A shear parameter adaptive control method and system for multi-hardness materials

Abstract

The present application relates to the field of industrial control technology, in particular to a shear parameter adaptive control method and system for multi-hardness materials, which comprises pre-pressing and applying a micro-disturbance signal to the material to be processed, extracting the complex impedance spectrum characteristics of the dynamic mechanical response to construct the frequency domain fingerprint of the internal heterogeneous structure; predicting the load mutation based on the fingerprint to generate a virtual control trajectory based on the shear knife holder displacement; using a predictive trajectory planner to calculate the feedforward torque compensation, superimposing it with the position feedback adjustment amount to generate the optimal blade gap and shear speed cooperative motion instruction adaptive to the current hardness; the present application realizes non-destructive detection of internal hardness mutation of the material through frequency domain fingerprint technology, and uses the feedforward compensation mechanism based on inverse dynamics to adjust the driving torque and mechanical gap synchronously before the shear mutation occurs, solving the problems of response lag and easy blade collapse in the processing of heterogeneous materials.

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, specifically to an adaptive control method and system for shear parameters of materials with varying hardness. Background Technology

[0002] In the shearing process of heterogeneous or multi-hardness materials, due to the strong time-varying, random and unobservable physical properties of the controlled object, existing open-loop or fixed-parameter control systems cannot respond to instantaneous load changes in real time, resulting in poor robustness of the control system, easy overload damage and jamming failure of the actuator, and low energy efficiency.

[0003] Specifically, since the non-homogeneous characteristics of the controlled object constitute a black box model, traditional control methods lack a feedforward prediction mechanism and can only rely on the feedback error signal after the controlled variable deviates significantly for adjustment. However, due to the inherent transmission delay and dynamic response bandwidth limitation of the actuator, this lagging feedback adjustment cannot establish an effective control quantity match within the millisecond-level transient window of the disturbance signal input, resulting in insufficient phase margin of the closed-loop control system and a serious deterioration in dynamic tracking performance.

[0004] The lack of predictive capability of this model directly leads to poor robustness of the control strategy and the inability to achieve optimal control: if high-gain or conservative control parameters are used to suppress potential maximum disturbances, the system will be in an inefficient operating state of overdamping or high energy consumption; if conventional control parameters are used, once a random parameter perturbation occurs, the control input will saturate or oscillate, leading to system instability or damage to the actuator due to overload.

[0005] To address this, an adaptive control method and system for shear parameters of materials with varying hardness is proposed. Summary of the Invention

[0006] The purpose of this invention is to provide an adaptive control method and system for shear parameters of materials with multiple hardness levels, which achieves adaptive control of shear parameters by predicting internal hardness abrupt changes.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] The material to be processed is pre-compressed, and a micro-amplitude perturbation signal is applied during pre-compression. The dynamic mechanical response of the micro-amplitude perturbation signal of the material to be processed is collected. The complex impedance spectrum characteristics of the dynamic mechanical response are extracted by fast Fourier transform. Based on the complex impedance spectrum characteristics, a frequency domain fingerprint of the heterogeneous structure distribution inside the material to be processed is constructed.

[0009] A virtual control generator is constructed to predict the load mutation time and peak intensity caused by the sudden change in material hardness based on the frequency domain fingerprint; the predicted load mutation time and peak intensity are converted into discrete feature nodes with the vertical stroke displacement of the shearing tool holder as the reference, and the discrete feature nodes are connected by a spline interpolation algorithm to generate a virtual control trajectory.

[0010] The virtual control trajectory is input to the predictive trajectory planner; the feedforward torque compensation required to overcome the resistance of hardness change is calculated, and the feedforward torque compensation is superimposed with the feedback adjustment based on the real-time position error to generate an optimal blade gap and shearing speed coordinated motion command sequence that adaptively matches the current hardness, driving the hydraulic servo valve and the gap adjustment motor to execute shearing parameter control.

[0011] Preferably, applying a micro-amplitude perturbation signal during preloading specifically involves:

[0012] The control pressure device applies and maintains a constant static preload to the material to be processed. After the static preload stabilizes, a sinusoidal scanning excitation signal with a linearly increasing frequency within a preset bandwidth is superimposed onto the drive system of the pressure device. The amplitude of the sinusoidal scanning excitation signal is set to a preset percentage of the static preload. The stress amplitude generated by the micro-amplitude disturbance signal is lower than the micro-yield strength threshold corresponding to the current hardness of the material to be processed. The dynamic mechanical response of the micro-amplitude disturbance signal of the material to be processed is collected.

[0013] Preferably, the dynamic mechanical response is the real-time excitation force time-domain signal and the real-time displacement deformation time-domain signal synchronously acquired by the sensor during the loading of the sinusoidal scanning excitation signal;

[0014] The real-time excitation force time-domain signal characterizes the actual driving force waveform applied to the material, and the real-time displacement deformation time-domain signal characterizes the following elastic deformation waveform generated by the material.

[0015] Preferably, the virtual control generator adopts a generative mapping model based on inverse dynamics. The specific construction process includes: establishing a nonlinear mapping relationship between the peak value of the complex impedance modulus in the frequency domain fingerprint and the shear strength distribution of the material to be processed along the thickness direction;

[0016] The inverse dynamic equation is constructed using the kinematic parameters of the shearing mechanism. The shearing intensity distribution is substituted into the inverse dynamic equation as a boundary condition. The theoretical resistance evolution curve required for the shearing tool to overcome the shearing intensity distribution under constant speed conditions is derived in reverse. The extreme points and their corresponding time-domain positions are extracted from the theoretical resistance evolution curve as the moment of change between the peak intensity and the load.

