Intelligent Monitoring Method and System for Machining of Precision Automotive Parts
By superimposing sinusoidal feedforward disturbances on the machine tool interpolation operation layer, the cutting signal components are extracted, the dynamic stiffness transfer coefficient is calculated, and a pure wear characteristic sequence is generated. This solves the misjudgment problem in long-stroke cutting monitoring and improves the accuracy and efficiency of precision machining.
Patent Information
- Application Number
- CN202611594725.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-15
- Publication Date
- 2026-11-17
AI Technical Summary
Existing long-stroke cutting monitoring cannot distinguish between gradual changes in workpiece dynamic stiffness and signal offsets caused by tool wear, which can easily lead to misjudgments and affect machining accuracy and efficiency.
Micrometer-level sinusoidal feedforward perturbation is superimposed on the machine tool interpolation operation layer. The cutting fundamental frequency and perturbation frequency response components are extracted through the phase-locked channel. The dynamic stiffness transfer coefficient is calculated to generate a pure wear feature sequence. Combined with multi-cycle trend verification and phase distortion detection, tool wear and local anomalies are identified.
It enables accurate differentiation between tool wear and workpiece stiffness changes, reduces misjudgments, ensures the monitoring accuracy of long-stroke precision machining, and reduces tool replacement losses.
Smart Images

Figure CN123103446A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of machining monitoring technology, specifically to an intelligent monitoring method and system for machining precision automotive parts. Background Technology
[0002] Long shaft components such as drive shafts and half-shafts in automotive precision parts are often formed segment by segment using external cylindrical cutting processes. These processes involve long machining strokes and stringent requirements for dimensional accuracy and surface quality. Online intelligent monitoring technology, by collecting multi-source sensor signals such as vibration, spindle current, and acoustic emission during the cutting process, and combining them with algorithmic models to identify tool wear and machining anomalies, has become an important technical means to control quality risks and optimize tool usage costs in mass production. Industry-wide technical optimizations largely revolve around feature extraction algorithms and judgment models for sensor signals, enhancing the identification effect of tool status by improving feature recognition and model accuracy.
[0003] In actual long-stroke cutting processes, as the tool feeds along the workpiece axis, the effective overhang length of the workpiece continuously changes with the cutting position. This corresponds to a gradual change in the dynamic structural stiffness of the workpiece, and the vibration response amplitude of the workpiece under cutting force also gradually changes with the stiffness. Existing monitoring solutions do not establish a signal correction mechanism for the gradual effect of workpiece dynamic stiffness changes. The signal change trend caused by stiffness changes highly overlaps with the signal change trend caused by normal tool wear, making it impossible for the judgment model to effectively distinguish between the two types of signal shifts. Current industry correction methods mostly use pre-constructed stiffness theory models. In actual production, clamping deviations and blank dispersion often lead to model mismatches, and the change patterns of the two factors are highly similar in amplitude dimension, making it impossible to achieve a traceable and complete separation at the mechanistic level. The closer the machining is to the end of the workpiece overhang, the more obvious the signal rise caused by the decrease in stiffness becomes, which is easily misjudged as increased tool wear. This can lead to unnecessary premature tool replacement and may also mask real local tool anomalies due to signal base shift, making it difficult to ensure the reliability of monitoring throughout the entire long-stroke precision machining process. Summary of the Invention
[0004] To address the shortcomings of existing technologies, this invention provides an intelligent monitoring method and system for the machining of precision automotive parts, which solves the problems of existing long-stroke cutting monitoring being unable to distinguish between the gradual change in workpiece dynamic stiffness and the signal offset caused by tool wear, and being prone to misjudgment.
[0005] To achieve the above objectives, the present invention provides a method for intelligent monitoring of precision machining of automotive parts, comprising the following steps:
[0006] S1. A sinusoidal feedforward disturbance is superimposed on the original interpolation command of the machine tool interpolation calculation layer to generate an interpolation command with disturbance and output to the servo system, and a disturbance reference signal is output synchronously; the frequency of the sinusoidal feedforward disturbance is a non-integer multiple of the cutting fundamental frequency, and the amplitude is on the order of micrometers.
[0007] S2. Relying on the machine tool's global synchronous clock, the cutting fundamental frequency component and the disturbance frequency response component are extracted through two independent phase-locked channels to obtain two time-aligned component signals.
[0008] S3. Combining the excitation amplitude parameters of the feedforward disturbance with the amplitude data of the disturbance response components, calculate the dynamic stiffness transfer coefficient of the current cutting position on an interpolation cycle, and generate a real-time stiffness sequence that changes synchronously with the feed position.
[0009] S4. Based on the real-time stiffness sequence, multiply the amplitude of the cutting fundamental frequency component with the dynamic stiffness transfer coefficient at the corresponding position to calculate the cutting force equivalent sequence and generate the pure wear characteristic sequence.
[0010] S5. Perform multi-period trend verification on the pure wear feature sequence and output the wear judgment result; at the same time, monitor the phase change of the disturbance response component, identify the phase distortion point and output the local anomaly judgment result.
[0011] Furthermore, step S1 specifically includes:
[0012] S11. Based on the perturbation parameters corresponding to the processing procedure and workpiece type, generate a sinusoidal perturbation sequence according to the preset perturbation frequency and amplitude;
[0013] S12. The sinusoidal perturbation sequence is superimposed on the position command of the feed axis to generate a perturbation interpolation command.
