Method for determining the delamination risk index of carbon fiber composite materials by ultrasonic vibration assisted drilling

Abstract

The application discloses a kind of carbon fiber composite material ultrasonic vibration assisted drilling layered risk index determination method.This method collects multi-source sensing signal in the process of drilling, and identifies exit window according to residual thickness;Build the critical layered load of exit that dynamically changes with residual thickness as evaluation benchmark;Equivalent dynamic thrust is extracted in exit window, and the overloading part is time-integrated to obtain exit action impulse based on critical load capacity;Thrust ratio, impulse ratio, acoustic emission and vibration anomaly are normalized and fused into layered risk index;Finally, the risk index is modified through the physical consistency evaluation of boundary constraint and monotonicity constraint, and the layered risk grade and early warning are output.The application quantifies the layered risk from the energy accumulation dimension, can distinguish transient signal fluctuation and sustained overload, overcome the defect that traditional peak force threshold is easy to misjudge, realize the real-time feedforward evaluation of exit layered risk.

Description

Technical Field

[0001] This invention relates to the field of machining and processing quality monitoring technology for carbon fiber composite materials, specifically a method for determining the risk index of layering in ultrasonic vibration-assisted drilling of carbon fiber composite materials. Background Technology

[0002] Carbon fiber composites, due to their high specific strength, high specific modulus, and excellent fatigue resistance, have been widely used in aerospace, high-end equipment, and rail transportation. Drilling is extensively used in assembly and connection processes. However, carbon fiber composites exhibit significant anisotropy, low interlaminar strength, and high brittleness, making them highly susceptible to delamination defects, especially at the exit stage. Exit delamination is one of the most detrimental drilling defects, severely reducing connection strength and structural reliability. While ultrasonic vibration-assisted drilling technology can improve processing quality to some extent, the high-frequency ultrasonic vibration generates superimposed transient pulse impacts at the extremely thin exit stage, making the assessment of exit delamination risk more complex.

[0003] In existing technologies, there are two main types of methods for evaluating borehole stratification. The first type is the stratification factor evaluation method based on post-drilling image measurement. This method is a post-detection method, which can only be found after the defect has formed and cannot provide real-time early warning and feedforward control during the processing. The second type is the online determination method based on a single axial force peak threshold. That is, a fixed force threshold is preset, and when the axial force peak exceeds the threshold, it is determined that there is a risk of stratification. However, this method has the following specific drawbacks: First, a fixed threshold cannot reflect the dynamic decay of the load-bearing capacity caused by the continuous reduction of the remaining thickness of the workpiece as the drill bit approaches the exit. In the early stage of drilling, the threshold is relatively low compared to the actual load-bearing capacity of the material, which can easily lead to false alarms. In the extremely thin stage near the exit, the threshold is relatively high compared to the already decayed load-bearing capacity of the material, which can easily lead to false alarms. Second, judging solely based on the instantaneous peak value of the force cannot distinguish between two completely different physical conditions: "instantaneous high-frequency signal fluctuation" and "continuous overload loading". Although the former has a high peak force, the duration of action is extremely short, and the accumulated fracture driving energy at the interlayer interface is insufficient to trigger delamination. Although the latter may have a similar peak force, the duration is long, and the accumulated energy is sufficient to drive crack propagation and lead to severe delamination. Third, the risk indicators obtained based on multi-source signal fusion calculation lack the constraint verification of physical laws. They may produce output results that contradict the physical mechanism due to sensor noise or signal fluctuations. For example, the risk may decrease when the remaining thickness decreases, which affects the credibility of decisions in engineering deployment. Summary of the Invention

[0004] The purpose of this invention is to provide a method for real-time assessment of exit stratification risk during drilling. This method needs to simultaneously satisfy: a critical load-bearing benchmark that is dynamically updated with the remaining thickness; a quantification mechanism that can distinguish between instantaneous fluctuations and continuous overloads from the perspective of time accumulation; and a verification capability that applies physical constraints to the output results to ensure the reliability of the judgment results.

[0005] To achieve the above objectives, the present invention proposes the following technical solution: a method for determining the risk index of layering in ultrasonic vibration-assisted drilling of carbon fiber composite materials, comprising the following steps:

[0006] S1: During the ultrasonic vibration-assisted drilling process of carbon fiber composite materials, multi-source sensing signals of the drilling process are collected. The multi-source sensing signals include at least axial force signals, vibration signals and acoustic emission signals.

[0007] S2: Determine the remaining thickness based on the real-time axial position of the drill bit and the total thickness of the workpiece, and identify the exit action window based on the comparison relationship between the remaining thickness and the preset exit threshold thickness;

[0008] S3: Based on the remaining thickness, layup parameters, clamping support parameters, and tool geometry parameters, construct the exit critical layer load capacity that dynamically changes with the remaining thickness;

[0009] S4: Within the exit action window, the equivalent dynamic thrust is extracted based on the multi-source sensor signal, and the overload portion of the equivalent dynamic thrust exceeding the exit critical layer load capacity is integrated over the time interval of the exit action window, with the exit critical layer load capacity as the benchmark, to obtain the exit action impulse.

[0010] S5: The ratio of the equivalent dynamic thrust to the critical stratification load at the exit, the ratio of the exit action impulse to the critical impulse benchmark, and the abnormal quantities of acoustic emission and vibration are normalized and fused to obtain the stratification risk index.

[0011] S6: Perform a physical constraint consistency assessment on the stratified risk index. The physical constraint consistency assessment includes at least boundary constraints and monotonicity constraints. Output the stratified risk level based on the assessed stratified risk index.

[0012] Furthermore, in this invention, the specific method for constructing the outlet critical stratification bearing capacity in step S3 is as follows:

[0013] The power function of the remaining thickness is used as the dominant attenuation term, the power function of the drill bit diameter is used as the scaling term, the reciprocal of the arctangent function of the drill bit tip angle is used as the geometric correction term, and the ply stiffness coefficient extracted from the laminate bending stiffness submatrix and the clamping boundary correction coefficient determined by the bottom support conditions are introduced as multiplicative correction factors. The terms are multiplied together to obtain the critical delamination bearing capacity at the outlet.

[0014] The ply stiffness coefficient is used to characterize the anisotropic anti-bending delamination capacity under different ply structures, and the outlet critical delamination load decreases monotonically as the remaining thickness decreases.

[0015] Furthermore, in this invention, the ply stiffness coefficient is extracted from the bending stiffness submatrix D in the ABD matrix of the laminate, and its expression is as follows:

[0016] The arithmetic square root of the product of longitudinal bending stiffness and transverse bending stiffness is added to the sum of the Poisson effect coupled bending stiffness and twice the torsional stiffness, and this sum is taken as the value of the ply stiffness coefficient.

[0017] Furthermore, in this invention, the method for determining the preset outlet threshold thickness in step S2 is as follows:

[0018] Calculate the first candidate value and the second candidate value respectively, and take the larger of the two values ​​as the preset exit threshold thickness; wherein, the first candidate value is the product of the ply critical coefficient and the single ply thickness; the second candidate value is the product of the geometric height influence coefficient and the drill bit tip angle geometric cutting height, wherein the drill bit tip angle geometric cutting height is determined by the product of half the drill bit diameter and the reciprocal of the tangent of the drill bit tip angle.

[0019] Furthermore, in this invention, the calculation method for the equivalent dynamic thrust in step S4 is as follows:

[0020] The mean value of axial force within the outlet action window is used as the basic term. The peak characteristic value of axial force, root mean square value, product of ultrasonic amplitude and ultrasonic frequency, and main shaft current fluctuation are used as weighted compensation terms. Each compensation term is multiplied by its corresponding weighting coefficient and then summed with the basic term to obtain the equivalent dynamic thrust.

[0021] The product of ultrasonic amplitude and ultrasonic frequency is explicitly coupled as an independent load component to characterize the additional contribution of ultrasonic vibration to the exit dynamic thrust.