[0017] Preferably, the specific process of converting the data into discrete feature nodes based on the vertical stroke displacement of the shearing tool holder is as follows:

[0018] The shearing speed value under constant speed conditions is multiplied by the predicted load change moment to obtain the corresponding vertical stroke displacement value of the shearing holder; the vertical stroke displacement value of the shearing holder is used as a reference position, and the peak intensity is used as the target load value at the reference position. The two are paired to form a set of feature data; multiple sets of feature data are arranged in ascending order of the vertical stroke displacement value of the shearing holder to form the discrete feature nodes.

[0019] Preferably, the predicted trajectory planner includes a dynamic prediction model of the shearing mechanism and a rolling optimization solver. The dynamic prediction model of the shearing mechanism establishes the correlation between the output torque of the hydraulic servo valve and the motion state of the shearing tool holder, and predicts the displacement response of the shearing tool holder.

[0020] The rolling optimization solver uses the virtual control trajectory as a reference benchmark. In each control cycle, it performs optimization calculations with the goal of minimizing the deviation between the predicted shearing blade displacement response and the virtual control trajectory, calculates the feedforward torque compensation required to balance the material hardness, and synchronously outputs the corresponding blade gap adjustment command.

[0021] Preferably, the specific process of generating the optimal blade gap and shearing speed coordinated motion command sequence that adaptively matches the current hardness is as follows:

[0022] A position closed-loop tracking controller is constructed to calculate the tracking error between the actual position of the shearing blade holder and the theoretical position set in the virtual control trajectory in real time, and to generate a feedback adjustment amount based on the tracking error;

[0023] The feedforward torque compensation amount and the feedback adjustment amount are superimposed to synthesize a total control amount, and the total control amount is converted into a current control signal to drive the hydraulic servo valve; at the same time, the blade gap adjustment command is parsed into a target position code and converted into a pulse control signal to drive the gap adjustment motor.

[0024] The current control signal and the pulse control signal are output within the same control cycle, enabling coordinated action of dynamic adjustment of the blade gap and variable speed cutting of the shearing tool holder.

[0025] An adaptive control system for shear parameters of materials with varying hardness includes:

[0026] Fingerprint construction module: Pre-compress the material to be processed, apply a micro-amplitude perturbation signal during pre-compression, collect the dynamic mechanical response of the micro-amplitude perturbation signal of the material to be processed, extract the complex impedance spectrum characteristics of the dynamic mechanical response through fast Fourier transform, and construct the frequency domain fingerprint of the heterogeneous structure distribution inside the material to be processed based on the complex impedance spectrum characteristics.

[0027] Trajectory generation module: Constructs a virtual control generator to predict the load change time and peak intensity caused by the sudden change in material hardness based on the frequency domain fingerprint; converts the predicted load change time and peak intensity into discrete feature nodes based on the vertical stroke displacement of the shearing blade holder, and connects the discrete feature nodes using a spline interpolation algorithm to generate a virtual control trajectory;

[0028] Feedforward compensation module: inputs the virtual control trajectory to the predictive trajectory planner; calculates the feedforward torque compensation amount required to overcome the resistance of hardness variation;

[0029] Parameter control module: It superimposes the feedforward torque compensation amount with the feedback adjustment amount based on the real-time position error to generate an optimal blade gap and shearing speed coordinated motion command sequence that adaptively matches the current hardness, drives the hydraulic servo valve and the gap adjustment motor, and executes shearing parameter control.

[0030] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0031] 1. This application constructs a frequency domain fingerprint of the material by applying a sinusoidal scanning micro-amplitude perturbation during the pre-compression stage and extracting the complex impedance spectrum features using fast Fourier transform. By utilizing the differences in frequency response of different hardness components to dynamic excitation, the originally closed, random, and unobservable internal hardness distribution of the material is non-destructively mapped into a quantifiable frequency domain feature signal. This effectively solves the blindness of traditional control methods in operating the internal structure of the material as a black box, enabling the control system to perceive the distribution law of the heterogeneous structure inside the material in advance before the shearing action begins, providing decisive prior data support for subsequent accurate predictive control.

[0032] 2. This application innovatively introduces a virtual control generator based on inverse dynamics, transforming the hardness mutation predicted by frequency domain fingerprinting into a virtual control trajectory based on the vertical stroke displacement of the shearing tool holder. A rolling optimization solver pre-calculates the feedforward torque compensation required to overcome the hardness mutation based on this trajectory. This allows for proactive output of a matching driving torque at the instant the tool contacts the hard point, rather than adjusting it only after the speed decreases. This feedforward compensation mechanism fundamentally overcomes the shortcomings of existing open-loop or simple PID feedback control, such as lag and large overshoot, when facing strong time-varying loads, significantly improving the dynamic response speed and robustness when dealing with extreme hardness changes.

[0033] 3. This application not only controls the shearing torque but also superimposes the feedforward torque and position feedback error, simultaneously generating the optimal tool clearance adjustment command. Within the same control cycle, the system can coordinately adjust the hydraulic servo valve (force / speed) and the clearance adjustment motor (geometric parameters) based on the hardness characteristics of the current position. This multi-variable collaborative mechanism ensures high precision and efficiency when cutting soft materials, while avoiding tool breakage or jamming when encountering extremely hard materials by increasing the clearance and torque. By upgrading the open-loop fixed-parameter shearing to a fully closed-loop adaptive collaborative shearing, the problem of unstable machining quality caused by the randomness of the physical properties of the controlled object is effectively solved, extending tool life. Attached Figure Description

[0034] Figure 1 This is a flowchart illustrating an adaptive control method for shear parameters in materials with varying hardness.

