An adaptive adjustment system for the machining process of a laser cutting rotary table

Abstract

This invention relates to the field of general control or adjustment system technology, and discloses an adaptive adjustment system for the machining process of a laser cutting rotary table. The system includes: an adjuster that synchronously acquires C-axis angle and displacement data before machining and performs harmonic analysis to identify the harmonic parameters of the workpiece; the adjuster performs quality control decisions based on higher-order harmonics and generates a feedforward compensation model using first- and second-order harmonic parameters; during machining, the model is queried, and Z-axis focus compensation signals and C-axis angular velocity compensation signals are output in real time. This invention enables the control system to actively identify the true geometric model of the workpiece before machining. Its feedforward compensation based on the true model avoids the lag of traditional feedback control and ensures the stability of the machining process.

Description

Technical Field

[0001] This invention relates to an adaptive adjustment system for the processing of a laser cutting rotary table, belonging to the field of general control or adjustment system technology. Background Technology

[0002] Currently, in the precision machining applications of laser cutting rotary tables, digital control systems are the technological foundation for achieving automation and high precision. The core function of such systems lies in trajectory tracking, which, based on a preset, idealized digital model, ensures through a servo closed-loop system that the physical actuator, such as the rotary table, precisely matches the motion state of that model. However, this control method lacks awareness of the machining process itself in its design. It assumes that the controlled object, i.e., the workpiece, is completely identical to the digital model. Therefore, it only guarantees the precise execution of motion but cannot predict or sense the process results caused by that motion. When this control method is applied to rotary cutting, the workpiece... Geometric deviations inherent in workpieces due to manufacturing tolerances or clamping, such as eccentricity and ellipticity, become a common disturbance, directly impacting the process results due to the lack of perceived process awareness. In this case, the faithful execution of ideal motion commands by the control system will lead to deviations in the process results due to the lack of process awareness. For example, a constant rotational angular velocity command applied to an elliptical workpiece will cause periodic fluctuations in the linear velocity of the workpiece surface, resulting in uncontrollable local energy density, such as overheating or incomplete cutting. At the same time, the eccentricity of the workpiece will also change the normal distance between its surface and the laser focus, causing focus drift.

[0003] Those skilled in the art have attempted to physically approximate the ideal model of a workpiece using high-precision physical fixtures and other methods. However, this approach is costly and cannot eliminate tolerances. It does not address the adjustment problem caused by the inconsistency between the digital model and the actual workpiece from the perspective of control logic. At the same time, most existing technological improvements focus on the mechanical structure of the tooling fixtures, neglecting the limitations of the control system itself. For example, Chinese invention patent CN120133774A discloses a plate and tube integrated laser cutting machine with an automatically adjusting tooling. This solution provides an adaptive component, which essentially still achieves automatic adjustment and clamping of different tubes through complex mechanical structures (such as electromagnetic slide rails and fixed block groups). This approach still attempts to passively adapt to the ideal shape of the workpiece through the tooling. It does not, and cannot, solve the dynamic geometric deviation problem caused by eccentricity and ellipticity during the rotation of the workpiece. Therefore, its control system still performs blind trajectory tracking based on an ideal model during processing, and cannot fundamentally eliminate the contradiction between the digital model and the actual workpiece.

[0004] Therefore, the technical problem to be solved by this invention is how to provide an adjustment system that can actively identify the real geometric model of the workpiece before processing and adaptively adjust its control commands based on the real model to reduce the gap between the digital model and the actual workpiece. Summary of the Invention

[0005] This invention provides an adaptive adjustment system for the processing of a laser cutting rotary table. Its main purpose is to solve the problem that existing control systems cannot actively identify the real geometric model of the workpiece before processing and adaptively adjust their control commands based on the real model, so as to reduce the gap between the digital model and the actual workpiece.

[0006] To achieve the above objectives, the present invention provides an adaptive adjustment system for the processing of a laser cutting rotary table. The system includes an adjuster electrically connected to the C-axis of the rotary table, a C-axis encoder, a displacement sensor, and a control interface between the rotary table and the laser cutting head. The adjuster is used to perform a system identification phase before executing the laser cutting processing adjustment phase. In the system identification phase, the adjuster executes the following rules:

[0007] Step 101: When controlling the C-axis rotation, simultaneously collect the real-time angle of the C-axis encoder and the real-time displacement data of the displacement sensor on the workpiece surface to obtain the real contour dataset of the workpiece.

[0008] Step 102: Perform a harmonic analysis operation on the actual contour dataset of the workpiece to identify a set of harmonic parameters;

[0009] Step 103: Extract the higher-order harmonic components from the harmonic parameters and compare the higher-order harmonic components with the preset workpiece quality acceptance threshold.

[0010] Step 104: When the higher-order harmonic components exceed the acceptable threshold for workpiece quality, terminate the processing adjustment stage.

[0011] Step 105: When the higher-order harmonic components are not greater than the acceptable threshold for workpiece quality, the amplitude and phase of the first-order harmonic in the harmonic parameters are extracted to be used as parameters characterizing workpiece eccentricity, and the amplitude and phase of the second-order harmonic in the harmonic parameters are extracted to be used as parameters characterizing workpiece ellipticity.

[0012] In the processing and adjustment stage: Step 106, based on the eccentricity parameter and ellipticity parameter, a feedforward compensation model related to the real-time angle of the C-axis is generated;

[0013] Step 107: During the laser cutting process, based on the real-time angle of the C-axis, query the feedforward compensation model, and calculate and output a Z-axis focus compensation signal and a C-axis angular velocity compensation signal to the control interface in real time.

[0014] Preferably, the displacement sensor is mounted on a micro-displacement actuator, and the system also includes a micro-force sensor for measuring the contact force between the displacement sensor and the workpiece; the regulator, during the system identification phase, is also used to initiate a force control closed loop, compare the contact force with a target contact force setpoint, and adjust the micro-displacement actuator in real time based on the comparison deviation to maintain a constant contact force, and the workpiece true contour dataset is obtained by synthesizing the displacement readings of the displacement sensor and the adjustment motion of the micro-displacement actuator.