[0022] Furthermore, in this invention, the specific calculation method for the outlet action impulse in step S4 is as follows:

[0023] Within the start and end time interval of the exit action window, the difference between the equivalent dynamic thrust and the critical bearing safety factor multiplied by the critical stratified bearing capacity of the exit is calculated time by time. When the difference is positive, the difference is taken; when it is negative, it is taken as zero. The truncated difference is integrally performed over the entire time interval of the exit action window to obtain the exit action impulse.

[0024] Furthermore, this invention also includes the step of constructing a damage sensitivity correction value:

[0025] The damage sensitivity coefficient is obtained by multiplying the tool wear characterization, temperature or thermal softening characterization, interlayer interface deterioration characterization, and adjacent ply angle mismatch characterization by the corresponding correction coefficients, summing them, and adding one.

[0026] The critical stratification load capacity at the outlet is reduced and corrected using the damage sensitivity coefficient to obtain the corrected critical stratification load capacity at the outlet. The corrected critical stratification load capacity at the outlet is then used to replace the original critical stratification load capacity at the outlet in subsequent risk index fusion calculations.

[0027] Furthermore, in this invention, in step S5, the normalization fusion has two optional mapping forms:

[0028] The first method is a linear weighted mapping form, in which each normalized quantity is multiplied by its corresponding weight coefficient and then summed to obtain the hierarchical risk index.

[0029] The second type is the exponential decay nonlinear mapping form, which involves multiplying each normalized quantity by its corresponding weight coefficient and summing the results to obtain the exponential parameter. This parameter is then substituted into the negative exponential form of the natural exponential function to calculate the stratified risk index.

[0030] The linear weighted mapping form is suitable for stable cutting conditions where the range of normalized values ​​is narrow, while the exponentially decaying nonlinear mapping form is suitable for severe cutting conditions with extreme singular values.

[0031] Furthermore, in this invention, the specific implementation of the monotonicity constraint in step S6 is as follows:

[0032] First, a sliding time window of a set width is used to smooth and denoise the original calculated hierarchical risk index to obtain a smoothed risk index.

[0033] Then, the historical extreme value envelope truncation function is introduced, and the larger value between the smoothed risk index and the historical maximum risk index from the start time of the exit action window to the current time is taken as the corrected current time stratified risk index.

[0034] This ensures that the layering risk index does not abnormally decrease under conditions of increased feed rate, accelerated tool wear, or reduced remaining thickness.

[0035] Furthermore, in this invention, step S6 further includes a logical threshold consistency check:

[0036] When the ratio of the equivalent dynamic thrust to the corrected exit critical stratification load is less than one, and the ratio of the exit action impulse to the critical impulse benchmark is less than one, and the acoustic emission anomaly does not show a continuous over-threshold jump, the stratification risk level is limited to no higher than the warning level.

[0037] When any of the ratios is greater than or equal to one, or when the amount of acoustic emission anomaly continues to exceed the threshold, the stratified risk level will be determined to be at least medium-high risk level.

[0038] Based on the magnitude of the risk index, the risk is divided into at least low risk, medium risk, high risk and extremely high risk levels, and corresponding output commands are given to keep the parameters unchanged, reduce the feed rate, reduce the feed rate and reduce the ultrasonic amplitude, and urgently stop the spindle and retract the tool.

[0039] Beneficial effects: The technical solution of this application has the following technical effects:

[0040] The present invention provides a method for determining the risk index of ultrasonic vibration-assisted drilling of carbon fiber composite materials for layering. By establishing an exit critical layering load capacity that dynamically changes with the remaining thickness as an evaluation benchmark, and using this dynamic benchmark as a reference, the time integral of the equivalent dynamic thrust exceeding the load capacity within the exit action window, i.e., the exit action impulse, is calculated. Then, multi-dimensional normalized indicators are integrated into a comprehensive risk index and physical constraints are applied for consistency evaluation. This method realizes real-time quantitative assessment and graded early warning of exit layering risk before the formation of macroscopic layering defects. It belongs to feedforward quality control and fundamentally overcomes the limitation of traditional post-detection methods that cannot intervene in a timely manner.

[0041] This invention utilizes a dynamic critical load capacity model to ensure that the evaluation benchmark decreases synchronously with increasing drilling depth and decreasing remaining thickness. This accurately matches the actual physical process of the sharp decline in load capacity of carbon fiber composites at the exit stage, overcoming the shortcomings of fixed threshold methods, such as false alarms in the early stage and false alarms at the end stage. The exit action impulse mechanism introduced in this invention quantifies the stratification risk based on the cumulative effect of overload force over time. It can fundamentally distinguish between "instantaneous harmless fluctuations with high peak force but extremely short duration" and "damaging overloads with excessive force values ​​and long durations," overcoming the technical pain point of easy misjudgment with a single peak force threshold. Comparative experiments show that under two conditions with identical peak thrust but different overload durations, the traditional single threshold method gives identical false judgment results, while the impulse coupling mechanism of this invention can accurately distinguish between the two and match the actual stratification situation.

[0042] Furthermore, this invention applies boundary constraints and monotonicity constraints to the risk index through physical constraint consistency assessment. This ensures that the risk index does not exhibit anomalous fluctuations that violate physical mechanisms when conditions such as decreasing remaining thickness and increased tool wear change during the exit stage. This guarantees the output reliability and decision-making credibility of the online monitoring system in engineering deployment. The method of this invention has the advantages of strong physical interpretability, good generalization ability, and online deployment capability. It can be directly applied to online quality monitoring and adaptive control of process parameters in ultrasonic vibration-assisted drilling of carbon fiber composite materials.

[0043] It should be understood that all combinations of the foregoing concepts and the additional concepts described in more detail below can be considered part of the inventive subject matter of this disclosure, provided that such concepts do not contradict each other.

[0044] The foregoing and other aspects, embodiments, and features of the teachings of the present invention will be more fully understood from the following description in conjunction with the accompanying drawings. Other additional aspects of the invention, such as features and / or beneficial effects of exemplary embodiments, will become apparent from the following description or may be learned through practice of specific embodiments according to the teachings of the present invention. Attached Figure Description

[0045] The accompanying drawings are not intended to be drawn to scale. In the drawings, each identical or nearly identical component shown in the various figures may be denoted by the same reference numeral. For clarity, not every component is labeled in each figure. Embodiments of various aspects of the invention will now be described by way of example and with reference to the accompanying drawings, wherein:

[0046] Figure 1 Here is the overall method flowchart;

[0047] Figure 2 A dynamic relationship diagram between the exit window and the remaining depth;

[0048] Figure 3 This is a diagram of the stratified risk index fusion model.

[0049] Figure 4 This is a schematic diagram showing the changes in force and strength during the drill bit exit stage.

[0050] Figure 5 This is a logic structure diagram of "critical load-impulse coupling";

[0051] Figure 6 The graph shows the results of the physical constraint monotonicity verification. Detailed Implementation

[0052] The embodiments of the invention will be described in detail below with reference to the accompanying drawings, such as... Figure 1This embodiment provides a method for determining the risk index of layer delamination in ultrasonic vibration-assisted drilling of carbon fiber composite materials. The core technical concept lies in establishing a three-layer closed-loop risk assessment system based on the physical evolution of interlayer delamination defects at the exit stage of ultrasonic vibration-assisted drilling of carbon fiber composite materials. This system uses "dynamic critical load capacity as the benchmark, overload impulse integral as the core metric, and physical constraint consistency assessment as the verification guarantee." This system can quantitatively characterize and provide graded early warnings of the risk trend of exit layer delamination in real time before macroscopic delamination defects actually form, thereby achieving feedforward quality control.

