Spring forming on-line control system and control method

Abstract

The present application relates to the technical field of precision spring manufacturing and automatic control, in particular to a spring forming online control system and control method; containing dynamics modeling, disturbance observation, impedance decoupling and trajectory correction module; the system calculates the disturbance torque by constructing the mechanical transmission dynamics reference model and collecting the servo parameters; the core is to decouple the material deformation impedance component representing the instantaneous rheological properties of the wire from the disturbance torque, and generate the position correction instruction based on the elastic-plastic compensation model to dynamically adjust the forming trajectory in the current cycle; the present application reuses the servo motor as a virtual probe, eliminates the hysteresis of visual detection, can effectively deal with the strength fluctuation of the wire, and significantly improves the product consistency.

Description

Technical Field

[0001] This invention relates to the field of precision spring manufacturing and automated control technology, specifically to an online control system and control method for spring forming. Background Technology

[0002] With the rapid development of modern precision manufacturing technology, spring forming processes have placed higher demands on product consistency and production efficiency. However, the random fluctuations in the physical properties of raw material wires pose significant challenges to the control of forming accuracy. Currently, traditional machine vision inspection systems or offline measurement methods are generally used to monitor spring dimensions. These methods are essentially post-processing inspections, exhibiting significant lag and blind spots, and are unable to capture and respond to changes in material properties at the moment the forming action occurs. Due to the difficulty in sensing fluctuations in wire tensile strength or hardness in real time, existing equipment cannot dynamically adjust the forming trajectory during processing to offset springback errors. This open-loop or delayed closed-loop control mode results in poor product dimensional stability, easily leading to batch defects due to differences in raw material batches, and failing to meet the quality control requirements of high-speed production scenarios.

[0003] Therefore, how to sense the rheological properties of materials in real time and perform dynamic geometric compensation during high-speed forming has become an urgent problem to be solved in this field. Summary of the Invention

[0004] To solve the above-mentioned technical problems, the present invention provides an online control system and control method for spring forming. Specifically, the technical solution of the present invention is as follows:

[0005] A spring forming online control method includes: constructing a mechanical transmission dynamics reference model, which describes the ideal dynamic behavior of the forming equipment under no-load conditions; real-time acquisition of servo motor operating parameters, including actual current value, actual position value, and actual speed value; calculating the disturbance torque using a disturbance observer, where the disturbance torque is the difference between the actual output torque converted from the actual current value and the theoretical reference torque output by the mechanical transmission dynamics reference model; decoupling the material deformation impedance component from the disturbance torque, where the material deformation impedance component characterizes the transient rheological properties of the wire during the forming process; acquiring a preset elastoplastic compensation model, and generating a position correction command in real time based on the material deformation impedance component and the elastoplastic compensation model; and superimposing the position correction command onto the target position command of the forming axis to dynamically adjust the forming trajectory within the current forming cycle.

[0006] Preferably, the method for constructing the mechanical transmission dynamics reference model includes: driving the molding equipment to perform no-load operation tests across the entire speed range to obtain no-load test data; extracting the friction characteristics and inertia characteristics of the mechanical transmission chain based on the no-load test data; using the friction characteristics and inertia characteristics to establish dynamic equations including Coulomb friction, viscous friction, and rotational inertia, and marking the dynamic equations as the mechanical transmission dynamics reference model; wherein, the calculation logic of the theoretical reference torque is as follows: inputting the actual velocity value and the actual acceleration value into the mechanical transmission dynamics reference model, and solving for the torque component used to overcome the resistance of the mechanical system itself.

[0007] Preferably, the method for calculating the disturbance torque using a disturbance observer includes: multiplying the actual current value by a preset motor torque constant to obtain the actual output torque; using the actual output torque in subsequent calculations; subtracting the theoretical reference torque from the actual output torque to obtain a preliminary disturbance signal; performing low-pass filtering on the preliminary disturbance signal to suppress high-frequency quantization noise, and marking the processed signal as the disturbance torque; wherein the bandwidth frequency of the disturbance observer is set to be higher than the wire forming characteristic frequency and lower than the current loop sampling frequency.

[0008] Preferably, the method for decoupling the material deformation impedance component from the disturbance torque includes: constructing a bandpass filter whose passband frequency range covers the characteristic frequency band of wire plastic deformation; inputting the disturbance torque into the bandpass filter to filter out low-frequency components caused by mechanical drift and high-frequency components caused by electromagnetic interference, thereby obtaining a cleaned torque signal; calculating the product of the cleaned torque signal and the actual velocity value, and integrating the product over time to obtain the instantaneous forming energy density; and labeling the instantaneous forming energy density as the material deformation impedance component; wherein the magnitude of the material deformation impedance component is positively correlated with the tensile strength of the wire.

