nozzle must maintain a certain pressure to push the melt into the mold cavity. When the melt in the mold cavity cools and cannot flow back, the cooling stage begins, and the plastic product is further cooled and shaped in the mold cavity. At the same time, the system enters the plasticizing stage: the hydraulic motor drives the screw to rotate, and the material is transported forward along the screw groove under the action of friction. At the same time, the material begins to melt and plasticize under the action of the machine cylinder heating ring, and changes into a viscous fluid stored in the storage room at the front of the screw, preparing for the next injection. After the plasticizing stage and the cooling stage are completed, the mold is opened, the ejection device is connected to make the product fall off, and the injection molding process is completed.

[0043] Wherein, the injection stage can be represented by a model constructed by the following equations: ; ; ; ; Wherein, is the cross-sectional area of the injection cylinder, is the cross-sectional area of the barrel, is the volume of the injection rod, is the volume of the material in the barrel, is the injection speed, is the flow rate of the hydraulic system, is the bulk modulus of the hydraulic fluid, is the bulk modulus of the nozzle material, is the injection cylinder pressure, is the nozzle pressure, is the nozzle radius, is the mass of the screw, is the injection position, is the power law index of the material melting, is the initial length of the screw, is the radius ratio of the screw to the nozzle, is the shear rate the corresponding melt viscosity, is the average flow rate of the material.

[0044] Due to different working environments and measurement conditions, modeling in actual environment is more complex, and it is more difficult to construct a mathematical model that accurately reflects the real situation. Therefore, the following nonlinear system model is constructed to represent the above injection stage model: ; wherein, is the order of the output variable, is the order of the input variable; is a nonlinear function.

[0045] Since the injection molding process has repeatability, an iteration index characterizing the number of iterations of the system can be introduced, so that the above nonlinear system model is further modeled as a nonlinear system model with iteration characteristics: .

[0046] The model-free adaptive iterative learning control strategy adopted in the embodiment of the present application does not need to rely on the accurate mathematical model of the injection molding process, but only relies on real-time system input and output data and iterative learning mechanism, so as to dynamically optimize the control strategy, adapt to the complex nonlinear characteristics in the injection molding process, and thus accurately control the injection speed.

[0047] Similarly, since the embodiment of the present application does not rely on the accurate mathematical model of a specific system, it is not only applicable to the injection speed control of the injection molding process, but also can be extended to other dynamic systems with nonlinear and repetitive characteristics, and has good universality.

[0048] Step 3: By introducing pseudo partial derivative , the nonlinear system model with iteration characteristics in step 2 is linearized on the iteration axis to obtain a compact format dynamic linearization model; In one possible implementation, the compact format dynamic linearization model established in step 3 includes: ; ; ; wherein, and respectively represent the control input change and the system output change under the operation of adjacent batches of the system; for the arbitrary sampling time and the iteration number , both satisfy , wherein, is a constant; all pseudo partial derivatives symbols are consistent, that is, , wherein, is constant.

[0049] Here, the derivation logic of the compactly formatted dynamic linearization model established in step 3 is: The nonlinear system model with iterative characteristics satisfies the assumption of the compactly formatted dynamic linearization method: Assumption 1: The partial derivative of the control input is continuous, that is, , is a positive number.

[0050] Assumption 2: The nonlinear system model with iterative characteristics satisfies the generalized Lipschitz condition on the iteration axis, that is, for , and , ; wherein , and is a constant; ; .

[0051] When the nonlinear system model with iterative characteristics satisfies assumptions 1 and 2, if , there exists a pseudo partial derivative (PPD) such that the injection process system is represented as a compactly formatted dynamic linearization model: ; Step 4: Based on the compactly formatted dynamic linearization model of step 3, the nonlinear system model with iterative characteristics of step 2 is derived into a compactly formatted dynamic linearization output model; In one possible implementation, based on the compactly formatted dynamic linearization model of step 3, the nonlinear system model with iterative characteristics of step 2 can be converted into a compactly formatted dynamic linearization output model: .

[0052] Specifically, by substituting into , we can obtain: .

