Method for improving production efficiency of high-efficiency spring steel wire

Abstract

The invention provides a method for improving the production efficiency of a high-efficiency spring steel wire, and relates to the technical field of spring steel wire production.The method comprises the following steps that multi-source physical quantities in the spring steel wire winding process are collected in real time, and the multi-source physical quantities comprise steel wire tension distribution, the bending rod curvature radius, a roller temperature field and an equipment vibration frequency spectrum; based on the obtained multi-source physical quantity, a torque instruction, a rotating speed instruction and a bending radius compensation amount are generated through a dynamic collaborative mapping function; according to the generated instruction, a servo motor set is driven to adjust the wire feeding speed and the angular speed of a roller, and meanwhile a piezoelectric trimmer is controlled to correct the bending radius; and adopting a reinforcement learning strategy to optimize control parameters on line, and dynamically updating a mapping function according to winding quality feedback. According to the method, the problem that the remodeling debugging time is long can be effectively solved, the wire breakage rate can be remarkably reduced, and the size consistency can be effectively improved.

Description

Technical Field

[0001] The invention relates to the technical field of spring steel wire production, and in particular to a method for improving the production efficiency of high-performance spring steel wire. Background Art

[0002] Currently, the high-efficiency production of spring steel wire faces three major technical bottlenecks in the current spring wire winding process: long changeover and commissioning times (traditional production lines require an average of 2 to 3 hours), high wire breakage rates (especially for small-diameter wires), and poor dimensional consistency (spiral angle fluctuations exceeding ±3°). The root cause lies in the conflict between rigidity parameter settings and dynamic disturbances: traditional PID control cannot adapt to time-varying disturbances such as material property fluctuations (e.g., ±15% deviation in hardness), equipment thermal deformation (roller displacement caused by a temperature rise of 0.1°C / min), and mechanical vibration (30Hz-50Hz resonance peaks).

[0003] The solutions in the existing related technologies have obvious limitations. For example, the limiting cylinder provided in the Chinese patent document with publication number CN119016645A can improve the molding consistency, but it lacks dynamic adjustment capabilities; the lateral traction mechanism provided in the Chinese patent document with publication number CN119346764A relies on mechanical compensation and has a slow response speed, while the support components in the traditional hot winding system (for example, the Chinese patent document with publication number CN116550905A) only solve the static stability problem.

[0004] Therefore, how to achieve real-time collaborative optimization of winding parameters has become an urgent problem to be solved. Summary of the Invention

[0005] In order to solve at least part of the technical problems in the related art, the present invention provides a method for improving the production efficiency of high-performance spring steel wire.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is: A method for improving the production efficiency of high-performance spring steel wire comprises the following steps: Step S1: Real-time acquisition of multi-source physical quantities during the spring wire winding process, including wire tension distribution, bending rod curvature radius, roller temperature field, and equipment vibration spectrum; Step S2: Based on the acquired multi-source physical quantities, a torque command, a speed command and a bending radius compensation are generated through a dynamic collaborative mapping function; Step S3: According to the instructions generated in step S2, the servo motor group is driven to adjust the wire feeding speed and the roller angular velocity, and the piezoelectric trimmer is controlled to correct the bending radius; Step S4: Use reinforcement learning strategy to optimize control parameters online, and dynamically update the mapping function of step S2 according to the winding quality feedback.

[0007] Optionally, in step S2, the construction of the dynamic collaborative mapping function includes: Establish a cross-domain coupling model of wire deformation, equipment dynamics, and thermodynamics: Where, is the plastic deformation energy density of the steel wire, in units of , is the time variable, is the thermal conductivity coefficient, in units of , is the Laplace operator, and its unit is , is the deformation conversion coefficient, is the von Mises equivalent stress, in MPa, is the logarithmic strain rate, in units of , is the heat exchange coefficient, is the ambient temperature in °C, The real-time temperature of the roller in °C.

