Ultrasonic welding quality online detection system and method

Abstract

The invention discloses an ultrasonic welding quality online detection system and method, and belongs to the technical field of laser non-contact measurement. According to the invention, the laser vibration measurement probe is rigidly coupled to the welding head to construct a follow-up-common-optical-path measurement unit, so that synchronous acquisition of vibration signals with zero mass loading is realized; further extracting an attenuation section after welding is finished from the signal, and inverting an equivalent modal damping ratio representing an interface energy dissipation state based on an exponential attenuation model; meanwhile, multi-dimensional physical indexes such as an energy transfer efficiency index Itrans and an input stability index Iin are calculated, and in combination with an energy absorption state index, collaborative fusion is carried out through a weighted fusion formula to obtain a comprehensive quality score; and finally, realizing quality attenuation early warning based on time sequence trend analysis of a scoring sequence. On-line sensing of the deep dynamic state in the welding process is achieved, the early defect recognition sensitivity and the system anti-interference capacity are remarkably improved, and the predictive maintenance capacity of quality degradation is achieved.

Description

Technical Field

[0001] This invention relates to the field of laser non-contact measurement technology, and in particular to an online ultrasonic welding quality inspection system and method. Background Technology

[0002] Ultrasonic welding is a technique that forms a strong bond between similar or dissimilar materials through high-frequency vibration. It boasts significant advantages such as a small heat-affected zone, high weld strength, no need for filler materials, and high efficiency and environmental friendliness, showing broad application prospects in numerous fields including electronics, aerospace, new energy vehicles, and medical equipment manufacturing. However, various defects that may occur during the welding process can directly affect product performance and lifespan, and even pose safety hazards. Due to the short cycle and narrow process window of ultrasonic welding, real-time monitoring of the welding process is challenging. Furthermore, the limited weld joint area and high sensitivity to various factors such as welding machine status, workpiece surface conditions, and welding parameters further complicate quality assessment.

[0003] Currently, while traditional testing methods such as tensile strength testing and contact resistance testing can directly reflect the quality and performance of welded joints, they are all post-weld inspection methods and are destructive, making it difficult to achieve comprehensive, non-destructive, and real-time quality monitoring. These methods have low inspection efficiency and are no longer suitable for the high quality control requirements of large-scale industrial production. Furthermore, existing technologies are mostly limited to fixed-point measurements, making it difficult to capture the dynamic evolution of the welding system in real time, resulting in insufficient information entropy for quality criteria. Fixed sensor layouts are susceptible to interference from operating condition drift and environmental noise coupling, leading to a high false positive rate. Traditional threshold discrimination only focuses on two-dimensional amplitude-frequency characteristics, failing to explore deeper physical invariants such as energy transfer efficiency, modal damping, and multiple vibration modes, resulting in low signal-to-noise ratios for early micro-defect responses. Moreover, most methods are offline and post-inspection, lacking the ability to predict quality degradation trends based on system dynamic characteristics. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides an online ultrasonic welding quality detection system and method. The invention constructs a follow-up-common optical path measurement unit by rigidly coupling a laser vibration probe to the ultrasonic welding head, achieving zero-mass loading and high-fidelity synchronous acquisition of vibration signals at the welding head-workpiece interface. Based on the acquired vibration velocity signals, an exponential decay model is used to fit the vibration decay segment after welding, and the equivalent modal damping ratio, which characterizes the energy dissipation at the interface, is retrieved online. Simultaneously, a system incorporating energy transfer efficiency indicators was constructed. I trans Energy input stability index I in and energy absorption state index The comprehensive quality score is obtained by fusing multiple physical indicators, including those from various dimensions, using a weighted fusion formula. Ultimately based on Time-series trend analysis of values ​​enables early warning of quality degradation. It significantly improves the sensitivity of early defect identification, greatly enhances the system's anti-interference capability under complex operating conditions, and realizes the leap from "post-event judgment" to "process early warning".