[0009] Preferably, the method for generating position correction commands in real time based on material deformation impedance components and an elasto-plastic compensation model includes: defining a sensing axis and a cooperating axis, wherein the sensing axis is a servo axis that detects changes in the material deformation impedance components, and the cooperating axis is a servo axis that performs geometric correction; searching for the springback compensation amount corresponding to the material deformation impedance components in the elasto-plastic compensation model; converting the springback compensation amount into a pulse increment of the cooperating axis, and marking the pulse increment as a position correction command; wherein, when the sensing axis is a wire feed axis, the cooperating axis is configured as a variable diameter axis or a pitch axis; when the sensing axis is a variable diameter axis, the cooperating axis is configured as a pitch axis; the generation and execution of the position correction command are completed within the same interpolation cycle.

[0010] Preferably, the method for constructing the elastoplastic compensation model includes: acquiring multiple sets of sample wires with different physical properties; recording the material deformation resistance component and geometric dimensional deviation of each set of sample wires under standard forming instructions; using the material deformation resistance component corresponding to each set of sample wires as input features and the corresponding geometric dimensional deviation as output labels to construct a mapping database; using a nonlinear regression algorithm to fit the data relationship in the mapping database to generate a compensation function; and labeling the compensation function as an elastoplastic compensation model; wherein, the elastoplastic compensation model is used to describe the nonlinear mapping relationship between wire hardness fluctuation and the required geometric correction amount.

[0011] Preferably, the method of superimposing the position correction command onto the target position command of the forming axis includes: obtaining the original motion command of the current interpolation cycle; determining whether the material deformation resistance component exceeds a preset dead zone threshold; if the material deformation resistance component exceeds the dead zone threshold, directly adding the position correction command to the end position of the original motion command; if the material deformation resistance component does not exceed the dead zone threshold, keeping the original motion command unchanged; and using a feedforward control strategy to adjust the speed feedforward gain and torque feedforward gain of the forming axis according to the position correction command to eliminate the dynamic lag caused by the position correction.

[0012] Preferably, it also includes: real-time monitoring of the spectral characteristics of the disturbance torque; if the amplitude of a specific frequency band in the spectral characteristics shows a monotonically increasing trend over time, a tool wear warning signal is generated; if the mean value of the disturbance torque undergoes a step change, a raw material batch change prompt signal is generated; and the tool wear warning signal and the raw material batch change prompt signal are sent to the human-machine interface.

[0013] An online control system for spring forming includes: a model building module for constructing a mechanical transmission dynamics reference model and extracting the frictional and inertial characteristics of the mechanical transmission chain; a data acquisition module for real-time acquisition of servo motor operating parameters, including current, position, and speed information; a disturbance observation module for calculating disturbance torque using a disturbance observer and decoupling the material deformation resistance component from the disturbance torque; a compensation calculation module for obtaining a preset elasto-plastic compensation model and calculating position correction commands based on the material deformation resistance component; and a motion control module for superimposing the position correction commands onto the target position commands of the forming shaft to drive the servo motor to perform dynamic compensation.

[0014] Preferably, the data acquisition module is configured in the current loop control loop of the servo driver, and the sampling frequency is not less than 16kHz; the disturbance observation module and the compensation calculation module are embedded in the field programmable gate array chip to achieve microsecond-level data processing and command response.

[0015] Compared with the prior art, the present invention has the following beneficial effects:

[0016] 1. This invention, by constructing a mechanical transmission dynamics reference model and combining it with a disturbance observer, can eliminate the friction and inertial torque of the machine itself in real time, thereby accurately decoupling the material deformation resistance component characterizing the transient rheological properties of the wire. This method reuses the servo motor as a highly sensitive virtual probe, which can complete the perception and feedback at the moment the tool contacts the wire. Compared with traditional machine vision inspection, this technology eliminates the lag and blind spot problems caused by image processing, effectively copes with the random fluctuations in the tensile strength of the wire in high-speed production, and significantly improves the consistency of the product.

[0017] 2. This invention employs a signal processing method based on energy domain analysis. By using bandpass filtering and power integration, it obtains the instantaneous forming energy density and accurately quantifies the material hardness change. Combined with a preset elasto-plastic compensation model, the system can directly map the nonlinear material rheological properties into geometric springback compensation and dynamically generate position correction commands within the current interpolation cycle. This mechanism ensures that even if there are local hardness inconsistencies in different batches of wire rod or wire, the equipment can automatically adjust the forming trajectory, avoiding the generation of batch scrap.