[0015] Preferably, the system further includes a process status sensor for real-time acquisition of a process status feedback signal during the processing adjustment stage; the regulator is also used to compare the process status feedback signal with a process status setpoint to generate a residual error signal; based on the residual error signal, a feedback compensation adjustment signal is calculated through a feedback control algorithm, and the regulator is also used to: perform algebraic summation on the feedback compensation adjustment signal and the Z-axis focus compensation signal output in step 107 to generate a final Z-axis focus compensation signal and output it to the control interface.

[0016] Preferably, the regulator is also used to perform: step 401, extracting higher-order harmonic components from the harmonic parameters, and generating a high-frequency feedforward compensation model based on the higher-order harmonic components; step 402, during the machining adjustment stage, querying the high-frequency feedforward compensation model according to the real-time angle of the C-axis, and calculating and outputting a laser power compensation signal to the control interface in real time. The laser power compensation signal is used to compensate for the machining energy fluctuations caused by the workpiece geometric deviations corresponding to the higher-order harmonic components.

[0017] Preferably, during the system identification phase, the regulator is also used to perform: step 501, extracting the DC component from the harmonic analysis calculation result to determine the measured average radius of the workpiece; step 502, comparing the measured average radius with a nominal radius to calculate a global dimensional deviation; and step 503, before the start of the machining adjustment phase, automatically correcting the machining coordinate system or tool compensation parameters of the control interface based on the global dimensional deviation.

[0018] Preferably, the Z-axis focus compensation signal is the instantaneous radius change synthesized by the regulator based on the eccentricity parameter and the ellipticity parameter, which is used to adjust the laser focus in real time to match the actual radius of the workpiece.

[0019] Preferably, the C-axis angular velocity compensation signal is the instantaneous tangent slope change synthesized by the regulator based on the eccentricity parameter and ellipticity parameter, which is used to dynamically adjust the angular velocity of the C-axis to ensure that the linear velocity of the laser point on the true contour surface of the workpiece remains constant.

[0020] Preferably, the higher-order harmonic components include third-order harmonic components and fourth-order harmonic components. In step 103, the regulator determines whether to execute step 104 or step 105 by determining whether the amplitude of the third-order harmonic component or the fourth-order harmonic component is greater than the acceptable threshold for workpiece quality.

[0021] Preferably, the process status sensor is an infrared temperature sensor or a photodiode, and the process status feedback signal is a signal characterizing the radiation intensity of the cutting molten pool.

[0022] Preferably, the displacement sensor is a contact LVDT probe or a laser displacement sensor.

[0023] Compared with the prior art, the beneficial effects of the present invention are:

[0024] 1. By synchronously acquiring real-time angle data of the rotary table and real-time displacement data of the displacement sensor during the system identification stage before processing, and performing harmonic analysis on the spatiotemporally bound dataset, the control system gains the ability to actively identify the real geometric model of the workpiece. This method transforms the control system from a passive executor that relies on an ideal model into an adjuster that can pre-cognize the actual physical deviations (such as eccentricity and ellipticity) of the controlled object, i.e., the workpiece, providing a realistic model input for the adaptive adjustment of subsequent control commands.

[0025] 2. Based on the identified disturbance model parameters, a corresponding feedforward compensation model is generated before processing. During the processing adjustment, the system anticipates and actively outputs compensation adjustment signals by querying the feedforward model based on the current real-time angle of the rotating shaft. This feedforward control logic based on a known model avoids the system lag that is inevitably introduced by traditional feedback control after sensing deviations such as changes in the molten pool and then performing calculations and adjustments. From the perspective of control principle, it avoids adjustment errors caused by response delays.

[0026] 3. The control logic constructed in this invention can output two physically independent compensation adjustment signals in parallel and collaboratively during the processing, based on the same set of geometric disturbance model parameters extracted by the system identification, namely eccentricity and ellipticity parameters: a Z-axis position compensation signal for stabilizing the focus and a C-axis angular velocity compensation signal for maintaining a constant linear velocity. This multivariable feedforward adjustment mechanism enables the control system to simultaneously and actively compensate for the two core process disturbances caused by workpiece geometric deviations: focus drift and linear velocity fluctuation. This ensures that the energy density of the laser on the workpiece surface remains stable throughout the processing. The stability and consistency of this control are difficult to achieve by adjusting only a single variable. Attached Figure Description

[0027] Figure 1 This is a flowchart of the control logic of the adaptive adjustment system of the present invention;

[0028] Figure 2 A bar chart comparing the compensation effects of different control modes of the present invention;

[0029] Figure 3 This is a schematic diagram of the functional architecture of the adaptive adjustment system of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining this invention and are not intended to limit the scope of protection of this invention.

[0031] This invention provides an adaptive adjustment system for the machining process of a laser cutting rotary table. Its core is an adjuster that operates within a control architecture, replacing the trajectory tracking logic in traditional digital control systems. The adjuster's workflow is divided into two mandatory, sequentially executed stages: a system identification stage before the laser is turned on, and a machining adjustment stage during the laser cutting process. The control objective of the system identification stage is to actively acquire the true geometric disturbance model of the workpiece, while the control objective of the machining adjustment stage is to perform feedforward compensation based on this true model to eliminate the deviation between the ideal model and the actual workpiece. In a specific control system implementation… Currently, the regulator is an independent control unit or a functional module integrated into the CNC system. It is electrically connected to the C-axis servo drive, C-axis encoder, and displacement sensor of the rotary table, as well as the Z-axis position control interface and C-axis speed input interface of the laser cutting head. The displacement sensor can be a contact LVDT probe or a high-precision laser displacement sensor, which is installed to detect the radial displacement of the workpiece during rotation. Traditional control systems face the challenge of assuming an ideal controlled object model and being blind to the process of geometric deviations commonly found in the workpiece. To overcome this control limitation, this system performs a system identification stage... In step 101, the regulator controls the C-axis to rotate one revolution at a constant low angular velocity, for example, set to 5-10 rpm. During the rotation, the data acquisition module inside the regulator reads the real-time angle θ_i from the C-axis encoder and the real-time displacement data R_i from the displacement sensor at a high sampling rate, for example, set to 1 kHz. The regulator aligns these two data streams based on the same timestamp to obtain a spatiotemporally strongly correlated real contour dataset of the workpiece D={(θ_1,R_1),(θ_2,R_2),...,(θ_n,R_n)}, which constitutes the input of the subsequent system identification algorithm. In step 102, the regulator immediately performs a harmonic analysis operation on the displacement sequence R_i in the acquired dataset D. In a preferred control implementation, this operation is a Fast Fourier Transform (FFT). The FFT operation transforms the time-domain contour data R_i to the frequency domain and outputs a set of complex harmonic parameters H_k, where k is the harmonic order. This set of harmonic parameters H_k characterizes the geometric composition of the workpiece contour, where the DC component H_0 corresponds to the average radius of the workpiece, the first harmonic H_1 corresponds to eccentricity, the second harmonic H_2 corresponds to ellipticity, and higher-order harmonics with k≥3 correspond to more complex shape deviations such as triangular or square shapes of the workpiece.