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: An online ultrasonic welding quality detection method, comprising the following steps: S1: using a laser vibration probe rigidly coupled to the ultrasonic welding head for follow-up measurement, non-contactly acquiring the normal vibration velocity signal of the welding head-weld seat coupling interface during the welding process in real time. v n ( t ); S2: The vibration velocity signal is processed using an adaptive filtering algorithm. v n ( t Preprocessing is performed to obtain the filtered signal. ; S3: Extract the filtered signal Root mean square value in RMS Frequency domain characteristics and vibration decay signals after welding. v ( t The envelope amplitude of the signal during the vibration decay stage is then analyzed using an exponential decay model. A ( t The equivalent modal damping ratio is obtained by fitting and inverting. The envelope amplitude A ( t )for: in, A 0 represents the initial amplitude of attenuation. oh n For the system's inherent frequency, t For discrete time sequence numbers; S4: Based on the vibration velocity signal, construct and weighted fuse multi-dimensional quality indicators to calculate the comprehensive quality score. The quality is determined according to the quality judgment rules; the multi-dimensional quality indicators include at least the energy input stability index. I in System stability indicators I sys Energy transfer efficiency index I trans and energy absorption state index The formula for the weighted fusion is: ,in, oh 1 、oh 2 、 oh 3 、oh 4 represents the weighting coefficients determined based on historical data optimization; S5: Comprehensive quality score based on continuous welding process The sequence is analyzed by calculating its moving average through a sliding window and examining its changing trend. When the overall quality score is detected... When a continuous downward trend is observed, a quality degradation warning is triggered.

[0006] Preferably, the follow-up measurement is achieved by rigidly fixing the laser vibration probe to the side of the welding head, and the measuring optical axis of the laser vibration probe is coaxial with the normal of the working surface of the welding seat.

[0007] Preferably, the formula for the adaptive filtering algorithm is as follows: in, This is the filtered signal; The original signal; For the first i The weighting coefficients of an adaptive filter; t For discrete time sequence numbers; Environmental vibration and noise; M This represents the order of the filter used during filtering.

[0008] Preferably, the weighting coefficients of the adaptive filter The formula for iterative updates is: in, m The convergence factor; This is the filtered signal; For the first i The weighting coefficients of an adaptive filter; t For discrete time sequence numbers; This refers to environmental vibration and noise.

[0009] Preferably, the root mean square value RMS The formula is: in, N This represents the number of sampling points; t For discrete time sequence numbers; v n ( t () represents the vibration velocity signal.

[0010] Preferably, the energy input stability index I in The formula is: I in = s RMS / m RMS ,in, s RMS For the signal of the welding vibration stabilization stage RMS Standard deviation m RMS The stable segment of the signal during the stable phase of welding vibration. RMS Mean.

[0011] Preferably, the system stability index I sys The formula is: in, AHead t For from the first t The duration is 10 ms Harmonic vibration amplitude extracted within the time window; t This represents the discrete-time sequence number.

[0012] The energy transfer efficiency index I trans The formula is: I trans = A / AHead ,in, A The filtered signal At ultrasonic working frequency f The amplitude at 0; AHead This reflects the harmonic amplitude of the welding head's own vibration.

[0013] Preferably, the steps of the quality judgment rule are as follows: Step 1: Determine the acceptance threshold based on qualified and defective samples from historical data. T good and defect threshold T defect ; Step 2: Calculate the overall quality score. Compared with the preset qualified threshold T good and defect threshold T defect Compare; if T good < If it is, then it is judged as "qualified"; if T defect < < T good If it is, it is judged as "suspicious" and an alert is triggered; if <T defect If so, it is judged as a "defect"; Step 3: Further based on the energy transfer efficiency index I trans With the energy absorption state index The combination relationship is used to determine the specific defect type.

[0014] An online ultrasonic welding quality inspection system, employing the aforementioned online ultrasonic welding quality inspection method, includes: a servo-common optical path measurement unit, comprising a laser vibration probe rigidly fixed to the welding head for acquiring the normal vibration velocity signal of the welding head-weld seat coupling interface during welding; and a data processing and analysis unit, signal-connected to the servo-common optical path measurement unit, comprising: a damping inversion module for performing equivalent modal damping ratio calculation based on an exponential decay model. Inversion calculation; envelope amplitude of the signal during the vibration attenuation stage. A ( t )for: in, A 0 represents the initial amplitude of attenuation. oh n For the system's inherent frequency, t For discrete time sequence numbers; The integrated assessment module is used to construct and weightedly integrate multi-dimensional quality indicators to calculate a comprehensive quality score. The formula for the weighted fusion is: ,in, oh 1 、 oh 2 、oh 3 、oh 4 represents the weighting coefficients determined based on historical data optimization; the quality assessment and early warning module is used to evaluate the overall quality score. The sequence is analyzed for time-series trends and an early warning is triggered.