[0018] 3. This invention proposes a control strategy that cross-couples the sensing axis and the cooperating axis. When the feed axis senses an impedance change, it can coordinate with the variable diameter axis or pitch axis for compensation, realizing three-dimensional adjustment of comprehensive parameters such as diameter, pitch, and free length. At the same time, in conjunction with a dynamic feedforward control strategy based on position correction commands, the system can synchronously adjust the speed feedforward gain and torque feedforward gain in real time according to the rate of change of the correction amount. This effectively eliminates the servo following error caused by sudden changes in position commands, ensuring trajectory smoothness and high dynamic tracking performance during high-speed correction.

[0019] 4. This invention is not only used for molding control, but also deeply explores the added value of disturbance torque signals; by monitoring the spectral characteristics and statistical mean of disturbance torque in real time, the system can identify the monotonically increasing trend of amplitude in specific frequency bands to warn of tool wear, and automatically identify batch changes of raw materials or welding points through step changes in the mean; this intelligent monitoring mechanism can identify production risks in advance, transforming traditional passive maintenance into proactive early warning based on status, further ensuring the continuity and safety of production. Attached Figure Description

[0020] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0021] Figure 1 This is a flowchart of the method of the present invention;

[0022] Figure 2 This is a structural diagram of the system of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0024] Example 1:

[0025] Please see Figure 1 A spring forming online control method includes: constructing a mechanical transmission dynamics reference model, which describes the ideal dynamic behavior of the forming equipment under no-load conditions; real-time acquisition of servo motor operating parameters, including actual current value, actual position value, and actual speed value; calculating the disturbance torque using a disturbance observer, where the disturbance torque is the difference between the actual output torque converted from the actual current value and the theoretical reference torque output by the mechanical transmission dynamics reference model; decoupling the material deformation impedance component from the disturbance torque, where the material deformation impedance component characterizes the transient rheological characteristics of the wire during the forming process; acquiring a preset elastoplastic compensation model, and generating a position correction command in real time based on the material deformation impedance component and the elastoplastic compensation model; and superimposing the position correction command onto the target position command of the forming axis to dynamically adjust the forming trajectory within the current forming cycle.

[0026] This embodiment details the core execution flow of the online control method for spring forming. This method aims to solve the technical problems in the prior art where the spring forming dimensions are unstable due to random fluctuations in the physical properties of the wire, and where traditional visual inspection has a lag. The system executes a model building step to construct a mechanical transmission dynamics reference model. This model is pre-stored in the non-volatile memory of the motion controller as a background noise benchmark for subsequent calculation of disturbance quantities. The system executes a data acquisition step to collect the operating parameters of the servo motor used to drive the forming axis in real time through the high-speed data bus of the servo driver. The actual current value is sourced from the servo driver's current loop sampling. Its physical meaning is a quantity characterizing the motor's output torque, and its unit is... ; The actual position value, sourced from motor encoder feedback, physically represents the angular displacement of the motor rotor, and is measured in units of... ; The actual speed value, derived from position differential or speed observer, physically represents the angular velocity of the motor rotor, and is measured in units of... The system calculates the disturbance torque using a disturbance observer. The core logic of this step is to use difference calculations to eliminate the torque required by the machine's own friction and inertia, thereby exposing changes in the external load. Based on this, the system performs a decoupling step to decouple the material deformation resistance component from the disturbance torque. Since the disturbance torque may contain high-frequency vibration noise or low-frequency thermal drift from the mechanical transmission chain, this step extracts pure material characteristics through frequency domain analysis. Then, the system obtains a preset elastoplastic compensation model and, based on the decoupled material deformation resistance component, queries the corresponding springback compensation amount in the model to generate a position correction command. The system performs a dynamic adjustment step, superimposing the position correction command onto the target position command of the forming shaft, and drives the motor to perform the action.

[0027] In this embodiment, under the scenario of high-speed cold heading of springs, the servo motor is reused as a highly sensitive virtual rheological probe, realizing millisecond-level mechanical perception and geometric correction. This solution completes closed-loop compensation at the moment the tool contacts the wire, effectively eliminating the problem of poor product consistency caused by fluctuations in the tensile strength of the steel wire, avoiding the lag and blind spots of traditional CCD visual inspection, and significantly improving the robustness of the equipment to batch differences of raw materials.