[0014] Furthermore, step S2 specifically includes:
[0015] S21. The first phase-locked channel uses the cutting fundamental frequency as the reference frequency to extract the fundamental frequency amplitude and phase data of the cutting vibration;
[0016] S22. The second phase-locked channel uses the frequency of the disturbance reference signal as the reference frequency to filter out the interference of cutting signals and environmental noise, and extract the vibration response component excited by the feedforward disturbance;
[0017] S23. The two phase-locked loop channels share the machine tool's global clock, maintaining synchronization accuracy consistent with the interpolation period, and outputting two component signals with synchronized timing.
[0018] Furthermore, in step S3, the dynamic stiffness transfer coefficient is calculated based on the structural dynamics transfer relationship, from the excitation amplitude of the feedforward disturbance and the disturbance response amplitude at the corresponding time. A stiffness calculation is completed once in each interpolation cycle, generating a continuous real-time stiffness sequence, which is updated synchronously with the feed position.
[0019] Furthermore, step S4 specifically includes:
[0020] S41. Retrieve the pre-stored stiffness-cutting force calibration mapping relationship. The mapping relationship is generated by fitting the calibration data under the new tool state and reflects the corresponding benchmark of vibration amplitude and cutting force under different stiffness levels.
[0021] S42. Substitute the currently measured dynamic stiffness transfer coefficient and the amplitude of the cutting fundamental frequency component into the mapping relationship to calculate the equivalent value of the cutting force at the current position;
[0022] S43. Arrange all cutting force equivalent values in the order of feed position to form a pure wear characteristic sequence.
[0023] Furthermore, in step S5, the multi-cycle trend verification adopts a sliding time window logic, taking the feature data of multiple consecutive interpolation cycles as the verification window, eliminating abnormal data that quickly falls back after a single point jump, and evaluating the wear level based on the continuously changing feature data; combining historical wear data to classify the verified feature quantities into levels, recording the rate of change of feature quantities, and marking abnormal wear states where the rate of change exceeds the preset range.
[0024] Furthermore, in step S5, phase distortion identification uses a sliding window to detect the instantaneous jump amplitude of the disturbance response phase. When the jump amplitude exceeds the preset fluctuation range, it is determined to be a local anomaly. The anomaly type is distinguished based on the phase recovery trend after the jump, and the anomaly point is mapped to the specific cutting position and tool rotation phase to complete the anomaly location.
[0025] Furthermore, the method also includes an adaptive adjustment step for the perturbation parameters:
[0026] The signal-to-noise ratio data of the current cutting condition parameters and disturbance response are read in real time, and the amplitude and frequency parameters of the feedforward disturbance are adjusted progressively. The adjustment range of each parameter is within the preset range. The disturbance frequency is kept away from the cutting fundamental frequency and its harmonic frequency band. The upper and lower limits of the disturbance amplitude are set by setting parameter safety thresholds to ensure that the disturbance amplitude is always within the allowable range of machining tolerance.
[0027] Furthermore, the method also includes a full-stroke stiffness field iteration step:
[0028] The verified stiffness data is stored according to workpiece type and clamping method to construct a full-stroke stiffness field database. The stiffness data in the database is statistically optimized periodically, and the stiffness reference value and fluctuation range at each position are updated. When machining the same type of workpiece, the stiffness field database is called as the initial reference, and the reference adaptation is completed by combining the real-time calibration results of the first stroke.
[0029] This invention also provides an intelligent monitoring system for the machining of precision automotive parts, comprising:
[0030] An interpolation period feedforward disturbance generation module is deployed on the machine tool interpolation calculation layer. It is used to generate sinusoidal feedforward disturbances and superimpose them onto the original interpolation command, output the disturbance-included interpolation command to the servo system, and synchronously output the disturbance reference signal to the acquisition end.
[0031] A dual-component phase-locked synchronous acquisition module is connected to the machine tool vibration sensing channel. Relying on the machine tool's global synchronous clock, it extracts the cutting fundamental frequency component and the disturbance frequency response component respectively, and outputs two time-aligned component signals.
[0032] A cycle-by-cycle dynamic stiffness calibration module is used to combine the excitation parameters of the feedforward disturbance with the measured response amplitude to calculate the dynamic stiffness transfer coefficient of the current cutting position on an interpolation cycle basis, and generate a real-time stiffness sequence.
[0033] The stiffness-normalized wear characteristic calculation module is used to calculate the cutting force equivalent based on real-time stiffness data and calibration mapping relationship, and generate a pure wear characteristic sequence.
[0034] The state determination module is used to perform multi-cycle trend verification on the wear feature sequence, identify the phase distortion of the disturbance response, and output the wear determination result and the local anomaly determination result respectively.
[0035] The disturbance parameter adaptive adjustment module is used to read cutting condition parameters and signal-to-noise ratio data, and gradually adjust the amplitude and frequency parameters of the feedforward disturbance.
[0036] The full-stroke stiffness field iteration module is used to classify and store stiffness data, build and update the full-stroke stiffness field database, and call the database to complete the benchmark adaptation before processing.