[0015] By adopting the above technical solution, the present invention has the following beneficial effects.

[0016] (1) This invention rigidly locks the miniature LDV probe to the welding head through a ceramic heat insulation sleeve. The optical axis is parallel to the welding head axis and coaxial with the welding seat normal, forming an integrated "sensor-welding head" follow-up rigid body and self-reference common optical path. This solves the technical problems of traditional contact sensors introducing additional mass load, fixed measurement having relative motion artifacts, and non-common optical path being greatly affected by environmental disturbances. It achieves zero mass loading and non-contact measurement, while eliminating relative motion errors. This ensures that the collected welding head-workpiece interface vibration signal has extremely high spatiotemporal synchronization and fidelity, laying the foundation for subsequent accurate analysis.

[0017] (2) This invention uses an adaptive filtering algorithm to filter the high-fidelity signal acquired by the servo-common optical path measurement unit, wherein the weighting coefficients of the filter are based on the formula: Iterative updates are performed to eliminate interference from common-mode vibration noise. Simultaneously, this invention focuses on the vibration decay stage after welding and fits the signal envelope based on an exponential decay model to online invert the equivalent modal damping ratio, which characterizes the energy dissipation at the interface. The damping ratio The energy transfer efficiency index extracted from the same signal I trans Closely related in physical mechanism, the system jointly characterizes the welding interface state from two dimensions: energy "dissipation" and "coupling". This enables the system to achieve an order-of-magnitude improvement in the sensitivity of identifying early micro-defects (such as micro-cold welds), effectively identify abnormal interface states with energy transfer efficiency below 5%, and perform preliminary diagnosis of defect types through the combination of index relationships.

[0018] (3) This invention calculates the comprehensive quality score by constructing and weighting multi-dimensional physical indicators. The multi-dimensional physical indicators include energy input stability indicators. I in System stability indicators I sys Energy transfer efficiency index I trans and energy absorption state index System stability indicators I sys It can monitor external mechanical disturbances, while energy input stability indicators I in and energy transfer efficiency index I trans These reflect the stability of energy input and transfer in the time and frequency domains, respectively, and are related to the energy absorption state index that reflects internal dissipation. Collaborative work together forms a three-dimensional, highly robust quality assessment system, significantly reducing the overall misjudgment rate of the system.

[0019] (4) This invention establishes a comprehensive quality score based on the aforementioned comprehensive quality score. The time-series trend early warning mechanism uses a sliding window calculation. The moving average was calculated and its continuous downward trend was analyzed. The effectiveness of this mechanism directly depends on a stable, low-volatility overall quality score. Value sequence. When detected When the value shows a statistically significant downward trend, the system can issue an early warning even if its absolute value is still within the acceptable threshold (e.g., about 200 solder joints before a batch of defects occur). This realizes a shift from a passive "acceptance / defect" judgment to a proactive "process performance prediction" early warning mode, providing a key decision window for predictive maintenance.

[0020] (5) This invention achieves synchronous acquisition of vibration signals with zero mass loading by rigidly coupling a laser vibration probe to the welding head to construct a follower-common optical path measurement unit; then, it extracts the attenuation segment after welding from the signal and inverts the equivalent modal damping ratio characterizing the energy dissipation state of the interface based on the exponential attenuation model. Simultaneously, calculate the energy transfer efficiency index. I trans Input stability index I in Multidimensional physical indicators, combined with energy absorption state indicators The overall quality score is obtained by synergistic fusion through a weighted fusion formula. Ultimately, quality degradation early warning is achieved based on time-series trend analysis of the scoring sequence. This enables online perception of the deep dynamic state of the welding process, significantly improving the sensitivity of early defect identification and the system's anti-interference capability, and providing predictive maintenance capabilities for quality degradation. Attached Figure Description

[0021] The following provides a detailed discussion of the manufacture and application of preferred embodiments of the present invention. However, it should be understood that the present invention provides many applicable inventive concepts that can be embodied in various specific environments. The specific embodiments discussed are merely illustrative of specific ways of manufacturing and using the present invention and do not limit the scope of the invention. For those skilled in the art, other drawings can be obtained from these drawings without any creative effort.