[0008] Optionally, the specific implementation of the dynamic collaborative mapping function includes: The torque gain coefficient and bending radius compensation coefficient are calculated based on Lyapunov stability theory: Where, is the torque gain coefficient, is the time-varying gain coefficient, in units of , is the attenuation coefficient, in units of , is the yield strength of steel wire, in MPa, is the bending radius compensation coefficient, is the space compensation coefficient, the unit is mm, is the thermal attenuation coefficient, in units of , is the absolute value of temperature deviation, that is , unit is ℃, is the second norm of the deformation energy gradient, is the characteristic stress, in MPa.

[0009] Optionally, when a sudden change in material hardness is detected, The torque gain The adjustment rate is controlled at within the scope; When the temperature deviation When the thermal expansion compensation mechanism is activated, the bending radius compensation coefficient Decays exponentially, and the decay rate satisfies .

[0010] Optionally, during the execution of step S3, vibration suppression is performed simultaneously: Use phase-compensation controller to generate vibration compensation signal: Where, is the vibration compensation signal, i.e. acceleration command, in units of , is the acceleration gain, in units of , is the total number of vibration modes, is the index value of the vibration mode, For the The acceleration amplitude of the order vibration mode, in units of , is the reference acceleration, in units of , is the imaginary unit, For the The angular frequency of the order vibration mode, in rad / s, For the The lag phase angle of the order vibration mode, in rad, is the cutoff frequency in Hz.

[0011] Optionally, in step S4, the reinforcement learning strategy specifically includes: Designing differentiated reward functions to drive multi-agent collaborative optimization: Where, For the moment The reward value, 、 and are weight coefficients, is the curvature radius deviation, in mm, is the reference length of the curvature radius, in mm, is the standard deviation of tension, in N, is the reference tension value, in N. is the rate of change of deformation energy, in units of , is the reference power density, in units of .

[0012] Optionally, in step S1, the vibration spectrum is collected by: Arrange the MEMS acceleration sensor array (no less than 8 measuring points) around the axial direction of the bending rod. Analyze the 30Hz-50Hz resonance peak using the improved Fast Fourier Transform algorithm: Where, For the frequency range The first The acceleration amplitude of the resonance peak, in units of , is the frequency, is the fast Fourier transform, is the time domain vibration signal, is the time variable, is the time window width of the Gaussian window.

[0013] Beneficial effects: 1. Through the above technical solution, the present invention can effectively solve the problem of long debugging time during model changeovers. Specifically, traditional production line model changes require manual parameter adjustment (taking 2-3 hours). However, the method of the present invention achieves automatic adaptation of process parameters through real-time multi-source physical quantity acquisition and closed-loop control of dynamic collaborative mapping functions. Through real-time data-driven decision-making (by continuously monitoring steel wire tension, curvature radius, temperature field and vibration spectrum, the system automatically generates torque, speed and bending compensation instructions without manual trial and error adjustment) and an online parameter optimization mechanism (a reinforcement learning strategy continuously updates the mapping function based on winding quality feedback, allowing the system to autonomously adapt to the production requirements of steel wires of different specifications, significantly shortening model changeover debugging time), it can effectively shorten model changeover debugging time.

[0014] Second, the present invention can significantly reduce the wire breakage rate. Specifically, to address the high wire breakage rate of small-diameter steel wire, the present invention establishes a dynamic disturbance active suppression mechanism through multi-dimensional state perception and real-time collaborative control. Multi-dimensional state perception involves synchronously collecting key physical quantities such as tension distribution (to prevent local overload) and vibration spectrum (to identify mechanical resonance risks) to detect wire breakage causes in advance. Real-time collaborative control involves dynamically adjusting the wire feed speed and roller angular velocity through a servo motor group, and combining it with a piezoelectric fine-tuner to instantly correct the bending radius, avoiding wire breakage caused by stress concentration or sudden vibration changes.

[0015] Third, the present invention can effectively improve dimensional consistency. Specifically, to address the issue of helix angle fluctuation exceeding ±3°, the present invention constructs a closed-loop precision control system through cross-parameter collaborative optimization (using a dynamic collaborative mapping function to synchronously process multiple sources of information, such as tension, curvature, temperature, and vibration, to generate mutually coupled compensation instructions, such as bend radius compensation, to ensure dynamic matching of various parameters) and continuous self-learning capabilities (using a reinforcement learning strategy to iteratively optimize the mapping function based on actual winding quality, enabling the system to maintain high precision despite interference such as material fluctuations and thermal deformation, for example, improving helix angle stability). This effectively improves dimensional consistency.