[0053] Based on the fundamental principles of fracture mechanics, delamination in carbon fiber composites is essentially an energy-driven crack propagation process: a crack will only propagate from initiation to penetration when the energy accumulated by the external load at the interlaminar interface exceeds the material's interlaminar fracture toughness. Therefore, judging delamination risk solely based on whether the peak force value at a single instant exceeds the limit is physically insufficient. The core insight of this invention lies in this: by introducing the exit impulse as a cumulative time factor, the assessment of delamination risk is elevated from "force exceeding the limit judgment" to "overload energy accumulation judgment," thereby distinguishing between "instantaneous harmless fluctuations" and "continuous damaging overloads" at the physical mechanism level, significantly improving the accuracy of assessment and the robustness of online monitoring.

[0054] The steps of the method of the present invention will be described in detail below with reference to specific process conditions and numerical parameters.

[0055] This embodiment uses a T700 / epoxy resin carbon fiber composite laminate as the machining object. The total workpiece thickness is 5.0 mm, the single-layer layup thickness is 0.125 mm, and the layup sequence is a multi-directional symmetrical layup structure. A 6 mm diameter carbide twist drill with a 118° apex angle and a cutting edge radius of approximately 15 μm is selected. The bottom adopts a standard suspended, unbacked clamping method. The spindle speed is set to 3000 r / min, the feed rate is 15 mm / min, the ultrasonic frequency is 20 kHz, and the ultrasonic amplitude is 10 μm.

[0056] The detailed implementation steps are as follows: Step S1: Multi-source sensor signal acquisition and preprocessing. The purpose of this step is to acquire sensor signals from multiple physical dimensions during the drilling process under a unified time reference, providing a complete, synchronous, and low-noise data source for subsequent steps. Since the formation of delamination at the borehole exit of carbon fiber composite materials is affected by the coupling of multiple factors such as axial force, high-frequency vibration impact, material micro-damage, and spindle load changes, a single sensor channel cannot fully capture the precursor features of delamination. Therefore, multi-source sensor signal acquisition is required.

[0057] During the drilling process, a hardware-level timestamp synchronization mechanism is established using a multi-channel data acquisition system to synchronously acquire the following signal channels: axial force signal (acquired by a piezoelectric force gauge), axial vibration acceleration signal (acquired by an acceleration sensor), acoustic emission signal (acquired by an acoustic emission sensor), and electric spindle current signal (acquired by a Hall current sensor).

[0058] Since the above multi-source signals have significant differences in frequency band distribution and sampling rate, the following preprocessing operations are required:

[0059] (I) Filtering Processing. For the axial force signal and spindle current signal, which reflect low-frequency dynamic mechanical characteristics, a Butterworth low-pass filter is used to remove high-frequency electromagnetic interference noise, with the low-pass cutoff frequency set to 1 kHz. For the acoustic emission signal, which reflects the high-frequency transient characteristics of microscopic damage within the material, a high-pass filter is used to remove low-frequency machine tool vibration interference, with the high-pass cutoff frequency set to 100 kHz. The filtered signal is then subjected to envelope detection processing to extract the acoustic emission energy envelope.

[0060] (II) Synchronization Processing. A unified zero-point timestamp is established using the multi-channel hardware triggering mechanism of the data acquisition card. The sampling rate of the axial force signal and the spindle current signal is set to 10 kHz; the sampling rate of the ultrasonic vibration signal is set to 100 kHz; and the original sampling rate of the acoustic emission signal is set to 2 MHz. The high-frequency acoustic emission energy envelope signal and vibration signal are downsampled by decimation or averaging and uniformly aligned to the 10 kHz sampling rate reference of the axial force signal to eliminate time delay mismatch between channels.

[0061] (III) Windowing Processing. A sliding analysis window is established based on the spindle rotation period. In this embodiment, the spindle speed is 3000 r / min, the single rotation period is 0.02 s, and the sliding window length is set to 3 rotation periods, i.e., 0.06 s. Within each sliding window, the mean, peak value, and root mean square value of each signal channel are extracted as timing inputs for subsequent algorithm steps.

[0062] The functional principle of this step lies in sensing the drilling process status from different physical dimensions through multi-source sensors. Axial force reflects the macroscopic cutting load level, vibration signals reflect the dynamic impact characteristics under ultrasonic superposition, acoustic emission signals reflect transient events such as microcrack initiation and fiber breakage within the material, and current signals reflect spindle load changes and tool wear status. After unified clock synchronization and normalization preprocessing, the multi-source signals can provide a multi-dimensional, high-fidelity data foundation for subsequent steps, overcoming the limitations of insufficient information from a single sensor.

[0063] Step S2: Exit Action Window Identification. The purpose of this step is to accurately define the time period during which the delamination risk truly begins to change significantly at the drill bit exit stage, i.e., the "exit action window." The delamination risk of carbon fiber composites is not uniformly present throughout the drilling process, but is concentrated in the final stage where the drill bit approaches the exit and the remaining un-drilled thickness becomes extremely thin. Accurately identifying the starting point of this window can both avoid triggering unnecessary risk calculations and false warnings during the safe steady-state drilling phase, and ensure that the assessment algorithm is activated immediately when entering the truly dangerous stage, without missing any critical risk precursor signals.

[0064] Calculate the remaining thickness at the current moment based on the drill bit's real-time axial feed depth and the workpiece's total thickness:

[0065] The meanings of the symbols in the formula are as follows: For the current moment The remaining thickness, in mm; In this embodiment, the total thickness of the workpiece is... mm; The axial feed depth of the drill bit at the current moment is expressed in mm. The real-time axial feed depth of the drill bit can be obtained through real-time feedback from the encoder of the CNC system, or directly measured by an external spindle displacement sensor.

[0066] When the remaining thickness is less than or equal to the preset outlet threshold thickness At that time, it is determined that the drill bit has entered the exit action window. The preset exit threshold thickness The calculation formula is based on the combined thickness of the single-layer layup of the workpiece and the geometric height of the drill tip:

[0067] The meanings of the symbols in the formula are as follows: In this embodiment, the thickness of a single layer is [not specified]. mm; The layup critical coefficient represents the minimum number of layups that the bottom of the workpiece should contain when the risk of delamination is significant. In this embodiment, it is taken as... ; In this embodiment, the drill bit diameter is... mm; In this embodiment, the drill bit tip angle is... ; The geometric height influence coefficient is taken in this embodiment. .

[0068] Substituting into the calculation, the first candidate value is mm, the second candidate value is Approximately mm (the exact value depends on the precise trigonometric function value), take the larger of the two values ​​as... In this embodiment, it is set mm is used as an exit-sensitive threshold for explanation (i.e., a typical case dominated by layup conditions). The remaining thickness is calculated in real-time. When the value reaches mm, the system triggers the exit action window and begins to activate the subsequent risk assessment algorithm.

[0069] The functional principle of this step is based on the following physical fact: when the drill bit approaches the exit point, the number of remaining un-drilled layers in the laminate is extremely small, and the local bending stiffness and anti-delamination ability of the workpiece at the drill bit tip decreases sharply. Exit threshold thickness. The two-condition maximum value rule ensures that regardless of the thickness of the laminate (determined by the layup factor), the maximum value is determined. (Control) or what angle of drill bit (based on geometric height coefficient) The control mechanism allows the trigger boundary of the exit action window to adaptively match the actual physical failure hazard area. This adaptive mechanism gives the method good generalization ability and makes it applicable to different material-tool combinations.

[0070] Step S3: Constructing the Critical Delamination Bearing Capacity at the Exit. The purpose of this step is to establish a dynamic critical delamination thrust model at the exit, which varies with the remaining thickness, to characterize the workpiece's "remaining anti-delamination capability" at the drill bit exit stage. This critical bearing capacity serves as the physical baseline for subsequent impulse integration; only thrust exceeding this baseline will inject the driving energy leading to delamination into the interlayer interface. Traditional methods use fixed empirical thresholds, which cannot reflect the rapid decay of bearing capacity as the remaining thickness decreases rapidly, easily leading to false alarms in the early stages and false alarms in the later stages. This invention fundamentally solves this problem by constructing a dynamic critical bearing capacity.