[0037] The machine tool control linkage output module is used to display real-time monitoring data and judgment results, trigger corresponding level of early warning prompts, and output linkage control commands to the machine tool control system.
[0038] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0039] This invention superimposes a micrometer-level sinusoidal feedforward perturbation onto the machine tool interpolation layer, simultaneously extracting the cutting fundamental frequency component and the perturbation frequency response component. It calculates the dynamic stiffness transfer coefficient of the current cutting position cycle by cycle, thereby performing position-synchronized stiffness correction on the cutting vibration signal. This isolates the influence of stiffness variations caused by changes in workpiece overhang length on the vibration amplitude, yielding pure wear characteristics related only to tool wear. It distinguishes between signal offsets caused by stiffness changes and tool wear at the mechanistic level, avoiding misjudgments of tool wear due to decreased workpiece end-effector stiffness and reducing unnecessary premature tool replacement. Simultaneously, it utilizes phase distortion detection of the perturbation response to locate local anomalies during the cutting process, eliminating the masking effect of stiffness base offset on abnormal signals. This ensures accurate monitoring throughout the long-stroke precision machining process and is suitable for online monitoring scenarios in the batch cutting of automotive precision long-shaft parts. Attached Figure Description
[0040] Figure 1 This is an overall flowchart of the intelligent monitoring method for machining precision automotive parts according to the present invention;
[0041] Figure 2 This is a flowchart of the dynamic stiffness calibration and pure wear characteristic calculation of the present invention;
[0042] Figure 3 This is a diagram of the intelligent monitoring system for machining precision automotive parts according to the present invention. Detailed Implementation
[0043] 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.
[0044] Please see Figures 1-2 This invention provides an intelligent monitoring method for the machining of precision automotive parts, comprising the following steps:
[0045] S1. A sinusoidal feedforward disturbance is superimposed on the original interpolation command of the machine tool interpolation calculation layer to generate an interpolation command with disturbance and output to the servo system, and a disturbance reference signal is output synchronously; the frequency of the sinusoidal feedforward disturbance is a non-integer multiple of the cutting fundamental frequency, and the amplitude is on the order of micrometers.
[0046] S2. Relying on the machine tool's global synchronous clock, the cutting fundamental frequency component and the disturbance frequency response component are extracted through two independent phase-locked channels to obtain two time-aligned component signals.
[0047] S3. Combining the excitation amplitude parameters of the feedforward disturbance with the amplitude data of the disturbance response components, calculate the dynamic stiffness transfer coefficient of the current cutting position on an interpolation cycle, and generate a real-time stiffness sequence that changes synchronously with the feed position.
[0048] S4. Based on the real-time stiffness sequence, multiply the amplitude of the cutting fundamental frequency component with the dynamic stiffness transfer coefficient at the corresponding position to calculate the cutting force equivalent sequence and generate the pure wear characteristic sequence.
[0049] S5. Perform multi-period trend verification on the pure wear feature sequence and output the wear judgment result; at the same time, monitor the phase change of the disturbance response component, identify the phase distortion point and output the local anomaly judgment result.
[0050] Specifically, this solution executes synchronously within the machine tool's interpolation and acquisition cycles. All calculations are completed within a single interpolation cycle, ensuring the accuracy of the correspondence between the monitoring results and the cutting position. The sinusoidal feedforward disturbance is superimposed on the original feed command in the form of position increments, with amplitude controlled at the micrometer level, ensuring it does not exceed the workpiece machining tolerance range and does not interfere with the normal cutting process. The cutting fundamental frequency is calculated from the spindle speed and the number of tool teeth, and is updated in real time as the spindle speed adjusts. The disturbance frequency always maintains a non-integer multiple relationship with the cutting fundamental frequency, avoiding spectral overlap and frequency coupling, and ensuring the independent extraction effect of the two components. The two phase-locked loops operate based on the same clock source, with the sampling trigger time strictly aligned with the interpolation output time, eliminating amplitude and phase calculation errors caused by timing deviations. The stiffness calculation for each interpolation cycle can be updated in real time following the tool feed position, accurately capturing the gradual stiffness change process caused by changes in the workpiece overhang length, without relying on a pre-established theoretical stiffness model. The cutting force equivalent sequence, corrected by the dynamic stiffness transfer coefficient, can isolate the signal amplitude shift caused by stiffness changes, retaining only the characteristic changes directly related to tool wear, thus distinguishing between gradual stiffness changes and wear-induced signal trends at the mechanistic level. Multi-cycle trend verification can filter out misjudgments caused by instantaneous interference, and phase distortion identification can locate local abnormal events in the cutting process. Together, they constitute a complete condition monitoring output. The vibration sensing channel uses an accelerometer installed at the machine tool spindle or tool post. The signal, after conditioning, is directly input into the phase-locked loop channel, eliminating the need for additional high-speed acquisition equipment. The entire monitoring process requires no additional excitation device; it can be implemented using only the machine tool's existing servo system and vibration sensing channel, making it suitable for the retrofitting and upgrading of existing mass-produced machine tools.
[0051] In a specific embodiment, step S1 specifically includes:
[0052] S11. Based on the perturbation parameters corresponding to the processing procedure and workpiece type, generate a sinusoidal perturbation sequence according to the preset perturbation frequency and amplitude;
[0053] S12. The sinusoidal perturbation sequence is superimposed on the position command of the feed axis to generate a perturbation interpolation command.