[0035] Figure 2 This is the closed-loop process of perception-prediction-compensation-execution in this invention;

[0036] Figure 3 This is a schematic diagram of an adaptive control system for shear parameters of materials with varying hardness. Detailed Implementation

[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] Example 1:

[0039] Please see Figures 1 to 3 This invention provides an adaptive control method and system for shear parameters of materials with multiple hardness levels. The technical solution is as follows:

[0040] An adaptive control method for shear parameters of materials with multiple hardnesses includes:

[0041] The material to be processed is pre-compressed, and a micro-amplitude perturbation signal is applied during pre-compression. The dynamic mechanical response of the micro-amplitude perturbation signal of the material to be processed is collected. The complex impedance spectrum characteristics of the dynamic mechanical response are extracted by fast Fourier transform. Based on the complex impedance spectrum characteristics, a frequency domain fingerprint of the heterogeneous structure distribution inside the material to be processed is constructed.

[0042] A virtual control generator is constructed to predict the load mutation time and peak intensity caused by the sudden change in material hardness based on the frequency domain fingerprint; the predicted load mutation time and peak intensity are converted into discrete feature nodes with the vertical stroke displacement of the shearing tool holder as the reference, and the discrete feature nodes are connected by a spline interpolation algorithm to generate a virtual control trajectory.

[0043] The virtual control trajectory is input to the predictive trajectory planner; the feedforward torque compensation required to overcome the resistance of hardness change is calculated, and the feedforward torque compensation is superimposed with the feedback adjustment based on the real-time position error to generate an optimal blade gap and shearing speed coordinated motion command sequence that adaptively matches the current hardness, driving the hydraulic servo valve and the gap adjustment motor to execute shearing parameter control.

[0044] The specific steps of applying a small-amplitude perturbation signal during preloading are as follows:

[0045] The control pressure device applies and maintains a constant static preload to the material to be processed. After the static preload stabilizes, a sinusoidal scanning excitation signal with a linearly increasing frequency within a preset bandwidth is superimposed onto the drive system of the pressure device. The amplitude of the sinusoidal scanning excitation signal is set to a preset percentage of the static preload. The stress amplitude generated by the micro-amplitude disturbance signal is lower than the micro-yield strength threshold corresponding to the current hardness of the material to be processed. The dynamic mechanical response of the micro-amplitude disturbance signal of the material to be processed is collected.

[0046] Specifically, a command is sent to the pressing device of the shearing machine to drive it to move downward and contact the surface of the material to be processed, apply and maintain a constant static pre-pressure, and monitor the pressure value in real time through a built-in pressure sensor. When the pressure fluctuation value is detected to converge to a preset error range and be maintained for a certain period of time, it is determined that the pressing device has entered a steady state. At this time, the excitation module is started to superimpose a sinusoidal scanning excitation signal with a frequency that increases linearly within a preset bandwidth onto the servo drive system of the pressing device. The frequency of the sinusoidal scanning excitation signal starts from the starting frequency and increases linearly with time to the ending frequency, covering multiple inherent frequency response ranges that the material to be processed may have.

[0047] During the application of the excitation signal, the energy intensity of the signal is strictly limited. The amplitude of the sinusoidal scanning excitation signal is set to a preset percentage of the static preload (5% in this embodiment). Through this small change in force loading, it is ensured that the stress amplitude generated by the micro-amplitude disturbance signal is lower than the micro-yield strength threshold corresponding to the current hardness of the material to be processed throughout the entire excitation cycle. The purpose of this limitation is to ensure that the material only undergoes elastic deformation when disturbed, and remains within the linear elastic range applicable to Hooke's Law, thereby avoiding damage to the material structure or interference with the accuracy of subsequent impedance analysis due to plastic deformation. While the disturbance is applied, the dynamic mechanical response of the micro-amplitude disturbance signal of the material to be processed is simultaneously acquired using a high-frequency force sensor and a micro-displacement sensor.

[0048] By superimposing a micro-amplitude sinusoidal scanning excitation signal on static pre-stress, a wide-frequency non-destructive detection of the internal heterogeneous structure of materials with varying hardness was achieved. On the one hand, by using linearly increasing frequency scanning, the dynamic impedance characteristics of the material at different frequency bands can be effectively excited and captured, avoiding the feature omissions that may occur with single-frequency detection. On the other hand, by strictly controlling the disturbance within the linear elastic range, the detection process is physically prevented from causing microscopic plastic damage or surface indentation to the material, ensuring the quality of the finished product. It also mathematically ensures the linearity of the mechanical response, significantly improving the signal-to-noise ratio and hardness recognition accuracy of subsequent frequency domain fingerprint extraction.

[0049] The dynamic mechanical response is the real-time excitation force time-domain signal and the real-time displacement deformation time-domain signal synchronously acquired by the sensor during the loading of the sinusoidal scanning excitation signal;

[0050] The real-time excitation force time-domain signal characterizes the actual driving force waveform applied to the material, and the real-time displacement deformation time-domain signal characterizes the following elastic deformation waveform generated by the material.

[0051] Specifically, when the sinusoidal scanning excitation signal begins to be applied to the pressure device drive system, a hardware-triggered interrupt mechanism is used to simultaneously activate the high-frequency dynamic force sensor installed on the end actuator of the pressure device and the high-precision micro-displacement sensor installed beside the mechanical kinematic pair. Under the control of a unified synchronous clock signal, these two sensors perform high-speed continuous sampling of physical quantities at a sampling rate much higher than the highest cutoff frequency of the sinusoidal scanning excitation signal (in this embodiment, it is set to more than 10 times the upper limit of the excitation frequency), thereby synchronously acquiring the real-time excitation force time-domain signal and the real-time displacement deformation time-domain signal.

[0052] The real-time excitation force time-domain signal records the force value sequence superimposed on the static preload reference, with amplitude and frequency dynamically changing over time, representing the actual driving force waveform applied to the material, and serving as the input excitation function; while the real-time displacement deformation time-domain signal records the instantaneous microscopic compression and rebound displacement generated in the vertical direction on the surface of the material to be processed under the action of the above-mentioned alternating excitation force, representing the following elastic deformation waveform generated by the material, and serving as the output response function to the excitation; after strict timestamp alignment and noise reduction processing of these two sets of time-domain signals, they are packaged and stored as dual-channel time series data, serving as the basic data source for subsequent calculation of complex impedance and extraction of frequency domain fingerprint;

[0053] By adopting hardware-level trigger interrupt and unified clock synchronization mechanism, microsecond-level synchronous acquisition of excitation force (input) and displacement deformation (output) is achieved, completely eliminating phase delay error between multi-channel signals; combined with high-rate oversampling strategy, signal aliasing and high-frequency feature loss are effectively prevented.

