Full-electric injection molding machine double-motor parallel injection control method based on cross coupling

Abstract

The invention relates to the technical field of injection molding machines, in particular to an all-electric injection molding machine double-motor parallel injection control method based on cross coupling, and the method comprises the following steps: a signal collection step: collecting the rotating speeds of a master servo motor and a slave servo motor, calculating a tracking error and a synchronization error, and establishing a cross coupling relationship; an error compensation step: performing state transformation, weight updating and output control quantity calculation on system deviation through an improved single neuron PID algorithm; and a speed adjusting step: inputting the calculated control quantity into a driver to realize the position balance of the double motors. According to the invention, tracking errors and synchronization errors are cooperatively processed through a cross coupling structure, and an improved single-neuron PID algorithm is matched, so that accurate regulation and control of multi-dimensional errors are realized, the synchronization precision of the master servo motor and the slave servo motor is greatly improved, the inhibition capability of the system on external disturbance is enhanced, and thus the position control precision and the overall stability in the injection molding process are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of injection molding machines, and in particular to a dual-motor parallel injection control method for a full-electric injection molding machine based on cross-coupling. Background Art

[0002] In the dual-motor parallel injection system of an all-electric injection molding machine, synchronization control accuracy and anti-disturbance capability are key indicators affecting the quality of the molded product. Traditional control methods mainly include master-command control, master-slave control, and virtual spindle control.

[0003] First, master control means that each motor in the injection drive system receives the same speed reference signal. Therefore, this method depends on the parameters of the drive device and motor. If there are certain differences between the drive devices or motors, complete synchronization cannot be achieved.

[0004] Secondly, in master-slave control, one motor in the system is the master, and the second motor is the slave. The slave receives the speed feedback from the master as the speed reference. Since there is no signal coupling between the master and slave motors, that is, there is no feedback from the slave speed to the master, the system response speed is limited by the response speed of the master.

[0005] An existing Chinese patent, application number CN103624946B, entitled "A Method for Controlling Dual-Motor Parallel Injection in an Electric Injection Molding Machine," proposes a method for controlling dual-motor parallel injection in an electric injection molding machine through master-slave control. The method uses a controller to issue torque and speed commands to the master servo drive, which in turn issues the master drive's real-time output torque signal and speed command to the slave axis as target torque and speed command parameters, thereby achieving torque superposition. However, this method fails to account for cross-coupling between the master and slave axes, making it susceptible to load fluctuations and resulting in significant synchronization errors. Furthermore, the method imposes speed parallelism through the addition of an external synchronous belt, which can easily cause mechanical overload and lead to structural failure. Summary of the Invention

[0006] In view of the defects in the prior art, the present invention provides a dual-motor parallel injection control method for a full-electric injection molding machine based on cross-coupling.

[0007] The present invention is achieved through the following technical solutions:

[0008] A dual-motor parallel injection control method for a fully electric injection molding machine based on cross-coupling includes signal acquisition steps: acquiring the speeds of the master and slave servo motors, calculating tracking error and synchronization error, and establishing a cross-coupling relationship;

[0009] Error compensation steps: Improve the single neuron PID algorithm to transform the system deviation, update the weight and calculate the output control quantity;

[0010] Speed ​​adjustment steps: Input the control quantity calculated above into the driver to achieve dual motor position balance.

[0011] In a preferred embodiment of the present invention, the synchronization error is a difference between the rotational speeds of the master and slave servo motors and a reference rotational speed.

[0012] In a preferred embodiment of the present invention, the tracking error is the difference in rotational speed between the master and slave servo motors.

[0013] In a preferred embodiment of the present invention, y(k) is obtained by weighting the tracking error, synchronization error, and set value of the motor speed:

[0014] y(k)=e*K1+e Sync *K2

[0015] Compared with the given set value r(k), the system deviation at time K is calculated to be e(k)=r(k)-y(k).

[0016] In a preferred embodiment of the present invention, the neuron input dimension n is set to 3, and the three state quantities required for neuron learning control are obtained by state transformation of e(k), namely x1(k), x2(k) and x3(k), and their calculation formulas are:

[0017]

[0018] In a preferred embodiment of the present invention, x1(k) is the error at the current moment; x2(k) is the error at the previous moment; and x3(k) is the error at the previous moment.