[0054] Specifically, the matching of disturbance parameters is accomplished by a preset parameter library, which stores recommended disturbance amplitudes and frequency ratios for different workpiece specifications and cutting processes. Before machining, the system reads the spindle speed, number of tool teeth, and workpiece diameter parameters for the current process, calculates the current cutting fundamental frequency, and then generates the disturbance frequency according to the preset ratio coefficient. The initial value of the disturbance amplitude is determined based on the workpiece machining tolerance grade. For example, for shaft parts with IT7 precision, the initial disturbance amplitude is set to 2μm, always controlled within one-tenth of the tolerance band, to ensure that it will not affect the final machined surface quality. The generation of the sinusoidal disturbance sequence is synchronized with the interpolation operation. Each interpolation cycle outputs a disturbance increment value, which is superimposed on the original position command of the corresponding feed axis. The superposition operation is performed at the end of the interpolation output, which will not interfere with the trajectory planning accuracy of the original interpolation algorithm, nor will it affect the original feed stability of the machine tool. For multi-axis linkage cutting scenarios, the disturbance is only superimposed on the feed axis corresponding to the main cutting force direction, reducing unnecessary axis-end disturbance output.
[0055] In one specific embodiment, step S2 specifically includes:
[0056] S21. The first phase-locked channel uses the cutting fundamental frequency as the reference frequency to extract the fundamental frequency amplitude and phase data of the cutting vibration;
[0057] S22. The second phase-locked channel uses the frequency of the disturbance reference signal as the reference frequency to filter out the interference of cutting signals and environmental noise, and extract the vibration response component excited by the feedforward disturbance;
[0058] S23. The two phase-locked loop channels share the machine tool's global clock, maintaining synchronization accuracy consistent with the interpolation period, and outputting two component signals with synchronized timing.
[0059] Specifically, both phase-locked loop (PLL) channels employ digital PLL amplification algorithms and operate in parallel within the computing unit of the FPGA or CNC system. The reference frequency of the first channel is calculated in real-time from the feedback signal of the spindle encoder and is synchronously adjusted to follow the fluctuations in spindle speed, ensuring it is always aligned with the current cutting fundamental frequency. The extracted amplitude and phase data directly reflect the system vibration response under cutting excitation. The reference signal of the second channel is directly taken from the synchronous output of the disturbance generation module, and its frequency is completely consistent with the feedforward disturbance. Through phase-sensitive detection and low-pass filtering, weak disturbance response components can be extracted from strong cutting background noise, resulting in a high signal-to-noise ratio. The sampling trigger signals for both channels are uniformly issued by the machine tool's global synchronization clock, with the trigger interval being exactly equal to the interpolation period. For example, when the interpolation period is 1ms, both channels synchronously complete sampling and calculation every 1ms. The two sets of output amplitude and phase data have completely consistent timestamps, ensuring the timing matching of the data in subsequent stiffness calculations and avoiding stiffness calculation deviations caused by time misalignment.
[0060] In a specific embodiment, in step S3, the dynamic stiffness transfer coefficient is calculated based on the structural dynamics transfer relationship, from the excitation amplitude of the feedforward disturbance and the disturbance response amplitude at the corresponding time. A stiffness calculation is completed once in each interpolation cycle, generating a continuous real-time stiffness sequence, which is updated synchronously with the feed position.
[0061] Specifically, the calculation of the dynamic stiffness transfer coefficient is derived from the amplitude-frequency response relationship of a single-degree-of-freedom system under harmonic excitation. Under steady-state harmonic excitation, the system response amplitude is linearly proportional to the excitation force amplitude, and the proportionality coefficient is determined by the current dynamic stiffness of the system. The feedforward disturbance is generated by the displacement command output by the servo system, and the corresponding excitation force can be obtained through the force-displacement conversion coefficient of the servo system. After simplifying the relationship between the excitation force and the response amplitude, the formula for calculating the dynamic stiffness transfer coefficient is obtained: In the formula, This is the dynamic stiffness transfer coefficient at the current cutting position, in N / m. The force-displacement conversion coefficient of the feed axis servo system is expressed in N / m and is obtained from the servo system's factory calibration parameters or on-site no-load calibration tests. The magnitude of the commanded displacement of the feedforward disturbance is expressed in meters (m). The displacement amplitude of the disturbance response component is expressed in meters (m).
[0062] At a specific feed position in actual machining, a set of typical operating parameters are selected for calculation. The command amplitude of the feedforward disturbance is... for m, force-displacement conversion coefficient of servo system for N / m, the displacement amplitude of the disturbance response extracted via the phase-locked channel. for Substituting m into the above formula, we can obtain the dynamic stiffness transfer coefficient at the current position. for N / m. The calculation process is completed once in each interpolation cycle, and the calculation result corresponds one-to-one with the current tool feed position, forming a continuous stiffness sequence with the tool feed. As the tool moves towards the end of the workpiece overhang, the effective cantilever length of the workpiece increases, the overall stiffness of the system decreases, and the value of the dynamic stiffness transfer coefficient will gradually decrease, truly reflecting the structural stiffness change at the cutting position.