[0022] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0023] The following provides a detailed discussion of the manufacture and application of preferred embodiments of the present invention. However, it should be understood that the present invention provides many applicable inventive concepts that can be embodied in various specific environments. The specific embodiments discussed are merely illustrative of specific ways of manufacturing and using the invention and do not limit the scope of the invention.

[0024] This invention addresses the problems of limited monitoring dimensions, insensitivity to early micro-defects, and lack of predictive maintenance capabilities in ultrasonic welding process quality monitoring, providing a complete and specific embodiment. For example... Figure 1The method for online detection of ultrasonic welding quality includes the following steps: S1: A laser vibration probe rigidly coupled to the ultrasonic welding head is used for follow-up measurement to non-contactly acquire the normal vibration velocity signal of the welding head-weld seat coupling interface in real time during the welding process. v n ( t The servo measurement is achieved by rigidly fixing the laser vibration probe to the side of the welding head, with the measuring optical axis of the laser vibration probe being coaxial with the normal of the welding base working surface. The acquisition time covers the entire welding process, including the three stages of vibration initiation, stabilization, and attenuation. The laser vibration probe is a miniature LDV probe, which is rigidly locked to the side of the welding head through a ceramic heat-insulating sleeve, forming an integrated rigid body of "sensor-welding head". This ensures that the optical axis of the laser vibration probe is parallel to the axis of the welding head and coaxial with the normal of the welding base surface. The welding head itself is then used as a light path reflection reference to construct a self-referenced common optical path, eliminating the optical path difference caused by air refraction and relative displacement, achieving zero-mass loading and non-contact measurement.

[0025] S2: The vibration velocity signal is processed using an adaptive filtering algorithm. v n ( t Preprocessing is performed to obtain the filtered signal. The formula for the adaptive filtering algorithm is as follows: in, This is the filtered signal; The original signal; t For discrete time sequence numbers; Environmental vibration and noise; For the first i The weighting coefficients of an adaptive filter; M The order of the filter used during filtering, and the weighting coefficients of the adaptive filter. The formula for iterative updates is: in, m The convergence factor (0 < m <2 / l max , l max (The input is the largest eigenvalue of the autocorrelation matrix). This is the filtered signal; For the first i The weighting coefficients of an adaptive filter; t For discrete time sequence numbers; x ( t () refers to environmental vibration noise.

[0026] S3: Extract the filtered signal Root mean square value in RMS Frequency domain characteristics and vibration decay signals after welding. v ( t The envelope amplitude of the signal during the vibration decay stage is then analyzed using an exponential decay model. A ( t The equivalent modal damping ratio is obtained by fitting and inverting. The envelope amplitude A ( t )for: in, A 0 represents the initial amplitude of attenuation. oh n For the system's inherent frequency, t Let be the discrete-time index. Taking the natural logarithm of both sides of the equation for the exponential decay model, we obtain a linear relationship: The equivalent modal damping ratio can be obtained through linear regression. Calculate the damping ratio of the workpiece signal, denoted as . .

[0027] The root mean square value RMS The formula is: Where N is the number of sampling points; t For discrete time sequence numbers; v n ( t () represents the vibration velocity signal.

[0028] The frequency features were extracted using a Fast Fourier Transform. FFT Extracting the main frequency f 0 and its amplitude A h : V ( f ) =FFT { v n ( t )}; ; ,in, argmax f The optimal clock frequency value is determined to achieve the maximum value. max f It represents the maximum amplitude across the entire frequency domain. f For frequency; v n ( t () represents the vibration velocity signal; FFT { vn ( t )} represents the vibration velocity signal v n ( t Perform a Fast Fourier Transform to convert it into a frequency signal.