[0016] 2. Other beneficial effects or advantages of the present invention will be described in detail in the specific implementation manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative labor.

[0018] in: Figure 1 The figure is a schematic flow chart of the steps of a method for improving the production efficiency of high-performance spring steel wire provided by an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0020] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0021] In addition, the terms "including" and "having" and any variations thereof mentioned in the description of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or devices. It should also be noted that in the embodiments of the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a concrete way.

[0022] The technical solution of the present invention is described in detail below with reference to the accompanying drawings.

[0023] like Figure 1 As shown, this embodiment provides a method for improving the production efficiency of high-performance spring steel wire, comprising the following steps: Step S1: Real-time acquisition of multi-source physical quantities during the spring wire winding process, including wire tension distribution, bending rod curvature radius, roller temperature field, and equipment vibration spectrum; Step S2: Based on the acquired multi-source physical quantities, a torque command, a speed command and a bending radius compensation are generated through a dynamic collaborative mapping function; Step S3: According to the instructions generated in step S2, the servo motor group is driven to adjust the wire feeding speed and the roller angular velocity, and the piezoelectric trimmer is controlled to correct the bending radius; Step S4: Use reinforcement learning strategy to optimize control parameters online, and dynamically update the mapping function of step S2 according to the winding quality feedback.

[0024] Through the above technical solution, the present invention can effectively solve the problem of long debugging time during model changeovers. Specifically, traditional production line model changes require manual parameter adjustment (taking 2-3 hours). However, the method of the present invention achieves automatic adaptation of process parameters through real-time multi-source physical quantity acquisition and closed-loop control of dynamic collaborative mapping functions. Through real-time data-driven decision-making (by continuously monitoring steel wire tension, curvature radius, temperature field and vibration spectrum, the system automatically generates torque, speed and bending compensation instructions without manual trial and error adjustment) and an online parameter optimization mechanism (a reinforcement learning strategy continuously updates the mapping function based on winding quality feedback, allowing the system to autonomously adapt to the production requirements of different specifications of steel wire, significantly shortening model changeover debugging time), it can effectively shorten model changeover debugging time.

[0025] Second, the present invention can significantly reduce the wire breakage rate. Specifically, to address the high wire breakage rate of small-diameter steel wire, the present invention establishes a dynamic disturbance active suppression mechanism through multi-dimensional state perception and real-time collaborative control. Multi-dimensional state perception involves synchronously collecting key physical quantities such as tension distribution (to prevent local overload) and vibration spectrum (to identify mechanical resonance risks) to detect wire breakage causes in advance. Real-time collaborative control involves dynamically adjusting the wire feed speed and roller angular velocity through a servo motor group, and combining it with a piezoelectric fine-tuner to instantly correct the bending radius, avoiding wire breakage caused by stress concentration or sudden vibration changes.

[0026] Third, the present invention can effectively improve dimensional consistency. Specifically, to address the issue of helix angle fluctuation exceeding ±3°, the present invention constructs a closed-loop precision control system through cross-parameter collaborative optimization (using a dynamic collaborative mapping function to synchronously process multiple sources of information, such as tension, curvature, temperature, and vibration, to generate mutually coupled compensation instructions, such as bend radius compensation, to ensure dynamic matching of various parameters) and continuous self-learning capabilities (using a reinforcement learning strategy to iteratively optimize the mapping function based on actual winding quality, enabling the system to maintain high precision despite interference such as material fluctuations and thermal deformation, for example, improving helix angle stability). This effectively improves dimensional consistency.

[0027] In one embodiment of the present invention, in step S2 of the present invention, the construction of the dynamic collaborative mapping function includes: Establish a cross-domain coupling model of wire deformation, equipment dynamics, and thermodynamics: Where, is the plastic deformation energy density of the steel wire, in units of , is the time variable, is the thermal conductivity coefficient, in units of , is the Laplace operator, and its unit is , is the deformation conversion coefficient, is the von Mises equivalent stress, in MPa, is the logarithmic strain rate, in units of , is the heat exchange coefficient, is the ambient temperature in °C, The real-time temperature of the roller in °C.