[0054] The virtual control generator adopts a generative mapping model based on inverse dynamics. The specific construction process includes: establishing a nonlinear mapping relationship between the peak value of the complex impedance modulus in the frequency domain fingerprint and the shear strength distribution of the material to be processed along the thickness direction.

[0055] The inverse dynamic equation is constructed using the kinematic parameters of the shearing mechanism. The shearing intensity distribution is substituted into the inverse dynamic equation as a boundary condition. The theoretical resistance evolution curve required for the shearing tool to overcome the shearing intensity distribution under constant speed conditions is derived in reverse. The extreme points and their corresponding time-domain positions are extracted from the theoretical resistance evolution curve as the moment of change between the peak intensity and the load.

[0056] Specifically, establishing a quantitative relationship between the frequency domain fingerprint and the material's mechanical properties includes constructing a nonlinear mapping relationship between the peak value of the complex impedance modulus in the frequency domain fingerprint and the shear strength distribution of the material to be processed along the thickness direction. This nonlinear mapping relationship is established by subjecting reference samples with known thickness distributions and hardness to the same micro-amplitude perturbation excitation and frequency domain fingerprint extraction to form a sample library. A multilayer perceptron machine learning model is then used, with the frequency domain fingerprint feature vector of each reference sample as input and its corresponding thickness-direction hardness distribution (obtained from measured data or metallographic microscopy) as the training target. Offline training of the mapping model is completed, and the output is a one-dimensional array whose length is equal to the number of thickness segments of the material. Each array element represents the relative shear strength value at that thickness position. Key features obtained in the previous steps are extracted from the frequency domain fingerprint, particularly the magnitude of the resonance peak value of the complex impedance modulus in a specific frequency band and its corresponding frequency offset. Based on this, shear strength profile data of the material to be processed in the longitudinal depth is output. This shear strength profile data accurately describes whether there are high-hardness inclusions or hardened layers inside the material, and the specific distribution location of these high-hardness regions in the thickness direction.

[0057] The inverse dynamics equation is constructed using the kinematic parameters of the shearing mechanism to read the mechanical structural parameters of the shearing machine, including the equivalent moment of inertia of the tool holder system, the blade rotation radius, the slider friction coefficient, and the lever arm length of the hydraulic cylinder. A rigid body dynamics model is established based on the Newton-Euler method or the Lagrange equation. In order to simply predict the load change caused by material properties, a virtual ideal cutting state is set, that is, assuming that the tool is shearing at a constant speed (zero acceleration). The shearing intensity distribution is multiplied by the real-time shearing cross-sectional area corresponding to the tool penetration depth to obtain the shearing resistance function that varies with displacement, and this function is substituted into the inverse dynamics equation as a boundary condition for inverse solution.

[0058] By constructing an inverse dynamics model that includes key mechanical parameters such as moment of inertia and friction coefficient, a precise physical mapping between the microscopic shear strength distribution of the material and the macroscopic hydraulic driving force of the equipment is established. This makes the generated control commands no longer based on empirical estimations, but precise values ​​that conform to physical laws, thus providing a highly accurate torque compensation benchmark for feedforward control.

[0059] Through the above reverse solution process, the theoretical resistance evolution curve required for the shearing tool to overcome the shear strength distribution under constant speed conditions is reverse-engineered. The theoretical resistance evolution curve simulates the theoretical driving force waveform required by the hydraulic system when the tool cuts the heterogeneous material at a constant speed. Numerical analysis is performed on the theoretical resistance evolution curve, and the peak position of the resistance rising sharply in the curve is identified using an extreme value search algorithm (in this embodiment, the first derivative zero-crossing point determination method). The extreme points and their corresponding time domain positions are extracted. These extreme points represent the key nodes when the tool is about to contact the material hardness abrupt layer (such as cemented carbide inclusions or welds). The system marks them as the peak strength and the moment of load abrupt change, respectively, as the core basis for generating feedforward compensation commands in the future.

[0060] This invention employs a generative mapping model based on inverse dynamics to establish a physical connection from microscopic frequency domain fingerprints to macroscopic mechanical loads. By nonlinearly mapping abstract complex impedance characteristics to specific shear strength distributions and combining this with mechanical kinematic parameters for inverse deduction, the exact location and peak intensity of the material's internal hardness abrupt change point in the time domain can be accurately predicted before the shearing action occurs. This feedforward prediction mechanism overcomes the lag defect of traditional feedback control in dealing with abrupt load changes, and can adjust hydraulic output and motion parameters in advance to specifically suppress the impact vibration caused by hardness changes. Thus, while ensuring the flatness of the sheared section, it significantly reduces the risk of tool chipping and equipment mechanical wear.

[0061] The specific process of converting the nodes into discrete feature nodes based on the vertical stroke displacement of the shearing tool holder is as follows:

[0062] The shearing speed value under constant speed conditions is multiplied by the predicted load change moment to obtain the corresponding vertical stroke displacement value of the shearing holder; the vertical stroke displacement value of the shearing holder is used as a reference position, and the peak intensity is used as the target load value at the reference position. The two are paired to form a set of feature data; multiple sets of feature data are arranged in ascending order of the vertical stroke displacement value of the shearing holder to form the discrete feature nodes.