[0071] Export critical layer load capacity The construction of is expressed in the following form:

[0072] The meanings of the symbols in the formula are as follows: This is the clamping boundary correction factor, reflecting the influence of the workpiece bottom support conditions on the delamination resistance. It is taken as [value missing] when suspended without backing. When there is backing support, the value is greater than 1; , which is the ply stiffness coefficient, extracted from the laminate bending stiffness submatrix, and is used to characterize the anisotropic anti-bending delamination capability under different ply structures; To support the state correction coefficient; The current remaining thickness, in mm; The drill bit diameter is in mm. The drill bit tip angle; For thickness calibration coefficient, this embodiment takes... ; For the diameter calibration coefficient, this embodiment takes... .

[0073] The specific values ​​of the thickness calibration factor *m* and the diameter calibration factor *n* are derived by fitting experimental data on the bending response of specific carbon fiber laminates. For different material systems and layup structures, the recommended value range for *m* is 1.0 to 2.0, and the recommended value range for *n* is 0.5 to 1.5. The clamping boundary correction factor *Cb* is selected based on the workpiece bottom support stiffness between 0.8 and 1.2. Where the bottom is completely suspended and there is clamp flexibility, *Cb* is below 1.0; for standard suspension without backing, *Cb* = 1.0; and for rigid backing support, *Cb* is above 1.0. Those skilled in the art, when faced with new material-tool-clamping combinations, can conduct a small number of exit bending response pre-tests on the same material, collect bending displacement and force response data under different remaining thicknesses, and obtain the calibration values ​​of *m* and *n* through linear fitting in a double logarithmic coordinate system.

[0074] Among them, the ply stiffness coefficient Bending stiffness submatrix from the ABD matrix of laminated plates Extracted from, its explicit expression formula is:

[0075] The meanings of the symbols in the formula are as follows: Longitudinal bending stiffness reflects the bending resistance of the laminate along the 0° direction; Transverse bending stiffness reflects the bending resistance of the laminate along the 90° direction; The Poisson effect is used to couple the bending stiffness, reflecting the coupling effect of bending deformation in the longitudinal and transverse directions; Torsional stiffness reflects the torsional resistance of the laminate under in-plane shear.

[0076] This explicit expression takes into account the anisotropic characteristics of laminates and can more accurately characterize the layered resistance benchmark under multi-ply structures than the traditional isotropic assumption.

[0077] In this embodiment, it is assumed that the combined constant of the ply stiffness coefficient and the support correction coefficient is... When the remaining thickness When the value is in mm, substitute it into the calculation:

[0078] The underlying physical principle of this step is that the anti-delamination ability of a laminate is essentially determined by its bending stiffness and residual thickness. According to classical laminate theory and the energy release rate criterion in fracture mechanics, when a concentrated thrust is applied to a thin plate, the interlaminar shear stress and normal stress caused by the bending deformation of the plate increase sharply as the plate thickness decreases. Specifically, the contribution of the residual thickness to the critical delamination thrust follows a power function relationship (exponential). This means that as the drill bit approaches the exit point and the remaining thickness decreases from 0.5 mm to 0.2 mm, the critical load capacity does not decrease linearly by half, but rather decays at a much faster rate. (Dynamic) The model accurately captures this nonlinear attenuation law, ensuring that the physical baseline for risk assessment remains consistent with the actual load-bearing capacity of the material. (Ply stiffness coefficient) The introduction of this feature enables the model to distinguish the differentiated effects of different ply sequences (e.g., quasi-isotropic ply and unidirectional ply) on delamination resistance, significantly improving the model's generalization applicability.

[0079] Step S4: Characterization of Exit Dynamic Load and Calculation of Exit Impulse. This step includes two sub-tasks: First, constructing an equivalent dynamic thrust, unifying the contributions of multiple load sources within the exit action window into a comprehensive mechanical characterization quantity; second, using the dynamic critical layered bearing capacity constructed in Step S3 as a benchmark, calculating the accumulation of the overload portion over time, i.e., the exit action impulse. The exit action impulse is the core metric for risk assessment in this invention. Its physical meaning lies in quantifying "the total driving force of loads exceeding the material's bearing capacity accumulated and input to the interlayer interface during the exit stage," capturing the energy conditions for crack propagation from the time integral dimension.

[0080] Within the export action window, define the equivalent dynamic thrust. as follows:

[0081] The meanings of the symbols in the formula are as follows: This represents the average axial force within the exit window, expressed in N, reflecting the steady-state thrust level during the cutting process. This is the peak characteristic quantity of axial force, expressed in N, reflecting transient impact intensity; This is the root mean square value of the axial force, in N, reflecting the total energy level of the force signal; Ultrasonic amplitude, in μm; The frequency is the ultrasonic frequency, measured in Hz. The product of these two values ​​characterizes the intensity of ultrasonic excitation and reflects the additional high-frequency impact contribution of ultrasonic vibration to the exit thrust. It is the fluctuation of spindle current, reflecting abnormal changes in cutting load; represents the weighting coefficients for each item.

[0082] In this embodiment, the weighting coefficient is taken as: , , , At a specific moment The features extracted by the sensor are: N, N, N, N (equivalent force dimension after normalization conversion). N. Substitute into the calculation:

[0083] In the equivalent dynamic thrust, the product of ultrasonic amplitude and ultrasonic frequency is explicitly coupled as an independent load component. The physical basis of this design is that during ultrasonic vibration-assisted drilling, the tool performs high-frequency reciprocating motion in the axial direction at the ultrasonic frequency, and each vibration cycle applies an additional pulse impact to the thin layer at the workpiece exit. The intensity of this pulse impact is proportional to the product of amplitude and frequency. If this term is ignored, the equivalent dynamic thrust will systematically underestimate the actual exit load under ultrasonic conditions, leading to an underestimation of risk.

[0084] Export effect impulse The calculation formula is:

[0085] The meanings of the symbols in the formula are as follows: This is the starting time of the export action window; This is the end time of the export window; This represents the equivalent dynamic thrust at the current moment. This represents the current dynamic critical stratified load capacity of the exit. As the critical load-bearing safety factor, this embodiment takes... Its function is to reserve a certain safety margin based on the actual critical value, so that the risk assessment can issue an early warning before the critical carrying capacity is completely exhausted. The truncation operator indicates that a positive overload contribution is generated only when the equivalent dynamic thrust exceeds the safe bearing threshold; the portion below the threshold is not included in the impulse.

[0086] In this embodiment, the safe carrying threshold is N. Current equivalent dynamic thrust N, the resulting overload difference is N. Assume this overload state occurs within the current integral infinitesimal time interval. If the current infinitesimal element remains stable within s, then its impulse contribution is:

[0087] The physical essence of the impulse at the exit point lies in the fact that interlaminar delamination is a crack propagation process. Whether a crack propagates depends on whether the energy release rate at the crack tip exceeds the interlaminar fracture toughness of the material. The energy release rate depends not only on the instantaneous magnitude of the force but also on the cumulative effect of the overload force over time. A brief spike force (e.g., a 2 ms electromagnetic interference pulse) may have a high instantaneous value, but the total energy input to the interlaminar interface is minimal and insufficient to drive crack propagation; while a moderate but sustained overload force (e.g., a 45 ms chip blockage) has enough accumulated energy to allow the crack to progress from initiation to penetration. The impulse integral is obtained through... The cutoff mechanism precisely "accumulates only the force-time area exceeding the safe bearing capacity," mathematically corresponding precisely to the physical process of "overload energy accumulation driving crack propagation" in fracture mechanics. This mechanism enables the system to fundamentally distinguish between "harmless instantaneous signal fluctuations" and "continuous overloads leading to stratification," and is the core technical means to achieve "no false alarms for instantaneous peak values ​​and no missed alarms for continuous overloads."