[0063] In one specific embodiment, step S4 specifically includes:
[0064] S41. Retrieve the pre-stored stiffness-cutting force calibration mapping relationship. The mapping relationship is generated by fitting the calibration data under the new tool state and reflects the corresponding benchmark of vibration amplitude and cutting force under different stiffness levels.
[0065] S42. Substitute the currently measured dynamic stiffness transfer coefficient and the amplitude of the cutting fundamental frequency component into the mapping relationship to calculate the equivalent value of the cutting force at the current position;
[0066] S43. Arrange all cutting force equivalent values in the order of feed position to form a pure wear characteristic sequence.
[0067] Specifically, the stiffness-cutting force calibration mapping relationship is completed during the first machining operation with a new tool. The calibration process selects 5 to 8 characteristic positions within the entire workpiece stroke, maintaining stable cutting conditions at each position. Vibration response data and cutting force reference data are simultaneously collected at the corresponding positions, and the mapping curve is obtained through polynomial fitting. The calibration data covers the main stiffness range within the entire workpiece stroke, ensuring that calculations at different positions are within the calibration range. The calculation of the cutting force equivalent follows the same dynamic relationship as the stiffness calculation, representing the reverse solution of the same amplitude response law: given the dynamic stiffness and cutting vibration amplitude at the current position, the magnitude of the cutting force that triggers the vibration is deduced. Considering the existence of multi-mode and damping nonlinearity in actual cutting systems, a calibration correction coefficient is introduced to compensate for the linear calculation results, ultimately yielding the cutting force equivalent calculation formula: In the formula, This represents the equivalent cutting force at the current position, expressed in Newtons (N). The displacement amplitude of the fundamental frequency component of the cutting element is expressed in meters (m). This is the dynamic stiffness transfer factor at the corresponding location, in N / m. The correction coefficient for the calibration mapping is a dimensionless parameter obtained from the new tool calibration data through least squares fitting. It is used to compensate for calculation errors caused by factors such as stiffness nonlinearity and sensor installation deviation.
[0068] Within an interval where stiffness changes approximately linearly, the correction coefficient approaches 1, and the cutting force equivalent can be simplified to the product of the vibration amplitude and the dynamic stiffness transfer coefficient. Using the stiffness parameters from the aforementioned location, if the current cutting fundamental frequency vibration amplitude... for m, with a correction factor of 1, can be substituted into the calculation formula to obtain the equivalent cutting force. The value is 20N. The corrected cutting force equivalent is only related to the magnitude of the actual cutting excitation and is not affected by changes in structural stiffness. When the tool wears, the cutting force increases, and the value of the cutting force equivalent will rise synchronously. The resulting characteristic sequence can directly reflect the true wear state of the tool. All calculation results are arranged in the order of feed positions, forming a pure wear characteristic sequence that corresponds one-to-one with the machining stroke, providing a data basis for subsequent wear determination.
[0069] In a specific embodiment, in step S5, the multi-cycle trend verification adopts a sliding time window logic, taking feature data from multiple consecutive interpolation cycles as the verification window, eliminating abnormal data that rapidly falls back after a single point jump, and evaluating the wear level based on the continuously changing feature data; combining historical wear data to classify the verified feature quantities into levels, recording the rate of change of feature quantities, and marking abnormal wear states where the rate of change exceeds a preset range.
[0070] Specifically, the length of the sliding time window is set based on the cutting feed rate and interpolation cycle. For example, data from 100 consecutive interpolation cycles can be selected as the verification window, which slides forward cycle by cycle as the machining progresses. This length setting can cover tens of times the cutting cycle, effectively filtering out instantaneous interference, without obscuring changes in wear trends due to an excessively long window. During verification, single-point jump data within the window is identified, i.e., data points where the feature quantity rises sharply in a single cycle and then falls rapidly in the following cycles. These are judged as anomalies caused by instantaneous interference such as chip breakage or random impact, and are removed from the feature sequence to avoid interfering with the wear trend judgment. The wear level classification refers to the full-life wear test data of similar tools, dividing the feature quantity into three levels: initial wear, normal wear, and rapid wear, each corresponding to a different processing strategy. The system also records the unit stroke change rate of the feature quantity. When the change rate exceeds the preset normal range, it is marked as an abnormal wear state. The value of the preset range is obtained from the wear test statistics of the same batch of tools, with 1.5 times the normal wear rate as the upper limit threshold. If it is lower than the lower limit threshold, it indicates that the tool is not participating in cutting normally.
[0071] In a specific embodiment, in step S5, phase distortion identification uses a sliding window to detect the instantaneous jump amplitude of the disturbance response phase. When the jump amplitude exceeds the preset fluctuation range, it is determined to be a local anomaly. The anomaly type is distinguished based on the phase recovery trend after the jump, and the anomaly point is mapped to the specific cutting position and tool rotation phase to complete the anomaly localization.