[0029] S4: Based on the vibration velocity signal, construct and weighted fuse multi-dimensional quality indicators to calculate the comprehensive quality score. The quality is determined according to the quality judgment rules, and the multi-dimensional quality indicators include at least the energy input stability index. I in System stability indicators I sys Energy transfer efficiency index I trans and energy absorption state index The formula for the weighted fusion is: ,in, oh 1 、oh 2 、 oh 3 、oh 4 represents the weighting coefficients determined based on historical data optimization.

[0030] The steps of the quality judgment rule are as follows: Step 1: Determine the acceptance threshold based on qualified and defective samples from historical data. T good and defect threshold T defect ; Step 2: Calculate the overall quality score. Compared with the preset qualified threshold T good and defect threshold T defect Compare; if T good < If it is, then it is judged as "qualified"; if T defect < < T good If it is, it is judged as "suspicious" and an alert is triggered; if < T defect If so, it is judged as a "defect"; Step 3: Further based on the energy transfer efficiency index I trans With the energy absorption state index The combination relationship is used to determine the specific defect type. I transToo high and A value slightly lower indicates "over-soldering". I trans Too low and The lack of noticeable change may indicate a "cold solder joint". I sys Too high and I trans A decrease may indicate "mechanical drift or welding head wear".

[0031] The coefficient of variation of the probe signal RMS is calculated within the stable welding section; this is the energy input stability index. I in The formula is: I in = s RMS / m RMS ,in, s RMS For the signal of the welding vibration stabilization stage RMS Standard deviation m RMS The stable segment of the signal during the stable phase of welding vibration. RMS Mean. Based on the energy input stability index. I in Quantifying the time-domain fluctuations of the relative energy between the welding head and the workpiece. I in The smaller the value, the more stable the input energy.

[0032] Specifically, the signal from the laser vibration probe is bandpass filtered from 0.5 to 2 kHz to extract the resonance amplitude of the welding head itself. AHead ; Calculate the adjacent 10 ms Time window length AHead Volatility, i.e., the system stability index I sys The formula is: in, AHead t For from the first t The duration is 10 ms Harmonic vibration amplitude extracted within the time window; t This represents the discrete-time sequence number. I sys It reflects the transient change in the coupling stiffness of the welding head-guide mechanism and is used to evaluate external disturbances and structural stability.

[0033] If the amplitude A of the main peak (ultrasonic frequency) of the probe signal near 20kHz is taken as the relative energy carrier, then the energy transfer efficiency index... I trans The formula is:I trans = A / AHead ,in, A The filtered signal At ultrasonic working frequency f The amplitude at 0; AHead This reflects the harmonic amplitude of the welding head's own vibration. I trans A decrease indicates an increase in the impedance at the welding head-workpiece interface, resulting in a reduction in energy coupling efficiency.

[0034] In addition, the logarithmic decrease is fitted using the relative signal envelope attenuation segment. Converted to equivalent damping ratio : ,by Average of qualified batches The normalized deviation is used to represent the energy absorption state index. : This measures the shift in damping characteristics of a workpiece due to material fusion. The larger the value, the more abnormal the energy dissipation.

[0035] Finally, the four indicators mentioned above are weighted and fused together, and the formula for the weighted fusion is as follows: ,in, oh 1 、oh 2 、oh 3 、oh 4 represents the weighting coefficients determined based on historical data optimization.

[0036] S5: Comprehensive quality score based on continuous welding process The sequence is analyzed by calculating its moving average through a sliding window and examining its changing trend. When the overall quality score is detected... When a continuous downward trend is observed, a quality degradation warning is triggered. At this time, even if the value is still within the acceptable range, the system will still issue a "quality degradation warning" alert, prompting the production line to perform welding head maintenance, pressure calibration, or process parameter optimization in advance, thus achieving a closed-loop improvement and early warning system from post-event detection to process warning.