[0028] Example 2:

[0029] The method for constructing a mechanical transmission dynamics reference model includes: driving the molding equipment to conduct no-load operation tests across the entire speed range to obtain no-load test data; extracting the friction and inertia characteristics of the mechanical transmission chain based on the no-load test data; using the friction and inertia characteristics, establishing dynamic equations that include Coulomb friction, viscous friction, and rotational inertia, and labeling the dynamic equations as the mechanical transmission dynamics reference model; wherein, the calculation logic of the theoretical reference torque is as follows: inputting the actual velocity value and the actual acceleration value into the mechanical transmission dynamics reference model, and solving for the torque component used to overcome the resistance of the mechanical system itself.

[0030] This embodiment further defines the construction details of the mechanical transmission dynamics reference model, employs the least squares method for parameter identification, and ensures the solvability of the model parameters; the system performs an unloaded calibration step, driving the forming equipment to conduct an unloaded operation test across the entire speed range with no wire installed on the equipment; the sampling period is set to... ,Record Groups of data samples, each group containing ,in, The system constructs a system of linear regression equations based on no-load test data. Wherein, observation vector Regression matrix for A matrix, whose structure is represented as:

[0031] ;

[0032] in, This is a sign function used to characterize the directional characteristics of frictional force;

[0033] Parameter vector to be identified Using the least squares formula The physical parameters of the mechanical transmission chain were calculated: Equivalent moment of inertia of the system ; : Coefficient of viscous friction ; Coulomb friction torque Using the above characteristics, a dynamic equation is established, and the theoretical reference torque is calculated. :

[0034] ;

[0035] in, This equation is labeled as the reference model for mechanical transmission dynamics.

[0036] Example 3:

[0037] The method for calculating disturbance torque using a disturbance observer includes: multiplying the actual current value by a preset motor torque constant to obtain the actual output torque; using the actual output torque in subsequent calculations; subtracting the theoretical reference torque from the actual output torque to obtain the preliminary disturbance signal; performing low-pass filtering on the preliminary disturbance signal to suppress high-frequency quantization noise, and marking the processed signal as the disturbance torque; wherein, the bandwidth frequency of the disturbance observer is set to be higher than the wire forming characteristic frequency and lower than the current loop sampling frequency.

[0038] This embodiment details the signal processing flow for calculating disturbance torque using a disturbance observer, focusing on the discretization of the filter; the system calculates the total electromagnetic torque, incorporating the actual current value... Axial components Multiply by the preset motor torque constant : The unit is The system acquires the initial disturbance signal and performs a subtraction operation: The signal obtained at this point contains both the actual disturbance and sampling noise. The system performs low-pass filtering on the initial disturbance signal, using a first-order Butterworth low-pass filter, whose discretization recursive formula is as follows:

[0039] ;

[0040] in, The sampling period is consistent with the sampling period defined in Example 2. The cutoff angular frequency is set; the bandwidth frequency is constrained as follows: ,in for Typical value is set to This is to suppress high-frequency quantization noise in the current loop while preserving the dynamic torque of material deformation; the above coefficient values ​​are examples pre-calculated based on specific sampling frequencies and cutoff frequencies; in practical applications, those skilled in the art should recalculate the corresponding filter discretization coefficients using the bilinear transform method according to the changes in the characteristic frequency of wire forming, in order to ensure accurate extraction of the material rheological signal.

[0041] Example 4:

[0042] The method for decoupling the material deformation impedance component from the disturbance torque includes: constructing a bandpass filter whose passband frequency range covers the characteristic frequency band of wire plastic deformation; inputting the disturbance torque into the bandpass filter to filter out low-frequency components caused by mechanical drift and high-frequency components caused by electromagnetic interference, thereby obtaining the cleaned torque signal; calculating the product of the cleaned torque signal and the actual velocity value, and integrating the product over time to obtain the instantaneous forming energy density; and labeling the instantaneous forming energy density as the material deformation impedance component; wherein the magnitude of the material deformation impedance component is positively correlated with the tensile strength of the wire.

[0043] This embodiment details an energy domain analysis method for decoupling the material deformation impedance component from the disturbance torque; the system constructs a bandpass filter, employing a second-order IIR filter structure, with the passband frequency range set to [value missing]. The bandpass filter covers the frequency band characteristic of wire plastic deformation; it adopts a second-order infinite impulse response (IIR) structure, and its discretized difference equation is defined as:

[0044] ;

[0045] in, The input disturbance torque, The output is the cleaned torque signal; the filter coefficients are calculated based on the bilinear transform method. In this embodiment, the sampling frequency is set. passband cutoff frequency The specific coefficient values ​​obtained from the calculation are set as follows:

[0046] ; to disturb torque The input to this filter is the cleaned torque signal. This step removes the DC component (0Hz) caused by mechanical drift and residual high-frequency electromagnetic interference; the system calculates the instantaneous forming energy density, i.e., the discrete integral of the instantaneous power within the forming action window:

[0047] ;

[0048] in, The system sampling period is and These indices correspond to the start and end times of a single forming action, such as the moment the cutter contacts and leaves the machine. The indices are automatically triggered by detecting the slope of changes in the feed axis's enable signal or position command. The integral result is then... This is denoted as the material strain resistance component; physically, its magnitude is primarily determined by the tensile strength of the wire. and wire diameter Decision, fulfilling the relationship The proportionality coefficient is defined as follows: Then there is It should be noted that the coefficients in the above proportional relationship can be obtained through pre-experimental calibration for wires of different specifications. In the actual online control process, the system mainly uses the relative change of instantaneous forming energy density relative to the standard sample as the control variable to offset the absolute value deviation caused by ambient temperature or sensor drift, and to ensure the robustness of the compensation logic.

[0049] Example 5:

[0050] The method for generating position correction commands in real time based on the material deformation impedance component and the elastoplastic compensation model includes: defining a sensing axis and a cooperating axis, where the sensing axis is a servo axis that detects changes in the material deformation impedance component, and the cooperating axis is a servo axis that performs geometric correction; finding the springback compensation amount corresponding to the material deformation impedance component in the elastoplastic compensation model; converting the springback compensation amount into a pulse increment of the cooperating axis, and marking the pulse increment as a position correction command; wherein, when the sensing axis is a wire feed axis, the cooperating axis is configured as a variable diameter axis or a pitch axis; when the sensing axis is a variable diameter axis, the cooperating axis is configured as a pitch axis; the generation and execution of the position correction command are completed within the same interpolation cycle.

[0051] This embodiment describes a cross-coupling control strategy based on generating position correction commands from material deformation impedance components. The system defines axis role logic: the sensing axis directly drives wire deformation and detects impedance changes, while the cooperating axis performs geometric corrections to offset errors. The system searches for the springback compensation amount corresponding to the current material deformation impedance component in the elastoplastic compensation model. In response to an increase in the perceived wire impedance component, the springback amount increases accordingly. The system converts this springback compensation amount into a pulse increment for the cooperating axis and marks it as a position correction command. During this process, the system executes a specific axis mapping strategy: in response to the sensing axis being the wire feeding axis (i.e., sensing an increase in the overall wire tensile strength), the system configures the variable diameter axis as the cooperating axis, outputting a command to reduce the winding diameter to compensate for the increased springback, while simultaneously configuring the pitch axis as the cooperating axis. To prevent frequent switching near the zero point, the system presets a switching dead zone threshold. Only when Only when the sensing axis is the wire feeding axis will the subsequent collaborative axis switching determination be performed; based on this, when the sensing axis is the wire feeding axis, the system will determine the switching based on the material deformation impedance component. Choosing the coordinating axis based on the changing trend: If If the localized stiffness of the wire increases, a variable diameter shaft is configured as a cooperating shaft, and a correction strategy of reducing the outer diameter of the spring is implemented to counteract the springback; if If the local hardness of the wire decreases, then the pitch axis is configured as the cooperating axis, and a correction strategy of increasing the pitch is implemented.

[0052] In response to the sensing axis being a variable diameter axis, the system configures the pitch axis as a cooperative axis; ensuring that the generation and execution of position correction commands are completed within the same interpolation cycle;

[0053] This embodiment breaks through the limitations of traditional single-axis control by using a cross-coupling mechanism of A-axis sensing and B-axis correction. In a multi-axis linkage spring machine system, this strategy achieves three-dimensional dynamic compensation, ensuring that when the material hardness fluctuates, it can not only maintain the dimension in a single dimension, but also coordinate to ensure the consistency of comprehensive geometric parameters such as spring diameter, pitch and free length.

[0054] Example 6:

[0055] The method for constructing the elastoplastic compensation model includes: acquiring multiple sets of sample wires with different physical properties; recording the material deformation resistance component and geometric dimensional deviation of each set of sample wires under standard forming instructions; using the material deformation resistance component corresponding to each set of sample wires as input features and the corresponding geometric dimensional deviation as output labels to construct a mapping database; using a nonlinear regression algorithm to fit the data relationship in the mapping database to generate a compensation function; and labeling the compensation function as an elastoplastic compensation model; wherein, the elastoplastic compensation model is used to describe the nonlinear mapping relationship between wire hardness fluctuation and the required geometric correction amount.