[0072] Specifically, the phase of the disturbance response remains relatively stable during normal cutting with a small fluctuation range. When local structural changes occur in the cutting area, the vibration transmission path of the system changes, and the vibration phase undergoes an instantaneous jump. Phase detection uses a sliding window with the same period as the stiffness calculation, with a window length set to 20 interpolation cycles. The phase change within the window is calculated cycle by cycle. The preset fluctuation range is determined by the phase statistics under normal cutting conditions, for example, the normal fluctuation range is ±5°. When the phase jump amplitude within a single window exceeds 20°, it is judged as a local anomaly. This threshold is set to 4 times the normal fluctuation amplitude, which can effectively identify real abnormal events while ensuring a low false detection rate. After a phase jump, if it gradually recovers to the original level after several cycles, it often corresponds to material defects such as hard spots or inclusions inside the workpiece. If the phase jump cannot recover and continues to deviate, it often corresponds to tool failure types such as tool chipping or tool tip breakage. The system maps the time of the anomaly to the current feed position and the spindle encoder rotation angle, achieving precise location of the anomaly, which facilitates subsequent workpiece re-inspection and tool inspection.
[0073] In one specific embodiment, the method further includes an adaptive adjustment step for the perturbation parameters:
[0074] The signal-to-noise ratio data of the current cutting condition parameters and disturbance response are read in real time, and the amplitude and frequency parameters of the feedforward disturbance are adjusted progressively. The adjustment range of each parameter is within the preset range. The disturbance frequency is kept away from the cutting fundamental frequency and its harmonic frequency band. The upper and lower limits of the disturbance amplitude are set by setting parameter safety thresholds to ensure that the disturbance amplitude is always within the allowable range of machining tolerance.
[0075] Specifically, the reading of operating parameters is synchronized with the interpolation cycle, mainly including parameters such as spindle speed, feed rate, and depth of cut. When the spindle speed changes, the cutting fundamental frequency changes synchronously, and the system automatically adjusts the disturbance frequency to always maintain the disturbance frequency as a non-integer multiple of the cutting fundamental frequency, while avoiding the harmonic bands of the cutting fundamental frequency to prevent spectral aliasing between the disturbance signal and the cutting signal. The signal-to-noise ratio (SNR) data is calculated in real time by the second phase-locked loop (PLL). When the SNR is low, the disturbance amplitude is appropriately increased to enhance the strength of the response signal; when the SNR is sufficient, the disturbance amplitude is appropriately decreased to reduce the impact on the machining process. Parameter adjustment adopts a gradual approach, with a single amplitude adjustment not exceeding 0.5 μm and a single frequency adjustment not exceeding 5% of the current frequency, avoiding system shocks caused by parameter abrupt changes. The system sets safe thresholds for the upper and lower limits of the amplitude, with the upper limit not exceeding one-tenth of the workpiece machining tolerance and the lower limit not lower than the minimum amplitude that the PLL can recognize, ensuring that the disturbance is always within a safe and effective range.
[0076] In one specific embodiment, the method further includes a full-stroke stiffness field iteration step:
[0077] The verified stiffness data is stored according to workpiece type and clamping method to construct a full-stroke stiffness field database. The stiffness data in the database is statistically optimized periodically, and the stiffness reference value and fluctuation range at each position are updated. When machining the same type of workpiece, the stiffness field database is called as the initial reference, and the reference adaptation is completed by combining the real-time calibration results of the first stroke.
[0078] Specifically, after each machining operation, the system stores the verified stiffness data from that operation into a database corresponding to the workpiece type and clamping method, categorized by feed position. The database is managed by workpiece model and clamping method, including common methods such as one clamp and one center, and two centers. Data storage granularity is set at 0.5mm intervals along the feed axis coordinates, with each node storing multiple sets of historical stiffness data. The system periodically performs statistical analysis on all data, calculating the mean stiffness and normal fluctuation range for each position, and updating the stiffness benchmark value. The statistical cycle can be set to once after machining 10 workpieces of the same type. As the number of machining batches increases, the amount of data in the database continuously expands, and the accuracy of the stiffness benchmark gradually improves. When machining the same type of workpiece with the same clamping method again, the system directly calls the stiffness field data in the database as the initial reference, without needing to re-calibrate the entire stroke. During the first 20mm stroke after machining begins, the system uses the real-time collected stiffness data to perform proportional adaptation correction on the benchmark value, further improving the monitoring accuracy in the initial machining stage. The database supports classification management of multiple workpiece types and multiple clamping methods, and is adapted to the production mode of mixed processing of multiple varieties on the production line.
[0079] Please see Figure 3 The present invention also provides an intelligent monitoring system for the machining of precision automotive parts, comprising:
[0080] An interpolation period feedforward disturbance generation module is deployed on the machine tool interpolation calculation layer. It is used to generate sinusoidal feedforward disturbances and superimpose them onto the original interpolation command, output the disturbance-included interpolation command to the servo system, and synchronously output the disturbance reference signal to the acquisition end.
[0081] A dual-component phase-locked synchronous acquisition module is connected to the machine tool vibration sensing channel. Relying on the machine tool's global synchronous clock, it extracts the cutting fundamental frequency component and the disturbance frequency response component respectively, and outputs two time-aligned component signals.
[0082] A cycle-by-cycle dynamic stiffness calibration module is used to combine the excitation parameters of the feedforward disturbance with the measured response amplitude to calculate the dynamic stiffness transfer coefficient of the current cutting position on an interpolation cycle basis, and generate a real-time stiffness sequence.
[0083] The stiffness-normalized wear characteristic calculation module is used to calculate the cutting force equivalent based on real-time stiffness data and calibration mapping relationship, and generate a pure wear characteristic sequence.