[0037] An online ultrasonic welding quality inspection system, employing the aforementioned online ultrasonic welding quality inspection method, includes: a servo-common optical path measurement unit, comprising a laser vibration probe rigidly fixed to the welding head for acquiring the normal vibration velocity signal of the welding head-weld seat coupling interface during welding; and a data processing and analysis unit, signal-connected to the servo-common optical path measurement unit, comprising: a damping inversion module for performing equivalent modal damping ratio calculation based on an exponential decay model. Inversion calculation; envelope amplitude of the signal during the vibration attenuation stage. A (t )for: in, A 0 represents the initial amplitude of attenuation. oh n For the system's inherent frequency, t For discrete time sequence numbers; The integrated assessment module is used to construct and weightedly integrate multi-dimensional quality indicators to calculate a comprehensive quality score. The formula for the weighted fusion is: ,in, oh 1 、oh 2 、 oh 3 、oh 4 represents the weighting coefficients determined based on historical data optimization; the quality assessment and early warning module is used to evaluate the overall quality score. The sequence is analyzed for time-series trends and an early warning is triggered.

[0038] The following is in conjunction with the appendix Figure 1 The hardware components and operating flow of the system and method of the present invention will be further described. An online ultrasonic welding quality detection system of the present invention includes a follow-up-common optical path measurement unit and a data processing and analysis unit. The follow-up-common optical path measurement unit rigidly fixes a laser vibration probe (a miniature LDV probe is used in this embodiment) to the side of the ultrasonic welding head through a ceramic heat-insulating sleeve. During installation, strict calibration is performed to ensure that the measuring optical axis of the probe is parallel to the axis of the welding head and ultimately coaxial with the normal of the welding seat working surface. This establishes a follow-up measurement link where the measurement coordinate system and the welding head motion coordinate system coincide in real time, achieving high-fidelity acquisition of the vibration velocity signal at the welding head-workpiece interface with zero mass loading and no relative motion artifacts.

[0039] The data processing and analysis unit employs an industrial computer, connected to the laser vibration probe via a high-speed data acquisition card. This unit runs dedicated software integrating a damping inversion module, a fusion evaluation module, and a quality judgment and early warning module. When implementing the method, the system is first started. During the welding process, the probe synchronously acquires the original vibration velocity signal at a sampling frequency of 100kHz. Next, the data processing and analysis unit activates a 32nd-order adaptive filter to process the signal provided by the frame vibration sensor. x ( t As a noise reference, according to the formula Online iterative update of filter weights, where the convergence factor m Set to 0.01, the real-time output filtered clean signal is the filtered signal. This adaptive filtering preprocessing step eliminates common-mode interference, while the servo measurement link ensures the intrinsic nature of the signal source. The combination of the two provides a high signal-to-noise ratio input for subsequent feature extraction.

[0040] Feature extraction includes damping inversion and multi-index calculation. The damping inversion module identifies the vibration decay phase after welding energy is turned off (typically lasting 5-15ms) and extracts the envelope amplitude of the signal during this phase. A ( t ), and based on the formula: Perform nonlinear least squares fitting, where oh n The equivalent modal damping ratio is obtained by inversion using the system's calibrated natural frequency (20kHz in this embodiment). Simultaneously, the fusion evaluation module calculates the root mean square value of the vibration signal during the welding stabilization phase (lasting 30ms in this embodiment). RMS And according to the formula I in = s RMS / m RMS Calculate the energy input stability index I in The amplitude at the 20kHz operating frequency was extracted by performing a Fast Fourier Transform on the filtered signal. A h The harmonic amplitude of the welding head was obtained by analyzing the signal in the 0.5-2kHz frequency band. AHead Then according to the formula I trans = A / AHead Calculate energy transfer efficiency index I trans The equivalent damping ratio obtained by inversion Average of qualified batches By comparison, the damping ratio offset, i.e., the energy absorption state index, is obtained. : .

[0041] The fusion evaluation module calls pre-stored weight coefficients (e.g., obtained through training with 500 sets of historical samples). oh 1 = 0.2, oh 2=0, oh 3 = 0.5 oh 4=0.3), according to the formula: Calculate the overall quality score During this process, the damping ratio (Reflecting energy dissipation) and energy transfer efficiencyI trans (Reflecting energy coupling) are not isolated parameters; they collectively characterize the complete dynamic state of the weld interface from a physical mechanism perspective. For example, when a "cold weld" occurs, poor interfacial bonding hinders energy transfer. I trans (significantly decreased), while the interfacial friction damping characteristics changed ( (Abnormal increase), this physical correlation allows the two indicators to cross-validate in the fusion model, significantly improving the accuracy and robustness of the diagnosis, which cannot be achieved by using only a single dimension of amplitude or frequency.