[0019] In a preferred embodiment of the present invention, according to the supervised Hebb learning rule, the update rule of the three weights w1(k), w2(k) and w3(k) of the input signal of the neuron PID is:

[0020]

[0021] Among them, w1(k), w2(k) and w3(k) are the weights of the integral, proportional and differential channels respectively, η i , η p ,η D They are the learning rates of integration, proportion and differentiation respectively, which allow the three weights of neuron PID to be adaptively adjusted with the system error e(k), output control quantity u(k) and input states x1(k), x2(k) and x3(k), thus realizing the self-optimization of traditional PID parameters.

[0022] The full name of the supervised Hebb learning rule is "supervised Hebb learning rule". It combines the unsupervised Hebb learning rule (synchronous excitation of neurons enhances connections) with the supervised Delta learning rule (introducing the error between expected output and actual output to guide learning). It allows the update of neural network connection weights to be based on both the correlation of neuronal activity and the guidance of error feedback. It is often used in scenarios that require adaptive adjustment, such as dual-motor control of injection molding machines. The formula generally reflects that the connection weight update is related to the product of input, output, and error.

[0023] In a preferred embodiment of the present invention, an error threshold ΔX and an output limit u are introduced. min 、u max , the output control quantity u(k) is calculated as follows:

[0024]

[0025] u(k)=clip(u(k-1)+Δu(k),u min ,u max )

[0026] u(k) is the output control quantity at time k. The output control quantity u(k) is updated incrementally. The control output at time k is obtained by adding the increment Δu to the control quantity u(k-1) at the previous moment to avoid system oscillation caused by sudden changes in the control quantity. u(k-1) is the control output at time k-1.

[0027] K is the control gain (algorithm parameter that needs to be debugged); Δu(k) is the control increment at time k; ω i (k) is the weight of the i-th control parameter (adaptively updated with iteration); x i (k) is the i-th input signal (such as error, error change rate); ρ is the error threshold switch, which determines whether to trigger specific control logic.

[0028] In a preferred embodiment of the present invention, when the system deviation e(k) is less than or equal to ΔX, no compensation value is output.

[0029] In a preferred embodiment of the present invention, before the step begins, the initial weight coefficients w1(k), w2(k) and w3(k) need to be determined; the error limit value ΔX needs to be determined to prevent the system from overfitting; a saturation limit needs to be added to prevent the parameter setting value from exceeding the system limit, and u is set. min ,u max The value of u(k) exceeds the maximum value u min Or less than the minimum value u min , then output the corresponding u min ,u max .

[0030] The beneficial effects of the present invention are:

[0031] 1. This application uses a cross-coupling structure to collaboratively process tracking error and synchronization error, and uses an improved single-neuron PID algorithm to achieve precise control of multi-dimensional errors, greatly improving the synchronization accuracy of the master and slave servo motors and enhancing the system's ability to suppress external disturbances, thereby improving the position control accuracy and overall stability of the injection molding process;

[0032] 2. This application utilizes a cross-coupling algorithm to coordinate the operating status of the dual motors in real time and dynamically compensate for synchronization errors. At the same time, by enhancing the robustness of the system, it ensures that high-precision injection molding control can be maintained under complex working conditions, providing reliable technical support for high-quality injection molding production. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a control flow chart of the cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to the present invention;

[0034] Figure 2 This is a control principle diagram of the cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to the present invention;

[0035] Figure 3 is a synchronization error curve diagram in the present invention;

[0036] Figure 4 This is the synchronization error curve of the traditional method;

[0037] Figure 5 This is an implementation scheme of the cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine of the present invention. DETAILED DESCRIPTION

[0038] The following detailed description of preferred embodiments of the present invention is provided in conjunction with the accompanying drawings to facilitate understanding of the advantages and features of the present invention by those skilled in the art, thereby providing a clearer and more precise definition of the scope of protection of the present invention. Directional terms used in the present invention, such as "up," "down," "front," "back," "left," "right," "top," and "bottom," are intended solely to refer to the directions of the accompanying drawings. Therefore, the directional terms used are intended to illustrate and facilitate understanding of the present invention and are not intended to limit the present invention.