[0028] In this way, by establishing a cross-domain coupling model of wire deformation-equipment dynamics-thermodynamics ( ), which quantifies material deformation, equipment thermal state, and mechanical dynamics in a unified manner, enabling the control system to synchronously respond to multi-physics field coupling effects and fundamentally improve the dynamic disturbance suppression capability. Quantify the heat conduction process (used to suppress roller displacement caused by temperature rise), term quantifies the deformation-stress conversion (to accommodate material hardness fluctuations), The term is used to quantify the environment-equipment heat exchange (compensating for sudden changes in ambient temperature), thereby providing a physical driving basis for generating torque commands, speed commands, and bending radius compensation, ensuring the coordinated matching of multiple commands. Furthermore, purely data-driven reinforcement learning is prone to falling into local optimality in industrial scenarios, and its decisions lack interpretability. The cross-domain coupling model of the present invention is a white-box model (not a black-box model) based on the laws of physics. Its output variables (such as plastic deformation energy density) provide reinforcement learning with interpretable optimization objectives (for example, directly related to wire breakage risk or dimensional error), accelerating strategy convergence and ensuring the industrial reliability of control logic.

[0029] The physical connotation of the cross-domain coupling model is explained below. In this embodiment, the cross-domain coupling model describes the spatiotemporal evolution of the plastic deformation energy density (Q) of the steel wire through a partial differential equation. Its three components correspond to three types of physical processes, among which, The term corresponds to the heat conduction process, that is, the heat diffusion in the contact area between the roller and the wire. The dynamic disturbance that this term opposes is the thermal deformation of the equipment (for example, the roller displacement caused by a temperature rise of 0.1°C / min). The term corresponds to the mechanical energy-to-thermal energy conversion process, that is, the work done by stress (σ) and strain rate (ϵ) during wire deformation is converted into thermal energy (β is the conversion coefficient). The dynamic disturbance that this term corresponds to is the material hardness fluctuation (for example, a stress mutation caused by a ±15% deviation); The term corresponds to the ambient heat exchange process, that is, the roller temperature ( ) and ambient temperature ( ) drives the heat loss / gain (γ is the exchange coefficient). The dynamic disturbance that this item corresponds to is the impact of workshop ambient temperature fluctuations on roller stability.

[0030] In one embodiment of the present invention, the specific implementation of the dynamic cooperative mapping function of the present invention may include: calculating the torque gain coefficient and the bending radius compensation coefficient based on Lyapunov stability theory: Where, is the torque gain coefficient, is the time-varying gain coefficient, in units of , is the attenuation coefficient, in units of , is the yield strength of steel wire, in MPa, is the bending radius compensation coefficient, is the space compensation coefficient, the unit is mm, is the thermal attenuation coefficient, in units of , is the absolute value of temperature deviation, that is , unit is ℃, is the second norm of the deformation energy gradient, is the characteristic stress, in MPa.

[0031] In this embodiment, first, in the existing related technology, when the material hardness changes suddenly or the temperature changes suddenly, the traditional PID control may cause "overshoot causing wire breakage" or "response lag causing dimensional deviation" (such as the helical angle fluctuation exceeds ±3°) due to the fixed gain parameter. (Based on material yield strength Dynamic scaling , so that the torque command transitions smoothly when the hardness changes suddenly, thus avoiding stress shock and wire breakage) and bending radius compensation coefficient (Introducing temperature deviation and deformation energy gradient , achieving coordinated compensation for thermal deformation and stress concentration, thereby suppressing dimensional drift) and building an adaptive gain mechanism. This effectively resolves the industrial control challenge of "the contradiction between dynamic response and stability."

[0032] Second, in this embodiment, the present invention is The integral form of ) accumulates historical deformation states to improve robustness to material fluctuations (e.g., hardness ±15%); at the same time, The exponential decay term ( ) quickly suppresses thermal interference (for example, the compensation attenuation rate can reach 15% / s for a 10°C temperature change), two orders of magnitude faster than traditional mechanical compensation. This allows for high precision while maintaining robustness.