[0042] The final quality assessment and trend prediction are based on scores. Proceed. The system will... With preset threshold (e.g.) T good =85, T defect =60) Compare and output the judgment of "Pass", "Suspicious" or "Defective". At the same time, the quality judgment and early warning module records the most recent 50 solder joints. The values ​​are used to calculate their moving averages and analyze the trend slope. When a downward trend is detected in the moving average of 20 consecutive solder joints and the slope exceeds a threshold (e.g., -0.5 / point), a "quality degradation warning" is triggered. At this point, the effectiveness of the trend warning directly depends on the stable, low-fluctuation values ​​produced by the aforementioned feature fusion. Value sequence. Without the coordinated fusion of the aforementioned multi-dimensional physical indicators to resist random interference, Q The value itself will be too noisy, which will cause the trend warning mechanism to fail or give false alarms.

[0043] To verify the effectiveness, this embodiment was used to test the ultrasonic welding process of aluminum alloy thin plates. Compared with the traditional monitoring method based on vibration amplitude threshold, the detection rate of micro-incomplete welds (welding area less than 90% of the standard) in this invention is increased from about 65% to over 95%. Under simulated conditions with strong environmental vibration, the false judgment rate of the traditional method rises to about 12%, while the false judgment rate of the method of this invention is stabilized below 2% due to the adoption of adaptive filtering and multi-index fusion collaborative anti-interference mechanism. In addition, the system of this invention successfully issued early warnings at about 200 weld points before the occurrence of batch incomplete weld defects caused by weld head wear, realizing true predictive maintenance. This invention solves the complex technical problems of low sensitivity, poor anti-interference ability, and lack of predictive ability in early defect identification. It achieves significant improvement in the early identification sensitivity and accuracy of welding defects (especially incomplete welds and over-welds), greatly enhances the anti-interference and robustness of the system under complex working conditions, and for the first time in this field, achieves a leap from "post-judgment" to "process early warning," providing a beneficial technical effect with direct basis for process optimization and predictive maintenance.

[0044] Although the specification has provided a detailed description, it should be understood that various changes, substitutions, and modifications can be made without departing from the spirit and scope of the invention as defined by the appended claims. Furthermore, the specific embodiments described are not intended to limit the scope of the invention, and those skilled in the art will readily understand based on this invention that existing or future-developed processes, machines, manufactures, compositions of matter, means, methods, or steps can perform substantially the same functions or achieve substantially the same results as the embodiments of the invention. Therefore, the appended claims are intended to include such processes, machines, manufactures, compositions of matter, means, methods, or steps within their scope.

Claims

1. An ultrasonic welding quality on-line detection method, characterized in that, Comprising the following steps: S1: through the laser vibration probe rigidly coupled to the ultrasonic welding head, the normal vibration velocity signal of the welding head-welding base coupling interface in the welding process is collected in real time non-contact v n ( t ); S2: pre-process the vibration velocity signal by using an adaptive filtering algorithm to obtain a filtered signal v n ( t )performing preprocessing on the vibration velocity signal ; S3: extracting the filtered signal the root mean square value in the signal RMS , the frequency domain feature and the vibration decay stage signal after the welding is finished v t , and the envelope amplitude A t of the vibration decay stage signal is fitted and inverted by an exponential decay model to obtain the equivalent modal damping ratio , and the envelope amplitude A t .​​​ wherein, A 0 is the initial amplitude of the decay, ω n is the system natural frequency, t is the discrete time index; S4: constructing and weighting fusion of multi-dimensional quality indexes based on the vibration velocity signal to calculate a comprehensive quality score , and performing quality determination according to a quality determination rule; the multi-dimensional quality indexes at least include an energy input stability index I in , a system stability index I sys , an energy transmission efficiency index I trans , and an energy absorption state index , and a formula of the weighting fusion is: , wherein, ω 1 、ω 2 、ω 3 、ω 4 are weight coefficients determined based on historical data optimization. S5: generating an overall quality score based on the continuous welding process sequence, by calculating its moving average with a sliding window and analyzing the trend of variation, when detecting a continuous decreasing trend of the overall quality score triggering a quality degradation warning when presenting a continuous decreasing trend.