[0084] The state determination module is used to perform multi-cycle trend verification on the wear feature sequence, identify the phase distortion of the disturbance response, and output the wear determination result and the local anomaly determination result respectively.
[0085] The disturbance parameter adaptive adjustment module is used to read cutting condition parameters and signal-to-noise ratio data, and gradually adjust the amplitude and frequency parameters of the feedforward disturbance.
[0086] The full-stroke stiffness field iteration module is used to classify and store stiffness data, build and update the full-stroke stiffness field database, and call the database to complete the benchmark adaptation before processing.
[0087] The machine tool control linkage output module is used to display real-time monitoring data and judgment results, trigger corresponding level of early warning prompts, and output linkage control commands to the machine tool control system.
[0088] Specifically, the interpolation cycle feedforward disturbance generation module runs in parallel with the original interpolation algorithm. After superimposing the commands at the interpolation output, it outputs the results to the servo drive and simultaneously sends a synchronous disturbance reference signal to the dual-component phase-locked loop synchronous acquisition module. The dual-component phase-locked loop synchronous acquisition module connects to the vibration sensor signal standardly equipped on the machine tool. The two calculated component signals are synchronously output to the cycle-by-cycle dynamic stiffness calibration module and the stiffness normalized wear feature calculation module. The real-time stiffness sequence output by the cycle-by-cycle dynamic stiffness calibration module is divided into two paths: one path is sent to the wear feature calculation module to participate in the cutting force equivalent calculation, and the other path is sent to the full-stroke stiffness field iteration module to complete data storage and update. The state determination module receives the pure wear feature sequence and phase data, completes the wear level determination and anomaly identification, and outputs the results to the machine tool control linkage output module. The disturbance parameter adaptive adjustment module reads the operating parameters and signal-to-noise ratio data in real time and feeds back the parameter adjustment commands to the disturbance generation module, forming a parameter adaptive closed loop. The machine tool control linkage output module interfaces with the machine tool's operating interface and control logic. It can directly display monitoring data and early warning information on the machine tool's operating panel, and can also output linkage commands such as feed hold and tool change prompts to the machine tool control system, realizing closed-loop linkage between monitoring and machining control.
[0089] In summary, this invention superimposes a micrometer-level sinusoidal feedforward perturbation onto the machine tool interpolation layer, simultaneously extracts the cutting fundamental frequency component and the perturbation frequency response component, and calculates the dynamic stiffness transfer coefficient of the current cutting position cycle by cycle. This allows for position-synchronized stiffness correction of the cutting vibration signal, eliminating the influence of stiffness variations caused by changes in workpiece overhang length on the vibration amplitude. This yields pure wear characteristics related only to tool wear, distinguishing between stiffness changes and tool wear at the mechanistic level. This avoids misjudgments of tool wear caused by decreased workpiece end stiffness and reduces unnecessary premature tool replacement. Simultaneously, phase distortion detection of the perturbation response locates local anomalies during the cutting process, eliminating the masking effect of stiffness base offset on abnormal signals. This ensures accurate monitoring throughout the long-stroke precision machining process and is suitable for online monitoring scenarios in the batch cutting of automotive precision long-shaft parts.
[0090] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0091] 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. An intelligent monitoring method for the machining of precision automotive parts, characterized in that, Includes the following steps: S1. A sinusoidal feedforward disturbance is superimposed on the original interpolation command of the machine tool interpolation calculation layer to generate an interpolation command with disturbance and output to the servo system, and a disturbance reference signal is output synchronously; the frequency of the sinusoidal feedforward disturbance is a non-integer multiple of the cutting fundamental frequency, and the amplitude is on the order of micrometers. S2. Relying on the machine tool's global synchronous clock, the cutting fundamental frequency component and the disturbance frequency response component are extracted through two independent phase-locked channels to obtain two time-aligned component signals. S3. Combining the excitation amplitude parameters of the feedforward disturbance with the amplitude data of the disturbance response components, calculate the dynamic stiffness transfer coefficient of the current cutting position on an interpolation cycle, and generate a real-time stiffness sequence that changes synchronously with the feed position. S4. Based on the real-time stiffness sequence, multiply the amplitude of the cutting fundamental frequency component with the dynamic stiffness transfer coefficient at the corresponding position to calculate the cutting force equivalent sequence and generate the pure wear characteristic sequence. S5. Perform multi-period trend verification on the pure wear feature sequence and output the wear judgment result; at the same time, monitor the phase change of the disturbance response component, identify the phase distortion point and output the local anomaly judgment result.
2. The intelligent monitoring method for machining precision automotive parts according to claim 1, characterized in that, The S1 step specifically includes: S11. Based on the perturbation parameters corresponding to the processing procedure and workpiece type, generate a sinusoidal perturbation sequence according to the preset perturbation frequency and amplitude; S12. The sinusoidal perturbation sequence is superimposed on the position command of the feed axis to generate a perturbation interpolation command.