[0033] In this embodiment, it should be noted that, in the present invention, the essence of the formulas for the torque gain coefficient and the bending radius compensation coefficient is to construct the derivative term of the Lyapunov function V, forcing the system energy to decrease over time. Specifically, The physical control logic includes: The term maps stress to , thus avoiding overload (when Approaching time saturation), The term is the forgetting factor, which is used to reduce the weight of historical errors and prevent integral saturation; The physical control logic includes: The term is used to suppress thermal drift (the greater the temperature deviation, the smaller the compensation amount). The term is the spatial gradient of deformation energy, and the high gradient area increases to compensate (to prevent local fracture).

[0034] In one embodiment of the present invention, when a sudden change in material hardness is detected, The torque gain The adjustment rate is controlled at range; when the temperature deviation When the thermal expansion compensation mechanism is activated, the bending radius compensation coefficient Decays exponentially, and the decay rate satisfies .

[0035] In this embodiment, first, in the existing related art, when the hardness of small diameter steel wire fluctuates by ±15%, the traditional PID is prone to torque overshoot (causing local stress to exceed the yield strength and break the wire) due to the fixed gain; while the present invention adjusts the rate by constraining the torque gain ( ), to achieve active buffering of stress impact, that is, when the hardness changes suddenly, Xiang Shi Nonlinear saturation, thus forcing , and its corresponding physical meaning is: limiting the torque change rate to the threshold of the plastic deformation rate of the steel wire to avoid brittle fracture.

[0036] Second, in the existing related technologies, temperature rise may cause the roller to accumulate thermal deformation, and the traditional support assembly cannot dynamically compensate; Exponential decay thermal compensation is activated when ), where the attenuation rate is used to make the bending compensation amount shrink accurately according to the thermal deformation law of the material.

[0037] In one embodiment of the present invention, during the execution of step S3 of the present invention, vibration suppression is performed synchronously: a phase offset controller is used to generate a vibration compensation signal: Where, is the vibration compensation signal, i.e. acceleration command, in units of , is the acceleration gain, in units of , is the total number of vibration modes, is the index value of the vibration mode, For the The acceleration amplitude of the order vibration mode, in units of , is the reference acceleration, in units of , is the imaginary unit, For the The angular frequency of the order vibration mode, in rad / s, For the The lag phase angle of the order vibration mode, in rad, is the cutoff frequency in Hz.

[0038] In this embodiment, in the existing related technology, the traditional production line has a resonance peak at 30Hz-50Hz (especially around the bending rod), which easily causes a sudden change in wire tension (increased wire breakage rate) and roller displacement (helix angle fluctuation exceeds ±3°); the method of the present invention generates an acceleration compensation signal through a phase hedge controller , which not only achieves accurate cancellation of resonance peaks (through Capturing the vibration phase and counteracting the energy with an anti-phase signal), and also enabling band-focused suppression ( The function concentrates the compensation on 30Hz-50Hz to avoid full-band interference), thus effectively solving the wire breakage and size fluctuation caused by 30Hz-50Hz mechanical resonance.

[0039] It should be noted that, for For the term, its physical meaning is: the complex exponential form represents the rotation vector of the vibration mode, and its mechanism for combating vibration is to generate an anti-phase signal with a phase difference of 180° from the original vibration to achieve peak-trough cancellation. In terms of terms, its physical meaning is: the time domain expression of the frequency domain rectangular window, and its mechanism of combating vibration is: focusing the compensation energy on frequency band to avoid high frequency noise interference. For the term, its physical meaning is: weighting the acceleration amplitude of the n-order vibration mode, and its mechanism of combating vibration is: preferentially suppressing the main mode with a high energy ratio (for example, the 30Hz peak or 50Hz peak described in the background technology). In terms of the term, its physical meaning is: acceleration gain (the negative sign indicates reverse compensation), and its mechanism for combating vibration is: setting it according to the thrust-voltage curve of the piezoelectric trimmer to ensure that the compensation force matches the vibration inertia force.

[0040] Taking an exemplary embodiment as an example, the implementation process of vibration suppression is described below.