[0032] Before the control system starts the processing flow, the incoming quality of the controlled object, i.e., the workpiece, must be assessed at the diagnostic level to avoid performing ineffective processing on workpieces that are already defective. Therefore, in step 103, the controller extracts the higher-order harmonic components from the harmonic parameters, specifically, the amplitudes of the third harmonic |H_3| and the fourth harmonic |H_4|. The controller compares these two amplitudes with a preset workpiece quality acceptance threshold Th_quality. Th_quality is a calibrable control parameter, and its determination procedure is as follows: by performing steps 101 and 102 on a batch of known qualified workpiece samples, the maximum fluctuation range of their |H_3| and |H_4| is statistically analyzed, and Th_quality is set based on this maximum range, for example, set to 1.2 times the maximum value of qualified samples. This threshold is stored in the controller. In step 104, when the controller determines that |H_3|>Th_quality or |H_4|>... When |>Th_quality, meaning the diagnostic system determines the workpiece is a triangular or square defective product, the regulator immediately terminates the subsequent processing adjustment stage and sends an alarm signal to the CNC system; in step 105, when the higher-order harmonic components are not greater than the workpiece quality acceptance threshold, i.e., the workpiece quality is qualified, the identification process continues to execute; at this time, the regulator extracts the amplitude |H_1| and phase ϕ_1 of the first-order harmonic H_1 identified in step 102, to use them as parameters characterizing the magnitude and direction of workpiece eccentricity; and extracts the amplitude |H_2| and phase ϕ_2 of the second-order harmonic H_2, to use them as parameters characterizing the magnitude and direction of the major and minor axes of the workpiece; after entering the processing adjustment stage, the regulator executes step 106; the challenge faced by the control system is how to use the identified eccentricity parameters, i.e., |H_1 The parameters |H_1 and ϕ_1, and the ellipticity parameters, i.e., |H_2| and ϕ_2, are converted into real-time control commands. To this end, the regulator generates a feedforward compensation model associated with the real-time C-axis angle based on these parameters. In a specific control implementation, this model is a compensation lookup table indexed by the C-axis angle θ, for example from 0° to 359.9° with a resolution of 0.1°. The lookup table pre-calculates and stores the two core feedforward manipulators corresponding to each angle θ_i: the instantaneous radius compensation R_comp (θ_i) and the instantaneous angular velocity factor V_factor (θ_i).

[0033] In step 106, a feedforward compensation model is generated. One specific engineering implementation procedure is as follows: The regulator uses the identified DC component H_0, the measured average radius, the first harmonic amplitude |H_1| and phase ϕ_1 of the eccentricity parameter, and the second harmonic amplitude |H_2| and phase ϕ_2 of the ellipticity parameter to reconstruct a mathematical model r(θ) of the instantaneous radius profile of a workpiece. This model uses the real-time C-axis angle θ as the only input variable, and its output instantaneous radius r(θ) is the algebraic superposition of the first and second harmonic functions with the DC component at the corresponding angle θ. The Z-axis focus compensation signal output by the regulator in step 107 is the reconstructed instantaneous radius r(θ) or the deviation Δr(θ) between r(θ) and a nominal radius. This signal is output to the Z-axis position control interface in real time. This is used to compensate for focus drift caused by eccentricity and ellipticity. In step 107, to generate the C-axis angular velocity compensation signal, the regulator further executes a constant linear velocity calculation procedure. The goal of this procedure is to keep the tangential linear velocity V_L of the laser point on the true contour surface of the workpiece constant. Specifically, the regulator obtains the instantaneous tangent slope r' at each C-axis angle based on the instantaneous radius contour model r(θ) reconstructed in the previous step through an internal real-time differential operation (e.g., calculating Δr / Δθ). (θ_i), and combined with the instantaneous radius r(θ_i), the arc length differential ds_i at that angle θ_i is calculated (i.e., the discretization calculation of ds_i=√(r(θ_i)_2+[r'(θ_i)]_2) dθ); finally, the C-axis angular velocity compensation signal MV_C is calculated as a velocity adjustment factor that is proportional to the target linear velocity V_L and inversely proportional to the instantaneous arc length differential ds_i or its corresponding instantaneous effective radius. This adjustment factor is output to the C-axis velocity setting interface to dynamically adjust the instantaneous angular velocity ω(θ) of the C-axis.