[0088] Step S5 (Optional Enhancement Step): Construction of Damage Sensitivity Correction. The purpose of this step is to construct a correction coefficient that comprehensively reflects the influence of factors such as tool wear, thermal softening due to temperature rise, interlayer interface deterioration, and layup angle mismatch on delamination tendency. This coefficient is used to reduce and correct the critical delamination load in Step S3. In actual long-sequence machining, tool wear leads to edge dulling, increasing the actual cutting force; the resin matrix softens under the action of cutting heat, reducing the interlayer bond strength; and the interlayer shear stress concentration at interfaces with abrupt changes in layup direction is more severe. All these factors reduce the actual delamination resistance of the workpiece. If the critical load is not corrected, the risk assessment will systematically underestimate the risk in the later stages of tool wear and under harsh working conditions.

[0089] Define damage sensitivity coefficient as follows:

[0090] The meanings of the symbols in the formula are as follows: This is a metric for tool wear, reflecting the amplification effect of tool wear on cutting force. It can be estimated in real time through cumulative cutting time mapping or frequency domain characteristic offset of acoustic emission signals. In this embodiment, we take... (Wear condition corresponding to 50 holes); Temperature or thermal softening is a characterization measure reflecting the degree of resin matrix softening caused by cutting heat. It can be estimated using a spindle power-driven heat input model. In this embodiment, we take... ; This is a characterization measure of interlayer interface degradation, reflecting the initial interface defect state of the workpiece. For standard new sheet metal... ; The characteristic quantity for the mismatch of adjacent ply angles is obtained by reading the ply design file, acquiring the absolute value of the angle between two adjacent fiber layers at the current cut interface and normalizing it by dividing by 90°. The weighting coefficients for each correction factor are taken in this embodiment. , , , (Simplified case).

[0091] The weighting coefficients of all the above correction factors meet the non-negativity requirement, that is, λ1, λ2, λ3, and λ4 are all greater than or equal to zero. The calibration method for each coefficient is consistent with the calibration method for the equivalent dynamic thrust weighting coefficient in step S4, that is, the contribution of each degradation factor (tool wear, cutting temperature, initial interface quality, and ply angle mismatch) in the orthogonal cutting test to the variance analysis of the exit delamination factor is proportionally allocated. In the simplified implementation scenario, if the processing object is a standard new plate and the influence of ply angle mismatch is not considered, λ3 and λ4 can be set to zero, and only the two main correction factors of tool wear and thermal effect are retained.

[0092] Substitute into the calculation:

[0093] The critical load-bearing capacity is reduced and corrected using the damage sensitivity coefficient:

[0094] In the formula: This is the corrected critical stratification load capacity at the outlet. Because... The revised critical load capacity is lower than the original value, indicating that under the conditions of tool wear and temperature rise, the actual thrust that the workpiece can safely withstand is reduced, and the risk of delamination increases accordingly.

[0095] The physical principle behind this step is that the critical delamination load essentially represents the "upper limit of the workpiece's resistance to delamination under ideal conditions," but in actual machining, there are various degradation factors that lower this upper limit. Tool wear dulls the cutting edge, generating a larger axial thrust under the same feed parameters, equivalent to the material needing to withstand a larger load; resin thermal softening reduces the fracture toughness of interlaminar bonding, equivalent to a decrease in the material's resistance to delamination; the greater the layup angle mismatch, the more severe the interlaminar shear stress concentration at the interface, and the easier it is to induce delamination. Damage sensitivity coefficient. These degradation factors are unified into a reduction factor for the critical load capacity, so that the risk assessment maintains physical accuracy across the entire operating range.

[0096] Step S5 continued: Hierarchical Risk Index Normalization and Fusion. The purpose of this step is to normalize and weightedly fuse the characteristic quantities of different physical dimensions—mechanical bearing capacity ratio, impulse accumulation ratio, acoustic emission anomaly, and vibration anomaly—to map them into a comprehensive hierarchical risk index between 0 and 1. The significance of multi-dimensional fusion is that single-dimensional indicators may become ineffective due to sensor blind spots or signal attenuation under specific operating conditions, while multi-dimensional cross-validation can significantly improve the reliability of risk assessment.

[0097] First, define the following four normalization quantities:

[0098] In the formula: Thrust ratio, characterizing the ratio of equivalent dynamic thrust to the corrected critical load capacity. This means that the thrust has been overloaded.

[0099] In the formula: For impulse ratio, To boost export performance, The critical impulse reference value (calibrated by a defect-free pre-test) is taken in this embodiment. N·s.

[0100] In the formula: This is an anomalous quantity of acoustic emission. This is a characteristic quantity of acoustic emission within the exit window. This is a reference value for acoustic emission.

[0101] In the formula: This is an abnormal vibration quantity. This is a characteristic quantity of vibration anomalies within the exit window. This is the vibration reference value.

[0102] This invention provides two forms of risk index fusion mapping. The first is a linear weighted mapping:

[0103]

[0104] In the formula: Let be the weight coefficient, and satisfy... This embodiment takes , , , This form is suitable for stable cutting conditions where the range of values ​​for each normalized sub-index is narrow and there are no extreme abnormal changes. It has low computational complexity and is easy to implement ultra-high frequency real-time calculations in low-level controllers with limited computing power.

[0105] The second type is an exponentially decaying nonlinear mapping:

[0106]

[0107] This approach is suitable for harsh cutting conditions where multi-source signals are susceptible to strong interference, resulting in extreme singular values. When a sub-index experiences a sudden increase of tens of times due to local material defects, the saturation asymptotic properties of the natural exponential function can effectively compress the impact of extreme values, preventing a single index anomaly from causing the overall risk index to severely exceed limits or triggering false warnings, thereby significantly improving the robustness of the system.

[0108] In practical engineering implementation, the automatic selection of the above two mapping forms can be achieved by setting quantitative statistical characteristic criteria. Specifically, the system can calculate each normalized sub-index (thrust ratio) within the sliding time window in real time. impulse ratio Acoustic emission anomalies Abnormal vibration The coefficient of variation of any sub-index is used. When the coefficient of variation of any sub-index exceeds a preset discrete threshold, or when the instantaneous value of an index undergoes a step change exceeding a set multiple of the historical stationary mean, the system determines that it has entered a severe cutting state and automatically switches to the exponentially decaying nonlinear mapping form; otherwise, the system defaults to the linear weighted mapping form. This quantitative criterion enables the algorithm to adaptively match the optimal fusion strategy during the processing.

[0109] This embodiment uses a linear weighted mapping method. Substituting the specific values ​​of each normalized quantity:

[0110]

[0111] The physical basis of multi-dimensional normalization fusion lies in the fact that the formation of layered defects is the result of the synergistic effect of multiple factors such as mechanical overload, energy accumulation, microscopic damage, and dynamic impact. Thrust ratio The impulse ratio reflects whether the instantaneous mechanical state is overloaded. Abnormal acoustic emission reflects whether overload energy has accumulated to a dangerous level. This reflects whether microcracks have already appeared inside the material, and the abnormal vibration amount. It reflects whether dynamic shocks have intensified abnormally. By integrating the four dimensions into a single index, even if the sensitivity of one sensing channel decreases due to operating conditions, other channels can still compensate to maintain the accuracy of risk assessment.

[0112] Step S6: Physical Constraint Consistency Assessment. The purpose of this step is to impose physical constraints on the original stratification risk index obtained from the fusion calculation, ensuring that the final output risk index does not violate the physical mechanism of stratification at the borehole exit of carbon fiber composite materials in terms of numerical behavior. Since the risk index is calculated based on weighted summaries of noisy multi-source sensor signals, the original calculated value may exhibit anomalous results that violate physical mechanisms, such as "increased feed rate but decreased risk" or "decreased remaining thickness but decreased risk index fluctuation," due to sensor fluctuations, signal spikes, or algorithm fitting errors. If this is not corrected, the system may generate misleading judgments during engineering deployment, undermining operators' confidence in risk warnings.

[0113] This step includes three levels of constraint and correction mechanisms:

[0114] Boundary constraints and corrections: the risk index satisfies the range constraints. When the original calculated value When, it was corrected to ;when When, it was corrected to This constraint ensures that the output of the risk index always has a physically interpretable probabilistic meaning.