3. The intelligent monitoring method for machining precision automotive parts according to claim 1, characterized in that, The S2 step specifically includes: S21. The first phase-locked channel uses the cutting fundamental frequency as the reference frequency to extract the fundamental frequency amplitude and phase data of the cutting vibration; S22. The second phase-locked channel uses the frequency of the disturbance reference signal as the reference frequency to filter out the interference of cutting signals and environmental noise, and extract the vibration response component excited by the feedforward disturbance; S23. The two phase-locked loop channels share the machine tool's global clock, maintaining synchronization accuracy consistent with the interpolation period, and outputting two component signals with synchronized timing.
4. The intelligent monitoring method for machining precision automotive parts according to claim 1, characterized in that, In step S3, the dynamic stiffness transfer coefficient is calculated based on the structural dynamics transfer relationship, from the excitation amplitude of the feedforward disturbance and the disturbance response amplitude at the corresponding time. A stiffness calculation is completed in each interpolation cycle, generating a continuous real-time stiffness sequence, which is updated synchronously with the feed position.
5. The intelligent monitoring method for machining precision automotive parts according to claim 1, characterized in that, The S4 step specifically includes: S41. Retrieve the pre-stored stiffness-cutting force calibration mapping relationship. The mapping relationship is generated by fitting the calibration data under the new tool state and reflects the corresponding benchmark of vibration amplitude and cutting force under different stiffness levels. S42. Substitute the currently measured dynamic stiffness transfer coefficient and the amplitude of the cutting fundamental frequency component into the mapping relationship to calculate the equivalent value of the cutting force at the current position; S43. Arrange all cutting force equivalent values in the order of feed position to form a pure wear characteristic sequence.
6. The intelligent monitoring method for machining precision automotive parts according to claim 1, characterized in that, In step S5, the multi-cycle trend verification adopts a sliding time window logic, taking the feature data of multiple consecutive interpolation cycles as the verification window, eliminating abnormal data that quickly falls back after a single point jump, and evaluating the wear level based on the continuously changing feature data; combining historical wear data to classify the verified feature quantities into levels, recording the rate of change of feature quantities, and marking abnormal wear states where the rate of change exceeds the preset range.
7. The intelligent monitoring method for machining precision automotive parts according to claim 1, characterized in that, In step S5, phase distortion identification uses a sliding window to detect the instantaneous jump amplitude of the disturbance response phase. When the jump amplitude exceeds the preset fluctuation range, it is determined to be a local anomaly. The anomaly type is distinguished based on the phase recovery trend after the jump, and the anomaly point is mapped to the specific cutting position and tool rotation phase to complete the anomaly location.
8. The intelligent monitoring method for machining precision automotive parts according to claim 1, characterized in that, The method also includes an adaptive adjustment step for the perturbation parameters: The signal-to-noise ratio data of the current cutting condition parameters and disturbance response are read in real time, and the amplitude and frequency parameters of the feedforward disturbance are adjusted progressively. The adjustment range of each parameter is within the preset range. The disturbance frequency is kept away from the cutting fundamental frequency and its harmonic frequency band. The upper and lower limits of the disturbance amplitude are set by setting parameter safety thresholds to ensure that the disturbance amplitude is always within the allowable range of machining tolerance.
9. The intelligent monitoring method for machining precision automotive parts according to claim 1, characterized in that, The method also includes a full-stroke stiffness field iteration step: The verified stiffness data is stored according to workpiece type and clamping method to construct a full-stroke stiffness field database. The stiffness data in the database is statistically optimized periodically, and the stiffness reference value and fluctuation range at each position are updated. When machining the same type of workpiece, the stiffness field database is called as the initial reference, and the reference adaptation is completed by combining the real-time calibration results of the first stroke.
10. An intelligent monitoring system for machining precision automotive parts, used to execute the intelligent monitoring method for machining precision automotive parts according to any one of claims 1 to 9, characterized in that, include: An interpolation period feedforward disturbance generation module is deployed on the machine tool interpolation calculation layer. It is used to generate sinusoidal feedforward disturbances and superimpose them onto the original interpolation command, output the disturbance-included interpolation command to the servo system, and synchronously output the disturbance reference signal to the acquisition end. A dual-component phase-locked synchronous acquisition module is connected to the machine tool vibration sensing channel. Relying on the machine tool's global synchronous clock, it extracts the cutting fundamental frequency component and the disturbance frequency response component respectively, and outputs two time-aligned component signals. A cycle-by-cycle dynamic stiffness calibration module is used to combine the excitation parameters of the feedforward disturbance with the measured response amplitude to calculate the dynamic stiffness transfer coefficient of the current cutting position on an interpolation cycle basis, and generate a real-time stiffness sequence. The stiffness-normalized wear characteristic calculation module is used to calculate the cutting force equivalent based on real-time stiffness data and calibration mapping relationship, and generate a pure wear characteristic sequence. The state determination module is used to perform multi-cycle trend verification on the wear feature sequence, identify the phase distortion of the disturbance response, and output the wear determination result and the local anomaly determination result respectively. The disturbance parameter adaptive adjustment module is used to read cutting condition parameters and signal-to-noise ratio data, and gradually adjust the amplitude and frequency parameters of the feedforward disturbance. The full-stroke stiffness field iteration module is used to classify and store stiffness data, build and update the full-stroke stiffness field database, and call the database to complete the benchmark adaptation before processing. The machine tool control linkage output module is used to display real-time monitoring data and judgment results, trigger corresponding level of early warning prompts, and output linkage control commands to the machine tool control system.