[0056] This embodiment discloses the offline construction and training process of the elastoplastic compensation model, and clarifies the specific regression model structure; the system performs the sample collection step to obtain... The sample wires were grouped together with different physical properties, i.e., different tensile strengths; key data for each group of sample wires were recorded when executing standard forming instructions. ,in, ; Input characteristics, namely the material deformation resistance components calculated above. ; Output label: This refers to the geometric dimensional deviations after molding, measured by precision measuring instruments, such as springback angle difference. A mapping database was constructed, and a nonlinear regression algorithm was used, specifically a cubic polynomial fitting. By minimizing the loss function Solving for the coefficient vector The polynomial function and its coefficients are labeled as an elasto-plastic compensation model; during system operation, the calculated values ​​will be displayed in real time. Substitute the polynomial into the formula to directly calculate the required geometric correction.

[0057] Example 7:

[0058] The method of superimposing the position correction command onto the target position command of the forming axis includes: obtaining the original motion command of the current interpolation cycle; determining whether the material deformation resistance component exceeds a preset dead zone threshold; if the material deformation resistance component exceeds the dead zone threshold, directly adding the position correction command to the end position of the original motion command; if the material deformation resistance component does not exceed the dead zone threshold, keeping the original motion command unchanged; and using a feedforward control strategy to adjust the speed feedforward gain and torque feedforward gain of the forming axis according to the position correction command to eliminate the dynamic lag caused by the position correction.

[0059] This embodiment details the method for superimposing position correction commands and the dynamic feedforward control strategy; the system acquires the original motion command of the current interpolation cycle. ; Determine whether the material deformation resistance component exceeds the preset dead zone threshold; In this step, the system presets the reference impedance value under standard forming conditions. and allowable fluctuation deviation The dead zone threshold is the value of the reference impedance. The dead zone threshold is usually set to a range of 1000 to 10000. The logic behind this range setting is to filter out random noise caused by power grid voltage microwave fluctuations or minor vibrations of the mechanical transmission chain, prevent the actuator from making unnecessary frequent corrections, and extend the service life of the equipment.

[0060] Calculate the current material deformation resistance components With reference impedance value The absolute value of the difference; in response to the absolute value of the difference exceeding the dead zone threshold. The system determines that the material deformation resistance component exceeds the dead zone threshold and directly adds the position correction command to the end position of the original motion command; in response to the absolute value of the difference not exceeding the dead zone threshold... If the system determines that the movement is within the acceptable range, it maintains the original motion command to prevent the system from becoming overly sensitive to minor fluctuations within the normal range. The system utilizes a feedforward control strategy to synchronously adjust the speed feedforward gain of the forming axis based on the rate of change of the position correction command. and torque feedforward gain The adjustment algorithm is as follows: calculate the first-order difference of the position correction command to obtain the correction speed. The gain is updated in real time according to the following formula:

[0061] ;

[0062] in, and The initial reference gain set for the system. and The preset dynamic response coefficients typically take the following values: and ; The first-order difference represents the position correction command, and its physical meaning is the superposition velocity of the compensating motion; and It has the reciprocal dimension of velocity, and its unit is: or To ensure the product term in the gain adjustment formula The value is dimensionless.

[0063] This embodiment solves the dynamic lag problem caused by sudden changes in position commands by introducing a dynamic feedforward mechanism. In high-speed production, this strategy ensures that the servo motor can not only accurately reach the corrected position, but also respond to the correction command quickly and without overshoot, thus guaranteeing the smoothness of the forming trajectory and high dynamic tracking performance.

[0064] Example 8:

[0065] It also includes: real-time monitoring of the spectral characteristics of disturbance torque; if the amplitude of a specific frequency band in the spectral characteristics shows a monotonically increasing trend over time, a tool wear warning signal is generated; if the mean value of the disturbance torque undergoes a step change, a raw material batch change prompt signal is generated; and the tool wear warning signal and the raw material batch change prompt signal are sent to the human-machine interface.

[0066] This embodiment utilizes disturbance torque to achieve intelligent monitoring of equipment health status; the system performs real-time Fast Fourier Transform on the disturbance torque to monitor its spectral characteristics; the system performs trend analysis; responding to the monotonically increasing trend of amplitude in a specific frequency band over time in the spectral characteristics, the system adjusts the current production speed accordingly. Unit: Quantities / minute, Calculating the fundamental frequency and define a specific frequency band as The third harmonic interval, among which, If the spectral amplitude within this specific frequency band It exhibits a statistically significant upward trend, namely, the slope of its linear fit with respect to the time window. satisfy The system determines that the tool is becoming dull and friction is increasing, generating a tool wear warning signal; simultaneously, the system monitors time-domain statistical characteristics; it responds to a step change in the statistical mean of the disturbance torque within a short period of time, for example, the deviation of the mean from the current steady-state disturbance torque mean jumps instantaneously by more than [a certain amount]. The system determines that there are solder joints in the current coil or that the coil has been replaced with a different batch of wire, and generates a raw material batch change prompt signal; the above signal is sent to the human-machine interface for display and alarm.