[0115] Monotonicity constraints and corrections, within the exit action window, under the condition that other factors remain unchanged, the risk evolution must conform to the following physical laws: as the feed rate increases, the risk index does not decrease; as tool wear intensifies, the risk index does not decrease; as the remaining thickness decreases, the risk index does not decrease; as the clamping support stiffness increases, the risk index does not increase.

[0116] To achieve the aforementioned monotonicity constraint, when a violation of the monotonically increasing law of risk index is detected, a joint correction strategy of "sliding window smoothing and historical extreme value envelope" is adopted. First, a width of... The original risk index is smoothed and denoised using a sliding time window to obtain a smoothed value:

[0117] In the formula: This represents the risk index smoothed by a sliding window at the current moment. The width of the sliding window is measured in the number of sampling points.

[0118] Subsequently, a historical extreme value envelope truncation function is introduced to update the risk index of the final output:

[0119] In the formula: This is the corrected risk index for the current moment; The start time of the self-exit action window Up to the current moment The historical maximum risk index value. The function of this envelope truncation function is to ensure that even if the instantaneous calculated value at the current moment drops temporarily due to high-frequency fluctuations of the sensor or local glitches at the interface, the corrected output will not be lower than the historical maximum value, thus forcibly preserving the historical maximum risk state and avoiding a sudden drop in risk that contradicts the physical failure mechanism in the ultra-thin stage.

[0120] Logical threshold consistency check, when thrust ratio impulse ratio And the anomalous amount of acoustic emission When there is no sustained threshold jump, the output risk level should not exceed the warning threshold to avoid false high-risk judgments caused by numerical coincidences in multi-source fusion. or If the abnormal acoustic emission level continues to exceed the threshold, the output should be at least at the medium-high risk level to avoid missed detection due to other indicators being low under clear overload conditions.

[0121] Based on the revised stratified risk index, the risk is divided into four levels:

[0122] Low-risk phase ( ): The current export capacity is sufficient, and the system will maintain the current process parameters unchanged.

[0123] Medium risk stage ( The output thrust has approached the critical load range, triggering a primary warning. The system adaptively reduces the feed rate by 10% to 20% to reduce the rate of buffer accumulation.

[0124] High-risk phase ( If the thrust is overloaded or there is a continuous high-energy acoustic emission anomaly, the system triggers an advanced warning, forcibly reducing the feed rate by 30% to 50%, while reducing the ultrasonic amplitude by 20% to 30% to weaken the local high-frequency impact.

[0125] Extremely high risk stage ( When macroscopic layering is about to occur or has already occurred, the system immediately triggers a spindle emergency stop and tool retraction command, and prompts for a forced tool change or the addition of an exit backing support when machining the next hole.

[0126] In this embodiment, the level threshold is calibrated as follows: , , The calculations obtained above ,satisfy The system determines that the current risk level is extremely high and immediately triggers the spindle emergency stop and tool retraction command.

[0127] The underlying principle of physical constraint consistency assessment lies in the fact that interlaminar damage in carbon fiber composites is an irreversible process. Once microcracks initiate at the interface, they will only propagate under the subsequent continuous exit thrust and will not heal on their own. Therefore, in the exit stage, as the remaining thickness continues to decrease and the material's load-bearing capacity continues to decline, the risk of delamination should only increase monotonically or remain unchanged, not decrease. The historical extreme value envelope mechanism mathematically guarantees this physical irreversibility. Sliding window smoothing eliminates meaningless jitter caused by high-frequency sensor noise before ensuring monotonicity, ensuring that the monotonicity constraint acts on the smoothed trend signal rather than the original noise signal, avoiding overly conservative judgments due to noise maxima. Logical threshold consistency verification serves as the final safety net, preventing misjudgments caused by numerical coincidences from the perspective of cross-validation of multiple physical quantities. The three-layer constraint mechanism is progressive, jointly ensuring the physical consistency, engineering credibility, and decision-making safety of the risk index output.

[0128] like Figure 2 As shown in the figure, with drilling time on the horizontal axis and remaining thickness on the vertical axis, the graph illustrates the continuous decreasing trend of the remaining thickness throughout the entire process from the drill bit entering the workpiece to drilling through and exiting the drill bit. The preset exit threshold thickness is marked with a horizontal dashed line. The location. When the remaining thickness curve crosses downwards from above. The corresponding horizontal line is the trigger moment of the exit action window. ; from then until the drill bit completely penetrates the workpiece (the remaining thickness is reduced to zero). This time interval is the exit action window. The figure also marks the shaded area within the exit action window, visually demonstrating the precise time period during which the risk assessment algorithm is activated. This figure demonstrates two key facts: first, the triggering timing of the exit action window is entirely determined by the real-time comparison between the remaining thickness and the exit threshold, possessing clear physical criteria rather than being set by human experience; second, the exit threshold thickness... The value of the threshold directly determines the lead time for risk assessment. A larger threshold triggers the window earlier and provides a more ample lead time for warnings, but also increases the computational load. Conversely, a smaller threshold triggers the window later, reducing the computational load but decreasing the warning margin. This invention determines the threshold using an adaptive rule that takes the maximum value of both the layup thickness and the tool geometry. A reasonable engineering balance was achieved between advance warning and computational efficiency.

[0129] like Figure 3 As shown in the diagram, this figure illustrates the normalization and fusion process of the hierarchical risk index in the form of a structural block diagram. The left side of the figure shows four input channels: thrust ratio... (Ratio of equivalent dynamic thrust to corrected critical load capacity), impulse ratio (Ratio of exit impulse to critical impulse baseline), acoustic emission anomaly (Ratio of acoustic emission characteristics within the exit window to reference value) and vibration anomaly quantity (The ratio of the vibration characteristic quantity within the exit window to the reference value). After weighting, the four normalized quantities are fused into a single hierarchical risk index through either a linear weighted mapping or an exponentially decaying nonlinear mapping. The figure demonstrates that the risk assessment of this invention does not rely on a simple threshold judgment of a single physical quantity, but rather cross-integrates four independent physical dimensions: mechanical load state, energy accumulation degree, precursory microscopic damage to materials, and dynamic impact anomalies. This multi-dimensional fusion mechanism ensures that even if a sensing channel experiences signal attenuation or temporary failure due to specific operating conditions, other channels can still compensate and maintain the accuracy of risk assessment, thereby significantly improving the robustness and reliability of the system.

[0130] like Figure 4 As shown, the graph uses the remaining thickness as the horizontal axis (decreasing from left to right, simulating the process of the drill bit gradually approaching the exit), and plots two key curves: one representing the equivalent dynamic thrust. The curves show the changes as drilling progress, and the other curve represents the dynamic exit critical stratification load. The decay curve as the remaining thickness decreases. During the early and middle stages of drilling, Much higher The existence of a sufficient safety margin between the two curves indicates that the workpiece's resistance to delamination at this point far exceeds the actual thrust, and the risk of delamination is extremely low. As the remaining thickness continues to decrease, The curve drops sharply in the form of a power function, while The curve shows an upward trend due to the deterioration of the cutting conditions at the exit stage (such as difficulty in chip removal and superimposed ultrasonic impact). When the two curves intersect, Exceed Entering the overload zone. This is marked by the shaded area in the diagram. Exceeding The integral of the shaded area over time within the region of (safety load threshold) is the exit impulse. This figure visually demonstrates two core arguments from the perspective of mechanical evolution: First, the critical bearing capacity must be dynamic. If a fixed threshold is used (represented by the horizontal dotted line in the figure), the threshold will be too low in the early stages, leading to a large area of ​​false alarms, while the threshold will be too high in the extremely thin stage, leading to missed alarms. A dynamic threshold, however, is crucial. The curve accurately tracks the decay trajectory of the material's actual load-bearing capacity; secondly, impulse (shadow area) rather than peak force is the correct measure of delamination risk. Under the same peak force, the size of the shadow area depends on the duration of overload. The larger the area, the more fracture driving energy is input to the interlaminar interface, and the higher the delamination risk.