[0039] like Figure 1-5 A dual-motor parallel injection control method for a fully electric injection molding machine based on cross coupling is shown, including

[0040] Step 1: Collect the speed of the master and slave servo motors, calculate the tracking error and synchronization error, and establish a cross-coupling relationship;

[0041] Relying on the high-precision encoder (such as an incremental encoder with a resolution of ≥2500 lines) configured in the injection molding machine's servo system, the controller synchronously collects the real-time speed feedback values ​​of the master and slave servo motors with a sampling period of no less than 1ms. The encoder converts the mechanical speed into an electrical signal by detecting the motor rotor pole position or grating pulses, and transmits it to the controller via a differential bus, ensuring the real-time and accuracy of the speed data.

[0042] Define the system speed setpoint and calculate the tracking error of each motor separately. Specifically, based on the encoder feedback, subtract the speed of each motor from the set speed to obtain the tracking error. The tracking error reflects the degree of deviation of a single motor's speed from the process setpoint and is directly related to the speed matching of the injection molding process.

[0043] To achieve dynamic synchronization of dual motors, a synchronization error is introduced to describe the speed difference between the motors. The specific operation method is to subtract the speed of different motors according to the feedback value of the encoder to obtain the synchronization error and realize cross coupling.

[0044] Unlike traditional master-slave control that only focuses on one-way tracking of "motor-set value", the cross-coupling mechanism establishes a two-way association of "master motor-slave motor" through synchronization error. The controller collaboratively processes the tracking error and the synchronization error (weighted fusion), upgrading the dual-motor control from "independent following" to "interlocking collaboration", effectively suppressing synchronization deviations caused by uneven load and differences in motor parameters.

[0045] Step 2: The controller calculates synchronization error compensation and tracking error compensation;

[0046] y(k) is obtained by weighting the tracking error, synchronization error, and set value of the motor speed:

[0047] y(k)=e*K1+e sync *K2

[0048] Compared with the given system speed setting value r(k), the system deviation at time K is calculated as follows:

[0049] e(k)=r(k)-y(k)

[0050] Among them, y(k) is the weighted fusion value of tracking error and synchronization error, reflecting the comprehensive deviation of "single motor tracking accuracy + dual motor synchronization"; e is the tracking error; e Sync is the synchronization error; K1 and K2 are weighted coefficients; r(k) is the target speed instruction of the injection molding process planning, which is dynamically generated by the injection molding machine control system according to the current process stage (injection, holding pressure, cooling, etc.);

[0051] Assuming the neuron input dimension n = 3, the state transformation of e(k) is performed to obtain the three state quantities required for neuron learning control, namely x1(k), x2(k) and x3(k), and their calculation formulas are respectively as follows:

[0052]

[0053] Among them, x1(k) is the error at the current moment; x2(k) is the error at the previous moment; x3(k) is the error at the previous moment;

[0054] If n is less than 3 (e.g., n=2), only the current error and rate of change can be obtained, without a deep description of the error accumulation or trend. This makes it difficult to cope with complex disturbances such as melt viscosity changes and multi-stage working condition switching during the injection molding process, resulting in insufficient control accuracy and robustness.

[0055] If n>3 (e.g. n=4), more historical error states will be introduced, increasing the amount of calculation and the difficulty of weight training, resulting in delayed controller response and easily reducing system stability due to overfitting, which does not meet the real-time control requirements of injection molding machines. However, n=3 strikes a balance between "state representation completeness" and "algorithm complexity", which can fully capture the dynamic characteristics of the error and ensure the efficient operation of the control algorithm. Therefore, the input dimension is generally 3;

[0056] According to the supervised Hebb learning rule, the three connection weights required for neuron learning control are obtained;

[0057] The update rule of the three weights w1(k), w2(k) and w3(k) of the input signal of the neuron PID is formula (3):

[0058]

[0059] Among them, w1(k), w2(k) and w3(k) are the weights of the integral, proportional and differential channels respectively, η i , η p ,η D The learning rates of integral, proportional, and differential are used to adjust the three weights of the neuron PID adaptively with the system error e(k), output control quantity u(k), and input states x1(k), x2(k), and x3(k), thus realizing the self-optimization of traditional PID parameters.