[0041] Step 1: Real-time vibration sensing: MEMS sensor array → Resonance peak analysis 、 、 (especially the 30Hz-50Hz segment); Step 2: Hedge signal synthesis: Calculate the compensation components of each mode (with additional π phase inversion), and by Convolution implements time-domain finite impulse response (FIR filtering); Step 3: Execute the acceleration command: →The piezoelectric trimmer generates reverse acceleration →offsets the roller vibration displacement.

[0042] In one embodiment of the present invention, in step S4 of the present invention, the reinforcement learning strategy specifically includes: designing a differentiated reward function to drive multi-agent collaborative optimization: Where, For the moment The reward value, 、 and are weight coefficients, is the curvature radius deviation, in mm, is the reference length of the curvature radius, in mm, is the standard deviation of tension, in N, is the reference tension value, in N. is the rate of change of deformation energy, in units of , is the reference power density, in units of .

[0043] In this embodiment, first, in the existing related technologies, the three major problems of long model change and debugging time, high wire breakage rate, and poor dimensional consistency constrain each other (for example, reducing the wire breakage rate requires reducing the bending radius, but this will aggravate dimensional fluctuations). However, the present invention uses a differentiated reward function to enable the system to autonomously balance the three major goals, avoiding the compromise design of manual parameter adjustment.

[0044] Second, in existing related technologies, traditional PID or static models require repeated trial and error adjustments (changing models takes 2-3 hours); while in the present invention, the reward function provides quantifiable instant feedback, for example, a single winding cycle can be calculated , hundreds of times faster than manual debugging; at the same time, 、 and Dynamically adjust the optimization direction (such as small diameter wire increase to prevent wire breakage).

[0045] Third, the reward function of the present invention can integrate the outputs of all submodules (e.g., Depends on Compensation effect, Reflecting the vibration suppression quality of the aforementioned scheme) to drive multi-agent learning of the global optimal strategy.

[0046] In this embodiment, for the differentiated reward function, Item is used to achieve dimensional accuracy priority (curvature radius deviation The smaller the value, the higher the reward. The exponential function is sensitive to small deviations, which is conducive to strengthening high-precision control). The term is used to implement tension stability penalty (the reward decreases sharply when the risk of wire breakage is high, and a direct gradient of the wire breakage risk is provided through the linear penalty term). The term is used to implement energy efficiency incentives (the reward increases when the deformation energy change rate is efficient).

[0047] In addition, it is understood that the weight coefficient can be adjusted according to actual needs, and the present invention does not specifically limit this. In an exemplary embodiment, 、 and The ratio is set to 5:3:1 to give priority to ensuring accuracy. Of course, when continuous wire breakage occurs, the ratio can be automatically increased. , to dynamically switch and optimize the center of gravity.

[0048] In one embodiment of the present invention, in step S1 of the present invention, the vibration spectrum is collected by arranging a MEMS acceleration sensor array (no less than 8 measuring points) along the axial direction of the bending rod. Analyze the 30Hz-50Hz resonance peak using the improved Fast Fourier Transform algorithm: Where, For the frequency range The first The acceleration amplitude of the resonance peak, in units of , is the frequency, is the fast Fourier transform, is the time domain vibration signal, is the time variable, is the time window width of the Gaussian window.

[0049] In this embodiment, first, in the existing related art, there are dense resonance peaks in the 30Hz-50Hz mechanical vibration (for example, 35Hz / 42Hz / 48Hz, etc.), and the traditional FFT causes the amplitude detection error to be greater than 15% due to spectrum leakage (affecting the subsequent suppression effect); in the present invention, a Gaussian window is used to improve the FFT, and the time domain attenuation characteristics of the Gaussian window are utilized ( ) Increase the spectrum resolution to 0.5Hz, which can effectively solve the problem of modal aliasing in traditional vibration detection.

[0050] Second, in the existing related technologies, conventional production lines are only equipped with 1-2 vibration sensors, which cannot capture the axial modal gradient of the bending rod (resulting in missed detection of local resonance); in the present invention, through the axial MEMS array (≥8 measuring points), not only can a large spatial sampling density be achieved and the standing wave mode can be reconstructed, but also the antinode position of the "bending rod resonance" described in the background technology can be effectively identified (that is, the high-risk area for broken wires), thereby effectively solving the problem of insufficient spatial coverage of vibration sensing.