[0131] like Figure 5As shown in the diagram, this modular logic block diagram illustrates the organic coupling relationship between the three core technical features of this invention. The diagram is presented in three layers: the bottom layer represents the dynamic critical layered load-bearing capacity. The building block (labeled "baseline layer") takes into account the remaining thickness, ply stiffness coefficient, clamping conditions, and tool geometry parameters, and outputs a dynamically updated critical load baseline; the intermediate layer is the exit impulse. The computation module (labeled "metric layer") receives data from the baseline layer. and from signal acquisition It performs overload truncation and time integration calculations, and outputs the impulse value; the top layer is the physical constraint consistency assessment module (labeled "verification layer"), which receives the fusion risk index from the metric layer. After performing boundary truncation, monotonicity envelope correction, and logical consistency verification, the final risk level is output. The three levels are connected by unidirectional arrows, representing strict data dependency chains. This diagram demonstrates the "dynamic..." Build "" The overload integral calculation, physical constraint consistency assessment, and the other two are not simply a stack of functions, but rather form an inseparable technical closed loop of "benchmark → metric → verification": the benchmark layer provides a physical reference for the metric layer, without dynamic... Without an impulse mechanism, the impulse integral will lose its benchmark for determining overload; the metric layer provides quantitative input to the verification layer, but without an impulse mechanism, the risk index will degenerate into a peak judgment that cannot distinguish between instantaneous fluctuations and sustained overload; the verification layer applies physical law filtering to the output of the metric layer, but without physical constraints, the numerical results of multi-source fusion may exhibit anomalies that violate the mechanism due to noise. The organic coupling of the three closed loops is the fundamental technical guarantee for achieving high accuracy and high robustness in online hierarchical risk assessment in this invention.

[0132] like Figure 6 As shown in the figure, the horizontal axis represents the time within the export action window, and the vertical axis represents the stratified risk index. Using the vertical axis as the ordinate, three curves were plotted for comparison: The first curve is the original calculated stratified risk index (without any constraints). This curve exhibits a jagged, non-monotonic characteristic with several brief downward fluctuations caused by high-frequency noise from the sensor or local interface burrs, interspersed with an overall upward trend; the second curve is the risk index curve after sliding window smoothing, where the high-frequency jaggedness is significantly suppressed, and the curve becomes smoother, but a few small local downward segments still exist; the third curve is the final output curve after joint correction by "sliding window smoothing + historical extreme value envelope truncation". This curve exhibits a strictly non-decreasing characteristic, monotonically increasing or remaining unchanged over time within the exit action window, without any decline. This figure visually demonstrates the effectiveness and necessity of the physical constraint consistency assessment mechanism: the anomalous decline in the original curve is physically unreasonable. In the exit stage, the remaining thickness will only continue to decrease, and the material's load-bearing capacity will only continue to decay; the stratified risk should not physically decline. If the anomalous decline of the original curve is directly output to the CNC system, the system may mistakenly deactivate the high-level warning when entering an extremely high-risk phase due to a brief drop in the index, causing serious safety hazards. The final curve, after joint correction, strictly adheres to the monotonicity of physical laws, eliminating spurious fluctuations caused by sensor noise and ensuring the reliability of the risk index output and the safety of decision-making in engineering deployment. Meanwhile, Figure 6 This also shows that both the sliding window smoothing and the historical extreme value envelope correction are indispensable: smoothing alone without envelope truncation cannot completely eliminate all local fallbacks; envelope truncation alone without smoothing may produce an overly conservative "height-locking" effect due to noise spikes. The combined use of the two ensures monotonicity while avoiding over-conservatism, achieving the optimal balance between robustness and sensitivity.

[0133] The steps of the method of the present invention are not simply a functional stack, but an organically coupled whole formed based on the physical evolution mechanism of carbon fiber composite material outlet stratification, with strict physical dependence and logical progression between the steps.

[0134] The multi-source signal acquisition in step S1 provides the data foundation for all subsequent steps. The exit action window identification in step S2 defines the time boundary for the entire risk assessment, ensuring that computational resources are focused on the truly dangerous exit phase. The dynamic critical hierarchical carrying capacity constructed in step S3 is the physical baseline for the impulse integral in step S4; without dynamic... As a benchmark, the impulse integral will lose its physical reference for "what to use as a standard to judge overload," degenerating into a meaningless force-time area calculation. The exit action impulse in step S4 is the core metric for risk quantification; if only the thrust ratio is retained... And remove impulse If the peak value is the same but the duration is different, it is impossible to distinguish between two operating conditions that have different durations, and thus impossible to distinguish between "instantaneous harmless fluctuations" and "continuous damaging overloads". The normalization fusion in step S5 integrates information from multiple physical dimensions into a single index, achieving multi-dimensional cross-validation. The physical constraint consistency assessment in step S6 applies physical law verification to the fused index to ensure that the output does not violate basic physical laws such as the irreversibility of damage.

[0135] The above-mentioned collaborative relationship can be summarized as: dynamic It provides a "ruler" (benchmark) for measurement. Integration completes the "weighing" (core) of measurement, while physical constraints realize the "calibration" (verification) of measurement. All three are indispensable and together form a complete technical closed loop from physical modeling to quantitative evaluation and then to logical verification.

[0136] The beneficial effects of this invention include: First, it enables quantitative risk assessment before macroscopic stratification at the exit point, representing a feedforward quality control approach, distinct from traditional post-detection methods. Second, by coupling the exit critical load-bearing capacity with the exit dynamic impulse, it overcomes the misjudgment problem caused by solely relying on the peak axial force threshold. Third, the introduction of a dynamic critical stratification thrust model driven by remaining thickness reflects the real physical process of rapid attenuation of workpiece load-bearing capacity as the drill bit approaches the exit point. Fourth, through physical constraint consistency assessment, it ensures the rationality of the risk index's variation with key variables, avoiding judgments contradicting the underlying mechanism. Fifth, the entire method balances mechanical interpretability and engineering deployability, and can be directly applied to online monitoring, adaptive control of process parameters, and tool life management.

[0137] The working principle of the method of this invention can be summarized as a closed-loop control link of eight stages: "sensing → positioning → benchmark establishment → measurement → correction → fusion → verification → decision".

[0138] In the sensing phase, the system synchronously collects mechanical, vibration, acoustic, and electrical signals from multiple sensors during the drilling process to obtain comprehensive status information at the exit stage. In the positioning phase, the system accurately locates the start time of the exit risk window by comparing the real-time remaining thickness with the adaptive exit threshold, avoiding wasting computing resources or generating false warnings during the safety phase.

[0139] In the benchmarking phase, the system, based on the theory of laminate bending stiffness and the fracture mechanics energy release rate criterion, constructs a critical delamination bearing capacity that dynamically decreases with the remaining thickness, providing a physical benchmark for overload determination that changes synchronously with the actual state of the material. In the impulse phase, the system uses the dynamic benchmark as a reference to calculate the cumulative integral of the equivalent dynamic thrust exceeding the safe bearing threshold within the exit time window, capturing the driving conditions for crack propagation from a time-energy dimension.

[0140] In the correction phase, the system reduces the critical load capacity based on degradation factors such as tool wear, temperature rise, and layup mismatch, ensuring that the risk assessment adapts to real-world conditions across all operating scenarios. In the fusion phase, the system normalizes and weights the four dimensions—mechanical ratio, impulse ratio, acoustic emission anomaly, and vibration anomaly—to output a comprehensive, stratified risk index.

[0141] In the verification phase, the system applies a three-layer physical law filter to the risk index: boundary constraints, monotonicity envelope constraints, and logical threshold consistency checks. This eliminates anomalous outputs caused by sensor noise and algorithm fitting errors, ensuring that the final result is physically self-consistent and engineering-reliable. In the decision-making phase, the system classifies the risk index based on the verified risk level and outputs graded handling instructions ranging from "keep parameters unchanged" to "spindle emergency stop and tool retraction," achieving real-time assessment and closed-loop control of the layered risk at the exit of ultrasonic vibration-assisted drilling of carbon fiber composite materials.