[0053] Step 5: By introducing a measurement disturbance , the compactly formatted dynamic linearization output model of step 4 is extended into an actual output model containing a measurement disturbance: ; wherein is the th iteration, At this moment, the measurement output including the measurement disturbance; the measurement disturbance is a random signal with the statistical characteristics , ,in is the mathematical expectation factor, is the variance factor, measuring the disturbance With the sampling time , the number of iterations , the measured output , the system output , the control input and the pseudo partial derivative Are not correlated, measurement disturbance The initial perturbation value is 0; Here, the measurement disturbance is a variety of high-frequency disturbances, which may be caused by external environmental factors or the sensor itself, or may be high-frequency signals caused by changes in system structure and parameters. These disturbances are not simulated, but real and can affect the sensor measurement results. The embodiment of the present application takes into account the impact of these measurement disturbances on the measurement results, so the measurement disturbance is added to the compact dynamic linearization output model. , thus obtaining the actual output model with measurement disturbance: .

[0054] The embodiment of the present application incorporates the measured disturbance into the model, so that the model can more realistically reflect the injection molding system, ensure that the control law designed based on the model can respond to the influence of the disturbance in a targeted manner, ensure the effectiveness of the control strategy in actual working conditions, and avoid error expansion or system instability caused by not considering the disturbance.

[0055] Step 6: Based on the expected output and the measured output , calculate the measurement error ; In one possible implementation, the measurement error is calculated in step 6. The methods include: ; in, is the expected output, representing the known optimal injection speed; For the iterations, The measured output at the time.

[0056] It should be noted that the desired output of this embodiment is the optimal injection speed curve that changes dynamically with time (e.g. Figure 5The control target of the present embodiment is thus the precise "tracking" of the optimal injection speed profile, with the focus on dynamically following the preset speed trajectory, rather than simply adjusting the injection speed to a fixed value.

[0057] Here, the optimal injection speed profile is further described as follows: From the general industry cognition and practical experience, when the molten material can be injected into the mold cavity at a stable and uniform rate at a certain specific injection speed, the benefits achieved are multi-dimensional and significant. On the one hand, it can ensure that the injection molded products fully meet the stringent standards set by the injection molding production in terms of specification accuracy, color consistency, and appearance integrity, greatly reducing the proportion of defective products; on the other hand, it can effectively promote the overall production efficiency of the injection molding process, maximizing resource utilization and production capacity release. The injection speed corresponding to this time is defined as the optimal injection speed in the industry, which is equivalent to the expected injection speed.

[0058] In the injection molding process, the screw continuously pushes the melt into the mold cavity by rotating and advancing, which can be divided into four stages. The first stage is the high-speed pushing stage, in which the screw pushes the melt at a faster speed to reduce heat loss, maintain good flowability of the melt, reduce heat loss during transportation, and ensure that the melt maintains appropriate temperature and rheological properties to smoothly enter the mold cavity. In the second stage, the mold cavity port deceleration stage, the screw pushing speed needs to be reduced in time when the melt front reaches the mold cavity port, so as to avoid splashing when the melt hits the mold cavity port at high speed, prevent material waste and uneven pressure inside the mold cavity, and ensure smooth injection molding process. In the third stage, the acceleration filling stage, the screw needs to speed up the pushing speed to achieve rapid and full filling of the mold cavity, prevent the melt from being affected by cooling and losing flowability, ensure that the melt fills every corner of the mold cavity in a short time, and ensure the integrity of the product forming. In the fourth stage, the second deceleration stage, the screw needs to reduce the pushing speed again, because if the injection speed is too fast, the melt injection amount will exceed the expectation, which may cause the product to be unqualified, such as generating flash to affect the size accuracy and appearance quality, and may also cause internal structural defects to reduce the mechanical properties, reducing the speed can accurately control the melt injection amount, and ensure that the product quality meets the standard.

[0059] Step 7: Construct a low-pass filter and pass the measurement error through the low-pass filter to process it as an error signal : ; wherein, the measurement error of step 6; is the low-pass filter transfer function for filtering out high-frequency disturbances in the measurement error ; wherein the low-pass filter transfer function is: ; wherein, is a pre-shift factor; is a filter parameter; is a filter order; the filter for satisfies ; The embodiment of the present application processes the measurement error into an error signal through a low-pass filter, filters out high-frequency disturbances in the measurement error, makes the error signal closer to the real deviation of the injection speed, and constructs a control law based on this, which can more accurately adjust the flow of the hydraulic system to correct the speed deviation, thereby significantly improving the tracking accuracy of the injection speed.