2. The ultrasonic welding quality on-line detection method according to claim 1, characterized in that, 、ω 3. The ultrasonic welding quality on-line detection method according to claim 1, characterized in that, The follow-me measurement is realized by rigidly connecting the laser vibration probe to the side of the welding head, and the measurement optical axis of the laser vibration probe is coaxial with the normal of the working surface of the welding seat. wherein, is the filtered signal; is the original signal; is the weight coefficient of the i th adaptive filter; t is the discrete time index; is the ambient vibration noise; M is the order of the filter used in filtering.

4. The ultrasonic welding quality on-line detection method according to claim 3, characterized in that, weight coefficients of the adaptive filter The formula for the iterative update is: wherein, The formula of the adaptive filtering algorithm is as follows: is a convergence factor; is a filtered signal; is a weight coefficient of the i th adaptive filter; t is a discrete time index; is an ambient vibration noise.

5. The ultrasonic weld quality on-line detection method of claim 1, wherein, the root mean square value μ The formula is: Wherein, N is the number of sampling points; t is a discrete time index; v n ( t ) is a vibration velocity signal.

6. The ultrasonic weld quality on-line detection method of claim 1, wherein, The energy input stability index I in The formula is: I in = RMS RMS σ RMS ,in, / μ RMS For the signal of the welding vibration stabilization stage σ Standard deviation RMS RMS The stable segment of the signal during the stable phase of welding vibration. μ Mean.

7. The ultrasonic weld quality on-line detection method of claim 1, wherein, The system stability index I sys The formula is: in, RMS t For from the first t The duration is 10 AHead Harmonic vibration amplitude extracted within the time window; t This represents the discrete-time sequence number.

8. The ultrasonic weld quality on-line detection method of claim 1, wherein, The energy transfer efficiency index I trans The formula is: I trans A / ms Wherein, A is the filtered signal The amplitude at the ultrasonic working frequency f 0; AHead is the harmonic amplitude reflecting the vibration of the welding head itself.​ 9. The ultrasonic weld quality on-line detection method of claim 1, wherein, AHead Step 1: determining a pass threshold based on the qualified samples and the defective samples in the historical data T good and a defective threshold T defect ; Step 2: The overall quality score is compared to a pre-set pass threshold and a defect threshold T good T defect If the overall quality score is greater than the pre-set pass threshold and less than the defect threshold, then the product is determined to be "passing" T good then the product is determined to be "passing"​​ If T defect If If T good then the decision is "suspicious" and a warning is initiated; if If T defect then the decision is "defective". Step 3: further according to the energy transfer efficiency index I trans in combination with the energy absorption state index to determine the specific defect type.

10. An ultrasonic welding quality on-line detection system using the ultrasonic welding quality on-line detection method according to any one of claims 1 to 9, characterized by, The steps of the quality judgment rule are as follows: Comprising: A follow-me-common-optical-path measurement unit, which comprises a laser vibration probe rigidly connected to the welding head, for collecting the normal vibration velocity signal of the welding head-welding seat coupling interface during the welding process; a damping inversion module for performing an inversion calculation of an equivalent modal damping ratio based on an exponential decay model; an envelope amplitude of the vibration decay phase signal ( A t ) is:​ wherein, A 0 is the initial amplitude of the decay, A data processing and analysis unit, which is in signal connection with the follow-me-common-optical-path measurement unit, and comprises: n is the system natural frequency, t is the discrete time index; a fusion evaluation module configured to construct and weight the fused multi-dimensional quality indicators to calculate a comprehensive quality score , the formula of the weighted fusion is , wherein, ω 1 ω 2 、ω 3 、 、ω ω 4 is a weight coefficient determined based on historical data optimization; A quality determination early warning module is configured to determine a comprehensive quality score The sequence is subjected to a time trend analysis and an early warning is triggered.

Images