[0063] Specifically, since the load abrupt change time is based on relative parameters in the time domain, in order to adapt to the servo characteristics of the shearing machine based on position control, a spatiotemporal mapping transformation is performed, that is, each predicted load abrupt change time is multiplied by the shearing speed value. Through this linear transformation calculation (the calculation formula is: displacement = speed × time), the corresponding vertical stroke displacement value of the shearing blade holder is obtained. The vertical stroke displacement value of the shearing blade holder physically represents the absolute physical coordinate of the blade holder in the Z-axis direction relative to the starting point when the shearing blade contacts the hardness abrupt change point inside the material.

[0064] Construct a multidimensional feature vector, taking the vertical stroke displacement value of the shearing blade holder as the reference position (horizontal axis) and the peak strength (i.e. the resistance extreme value extracted in the previous step) as the target load value at the reference position (vertical axis). Pair the two to form a set of feature data. Repeat this pairing process until all extracted mutation points are processed.

[0065] To meet the requirement of monotonicity of the input data for subsequent trajectory planning algorithms, multiple sets of feature data are arranged in ascending order of the vertical stroke displacement of the shearing tool holder. This sorting operation transforms the disordered time points into an ordered spatial sequence distributed along the tool's entry path, forming the discrete feature nodes. This sequence is not merely a simple set of data points; it essentially constitutes a digital spatial map of the hardness distribution within the material to be processed, providing precise geometric control points for generating a continuous and smooth virtual control trajectory through spline interpolation.

[0066] By accurately mapping time-domain load prediction to spatial-domain tool holder stroke displacement, a time-independent spatial distribution map of material hardness was successfully constructed. This spatiotemporal transformation mechanism effectively eliminates the sensitivity of traditional time-based control to fluctuations in actual shearing speed, ensuring that even under conditions of dynamic adjustment of shearing speed, the feedforward torque compensation command can still maintain strict zero-phase alignment with the material's internal hardness abrupt change layer in physical location. At the same time, through the orderly arrangement and monotonicity processing of feature nodes, geometric constraints conforming to mathematical logic are provided for subsequent trajectory planning, avoiding oscillations or singularities in the interpolation algorithm, and significantly improving the robustness and control accuracy of the adaptive control system under complex unsteady conditions.

[0067] The predicted trajectory planner includes a dynamic prediction model of the shearing mechanism and a rolling optimization solver. The dynamic prediction model of the shearing mechanism establishes the correlation between the output torque of the hydraulic servo valve and the motion state of the shearing tool holder, and predicts the displacement response of the shearing tool holder.

[0068] The rolling optimization solver uses the virtual control trajectory as a reference benchmark. In each control cycle, it performs optimization calculations with the goal of minimizing the deviation between the predicted shearing blade displacement response and the virtual control trajectory, calculates the feedforward torque compensation required to balance the material hardness, and synchronously outputs the corresponding blade gap adjustment command.

[0069] The dynamic prediction model of the shearing mechanism is established based on the physical characteristics of the hydraulic and mechanical systems. It uses a discrete-time state-space equation to establish the correlation between the output torque of the hydraulic servo valve and the motion state of the shearing holder. In this model, the real-time displacement, velocity, and acceleration of the shearing holder are defined as state variables, and the control current of the hydraulic servo valve is defined as the control input variable. Combining physical parameters such as the system's equivalent mass, viscous damping coefficient, and the bulk modulus of the hydraulic oil, matrix operations are used to simulate the dynamic behavior of the system. This allows for the mathematical prediction of the evolution trend of the shearing holder's displacement response within a finite future time domain, given a specific input.

[0070] The rolling optimization solver employs a model predictive control algorithm architecture, reading the virtual control trajectory as a reference into a buffer. Within each control cycle (set to 1 millisecond in this embodiment), the solver sets the prediction and control time domains for backward rolling. At the current sampling moment, based on the current state feedback, the solver uses the dynamic prediction model to perform multi-step iterative deduction, constructing a quadratic programming mathematical problem. The core of this mathematical problem lies in constructing an objective function, which aims to minimize the deviation between the predicted shear holder displacement response and the virtual control trajectory. Simultaneously, a constraint term for the control increment is introduced to ensure... The system stability is ensured; the quadratic programming mathematical problem is: to construct a minimization objective function containing two weighted terms: the first part is the tracking error term, which is the weighted sum of the squares of the deviations between the shear holder displacement response and the corresponding theoretical position in the virtual control trajectory at all predicted time points within the set prediction time domain, used to ensure control accuracy; the second part is the control stability term, which is the weighted sum of the squares of the change in control command (i.e., control increment) between adjacent control cycles within the set control time domain, used to suppress drastic fluctuations in control input; the solver aims to find an optimal control increment sequence that minimizes the sum of the above two parts;

[0071] The constraint of the control increment is: the allowable output control increment value in each control cycle is strictly limited to the preset maximum positive increment threshold and maximum negative increment threshold; that is, the change amplitude of the hydraulic servo valve drive signal in a single adjustment must not exceed the smoothness limit that the physical system can withstand, so as to prevent hydraulic shock or system oscillation due to excessive adjustment.

[0072] By utilizing multi-step iterative deduction and quadratic programming algorithms, predictive high-precision tracking of virtual control trajectories is achieved, effectively suppressing control input oscillations that may be caused by pursuing rapid response. This ensures both the dynamic response speed of the system and the operational stability and robustness of the actuator.

[0073] By solving the above optimization problem, a set of future optimal control input sequences is obtained, and only the first element of the sequence is taken as the control command at the current moment. The command is parsed into two parts. One part is to calculate the feedforward torque compensation required to balance the material hardness, which represents the additional driving force that the hydraulic system needs to output in advance to counteract the upcoming resistance of sudden hardness change. The other part is to query the preset shear force-optimal gap process database based on the mapping relationship between the calculated torque magnitude and the current material position, and synchronously output the corresponding blade gap adjustment command. These two sets of commands are sent synchronously to the underlying actuator to ensure that the tool has sufficient cutting force and is in the optimal shear gap state at the moment of contact with the hard point.