[0060] The filtered error signal of the embodiment of the present application reduces the control input fluctuation caused by high-frequency interference, makes the action of the hydraulic system, injection screw rod and other actuators more stable, enhances the stability of the system in the key stage of the injection molding process, thereby adapting to the nonlinear and repetitive characteristics of the injection molding process, and guarantees the consistency of the quality of plastic products.

[0061] The filter of the embodiment of the present application can not only filter out high-frequency measurement disturbances caused by external environmental factors or the sensor itself, but also eliminate high-frequency signals caused by changes in system structure and parameters, thereby effectively suppressing the adverse effects of high-frequency disturbances and achieving better tracking control effect.

[0062] Step 8: constructing a pseudo partial derivative estimation algorithm and calculating a pseudo partial derivative estimation value ; In a possible implementation, the step of constructing a pseudo partial derivative estimation algorithm and calculating a pseudo partial derivative estimation value in step 8 includes the following steps: S801: establishing a pseudo partial derivative estimation function and determining a quantitative index of estimation error; the pseudo partial derivative estimation function includes: ; wherein, is an objective function; is a weight coefficient; is an estimation value of the pseudo partial derivative; is a measurement output change amount, and the calculation formula is: ; is a control input change amount, and the calculation formula is: ; S802: obtaining the pseudo partial derivative estimation algorithm based on the pseudo partial derivative estimation function and the optimal condition , and calculating the pseudo partial derivative estimation value ; the pseudo partial derivative estimation algorithm comprises: ; wherein, is a step factor, for facilitating adjustment of the pseudo partial derivative and enhancing flexibility of the algorithm; is a weight coefficient; S803: a reset algorithm is established to ensure that the pseudo partial derivative estimation algorithm has good tracking performance; the reset algorithm comprises: ; if or or ; wherein, is the pseudo partial derivative initial value obtained in step 1 ; is a reset threshold, specifically ; is a sign function.

[0063] Step 9: based on the error signal of step 7 and the pseudo partial derivative estimation value of step 8, a control law with a filter is constructed to calculate the control input , and the control input is sent to the injection molding system, and the measured output after running is collected; wherein, the control law with a filter is: ; wherein, is a step factor, for making the control law with a filter more general; is a weight factor; In one possible implementation, the step of constructing a control law with a filter based on the error signal of step 7 and the pseudo partial derivative estimation value of step 8 in step 9 comprises: S901: a control input target function is established: ; wherein, is a target function; is a weight factor; S902: based on the control input target function and the optimal condition , a control law is obtained: ; wherein, is a step factor, for making the control law more general; is a weight factor; S903: based on the control law, the error signal of step 7 and the pseudo partial derivative estimate value of step 8 , the control law with filter is established, for calculating the control input .

[0064] Step 10: comparing the sampling time with the maximum sampling time of the single iteration , if , returning to step 6 and updating the measurement output of step 6 to the post-operation measurement output collected in step 9 ; otherwise, step 11 is executed; Step 11: collecting the measurement errors of all time points in the current iteration, comparing the absolute value of the maximum measurement error with an error threshold value , if , returning to step 6, otherwise, sending the final control input to the injection molding system and ending the iterative learning process.

[0065] The absolute value of the maximum measurement error can reflect the worst control accuracy in the current iteration; if the value is less than or equal to the error threshold value , it means that the deviation of all time points in the current iteration is within the acceptable range, avoiding product defects caused by the average error being qualified but the deviation of local time points exceeding the standard; here, the error threshold value is specifically .

[0066] As shown in Figure 3 , based on the same application concept, the embodiment of the present application also provides a filter-based model-free adaptive iterative learning injection speed control system, characterized in that it comprises an acquisition module, a storage module, a calculation module, a comparison module and an output module; the acquisition module is used to acquire the maximum sampling time of a single iteration , a pseudo partial derivative initial value , an initial control input , initial system output data , an expected output and a measurement output ; the storage module is used to store related data; the calculation module is used to calculate a measurement error and a pseudo partial derivative estimate value , control input ; the calculation module comprises a filter processing unit configured to process the measurement error into an error signal by a low-pass filter; the comparison module is configured to compare the sampling time with the maximum sampling time of a single iteration, and compare the absolute value of the maximum measurement error with an error threshold value; and the output module is configured to send the final control input to the injection molding system.