[0060] The supervised Hebb learning rule, short for "supervised Hebb learning rule," combines the unsupervised Hebb learning rule (synchronous neuronal excitation reinforces connections) with the supervised Delta learning rule (introducing the error between expected and actual outputs to guide learning). This allows the neural network connection weight updates to be based on both the correlation of neuronal activity and the guidance of error feedback. It is often used in scenarios requiring adaptive adjustment, such as dual-motor control of injection molding machines. The formula generally states that the connection weight update is related to the product of input, output, and error.

[0061] Introducing error threshold ΔX and output limit u min 、u max , the output control quantity u(k) is calculated as formula (4):

[0062]

[0063] u(k) is the output control quantity at time k. The output control quantity u(k) is updated incrementally. The control output at time k is obtained by adding the increment Δu to the control quantity u(k-1) at the previous moment to avoid system oscillation caused by sudden changes in the control quantity. u(k-1) is the control output at time k-1.

[0064] The increment Δu is generated by error weighted fusion + control gain adjustment;

[0065] K is the control gain (algorithm parameter that needs to be debugged); Δu(k) is the control increment at time k; ω i (k) is the weight of the i-th control parameter (adaptively updated with iteration); x i (k) is the i-th input signal (such as error, error change rate); ρ is the error threshold switch, which determines whether to trigger specific control logic.

[0066] Formula (5):

[0067]

[0068] ρ is the error threshold switch, which is determined by the relationship between the system error e(i) and the threshold ΔX. ρ is used to switch the control logic, such as whether to enable error compensation;

[0069] e(i) is the system error, ΔX is the preset error threshold (generally 0.005 based on experience);

[0070] In order to protect the servo drive and motor hardware, an output limiter u is introduced. min 、u max ;

[0071] Formula (6):

[0072] u(k)=clip(u(k-1)+Δu(k),umin ,u max )

[0073] u min Generally 0, u max 270, to prevent the control amount from exceeding the limit;

[0074] The control quantity u(k) at the kth moment is calculated by adding the control quantity increment Δu(k) to the control quantity u(k-1) at the previous moment, and then limiting it to [u min ,u max ], to avoid exceeding the control quantity limit.

[0075] u(k) is the output control quantity at time k, u(k-1) is the control output at time k-1, Δu(k) is the control increment at time k, u min is the minimum limit of the control increment, u max To control the maximum limit of the increment, clip is a limiting function to ensure that the result is within the set range.

[0076] When the system deviation e(k) is less than or equal to ΔX, no compensation value is output.

[0077] Step 3: Use u(k) calculated in the second step as the motor speed adjustment control variable and input it to the driver to maintain the position balance between the dual motors.

[0078] Before the step begins, the initialization of key parameters and the configuration of constraints need to be completed to lay a solid foundation for the stable operation of the control algorithm; the initial weight coefficients w1(k), w2(k) and w3(k) need to be determined. w1(k), w2(k) and w3(k) are the "prior knowledge" of neuron learning control, corresponding to the error state quantities x1(k), x2(k) and x3(k) respectively; the initial coefficients K1 and K2 need to be determined, corresponding to the tracking error compensation coefficient and the synchronization error compensation coefficient respectively.

[0079] The error limit value ΔX needs to be determined to prevent the system from overfitting; saturation limit needs to be added to prevent the parameter setting value from exceeding the system limit. Set u min ,u max The value of u(k) exceeds the maximum value u min Or less than the minimum value u min , then output the corresponding u min ,u max .

[0080] According to experience, ΔX is generally set to 0.005, and u is set to min is 0, u max It is 270.

[0081] According to experience, the initial weights w1(k), w2(k), and w3(k) are generally set to 0.03, 0.05, and 0.001, respectively.

[0082] According to experience, the initial weights K1 and K2 are generally set to 0.1 and 0.05 respectively.

[0083] According to the logic of "initialization parameters → signal acquisition → error calculation → state transformation → weight update → compensation fusion → limit constraint → output control", dual-motor synchronous injection control is realized.