[0074] By utilizing the MPC-based rolling optimization mechanism, the response lag of the hydraulic system is effectively overcome, and the control mode is fundamentally transformed from post-correction to pre-compensation. Through millisecond-level feedforward torque compensation and synchronous coordinated adjustment of the blade gap, the tool is ensured to maintain constant speed and optimal process when dealing with sudden changes in hardness, thereby significantly improving the quality of the shearing section and effectively reducing the risk of mechanical wear and chipping.

[0075] The specific process of generating the optimal blade gap and shearing speed coordinated motion command sequence that adaptively matches the current hardness is as follows:

[0076] A position closed-loop tracking controller is constructed to calculate the tracking error between the actual position of the shearing blade holder and the theoretical position set in the virtual control trajectory in real time, and to generate a feedback adjustment amount based on the tracking error;

[0077] The feedforward torque compensation amount and the feedback adjustment amount are superimposed to synthesize a total control amount, and the total control amount is converted into a current control signal to drive the hydraulic servo valve; at the same time, the blade gap adjustment command is parsed into a target position code and converted into a pulse control signal to drive the gap adjustment motor.

[0078] The current control signal and the pulse control signal are output within the same control cycle, enabling coordinated action of dynamic adjustment of the blade gap and variable speed cutting of the shearing tool holder.

[0079] Specifically, a position closed-loop tracking controller is constructed in the underlying motion controller. The closed-loop tracking controller reads the feedback signal of the high-precision displacement sensor installed on the shearing machine slider at a high-frequency sampling rate (1kHz in this embodiment) to obtain the actual position of the shearing blade holder. It then compares the actual position with the theoretical position corresponding to the current moment in the virtual control trajectory generated in the previous step in real time and calculates the deviation value between the two as the tracking error. The controller calculates the tracking error according to preset proportional, integral, and derivative gain parameters to generate a feedback adjustment amount. This adjustment amount aims to eliminate the steady-state position deviation caused by non-modeled disturbances such as friction and hydraulic oil temperature drift.

[0080] The composite control signal is processed by superimposing the feedforward torque compensation amount and the feedback adjustment amount. The calculation formula is: total control torque = feedforward compensation amount + feedback adjustment amount. The calculated total control amount is converted into a current control signal to drive the hydraulic servo valve through the fieldbus interface. The current signal directly determines the opening size and direction of the servo valve core, thereby accurately controlling the driving force and speed of the hydraulic cylinder.

[0081] In the parallel processing thread, the blade gap adjustment command is parsed into a target position code; based on the mechanical transmission ratio of the shearing machine blade holder gap adjustment mechanism, the target physical gap value is converted into the rotary encoder value required by the servo motor, and further converted into a pulse control signal to drive the gap adjustment motor;

[0082] To ensure strict synchronization between mechanical compensation and geometric adjustment in time, the synchronous clock interrupt function of the real-time operating system is used to output the current control signal and the pulse control signal to the hydraulic servo amplifier and motor driver within the same control cycle. Through this hard real-time synchronization mechanism, the system realizes the coordinated action of dynamic adjustment of blade gap and variable speed cutting of the shearing tool holder. That is, at the instant when the tool contacts different hardness levels of the material, the hydraulic system automatically outputs a matching variable force and variable speed drive, while the tool holder mechanism synchronously fine-tunes to the optimal lateral gap, thereby ensuring that the process parameters throughout the shearing process are always in an optimal matching state.

[0083] By constructing a composite control architecture of feedforward torque compensation and feedback position correction, and combining it with a multi-axis synchronous output mechanism, the technical challenges of traditional shearing machines being unable to handle heterogeneous materials and having fixed gaps have been solved. The superposition of feedforward and feedback ensures that the system can quickly respond to load impacts caused by sudden changes in hardness and eliminate accumulated errors. The millisecond-level synchronous coordination of hydraulic drive and motor gap adjustment adjusts the cutting force and meshing gap in real time according to the hardness changes of the material at every micrometer depth. This ensures efficient shearing while maximizing the perpendicularity and smoothness of the cut surface and extending the service life of the core components of the equipment.

[0084] This invention achieves transparent, non-destructive identification of the heterogeneous structure inside the material to be processed by introducing micro-amplitude disturbance detection and frequency domain fingerprint construction technology during the pre-compression stage. This effectively solves the technical problem that traditional shearing processes cannot predict sudden changes in the hardness of the material. Furthermore, by utilizing a feedforward torque compensation mechanism based on virtual control trajectory, the traditional post-feedback correction is upgraded to pre-predictive compensation. This mechanism can automatically match the optimal driving torque and blade clearance within a millisecond window before the tool contacts the hard point, completely eliminating the mechanical impact and system response lag caused by sudden load changes. This adaptive and cooperative control strategy significantly improves the flatness and perpendicularity of the sheared section of complex materials while effectively reducing the risk of tool chipping and greatly extending the service life of the core components of the shearing equipment.