[0067] Based on the same application concept, the embodiments of the present application also provide a computer device, comprising at least one processor; and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the instructions to implement the above method.

[0068] Based on the same application concept, the embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the above method.

[0069] Convergence analysis is performed as follows: 1. Prove that the error converges necessarily: The calculation formula of the tracking error is as follows: ; the formula (1) can be obtained by substituting the dynamic linearization output model in the tight format into the control law with the filter: ; The formula (2) can be obtained from : ; Wherein, is a positive constant.

[0070] The absolute value of both sides of the formula (1) is taken, and the formula (3) can be obtained in combination with the formula (2): ; Wherein, converges to 0 in a limited time interval , when approaches infinity, is bounded, which means that when the injection molding system output contains measurement disturbance, converges and is bounded; therefore, the embodiments of the present application have strong robustness, and the injection molding process obtains good tracking output performance.

[0071] ​​The above proof process proves from the algebraic convergence level that the tracking error of the injection molding process containing the measurement disturbance can converge to zero along the iteration axis point by point, that is, it proves that "the error must converge"; 2. Prove that the convergence is not destroyed by random disturbance: Taking the square of both sides of formula (1) can obtain formula (7): Formula (4) can be obtained: ; Therefore, formula (5) can be obtained: ; Taking the expectation of both ends of formula (5) can obtain formula (6): ; Taking the square of both sides of formula (1) can obtain formula (7): ; From It is known that From formula (2), it can be obtained that Therefore, for all There exists a positive constant Satisfying formula (8): ; From formula (7) and formula (8), formula (9) can be obtained: ; Taking the expectation of both ends of formula (9) can obtain formula (10): ; Since ; And Formula (11) can be obtained: ; From formula (10) and formula (11), formula (12) can be obtained: ; From the above derived And It can be obtained that: even if there is a measurement disturbance, the mathematical expectation and variance of the error can still converge to zero by the embodiments of the application, which guarantees the statistical optimality and disturbance robustness, that is, it proves that "the convergence is not destroyed by random disturbance".

[0072] It should be noted that for the low-pass filter, when At this time, this is only an ideal case; in the actual application process, due to the influence of various factors, there will inevitably be a certain degree of filter error, although the error is relatively small; in view of this, the variance of the output error will not be equal to zero in a strict sense, but will tend to a very small positive number; but even if the above-mentioned actual situation exists, compared with the general model-free adaptive iterative learning control algorithm, the embodiment of the application still shows significant advantages in the comprehensive performance of trajectory tracking accuracy, disturbance suppression ability and the like.

[0073] So far, the convergence analysis is over, and the following simulation verification is carried out: It should be noted that, in order to simplify the model of the injection stage, the following state variables are introduced: The model of the injection stage is simplified to the following state equation: ; Since the injection stage belongs to a nonlinear system, the model of the injection stage given by the embodiment of the application is only used to generate input and output data. The remaining parameters in the model are known system parameters, and the parameter size and physical meaning represented by them are shown in the following table 1:

[0074] Table 1

[0075] The following simulation experiment is carried out by means of the MATLAB platform, aiming to verify the effectiveness of the embodiment of the application in the injection speed trajectory tracking. According to the state equation, the injection speed system of the injection molding machine belongs to a single-input single-output system; based on this, a mathematical model of the injection speed is constructed in Simulink; in addition, in order to simulate the measurement disturbance existing in the actual environment, white noise with a frequency range of 500-600Hz is introduced into the system, so as to verify the robustness of the embodiment of the application.

[0076] In the MATLAB simulation process, the controller parameters involved are set as follows: initial control input initial system output data maximum sampling time of a single iteration the sampling time is and the filter is set to: ;​​​​​​​​​ wherein the filter satisfies under the condition that .

[0077] The results of the simulation verification are as follows: Figure 6 The comparison of the maximum tracking error (maximum learning error) curves in the iteration domain of the embodiment of the application (MFAILC scheme with filter) and the general model-free adaptive iterative learning control scheme (MFAILC scheme) is shown; it can be seen from Figure 6 that after the 40th iteration learning, the tracking error of the embodiment of the application is significantly reduced, and the convergence speed is faster than that of the general model-free adaptive iterative learning control scheme; therefore, the convergence performance of the embodiment of the application is superior to that of the general model-free adaptive iterative learning control scheme.