[0034] In step 107, the laser is turned on, the CNC system begins executing the cutting path, and the adjustment system of this invention runs in parallel. The regulator queries the feedforward compensation model in real time based on the current real-time angle θ_current fed back by the C-axis encoder. To solve the focus mismatch problem caused by workpiece geometric deviation, the regulator calculates and outputs the first compensation signal, namely the Z-axis focus compensation signal. This signal is the instantaneous radius change synthesized by the regulator based on the eccentricity parameter and ellipticity parameter. Its calculation method is either to use the formula MV_Z(θ_current) = R_comp (θ_current), where R_comp is the instantaneous radius value of H_1 and H_2 superimposed at the angle θ_current; this MV_Z signal is output to the Z-axis position control interface of the laser cutting head and superimposed on the Z-axis command of the CNC to ensure that the laser focus is always precisely fixed on the actual physical surface of the workpiece; at the same time, in order to solve the linear velocity mismatch problem caused by constant angular velocity acting on a non-circular contour, the regulator calculates and outputs a second compensation signal, namely the C-axis angular velocity compensation signal; this signal is calculated by the regulator based on the instantaneous tangential slope change synthesized by the eccentricity parameter and ellipticity parameter, and its calculation method is either by directly querying from the lookup table, MV_C (θ_current) = V_factor (θ_current); this V_factor is a speed adjustment factor used to ensure that the tangential linear velocity V_linear of the workpiece surface remains constant at any angle θ_current; this MV_C signal, for example as a speed percentage adjustment amount, is output to the C-axis speed setting interface of the rotary table to dynamically adjust the instantaneous angular velocity of the C-axis.

[0035] When the feedforward compensation model is deployed for the first time or when the workpiece type is switched, the regulator can also be set to execute an engineering procedure for model verification and gain tuning. This procedure includes: enabling the Z-axis focus compensation signal and the C-axis angular velocity compensation signal without turning on the laser, and calling an external high-precision laser displacement sensor, whose function is as described in Example 2, as a third-party measurement tool, to simultaneously collect the C-axis angle, the actual compensation residual error of the Z-axis, and the actual linear velocity fluctuation of the workpiece surface; the regulator will perform correlation analysis between the collected residual error and linear velocity fluctuation data and the harmonic parameters identified in step 102, and automatically or prompt the engineer to fine-tune the amplitude gain or phase lead of the feedforward compensation model accordingly. For example, if there is still a component in the Z-axis residual error that is in the same frequency as the C-axis rotation, the compensation amplitude of the first harmonic in MV_Z will be adjusted accordingly until the feedback error of the third-party measurement tool is suppressed within a preset engineering acceptance threshold, for example, within 3-5μm as shown in Example 2; in an enhanced implementation, to solve the dimensional error where the average size of the workpiece is inconsistent with the nominal size of the CNC program. To address the error, the regulator also performs step 501 after the harmonic analysis calculation is completed, which involves extracting the DC component H_0 from the harmonic analysis result. The value of H_0 is the measured average radius R_avg of the workpiece. In step 502, the regulator compares this R_avg with the nominal radius R_nominal stored in the CNC system and calculates a global dimensional deviation ΔR = R_avg - R_nominal. In step 503, before the machining adjustment stage begins, the regulator automatically writes this value into the global coordinate system offset register of the CNC system, such as G54 or the tool radius compensation register, thereby achieving adaptive calibration of the absolute dimension of the workpiece. In another enhanced implementation, to compensate for high-frequency disturbances caused by high-order geometric deviations such as triangular and square shapes, which the mechanical servo system's Z-axis and C-axis cannot track due to insufficient control bandwidth, the regulator also extracts the high-order harmonic components, H_3, H_4, ..., from the harmonic parameters in step 401, and generates a high-frequency feedforward compensation model R_(high_freq) based on these components. (θ); Since the mechanical axis cannot respond, the regulator maps this high-frequency disturbance to the high-bandwidth energy actuator, i.e., the laser power; In step 402, during the processing adjustment stage, the regulator queries the high-frequency model R_(high_freq) (θ_current) based on the real-time C-axis angle θ_current, and calculates and outputs a laser power compensation signal MV_Power in real time; The MV_Power signal is used to instantaneously reduce the power at high-order peaks, such as corners, and instantaneously increase the power at troughs, thereby actively compensating for the high-frequency fluctuations in processing energy caused by high-order geometric deviations.

[0036] To address the residual errors caused by unmodeled disturbances dynamically generated during processing, such as thermal deformation, that cannot be covered by the feedforward model, this system may also include a feedforward-feedback hybrid control architecture. In this implementation, the system further includes a process status sensor, which can be a non-contact infrared temperature sensor or a photodiode, used to align with the vicinity of the molten pool and acquire a process status feedback signal PV_feedback characterizing the radiation intensity of the molten pool in real time. The controller further compares this PV_feedback with a process status setpoint SP_feedback characterizing the ideal cutting state, and generates a residual error signal E_residual=SP_feedback in real time. -PV_feedback; Based on E_residual, the regulator calculates a feedback compensation regulation signal MV_feedback using a feedback control algorithm, such as a low-gain PID controller. Finally, the regulator algebraically sums this MV_feedback with the Z-axis focus compensation signal MV_(Z_feedforward) output based on the feedforward model in step 107, i.e., MV_(Z_final) = MV_(Z_feedforward) + MV_(Z_feedback), and outputs this synthesized final Z-axis focus compensation signal to the control interface, thereby suppressing unmodeled dynamics of the system in real time while performing feedforward compensation. To address the observer effect problem—where the measured contact force of a contact displacement sensor can cause elastic deformation of the workpiece and contaminate the identification model—in high-precision identification scenarios involving easily deformable, thin-walled workpieces, this system employs a force-controlled closed-loop identification scheme. In this implementation, a displacement sensor, such as an LVDT probe, is mounted on a micro-displacement actuator, such as a piezoelectric ceramic PZT, which is responsible for the micrometer-level feed Z_probe of the probe along the Z-axis. The system also includes a micro-force sensor for real-time measurement of the contact force PV_Force between the probe and the workpiece. During the system identification phase, i.e., step 101, the regulator initiates a force-controlled closed loop in parallel: the measured contact force PV_Force... The contact force is compared with a target contact force setpoint SP_Force close to zero, for example, 0.005N, and based on the comparison deviation E=SP_Force-PV_Force, the Z_probe axis position of the micro-displacement actuator is adjusted in real time through a force-controlled PID algorithm to maintain a constant contact force. In this force-controlled floating tracking process, the true contour of the workpiece is obtained through synthesis, that is, the displacement data R_i in the true contour dataset of the workpiece is reconstructed as: R_i=R_(LVDT_i) (displacement reading of LVDT sensor)+Z_(probe_i) adjustment motion of micro-displacement actuator. The regulator then performs harmonic analysis in step 102 on this reconstructed, uncontaminated, high-fidelity dataset.