[0067] This embodiment adds value to the control system beyond molding, enabling predictive maintenance based on process data. By deeply mining the characteristic information in the torque signal, this solution can identify tool failure risks and material mutations in advance, effectively avoiding batch scrap production caused by equipment deterioration or raw material abnormalities.

[0068] Example 9:

[0069] Please see Figure 2 An online control system for spring forming includes: a model building module for constructing a mechanical transmission dynamics reference model and extracting the friction and inertia characteristics of the mechanical transmission chain; a data acquisition module for real-time acquisition of servo motor operating parameters, including current, position, and speed information; a disturbance observation module for calculating disturbance torque using a disturbance observer and decoupling the material deformation resistance component from the disturbance torque; a compensation calculation module for obtaining a preset elasto-plastic compensation model and calculating position correction commands based on the material deformation resistance component; and a motion control module for superimposing the position correction commands onto the target position commands of the forming shaft to drive the servo motor to perform dynamic compensation.

[0070] The data acquisition module is configured in the current loop control loop of the servo driver, with a sampling frequency of not less than 16kHz; the disturbance observation module and the compensation calculation module are embedded in the field programmable gate array chip to achieve microsecond-level data processing and command response.

[0071] This embodiment describes the hardware architecture and module configuration of the online spring forming control system for implementing the above method. At the hardware level, the system includes a model building module, a data acquisition module, a disturbance observation module, a compensation calculation module, and a motion control module. The data acquisition module is configured in the current loop control circuit of the servo driver, and its sampling frequency is set to be no less than [a certain value]. That is, the period is less than ; The sampling frequency, derived from the hardware clock configuration, physically represents the data update rate, measured in Hz. The disturbance observation module and compensation calculation module do not run on the general-purpose CPU layer but are embedded within the system's field-programmable gate array (FPGA) chip. Utilizing the parallel processing capabilities of the FPGA, the aforementioned complex floating-point operations, including disturbance observer calculation, filtering, integration, and interpolation, are all completed within microseconds. The motion control module is responsible for instruction synthesis and pulse output, driving the servo motor to perform dynamic compensation.

[0072] This embodiment eliminates the uncertainty and delay caused by traditional bus communication and CPU interrupts through the deep integration of FPGA hardware acceleration and the bottom-level sampling of the current loop. The architecture ensures that the entire control loop has extremely high real-time performance, meets the stringent control requirements of high-speed spring machines, such as 600 cycles per minute, and realizes the industrial application of computing power sinking and edge intelligence.

[0073] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for online control of spring forming, characterized in that, include: A mechanical transmission dynamics reference model is constructed to describe the ideal dynamic behavior of the molding equipment under no-load conditions. The system collects servo motor operating parameters in real time, including actual current, position, and speed values. It calculates disturbance torque using a disturbance observer; the disturbance torque is the difference between the actual output torque converted from the actual current value and the theoretical reference torque output by the mechanical transmission dynamics reference model. It decouples the material deformation impedance component from the disturbance torque; this component characterizes the transient rheological properties of the wire during the forming process. It acquires a preset elasto-plastic compensation model and generates position correction commands in real time based on the material deformation impedance component and the elasto-plastic compensation model. The position correction commands are then superimposed onto the target position command of the forming axis to dynamically adjust the forming trajectory within the current forming cycle. The method for decoupling the material deformation impedance component from the disturbance torque includes: constructing a bandpass filter whose passband frequency range covers the characteristic frequency band of wire plastic deformation; inputting the disturbance torque into the bandpass filter to filter out low-frequency components caused by mechanical drift and high-frequency components caused by electromagnetic interference, thereby obtaining the cleaned torque signal; calculating the product of the cleaned torque signal and the actual velocity value, and integrating the product over time to obtain the instantaneous forming energy density; and labeling the instantaneous forming energy density as the material deformation impedance component; wherein the magnitude of the material deformation impedance component is positively correlated with the tensile strength of the wire. The method for superimposing the position correction command onto the target position command of the forming axis includes: obtaining the original motion command of the current interpolation cycle; determining whether the material deformation impedance component exceeds a preset dead zone threshold; if the material deformation impedance component exceeds the dead zone threshold, directly adding the position correction command to the end position of the original motion command; if the material deformation impedance component does not exceed the dead zone threshold, keeping the original motion command unchanged; and using a feedforward control strategy to adjust the speed feedforward gain and torque feedforward gain of the forming axis according to the position correction command to eliminate the dynamic lag caused by the position correction.