[0078] Figure 7 The comparison of the maximum tracking error curves in the iteration domain of the embodiment of the application under different measurement noise variances is shown; it can be seen from Figure 7 that the smaller the measurement noise variance is, the more significant the convergence performance of the embodiment of the application is improved, and even if the measurement noise variance is large, the tracking error can be ensured to converge on the iteration axis.

[0079] Figure 8 The injection speed tracking curves under different control schemes are shown; in the injection molding process system, PID control is a relatively common control method, and the structure of the PID controller is wherein , , ; it can be seen from Figure 8 that after 50 iterations of learning, the embodiment of the application has obvious advantages in injection speed tracking accuracy compared with the general MFAILC scheme and the PID control scheme.

[0080] Figure 9 The injection speed tracking error of the embodiment of the application and the PID control scheme is shown; it can be seen from Figure 9 and calculation that at the 50th iteration learning, the average error value of the tracking error of the embodiment of the application is , and the average error value of the tracking trajectory error of the injection molding system under the PID control is ; it can be seen that the average error of the tracking error of the embodiment of the application is significantly smaller than the average error of the tracking error under the PID control after 50 iterations, which fully embodies the excellent tracking performance of the embodiment of the application.

[0081] The above experimental results clearly show that the embodiments of the present application can efficiently inhibit the adverse effects of measurement disturbance on the injection speed, effectively reduce the tracking deviation, so that the injection speed can closely and stably follow the preset trajectory, thereby ensuring that the injection molding process can maintain a high-precision running state under complex interference, and providing a strong guarantee for the high quality and stability of injection molding production.

[0082] In the embodiments of the present application, the computer program can also execute other machine-readable instructions when executed by the processor to perform the methods described in the embodiments, and the specific method steps and principles are described in the embodiments, which will not be described in detail here.

[0083] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other manners. The described device embodiments are only schematic. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some communication interfaces, devices or units, and can be electrical, mechanical or other forms.

[0084] The units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments.

[0085] In addition, each functional unit in the embodiments of the present application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0086] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0087] It should be noted that similar reference numbers and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. In addition, the terms "first", "second", "third" and the like are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0088] Finally, it should be noted that the above-described embodiments are only specific implementations of the present application, which are used to illustrate the technical solutions of the present application, but not to limit them. The protection scope of the present application is not limited thereto. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily think of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed by the present application, or make equivalent replacements to some of the technical features. These modifications, changes or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application. They should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A filter-based model-free adaptive iterative learning injection speed control method, characterized in that: include: Step 1: Get the maximum sampling time of a single iteration , initial value of pseudo partial derivative , initial control input for system operation and initial system output data ; in, is the control input, indicating the iterations, The flow rate of the hydraulic system during the injection process at all times; is the system output, indicating the iterations, Injection speed during the injection process; is the sampling time, ; is the number of iterations, ; Step 2: Based on the input and output data, the injection process is modeled as a nonlinear system model with iterative characteristics; Step 3: By introducing pseudo partial derivatives , linearizing the nonlinear system model with iterative characteristics in step 2 on the iterative axis to obtain a compact dynamic linearized model; Step 4: Based on the compact dynamic linearization model of step 3, derive the nonlinear system model with iterative characteristics of step 2 into a compact dynamic linearization output model; Step 5: By introducing a measurement perturbation , expand the compact dynamic linearized output model in step 4 into an actual output model with measurement disturbance: ； in, For the iterations, At this moment, the measurement output including the measurement disturbance; the measurement disturbance is a random signal with the statistical characteristics , ,in is the mathematical expectation factor, is the variance factor, measuring the disturbance With the sampling time , the number of iterations , the measured output , the system output , the control input and the pseudo partial derivative Are not correlated, measurement disturbance The initial perturbation value is 0; Step 6: Based on the expected output and the measured output , calculate the measurement error ; Step 7: Construct a low-pass filter and pass the measurement error through the low-pass filter Processed as error signal : ； in, is the measurement error of step 6; is the low-pass filter transfer function, used to filter out the measurement error High-frequency disturbances in Wherein, the low-pass filter transfer function for: ； in, is the forward shift factor; is the filter parameter; is the filter order; filter for satisfy ; Step 8: Construct the pseudo partial derivative estimation algorithm and calculate the pseudo partial derivative estimate ; Step 9: Based on the error signal of step 7 and the pseudo partial derivative estimate of step 8 , construct a control law with a filter, calculate the control input , the control input Send to the injection molding system and collect the measured output after operation ; Wherein, the control law with the filter is: ； in, is the step size factor, , used to make the control law with the filter more general; is the weight factor; Step 10: The sampling time The maximum sampling moment of the single iteration For comparison, if , then return to step 6 and output the measurement in step 6 Updated to the measured output after the run collected in step 9 ; Otherwise, go to step 11; Step 11: Collect the measurement errors at all times in this iteration, and calculate the absolute value of the maximum measurement error. and error threshold For comparison, if , then return to step 6, otherwise the final control input Sent to the injection molding system, ending the iterative learning process.