[0037] Example 1: In a specific high-requirement processing scenario, a laser cutting rotary table is used to precision process a batch of thin-walled metal tubes with a nominal outer diameter of 3.0 mm and a wall thickness of 0.05 mm. The operating conditions of the control system for this type of workpiece are affected by two main disturbance factors: First, the inherent eccentricity and ellipticity tolerances of the workpiece, combined with the elastic deformation introduced during the clamping process, form an unknown static geometric disturbance; Second, due to the extremely thin wall thickness, the high energy injected during laser processing will trigger dynamic-thermal deformation disturbances. The operator loads a thin-walled tube onto the C-axis chuck of the rotary table. After receiving the start command, the control system executes the control logic of the system identification stage. For this type of thin-walled, easily deformable workpiece, the regulator activates a force-controlled closed-loop identification scheme, controlling a micro-displacement actuator equipped with a micro-force sensor and a displacement sensor to bring the displacement sensor probe close to the workpiece surface. The force control closed loop is activated, comparing the measured contact force PV_Force with a target contact force setpoint SP_Force of 0.005N. The force control PID algorithm adjusts the Z_probe axis position of the micro-displacement actuator in real time to maintain a constant contact force. Subsequently, the regulator controls the C-axis to rotate at a constant low angular velocity of 8 rpm for one revolution. During this process, the force control closed loop continues to run, and the Z_probe axis of the micro-displacement actuator dynamically moves forward and backward with the contour of the workpiece. The regulator simultaneously collects the real-time angle of the C-axis encoder, the displacement reading R_(LVDT_i) of the displacement sensor, and the adjustment motion Z_(probe_i) of the micro-displacement actuator. Based on the synthesis rule of R_i=R_(LVDT_i)+Z_(probe_i), a dataset of the true contour of the workpiece that is not contaminated by the measured force is generated.

[0038] The regulator performs harmonic analysis on the dataset, and the analysis results enter the quality control decision process. The regulator compares the higher-order harmonic components |H_3| and |H_4| with the preset workpiece quality acceptance threshold. In this scenario, the judgment result is qualified, and the identification process continues. The regulator performs dimensional adaptive calibration, extracts the DC component H_0 to determine the measured average radius R_avg of the workpiece as 1.508mm, and compares it with the nominal radius R_nominal in the CNC program, which is 1.500mm, to calculate the global dimensional deviation ΔR = +0.0.08mm; The regulator automatically writes this ΔR value into the machining coordinate system offset register of the CNC system, completing the absolute dimension calibration of the machining datum; The regulator extracts the eccentricity parameters |H_1|, ϕ_1 and the ellipticity parameters |H_2|, ϕ_2, and generates a feedforward compensation model based on these parameters that have been calibrated on the absolute dimension datum. The control system begins to execute the control logic of the machining adjustment stage, and the laser is turned on; At this time, the static geometric disturbances and global dimension deviations that the control system needs to deal with have been modeled by the identification system; The regulator queries the feedforward compensation model based on the real-time C-axis angle θ_current and outputs two compensation adjustment signals: one is the Z-axis focus compensation signal MV_(Z_feedforward), which is superimposed on the Z-axis coordinate system that has been corrected by ΔR, to compensate for the instantaneous radius change caused by |H_1| and |H_2|, so that the focus follows the workpiece surface; the other is the C-axis angular velocity compensation signal MV_C, which dynamically adjusts the C-axis angular velocity to compensate for the change caused by |H_2|. The resulting change in tangent slope keeps the linear velocity of the laser action point on the workpiece surface constant. During processing, the control system needs to suppress the second type of disturbance, namely the dynamic-thermal deformation caused by the heating of the thin-walled tube. This unmodeled disturbance causes the feedforward model MV_(Z_feedforward) to fail to cover the slow Z-axis drift on the workpiece surface. At this time, the feedforward-feedback hybrid control architecture in the regulator starts to operate. The infrared temperature sensor installed near the cutting head acts as a process status sensor, acquiring the PV_feedback signal in real time. This signal deviates from the process status setpoint SP_feedback due to the focus drift caused by thermal deformation, generating a residual error signal E_residu. The feedback control algorithm (PID) within the regulator calculates a feedback compensation regulation signal MV_feedback based on E_residual. The final output Z-axis focus compensation signal of the regulator is the algebraic sum of two control components, i.e., MV_(Z_final) = MV_(Z_feedforward) + MV_(Z_feedback). This hybrid control architecture achieves separation of control bandwidth in the control logic: the feedforward loop (MV_(Z_feedforward)) is used to compensate for identified high-gain geometric disturbances, while the feedback loop (MV_feedback) is used in a low-gain manner to suppress unmodeled, slowly changing thermal disturbances.