[0051] The above are only specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for improving the production efficiency of high-performance spring steel wire, characterized in that: The steps include: Step S1: Real-time acquisition of multi-source physical quantities during the spring wire winding process, including wire tension distribution, bending rod curvature radius, roller temperature field, and equipment vibration spectrum; Step S2: Based on the acquired multi-source physical quantities, a torque command, a speed command and a bending radius compensation are generated through a dynamic collaborative mapping function; Step S3: According to the instructions generated in step S2, the servo motor group is driven to adjust the wire feeding speed and the roller angular velocity, and the piezoelectric trimmer is controlled to correct the bending radius; Step S4: Use reinforcement learning strategy to optimize control parameters online, and dynamically update the mapping function of step S2 according to the winding quality feedback.

2. The method for improving the production efficiency of high-performance spring steel wire according to claim 1, characterized in that: In step S2, the construction of the dynamic collaborative mapping function includes: Establish a cross-domain coupling model of wire deformation, equipment dynamics, and thermodynamics: Where, is the plastic deformation energy density of the steel wire, in units of , is the time variable, is the thermal conductivity coefficient, in units of , is the Laplace operator, and its unit is , is the deformation conversion coefficient, is the von Mises equivalent stress, in MPa, is the logarithmic strain rate, in units of , is the heat exchange coefficient, is the ambient temperature in °C, The real-time temperature of the roller in °C.

3. The method for improving the production efficiency of high-performance spring steel wire according to claim 2, characterized in that: The specific implementation of the dynamic collaborative mapping function includes: The torque gain coefficient and bending radius compensation coefficient are calculated based on Lyapunov stability theory: Where, is the torque gain coefficient, is the time-varying gain coefficient, in units of , is the attenuation coefficient, in units of , is the yield strength of steel wire, in MPa, is the bending radius compensation coefficient, is the space compensation coefficient, the unit is mm, is the thermal attenuation coefficient, in units of , is the absolute value of temperature deviation, that is , unit is ℃, is the second norm of the deformation energy gradient, is the characteristic stress, in MPa.

4. The method for improving the production efficiency of high-performance spring steel wire according to claim 3, characterized in that: When a sudden change in material hardness is detected, The torque gain The adjustment rate is controlled at within the scope; When the temperature deviation When the thermal expansion compensation mechanism is activated, the bending radius compensation coefficient Decays exponentially, and the decay rate satisfies .

5. The method for improving the production efficiency of high-performance spring steel wire according to claim 2, characterized in that: During the execution of step S3, vibration suppression is performed synchronously: Use phase-compensation controller to generate vibration compensation signal: Where, is the vibration compensation signal, i.e. acceleration command, in units of , is the acceleration gain, in units of , is the total number of vibration modes, is the index value of the vibration mode, For the The acceleration amplitude of the order vibration mode, in units of , is the reference acceleration, in units of , is the imaginary unit, For the The angular frequency of the order vibration mode, in rad / s, For the The lag phase angle of the order vibration mode, in rad, is the cutoff frequency in Hz.

6. The method for improving the production efficiency of high-performance spring steel wire according to claim 5, characterized in that: In step S4, the reinforcement learning strategy specifically includes: Designing differentiated reward functions to drive multi-agent collaborative optimization: Where, For the moment The reward value, 、 and are weight coefficients, is the curvature radius deviation, in mm, is the reference length of the curvature radius, in mm, is the standard deviation of tension, in N, is the reference tension value, in N. is the rate of change of deformation energy, in units of , is the reference power density, in units of .

7. The method for improving the production efficiency of high-performance spring steel wire according to any one of claims 1 to 6, characterized in that: In step S1, the vibration spectrum is collected by: Arrange the MEMS acceleration sensor array (no less than 8 measuring points) around the axial direction of the bending rod. Analyze the 30Hz-50Hz resonance peak using the improved Fast Fourier Transform algorithm: Where, For the frequency range The first The acceleration amplitude of the resonance peak, in units of , is the frequency, is the fast Fourier transform, is the time domain vibration signal, is the time variable, is the time window width of the Gaussian window.

Images