[0142] The fundamental reason why the method of this invention can produce the above-mentioned technical effects is that it grasps the physical essence of delamination formation in carbon fiber composite materials at the exit point. Delamination is an energy-driven interlaminar crack propagation process, rather than a simple event of excessive force. The dynamic critical load capacity accurately tracks the material's load-bearing capacity decay law with drilling depth, the exit impulse captures the cumulative effect of overload energy over time, and physical constraints ensure that the assessment results adhere to the principle of irreversible damage. These three elements form a rigorous logical closed loop at the physical mechanism level, giving the risk index a clear physical correspondence and interpretability, rather than a black box output generated by data fitting. Therefore, it can maintain stable assessment accuracy and engineering applicability under different layups, different tool conditions, and different processing parameters.

[0143] While the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the invention. Those skilled in the art can make various modifications and refinements without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

Claims

1. A method for determining the risk index of layering in ultrasonic vibration-assisted drilling of carbon fiber composite materials, characterized in that, Includes the following steps: S1: During the ultrasonic vibration-assisted drilling process of carbon fiber composite materials, multi-source sensing signals of the drilling process are collected. The multi-source sensing signals include at least axial force signals, vibration signals and acoustic emission signals. S2: Determine the remaining thickness based on the real-time axial position of the drill bit and the total thickness of the workpiece, and identify the exit action window based on the comparison relationship between the remaining thickness and the preset exit threshold thickness; S3: Based on the remaining thickness, layup parameters, clamping support parameters, and tool geometry parameters, construct the exit critical layer load capacity that dynamically changes with the remaining thickness; S4: Within the exit action window, the equivalent dynamic thrust is extracted based on the multi-source sensor signal, and the overload portion of the equivalent dynamic thrust exceeding the exit critical layer load capacity is integrated over the time interval of the exit action window, with the exit critical layer load capacity as the benchmark, to obtain the exit action impulse. S5: The ratio of the equivalent dynamic thrust to the critical stratification load at the exit, the ratio of the exit action impulse to the critical impulse benchmark, and the abnormal quantities of acoustic emission and vibration are normalized and fused to obtain the stratification risk index. S6: Perform a physical constraint consistency assessment on the stratified risk index. The physical constraint consistency assessment includes at least boundary constraints and monotonicity constraints. Output the stratified risk level based on the assessed stratified risk index.

2. The method according to claim 1, characterized in that, In step S3, the specific method for constructing the critical stratified bearing capacity at the outlet is as follows: The power function of the remaining thickness is used as the dominant attenuation term, the power function of the drill bit diameter is used as the scaling term, the reciprocal of the arctangent function of the drill bit tip angle is used as the geometric correction term, and the ply stiffness coefficient extracted from the laminate bending stiffness submatrix and the clamping boundary correction coefficient determined by the bottom support conditions are introduced as multiplicative correction factors. The terms are multiplied together to obtain the critical delamination bearing capacity at the outlet. The ply stiffness coefficient is used to characterize the anisotropic anti-bending delamination capacity under different ply structures, and the outlet critical delamination load decreases monotonically as the remaining thickness decreases.

3. The method according to claim 2, characterized in that, The ply stiffness coefficient is extracted from the bending stiffness submatrix D in the ABD matrix of the laminate, and its expression is as follows: The arithmetic square root of the product of longitudinal bending stiffness and transverse bending stiffness is added to the sum of the Poisson effect coupled bending stiffness and twice the torsional stiffness, and this sum is taken as the value of the ply stiffness coefficient.

4. The method according to claim 1, characterized in that, In step S2, the preset outlet threshold thickness is determined as follows: Calculate the first candidate value and the second candidate value respectively, and take the larger of the two values ​​as the preset exit threshold thickness; wherein, the first candidate value is the product of the ply critical coefficient and the single ply thickness; the second candidate value is the product of the geometric height influence coefficient and the drill bit tip angle geometric cutting height, wherein the drill bit tip angle geometric cutting height is determined by the product of half the drill bit diameter and the reciprocal of the tangent of the drill bit tip angle.

5. The method according to claim 1, characterized in that, In step S4, the equivalent dynamic thrust is calculated as follows: The mean value of axial force within the outlet action window is used as the basic term. The peak characteristic value of axial force, root mean square value, product of ultrasonic amplitude and ultrasonic frequency, and main shaft current fluctuation are used as weighted compensation terms. Each compensation term is multiplied by its corresponding weighting coefficient and then summed with the basic term to obtain the equivalent dynamic thrust. The product of ultrasonic amplitude and ultrasonic frequency is explicitly coupled as an independent load component to characterize the additional contribution of ultrasonic vibration to the exit dynamic thrust.

6. The method according to claim 1, characterized in that, In step S4, the specific calculation method for the exit impulse is as follows: Within the start and end time interval of the exit action window, the difference between the equivalent dynamic thrust and the critical bearing safety factor multiplied by the critical stratified bearing capacity of the exit is calculated time by time. When the difference is positive, the difference is taken; when it is negative, it is taken as zero. The truncated difference is integrally performed over the entire time interval of the exit action window to obtain the exit action impulse.

7. The method according to claim 1, characterized in that, It also includes the step of constructing the damage sensitivity correction factor: The damage sensitivity coefficient is obtained by multiplying the tool wear characterization, temperature or thermal softening characterization, interlayer interface deterioration characterization, and adjacent ply angle mismatch characterization by the corresponding correction coefficients, summing them, and adding one. The critical stratification load capacity at the outlet is reduced and corrected using the damage sensitivity coefficient to obtain the corrected critical stratification load capacity at the outlet. The corrected critical stratification load capacity at the outlet is then used to replace the original critical stratification load capacity at the outlet in subsequent risk index fusion calculations.

8. The method according to claim 1, characterized in that, In step S5, the normalization fusion has two optional mapping forms: The first method is a linear weighted mapping form, in which each normalized quantity is multiplied by its corresponding weight coefficient and then summed to obtain the hierarchical risk index. The second type is the exponential decay nonlinear mapping form, which involves multiplying each normalized quantity by its corresponding weight coefficient and summing the results to obtain the exponential parameter. This parameter is then substituted into the negative exponential form of the natural exponential function to calculate the stratified risk index. The linear weighted mapping form is suitable for stable cutting conditions where the range of normalized values ​​is narrow, while the exponentially decaying nonlinear mapping form is suitable for severe cutting conditions with extreme singular values.

9. The method according to claim 1, characterized in that, In step S6, the specific implementation of the monotonicity constraint is as follows: First, a sliding time window of a set width is used to smooth and denoise the original calculated hierarchical risk index to obtain a smoothed risk index. Then, the historical extreme value envelope truncation function is introduced, and the larger value between the smoothed risk index and the historical maximum risk index from the start time of the exit action window to the current time is taken as the corrected current time stratified risk index. This ensures that the layering risk index does not abnormally decrease under conditions of increased feed rate, accelerated tool wear, or reduced remaining thickness.

10. The method according to claim 1, characterized in that, Step S6 further includes logical threshold consistency verification: When the ratio of the equivalent dynamic thrust to the corrected exit critical stratification load is less than one, and the ratio of the exit action impulse to the critical impulse benchmark is less than one, and the acoustic emission anomaly does not show a continuous over-threshold jump, the stratification risk level is limited to no higher than the warning level. When any of the ratios is greater than or equal to one, or when the amount of acoustic emission anomaly continues to exceed the threshold, the stratified risk level will be determined to be at least medium-high risk level. Based on the magnitude of the risk index, the risk is divided into at least low risk, medium risk, high risk and extremely high risk levels, and corresponding output commands are given to keep the parameters unchanged, reduce the feed rate, reduce the feed rate and reduce the ultrasonic amplitude, and urgently stop the spindle and retract the tool.

Images