[0039] Example 2: This example objectively verifies the effectiveness of the adaptive adjustment system of the present invention in suppressing two types of control disturbances caused by workpiece geometric deviations: focus drift and linear velocity fluctuation, compared to traditional control methods. The test platform is based on a laser cutting rotary table, whose control system can switch between three operating modes. The C-axis servo system of the test platform has an angular resolution of 0.001°, and the Z-axis servo system has a position resolution of 1μm. To simulate the adaptability of the control system under different disturbance intensities, three sets of test samples with different geometric deviation characteristics were prepared, all with a nominal diameter of 10.0mm: Sample A is the reference sample, with eccentricity and ellipticity both less than 5μm; Sample B is the biased sample... The sample C is an elliptic sample with an eccentric first harmonic H_1 amplitude of 0.082 mm and an ellipticity of less than 5 μm; the sample C is an elliptic sample with an eccentricity of less than 5 μm and an ellipticity second harmonic H_2 amplitude of 0.065 mm. Three control modes were set up for comparison: Control group 1 (trajectory tracking): using standard CNC control, the controller only executes motion based on an ideal 10.0 mm diameter circular digital model, blindly ignoring the actual geometric deviations of the workpiece; Control group 2 (single-variable feedback): using a simulated Z-axis servo feedback system, which can measure and feedback the Z-axis focus deviation in real time, but its control logic is unaware of the C-axis angular velocity and cannot compensate for linear velocity fluctuations; the present invention's sample group is adaptive. Control measures were implemented using the adaptive adjustment system of this invention. Before processing, a system identification phase was performed, and a dual-channel feedforward compensation model based on H_1 and H_2 parameters for Z-axis focus compensation and C-axis angular velocity compensation was activated. In the experiment, three control modes were used to process three sets of samples. The nominal C-axis angular velocity set by the CNC system was used to generate a constant linear velocity of 10.0 mm / s over a 10.0 mm diameter. An external high-precision laser displacement sensor (sampling frequency 10 kHz) was used as a third-party measurement tool to monitor in real time the deviation of the normal distance between the laser focus and the actual surface of the workpiece (i.e., Z-axis focus compensation error) during processing. Simultaneously, the real-time C-axis angular velocity was recorded to calculate the actual surface velocity of the workpiece. The fluctuation rate of linear velocity (i.e., C-axis velocity error) is analyzed as follows: Test data from sample A shows that when the workpiece geometric deviation is extremely small, both the Z-axis focus error and C-axis velocity error in the three control modes are at low levels. Test data from sample B (eccentricity) shows that the Z-axis focus error of control group 1 (conventional CNC trajectory tracking) at 82.5 μm is consistent with the workpiece eccentricity amplitude of 0.082 mm, indicating that its control system lacks compensation capability. Control group 2's Z-axis feedback can suppress the Z-axis focus error to 9.2 μm, but is still limited by the response delay of the feedback loop. The sample group of this invention obtains H1 = 0.082 mm through identification, and its feedforward compensation reduces the Z-axis focus error to 3.1μm; Under this condition, due to the small ellipticity, the C-axis velocity error in all three modes is very low; the experimental data on the C-axis ellipticity of the sample highlights the differences in control logic; Control group 1 (conventional CNC trajectory tracking) produces huge errors of 65.4μm and 12.8% in both the Z-axis and C-axis control dimensions; Control group 2's Z-axis feedback control logic is only for the Z-axis, thus suppressing the Z-axis focus error to 8.7μm, but its control system is blind to the geometric feature of ellipticity and its influence on linear velocity, resulting in a C-axis velocity error of 12.7%, almost the same as Control group 1, and the control objective fails; In the system identification stage of this invention, as shown in Table 1, H1=0.004mm and H2=0.065mm were identified, and the regulator generated a feedforward compensation model based on this real geometric model; In the machining adjustment stage, this model is called by the control system, and the Z-axis focus compensation signal is output in parallel to compensate for the projection of H1 and H2 on the Z-axis and the C-axis angular velocity compensation signal is output to compensate for the influence of H2 on the linear velocity.

[0040] Example 3: The test platform is the same as in Example 2. A test sample D with high-order harmonic deviation is selected, with a nominal diameter of 10.0 mm. The identification stage of the present invention is used to identify sample D, obtain its true contour dataset, and perform harmonic analysis. The identification results are: the amplitude of the first harmonic H1 is less than 0.005 mm, the amplitude of the second harmonic H2 is less than 0.005 mm, but the amplitude of the fourth harmonic H4 reaches 0.015 mm. Based on this same identification result, two control modes are set up for comparative testing. The control group 3 adopts a mechanical full compensation method. Its regulator generates a full-band feedforward compensation model based on all identified harmonic parameters (H1, H2, H4), and only uses the Z-axis focus compensation signal and C The angular velocity compensation signal of the axis is used to perform pure mechanical servo compensation for all geometric deviations. The sample of this invention adopts the control bandwidth separation logic of the specific implementation method. Its regulator splits the identification results and generates a first feedforward compensation model based on the low-order harmonic parameters H1 and H2. It is used to calculate and output MV_Z and MV_C to the mechanical servo interface. At the same time, it generates a second feedforward compensation model based on the high-order harmonic parameter H4. It is used to calculate and output a laser power compensation signal MV_Power to the laser energy control interface in real time. During the processing and adjustment stage, the two control modes process the sample D respectively. The nominal angular velocity of the C-axis is the same as in Example 2. A high-resolution profilometer is used to measure the micro-contour of the cutting edge and calculate the residual tracking error of the contour.

[0041] Control group 3 attempted to use Z-axis and C-axis servo systems to track the 15m amplitude of the H4 disturbance, which is 4 times the C-axis rotation frequency. Due to the physical inertia and limited control bandwidth of the servo system, its mechanical actuator could not effectively respond to this high-frequency command, resulting in a residual tracking error of up to 12.6μm for H4, with almost no effective compensation. Furthermore, the servo system exhibited observable high-frequency jitter during operation due to its attempt to track the high-frequency command. The control logic of the sample group in this invention avoids the physical limitation of mechanical bandwidth, allocates low-frequency disturbances (H1, H2) to the mechanical actuators (MV_Z, MV_C) for compensation, and simultaneously remaps the high-frequency disturbance (H4) to the laser power MV_Power of the energy actuator for compensation.