2. The filter-based model-free adaptive iterative learning injection speed control method according to claim 1, characterized in that: The nonlinear system model with iterative characteristics modeled in step 2 includes: ； in, is the order of the output variable, is the order of the input variable.

3. The filter-based model-free adaptive iterative learning injection speed control method according to claim 1, characterized in that: The compact dynamic linearization model established in step 3 includes: ； ； ； in, and Respectively represent the control input change and system output change under adjacent batches of the system; for any sampling time and the number of iterations , all satisfy ,in, is a constant; all pseudo-partial derivatives The symbols remain the same, that is, ,in, is a constant.

4. The filter-based model-free adaptive iterative learning injection speed control method according to claim 1, characterized in that: The compact dynamic linearization output model derived in step 4 includes: 。 5. The filter-based model-free adaptive iterative learning injection speed control method according to claim 1, characterized in that: The measurement error is calculated in step 6 The methods include: ； in, is the expected output, representing the known optimal injection speed; For the iterations, The measured output at the time.

6. The filter-based model-free adaptive iterative learning injection speed control method according to claim 1, characterized in that: In step 8, a pseudo partial derivative estimation algorithm is constructed and the pseudo partial derivative estimation value is calculated. The steps include: Establish a pseudo partial derivative estimation function and determine a quantitative index of the estimation error; the pseudo partial derivative estimation function includes: ;in, is the objective function; is the weight coefficient; is the estimated value of the pseudo partial derivative; To measure the output change, the calculation formula is: ; To control the input variation, the calculation formula is: ; Estimation function and optimal conditions based on the pseudo partial derivatives , obtain the pseudo partial derivative estimation algorithm, and calculate the pseudo partial derivative estimation value ; The pseudo partial derivative estimation algorithm includes: ;in, is the step size factor, , used to facilitate the adjustment of pseudo partial derivatives and enhance the flexibility of the algorithm; is the weight coefficient; A reset algorithm is established to ensure that the pseudo partial derivative estimation algorithm has good tracking performance; the reset algorithm includes: ； like or or ; in, is the initial value of the pseudo partial derivative obtained in step 1 ; To reset the threshold; is a symbolic function.

7. The filter-based model-free adaptive iterative learning injection speed control method according to claim 1, characterized in that: The error signal in step 9 is based on step 7. and the pseudo partial derivative estimate of step 8 , the steps to construct a control law with a filter include: Establish the control input objective function: ； in, is the objective function; is the weight factor; Based on the control input objective function and optimal conditions , and the control law is obtained: ; in, is the step size factor, , used to make the control law more general; is the weight factor; Based on the control law, the error signal of step 7 and the pseudo partial derivative estimate of step 8 , establish the control law with filter to calculate the control input .

8. A filter-based model-free adaptive iterative learning injection speed control system, characterized in that: include: Acquisition module: used to obtain the maximum sampling time of a single iteration , initial value of pseudo partial derivative , initial control input , initial system output data , expected output and measurement output ; Storage module: used to store relevant data; Calculation module: used to calculate measurement error , pseudo partial derivative estimates , control input The calculation module includes: a filtering processing unit, the filtering processing unit is used to pass the measurement error through a low-pass filter Processed as error signal ; Comparison module: used to compare sampling time The maximum sampling time of a single iteration Compare and compare the absolute value of the maximum measurement error and error threshold Make comparisons; Output module: used to input the final control Sent to the injection molding system.

9. A computer device, characterized in that: include: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the one processor, and the instructions are executed by the at least one processor so that the at least one processor implements the steps of the method according to any one of claims 1 to 7 when executing the instructions.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images