[0085] Example 2:

[0086] This embodiment takes the shearing of high-strength steel plates containing laser-welded seams in a scrapped car dismantling production line as an example. Since the hardness of the weld area is much higher than that of the base material, and its distribution is random, traditional shearing methods are prone to causing tool chipping or jamming. This system solves this problem through the collaborative work of the following four modules: Fingerprint Construction Module: Pre-compresses the material to be processed, applies a micro-amplitude perturbation signal during pre-compression, collects the dynamic mechanical response of the micro-amplitude perturbation signal of the material to be processed, extracts the complex impedance spectrum characteristics of the dynamic mechanical response through fast Fourier transform, and constructs a frequency domain fingerprint of the heterogeneous structure distribution inside the material to be processed based on the complex impedance spectrum characteristics;

[0087] Trajectory generation module: Constructs a virtual control generator to predict the load change time and peak intensity caused by the sudden change in material hardness based on the frequency domain fingerprint; converts the predicted load change time and peak intensity into discrete feature nodes based on the vertical stroke displacement of the shearing blade holder, and connects the discrete feature nodes using a spline interpolation algorithm to generate a virtual control trajectory;

[0088] Feedforward compensation module: inputs the virtual control trajectory to the predictive trajectory planner; calculates the feedforward torque compensation amount required to overcome the resistance of hardness variation;

[0089] Parameter control module: It superimposes the feedforward torque compensation amount with the feedback adjustment amount based on the real-time position error to generate an optimal blade gap and shearing speed coordinated motion command sequence that adaptively matches the current hardness, drives the hydraulic servo valve and the gap adjustment motor, and executes shearing parameter control;

[0090] The fingerprint construction module first controls the hydraulic pressing foot of the heavy-duty gantry shear to press down, fixing the high-strength steel plate to be sheared with a static preload of 20MPa. After the pressure stabilizes, the fingerprint construction module activates the piezoelectric ceramic exciter integrated inside the pressing foot, applying a sinusoidal scanning micro-amplitude perturbation signal with a frequency linearly increasing from 50Hz to 500Hz to the steel plate. The amplitude is controlled at 5% of the static preload (i.e., 1MPa) to ensure that no plastic indentation is generated. At the same time, the real-time excitation force and displacement deformation response of the steel plate to the perturbation are collected synchronously by a high-frequency force sensor and a laser displacement sensor. The data is processed by fast Fourier transform, and the system analysis finds that a significant complex impedance modulus peak appears in a specific high-frequency band of the spectrum. The peak feature constitutes the frequency domain fingerprint for identifying the location of hidden high-hardness welds inside the steel plate.

[0091] The trajectory generation module receives the aforementioned frequency domain fingerprint data and analyzes it using a built-in generative mapping model based on inverse dynamics. It identifies the complex impedance peak corresponding to the laser weld hardening layer at Z=15mm to Z=18mm along the steel plate thickness direction. This hardness mutation is mapped to a sharp increase in theoretical shear resistance, predicting a load peak intensity up to three times the normal value. Subsequently, this time-based load mutation prediction, combined with a set shearing speed of 50mm / s, is transformed into discrete feature nodes based on the vertical stroke displacement of the shearing tool holder. These nodes are smoothly connected using a spline interpolation algorithm to generate a virtual control trajectory containing the expected load mutation, describing where the tool holder will encounter weld resistance throughout its entire downward stroke.

[0092] The feedforward compensation module inputs the generated virtual control trajectory into the predictive trajectory planner. Using a rolling optimization solver, it predicts the tool holder state forward within each 1-millisecond control cycle. Calculations show that if the original hydraulic input is maintained, the tool holder speed will drop instantaneously when it contacts the weld at Z=15mm. Therefore, the required feedforward torque compensation to balance the sudden change in weld hardness is calculated, meaning the hydraulic system pressure needs to be increased in advance when the tool holder reaches Z=14.8mm. Simultaneously, based on the process database, the blade clearance needs to be temporarily adjusted from the current 0.5mm to 0.8mm to prevent chipping, and a corresponding blade clearance adjustment command is output accordingly.

[0093] During the shearing process, the parameter control module superimposes the calculated feedforward torque compensation amount with the real-time error adjustment amount fed back by the position sensor to generate a total control current to drive the hydraulic servo valve. This causes the oil cylinder to generate high pressure at the moment of contact with the weld, maintaining a constant shearing speed. Within the same control cycle, the blade gap adjustment command is converted into a pulse signal, driving the gap adjustment motor to quickly fine-tune the position of the tool holder. The tool holder descends at high speed, and at the moment of cutting into the weld, the hydraulic power is automatically enhanced and the blade gap is automatically slightly expanded. After cutting off the weld, the original parameters are quickly restored, realizing adaptive and stable shearing under variable hardness conditions.

[0094] This embodiment addresses the extreme condition of high-strength steel coexisting with laser-welded seams in end-of-life vehicle dismantling. It utilizes non-destructive micro-amplitude perturbation frequency domain fingerprinting technology to achieve precise visualization and prediction of hidden weld locations, effectively solving the blindness inherent in traditional shearing methods when faced with random hardness changes. By employing a feedforward compensation and gap collaborative control mechanism, high-pressure bursts and gap expansion are prepared within a millisecond window before the tool contacts the hardened layer. This ensures sufficient energy for cutting welds with three times the load strength while optimizing the physical gap to avoid tool chipping or jamming caused by hard-on-hard contact. This significantly improves the processing efficiency for heterogeneous scrap steel and extends the service life of core components.

[0095] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for adaptive control of shear parameters for multi-hardness materials, characterized in that, The method comprises the following steps: pre-pressing the material to be processed, applying a micro-perturbation signal during pre-pressing, collecting the dynamic mechanical response of the material to be processed to the micro-perturbation signal, extracting the complex impedance frequency spectrum characteristics of the dynamic mechanical response by fast Fourier transform, and constructing the frequency domain fingerprint of the internal heterogeneous structure distribution of the material to be processed based on the complex impedance frequency spectrum characteristics; constructing a virtual control generator to predict the load mutation time and peak intensity caused by the material hardness mutation based on the frequency domain fingerprint; the virtual control generator adopts a generative mapping model based on inverse dynamics, and the specific construction process comprises: establishing a nonlinear mapping relationship between the complex impedance modulus peak in the frequency domain fingerprint and the shear strength distribution of the material to be processed along the thickness direction; inverse dynamics equation is constructed by using the kinematic parameters of the shearing mechanism, the shear strength distribution is substituted into the inverse dynamics equation as a boundary condition, the theoretical resistance evolution curve of the shear cutter required to overcome the shear strength distribution under constant speed working condition is deduced reversely, and the extreme point and its corresponding time domain position are extracted from the theoretical resistance evolution curve as the peak intensity and load mutation time; the predicted load mutation time and peak intensity are converted into discrete feature nodes with the vertical stroke displacement of the shear tool holder as the reference, the discrete feature nodes are connected by using a spline interpolation algorithm, and a virtual control trajectory is generated; the virtual control trajectory is input into a predictive trajectory planner; the feedforward torque compensation required to overcome the hardness change resistance is calculated, and the feedforward torque compensation and the feedback adjustment based on the real-time position error are superimposed to generate an optimal blade gap and shear speed cooperative motion instruction sequence that matches the current hardness, which drives the hydraulic servo valve and the gap adjusting motor to execute the shear parameter control.