[0042] Example 4: This example combines Figures 1 to 3 This describes an adaptive adjustment system for the machining process of a laser cutting rotary table, such as... Figure 1 As shown, the process begins with system identification. Real-time angle data from the C-axis encoder and real-time displacement data from the displacement sensor are simultaneously collected to construct a dataset of the workpiece's true contour. The controller performs harmonic analysis on this dataset, generating a set of harmonic parameters. This is followed by quality control decision-making, where higher-order harmonic components are compared to a preset workpiece quality acceptance threshold. If the higher-order harmonic components exceed the threshold, the workpiece is deemed unqualified and the adjustment is terminated. If the higher-order harmonic components do not exceed the threshold, the process proceeds to the key parameter extraction step, extracting the first-order harmonic representing eccentricity and the second-order harmonic representing ellipticity. Based on the extracted eccentricity and ellipticity parameters, the controller generates a feedforward compensation model. During machining, the system adjusts the C-axis contour according to the actual... The model is queried and feedforward calculations are performed to output real-time Z-axis focus compensation signals and C-axis angular velocity compensation signals. These signals are used to adjust the Z-axis focus of the laser cutting head and the C-axis angular velocity of the rotary table, respectively. Simultaneously, a parallel feedback control loop can be activated. Process status feedback signals are provided by process status sensors such as infrared temperature sensors. These signals are compared with the given process status values ​​to generate residual error signals. Feedback control algorithms such as PID generate feedback compensation adjustment signals based on these signals. Finally, the feedback compensation adjustment signals and the feedforward Z-axis focus compensation signals are algebraically summed to generate the final Z-axis focus compensation signal, which together acts on the Z-axis adjustment of the laser cutting head.

[0043] like Figure 2As shown in the figure, the straight bar chart corresponds to the left vertical axis, representing the Z-axis focus compensation error in μm, and the diagonal bar chart corresponds to the right vertical axis, representing the C-axis speed error in %. The experimental results show that, due to the blindness to workpiece geometric deviations, the control group 1 (conventional CNC) has the highest Z-axis focus compensation error of 82.5 μm, and its C-axis speed error is also at a relatively high level of 0.8%. Although the Z-axis feedback of control group 2 can suppress the Z-axis focus compensation error to a lower level of 9.2 μm, its control logic cannot compensate for linear speed fluctuations, resulting in the C-axis speed error remaining the same as that of control group 1 at 0.8%. The sample group of this invention... Figure 1 The system identification and dual-channel feedforward compensation logic suppresses the Z-axis focus compensation error to a minimum level of 3.1μm, while reducing the C-axis velocity error by 0.3%; Figure 3 As shown, the system uses a regulator as the core control unit. The regulator integrates a data acquisition and synchronization module, a system identification module, a feedforward compensation module, and an optional feedback control module. During the system identification phase, the regulator connects to a C-axis encoder through the data acquisition and synchronization module to obtain angle signals, and connects to a displacement sensor to obtain workpiece displacement data. During the processing adjustment phase, the regulator outputs a Z-axis compensation signal to the Z-axis of the laser cutting head and a C-axis speed compensation signal to the C-axis of the rotary table through the feedforward compensation module.

[0044] Example 5: This example illustrates a standardized engineering calibration procedure for determining the acceptable quality threshold Th_quality of a workpiece. The control objective of this parameter calibration is to provide a reproducible and statistically based judgment benchmark for the diagnostic quality control decision logic in step 103, in order to distinguish between qualified workpieces and defective workpieces with high-order shape deviations. The initial state of the calibration process is defined as follows: The test environment is a laser cutting rotary table equipped with the adaptive adjustment system of this invention, and its system identification function, i.e., steps 101 and 102, has been debugged. The test objects include two groups of workpiece samples: the first group of grid samples contains 50 high-precision... The first set of workpieces was independently calibrated and confirmed by a coordinate measuring machine (CMM) to have third and fourth harmonic amplitudes less than 3.5 μm. The second set of defective samples consisted of 10 workpieces whose third or fourth harmonic amplitudes were confirmed to be between 10 μm and 20 μm by the CMM. The calibration procedure followed these control steps: The controller performed step 101 of the system identification phase on each of the 60 workpiece samples, i.e., controlling the C-axis rotation and synchronously acquiring angle and displacement data, and step 102, i.e., performing harmonic analysis on the actual contour dataset of the workpiece. Then, for the harmonic analysis results of each sample, the controller automatically extracted the third harmonic amplitude |H_3| and the fourth harmonic amplitude |H_4|. The larger of these two values ​​is recorded as the higher-order shape deviation E_higher for that sample. After identifying all samples, the regulator performs statistical analysis on the two sets of data. The results show that the mean of the 50 E_higher values ​​in the first group (qualified sample group) is 2.1 μm, the standard deviation is 0.4 μm, and none of the observations exceed 3.2 μm, which is recorded as Max_pass=3.2 μm. The minimum of the 10 E_higher values ​​in the second group (defective sample group) is 10.8 μm, which is recorded as Min_fail=10.8 μm. Finally, the regulator executes the threshold setting logic, which is used to balance the false positive rate, i.e., classifying qualified samples as defective. The defective product and the failure rate are used to determine whether a defective product is qualified. In this procedure, the logic is set as Th_quality = Max_pass × K, where K is a safety factor greater than 1.0, set to 1.3. Based on this, Th_quality = 3.2μm × 1.3 = 4.16μm. The regulator rounds the calculated value Th_quality to 4.2μm, solidifies and stores it as its internal workpiece quality acceptance threshold parameter, and uses it for diagnostic quality control comparison in step 103 of subsequent CNC batch processing. This parameter 4.16μm or 4.2μm is higher than the maximum deviation of 3.2μm for all qualified samples and lower than the minimum deviation of 10.8μm for all defective samples.

[0045] Example 6: This example is another alternative implementation of the adaptive control system. Its main difference from Examples 1 to 3 and 5 is that in step 101, a non-contact laser displacement sensor is used to adapt to certain workpiece materials or processing control environments that are not suitable for contact. In this control system, the controller faces a specific control challenge that varies depending on the working principle of the sensor: the measurement readings of the laser displacement sensor are easily affected by factors such as workpiece surface roughness, tilt angle changes, and uneven surface reflectivity of the material, resulting in high-frequency measurement noise being mixed into the collected true contour data of the workpiece. If the controller performs the harmonic analysis operation in step 102 on this noisy data without processing, the measurement noise will be amplified in the frequency domain, especially contaminating the identification accuracy of higher-order harmonic components (H_3H_4...) in steps 103 and 401, thereby interfering with the accuracy of diagnostic quality control decisions or hybrid actuator compensation.