[0084] In this embodiment, refer to Figure 2 and Figure 5 The specific control principle includes a controller (here, a PLC), two drivers, and corresponding master and slave servo motors. The master and slave servo motors are equipped with encoders for real-time position feedback. The controller communicates with the master and slave servo motors via an EtherCAT bus. The two servo motors are connected to the injection mechanism via a transmission mechanism, and the injection mechanism is driven by the synchronous action of the two motors.

[0085] The specific steps include: the controller generates the speed values ​​of the master and slave servo motors according to the injection molding process. The next step is to collect the real-time speeds of the master and slave servo motors through the Ethercat bus, run the "single neuron PID + supervised Hebb learning" algorithm, calculate and output the control quantity u(k), and finally, dynamically adjust the control parameters based on the feedback information to ensure the synchronization of the injection mechanism movement.

[0086] It should be noted that the parts not covered by the present invention are the same as the existing technology or can be implemented by using the existing technology; the various drives in the present invention can be implemented by using corresponding power structures such as cylinders, oil cylinders, electric cylinders, motors, etc. in combination with connecting rods, guide rods, etc., and are not limited to the description in the specification and the structure in the drawings.

[0087] In the description of the embodiments of the present invention, unless otherwise expressly specified or limited, the terms "installed," "connected," "connected," "disposed," and "provided with" should be understood in a broad sense. For example, they may refer to fixed, detachable, or integral connections; mechanical or electrical connections; direct or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention in specific contexts.

[0088] The above-described embodiments merely illustrate several implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A dual-motor parallel injection control method for a fully electric injection molding machine based on cross-coupling, characterized in that: include: Signal acquisition steps: Collect the speed of the master and slave servo motors, calculate the tracking error and synchronization error, and establish a cross-coupling relationship; Error compensation steps: Improve the single neuron PID algorithm to transform the system deviation, update the weight and calculate the output control quantity; Speed ​​adjustment steps: Input the control quantity calculated above into the driver to achieve dual motor position balance.

2. The method for controlling dual-motor parallel injection of a fully electric injection molding machine based on cross-coupling according to claim 1, characterized in that: The synchronization error is the difference between the rotation speed of the master and slave servo motors and the reference rotation speed.

3. The cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to claim 2, characterized in that: The tracking error is the difference in speed between the master and slave servo motors.

4. The method for controlling dual-motor parallel injection of a fully electric injection molding machine based on cross-coupling according to claim 1, characterized in that: y(k) is obtained by weighting the tracking error, synchronization error, and set value of the motor speed: y(k)=e*K1+e Sync *K2 Compared with the given set value r(k), the system deviation at time K is calculated to be e(k)=r(k)-y(k).

5. The cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to claim 4, characterized in that: Set the neuron input dimension n = 3, and transform e(k) into a state to obtain the three state quantities required for neuron learning control, namely x1(k), x2(k) and x3(k), and their calculation formulas are:

6. The cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to claim 5, characterized in that: The x1(k) is the error at the current moment; x2(k) is the error at the previous moment; and x3(k) is the error at the previous moment.

7. The cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to claim 5, characterized in that: According to the supervised Hebb learning rule, the update rules of the three weights w1(k), w2(k) and w3(k) of the input signal of the neuron PID are: Among them, η i , η p ,η D They are the learning rates of integration, proportion, and differentiation respectively.

8. The cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to claim 7, characterized in that: Introducing error threshold ΔX and output limit u min 、u max , the output control quantity u(k) is calculated as follows: u(k)=clip(u(k-1)+Δu(k),u min ,you max )。 9. The cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to claim 8, characterized in that: When the system deviation e(k) is less than or equal to ΔX, no compensation value is output.

10. The cross-coupling-based dual-motor parallel injection control method for an all-electric injection molding machine according to claim 9, characterized in that: Before starting the step, the initial weight coefficients w1(k), w2(k) and w3(k) need to be determined; the error limit value ΔX needs to be determined to prevent the system from overfitting; saturation limit needs to be added to prevent the parameter setting value from exceeding the system limit, and u min ,u max The value of u(k) exceeds the maximum value u min Or less than the minimum value u min , then output the corresponding u min ,u max .

Images