2. The method of claim 1, wherein, The micro-perturbation signal applied during pre-pressing is specifically: controlling the pressure device to apply and maintain a constant static pre-pressure to the material to be processed, superimposing a sinusoidal scanning excitation signal with linearly increasing frequency within a preset bandwidth to the driving system of the pressure device after the static pre-pressure is stable; the amplitude of the sinusoidal scanning excitation signal is set to be a preset percentage of the static pre-pressure, the stress amplitude generated by the micro-perturbation signal is lower than the micro-yield strength threshold corresponding to the current hardness of the material to be processed, and the dynamic mechanical response of the material to the micro-perturbation signal is collected.

3. The method of claim 2, wherein, The dynamic mechanical response is the real-time excitation force time domain signal and the real-time displacement deformation time domain signal collected by the sensor during the loading of the sinusoidal scanning excitation signal; wherein, the real-time excitation force time domain signal represents the actual driving force waveform applied to the material, and the real-time displacement deformation time domain signal represents the follow-up elastic deformation waveform of the material.

4. The method of claim 1, wherein, The conversion into discrete feature nodes with the vertical stroke displacement of the shear tool holder as the reference is specifically: Obtaining the shear speed value under the constant speed working condition, multiplying the predicted load mutation time by the shear speed value to convert the corresponding shear tool holder vertical stroke displacement value; taking the shear tool holder vertical stroke displacement value as the reference position, taking the peak intensity as the target load value at the reference position, and pairing them to form a set of characteristic data; arranging multiple sets of characteristic data in order from small to large according to the shear tool holder vertical stroke displacement value to form the discrete characteristic nodes.

5. The method of claim 1, wherein, The predicted trajectory planner includes a dynamics prediction model of the shearing mechanism and a rolling optimization solver, the dynamics prediction model of the shearing mechanism establishes the correlation between the output torque of the hydraulic servo valve and the motion state of the shearing tool holder, and predicts the displacement response of the shearing tool holder; The rolling optimization solver takes the virtual control trajectory as a reference benchmark, and in each control cycle, optimizes the calculation to minimize the deviation between the predicted displacement response of the shearing tool holder and the virtual control trajectory, calculates the required feedforward torque compensation for balancing the material hardness, and synchronously outputs the corresponding blade gap adjustment instruction.

6. The method of claim 5, wherein the shear parameter adaptive control method is oriented to a multi-hardness material, and The specific process of generating the optimal blade gap and shear speed cooperative motion instruction sequence that adaptively matches the current hardness is as follows: A position closed-loop tracking controller is constructed to calculate the tracking error between the actual position of the shearing tool holder and the theoretical position set in the virtual control trajectory in real time, and a feedback adjustment amount is generated based on the tracking error; The feedforward torque compensation and the feedback adjustment amount are superimposed to synthesize the total control amount, and the total control amount is converted into a current control signal for driving the hydraulic servo valve; at the same time, the blade gap adjustment instruction is analyzed into a target position code, which is converted into a pulse control signal for driving the gap adjustment motor; The current control signal and the pulse control signal are output in the same control cycle, and the blade gap dynamic adjustment and the shearing tool holder variable speed cutting are cooperatively actuated.

7. A shear parameter adaptive control system for multi-hardness materials, characterized by, It includes: A fingerprint construction module: pre-pressing the material to be processed, applying a micro disturbance signal during pre-pressing, collecting the dynamic mechanical response of the material to be processed to the micro disturbance signal, extracting the complex impedance frequency spectrum features of the dynamic mechanical response by fast Fourier transform, and constructing the frequency domain fingerprint of the internal heterogeneous structure distribution of the material to be processed based on the complex impedance frequency spectrum features; A trajectory generation module: a virtual control generator is constructed to predict the load mutation time and peak intensity caused by material hardness mutation based on the frequency domain fingerprint; The virtual control generator adopts a generative mapping model based on inverse dynamics, and the specific construction process includes: establishing a nonlinear mapping relationship between the complex impedance modulus peak in the frequency domain fingerprint and the shear strength distribution of the material to be processed along the thickness direction; An inverse dynamics equation is constructed using the kinematic parameters of the shearing mechanism, the shear strength distribution is substituted into the inverse dynamics equation as a boundary condition, the theoretical resistance evolution curve required for the shearing tool to overcome the shear strength distribution under constant speed working condition is inversely deduced, and the extreme point and its corresponding time domain position are extracted from the theoretical resistance evolution curve as the peak intensity and load mutation time. The predicted load mutation time and peak intensity are converted into discrete characteristic nodes based on the vertical displacement of the shear blade holder, a spline interpolation algorithm is used to connect the discrete characteristic nodes, and a virtual control trajectory is generated; A feedforward compensation module: the virtual control trajectory is input into a predictive trajectory planner; a feedforward torque compensation amount required to overcome the hardness change resistance is calculated; A parameter control module: the feedforward torque compensation amount and a feedback adjustment amount based on real-time position error are superimposed to generate an optimal blade gap and shear speed cooperative motion instruction sequence that matches the current hardness, which drives the hydraulic servo valve and the gap adjustment motor to perform shear parameter control.

Images