[0046] To address this control challenge, in this embodiment, before executing the harmonic analysis calculation in step 102, the regulator is forced by the control logic to insert a data preprocessing sub-step. The control logic of this sub-step is as follows: the regulator applies a digital low-pass filter to the original workpiece true contour dataset R_i acquired in step 101. This filter can be a Butterworth filter with a set cutoff frequency or a Chebyshev filter. Its control objective is to filter out measurement noise that is physically much higher than the highest-order geometric deviation frequency of the workpiece (such as H4). The regulator then sends this noise-reduced contour dataset to the harmonic analysis calculation module in step 102. This data preprocessing step is a built-in control logic of the control system to adapt to the physical characteristics of the laser displacement sensor, a specific identification tool, and is used to ensure the control effectiveness of the system in the identification stage under different sensor configurations.

[0047] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A system for adaptive adjustment of a machining process of a laser-cut rotary table, characterized in that, The system comprises a regulator electrically connected to a C-axis of a rotary table, a C-axis encoder of the C-axis, a displacement sensor, and a control interface of the rotary table and a laser cutting head, the regulator being configured to perform a system identification phase before a processing adjustment phase for performing laser cutting, in which the regulator performs the following rules: In step 101, the real-time angle of the C-axis encoder and the real-time displacement data of the displacement sensor on the workpiece surface are synchronously collected during the rotation of the C-axis to obtain a workpiece real profile data set; In step 102, a harmonic analysis operation is performed on the workpiece real profile data set to identify a set of harmonic parameters; In step 103, the high-order harmonic component in the harmonic parameters is extracted, and the high-order harmonic component is compared with a preset workpiece quality acceptance threshold value; In step 104, when the high-order harmonic component is greater than the workpiece quality acceptance threshold value, the processing adjustment phase is terminated; In step 105, when the high-order harmonic component is not greater than the workpiece quality acceptance threshold value, the amplitude and phase of the first-order harmonic in the harmonic parameters are extracted as parameters representing the eccentricity of the workpiece, and the amplitude and phase of the second-order harmonic in the harmonic parameters are extracted as parameters representing the ovality of the workpiece; In the processing adjustment phase: in step 106, a feedforward compensation model associated with the real-time angle of the C-axis is generated based on the eccentricity parameters and the ovality parameters; In step 107, during the laser cutting process, the feedforward compensation model is queried according to the real-time angle of the C-axis, and a Z-axis focal point compensation signal and a C-axis angular velocity compensation signal are calculated and output to the control interface in real time.

2. The adaptive adjustment system for a laser cutting process of a rotating table according to claim 1, wherein, The displacement sensor is installed on a micro-displacement actuator, and the system further comprises a micro-force sensor for measuring the contact force between the displacement sensor and the workpiece; the regulator is further configured to start a force control closed loop in the system identification phase, compare the contact force with a target contact force given value, and adjust the micro-displacement actuator in real time based on the comparison deviation to maintain the contact force constant, and the workpiece real profile data set is obtained by synthesizing the displacement reading of the displacement sensor and the adjustment motion amount of the micro-displacement actuator.

3. The adaptive adjustment system for laser cutting of a rotating table according to claim 1, characterized in that, The system further comprises a process state sensor for acquiring a process state feedback signal in real time during the processing adjustment phase; the regulator is further configured to compare the process state feedback signal with a process state given value to generate a residual error signal; Based on the residual error signal, a feedback compensation adjustment signal is calculated through a feedback control algorithm, and the regulator is further configured to: algebraically sum the feedback compensation adjustment signal and the Z-axis focal point compensation signal output in step 107 to generate a final Z-axis focal point compensation signal and output to the control interface.

4. The adaptive process adjustment system for laser cutting of a rotating worktable according to claim 1, wherein, The regulator is further configured to perform: step 401, extracting a high-order harmonic component in the harmonic parameter, and generating a high-frequency feedforward compensation model based on the high-order harmonic component; and step 402, in the machining adjustment stage, querying the high-frequency feedforward compensation model according to the real-time angle of the C-axis, and calculating and outputting a laser power compensation signal to the control interface in real time, the laser power compensation signal being used to compensate for the machining energy fluctuation caused by the workpiece geometric deviation corresponding to the high-order harmonic component.

5. The adaptive adjustment system for laser cutting of a rotating worktable according to claim 1, wherein, In the system identification stage, the regulator is further configured to perform: step 501, extracting a DC component in the harmonic analysis operation result to determine a measured average radius of the workpiece; step 502, comparing the measured average radius with a nominal radius to calculate a global dimensional deviation; and step 503, before the machining adjustment stage starts, automatically correcting the machining coordinate system or the tool compensation parameter of the control interface based on the global dimensional deviation.

6. The adaptive process adjustment system for laser cutting of a rotating worktable of claim 1, wherein, The Z-axis focal point compensation signal is a change in instantaneous radius synthesized by the regulator based on the eccentricity parameter and the ellipticity parameter, and is used to adjust the laser focal point in real time to match the real radius of the workpiece.

7. The adaptive process adjustment system for laser cutting of a rotating worktable according to claim 1, wherein, The C-axis angular velocity compensation signal is a change in instantaneous tangent slope synthesized by the regulator based on the eccentricity parameter and the ellipticity parameter, and is used to dynamically adjust the angular velocity of the C-axis to ensure that the linear velocity of the laser spot on the real profile surface of the workpiece remains constant.

8. The adaptive process adjustment system for laser cutting of a rotating worktable according to claim 1, wherein, The high-order harmonic component includes a third-order harmonic component and a fourth-order harmonic component, and in step 103, the regulator determines whether to perform step 104 or step 105 by determining whether the amplitude of the third-order harmonic component or the fourth-order harmonic component is greater than a workpiece quality acceptance threshold.

9. The adaptive adjustment system for a laser cutting process of a rotating table according to claim 3, wherein, The process state sensor is an infrared temperature measurement sensor or a photodiode, and the process state feedback signal is a signal representing the radiation intensity of the cutting molten pool.

10. The adaptive process adjustment system for laser cutting of a rotating worktable of claim 1, wherein, The displacement sensor is a contact LVDT probe or a laser displacement sensor.

Images