2. The online control method for spring forming according to claim 1, characterized in that, The method for constructing a mechanical transmission dynamics reference model includes: driving the molding equipment to perform no-load operation tests across the entire speed range to obtain no-load test data; extracting the friction and inertia characteristics of the mechanical transmission chain based on the no-load test data; using the friction and inertia characteristics, establishing dynamic equations that include Coulomb friction, viscous friction, and rotational inertia, and labeling the dynamic equations as the mechanical transmission dynamics reference model; wherein, the calculation logic of the theoretical reference torque is as follows: inputting the actual velocity value and the actual acceleration value into the mechanical transmission dynamics reference model, and solving for the torque component used to overcome the resistance of the mechanical system itself.

3. The online control method for spring forming according to claim 1, characterized in that, The method for calculating the disturbance torque using a disturbance observer includes: multiplying the actual current value by a preset motor torque constant to obtain the actual output torque; using the actual output torque in subsequent calculations; subtracting the theoretical reference torque from the actual output torque to obtain the preliminary disturbance signal; performing low-pass filtering on the preliminary disturbance signal to suppress high-frequency quantization noise, and marking the processed signal as the disturbance torque; wherein the bandwidth frequency of the disturbance observer is set to be higher than the wire forming characteristic frequency and lower than the current loop sampling frequency.

4. The online control method for spring forming according to claim 1, characterized in that, The method for generating position correction commands in real time based on material deformation impedance components and an elastoplastic compensation model includes: defining a sensing axis and a cooperating axis, wherein the sensing axis is a servo axis that detects changes in the material deformation impedance components, and the cooperating axis is a servo axis that performs geometric correction; searching for the springback compensation amount corresponding to the material deformation impedance components in the elastoplastic compensation model; converting the springback compensation amount into a pulse increment of the cooperating axis, and marking the pulse increment as a position correction command; wherein, when the sensing axis is a wire feed axis, the cooperating axis is configured as a variable diameter axis or a pitch axis; when the sensing axis is a variable diameter axis, the cooperating axis is configured as a pitch axis; the generation and execution of the position correction command are completed within the same interpolation cycle.

5. The online control method for spring forming according to claim 4, characterized in that, The method for constructing the elastoplastic compensation model includes: acquiring multiple sets of sample wires with different physical properties; recording the material deformation resistance component and geometric dimensional deviation of each set of sample wires under standard forming instructions; using the material deformation resistance component corresponding to each set of sample wires as input features and the corresponding geometric dimensional deviation as output labels to construct a mapping database; using a nonlinear regression algorithm to fit the data relationship in the mapping database to generate a compensation function; and labeling the compensation function as an elastoplastic compensation model; wherein, the elastoplastic compensation model is used to describe the nonlinear mapping relationship between wire hardness fluctuation and the required geometric correction amount.

6. The online control method for spring forming according to claim 1, characterized in that, Also includes: Real-time monitoring of the spectral characteristics of disturbance torque; If the amplitude of a specific frequency band in the spectral characteristics shows a monotonically increasing trend over time, a tool wear warning signal is generated. If the mean value of the disturbance torque changes abruptly, a raw material batch change prompt signal is generated; the tool wear warning signal and the raw material batch change prompt signal are sent to the human-machine interface.

7. A spring forming online control system for implementing the spring forming online control method as described in any one of claims 1-6, characterized in that, include: The model building module is used to build a reference model of mechanical transmission dynamics and extract the friction and inertia characteristics of the mechanical transmission chain; The data acquisition module is used to collect the operating parameters of the servo motor in real time, including current, position and speed information; The disturbance observation module is used to calculate the disturbance torque using a disturbance observer and decouple the material deformation resistance component from the disturbance torque; The compensation calculation module is used to obtain the preset elastic-plastic compensation model and calculate the position correction command based on the material deformation resistance component; the motion control module is used to superimpose the position correction command onto the target position command of the forming axis and drive the servo motor to perform dynamic compensation.

8. The online spring forming control system according to claim 7, characterized in that, The data acquisition module is configured in the current loop control circuit of the servo driver, with a sampling frequency of not less than 16kHz; the disturbance observation module and the compensation calculation module are embedded in the field programmable gate array (FPGA) chip to achieve microsecond-level data processing and command response.

Images