CN-122007703-A - Laser welding quality real-time evaluation method based on data analysis

CN122007703ACN 122007703 ACN122007703 ACN 122007703ACN-122007703-A

Abstract

The application relates to the technical field of laser welding quality monitoring, in particular to a real-time laser welding quality assessment method based on data analysis. The method comprises the steps of synchronously collecting reflected light signals of a welding keyhole area and plasma radiation signals above a molten pool at the same sampling frequency in a laser welding process, respectively performing trending treatment on the two paths of signals to extract high-frequency fluctuation components, calculating cross correlation functions of the two paths of high-frequency fluctuation components based on a sliding time window to generate a coupling stability index, generating a dynamic judgment threshold value based on the energy fluctuation variance of the plasma radiation signals, wherein the larger the energy fluctuation variance is, the lower the threshold value is, and comparing the coupling stability index with the dynamic judgment threshold value to judge the keyhole instability defect. According to the application, through dynamic coupling analysis of the reflected light of the keyhole and the plasma radiation signal, real-time assessment of welding quality is realized, and welding defects can be accurately identified without complex sensor fusion.

Inventors

PENG XIAOAN
WANG LIXIANG
NI YUWEN
LUO WENFENG

Assignees

东莞市杉达金属制品有限公司

Dates

Publication Date: 20260512
Application Date: 20260311

Claims (10)

1. The laser welding quality real-time evaluation method based on data analysis is characterized by comprising the following steps of: In the laser welding process, synchronously collecting a reflected light signal from a welding keyhole area and a plasma radiation signal from above a molten pool; respectively performing trending treatment on the reflected light signal and the plasma radiation signal, and extracting a high-frequency fluctuation component representing a welding transient process; calculating a cross-correlation function between the high-frequency fluctuation component of the reflected light signal and the high-frequency fluctuation component of the plasma radiation signal based on a sliding time window, and generating a coupling stability index; Generating a dynamic decision threshold based on an energy fluctuation variance of the plasma radiation signal within a current time window, wherein the dynamic decision threshold decreases with increasing energy fluctuation variance; and comparing the coupling stability index with the dynamic judgment threshold value to judge whether the key hole instability defect occurs in the current welding.
2. The method for evaluating the quality of laser welding based on data analysis according to claim 1, wherein the acquisition of the reflected light signal comprises: Collecting a reflected light signal representing the intensity of reflected light of a welding keyhole area through a coaxial light path; The collection of the plasma radiation signal comprises the step of collecting the plasma radiation signal representing the radiation light intensity of the plasma cloud above the molten pool through a paraxial light path.
3. The method for evaluating the quality of laser welding based on data analysis according to claim 1, wherein said trending process comprises: And respectively applying a high-pass digital filter to the reflected light signal and the plasma radiation signal, or removing a direct current component and a low-frequency trend term of the signal by moving average subtraction.
4. The method for real-time assessment of laser welding quality based on data analysis according to claim 1, wherein the calculation of the coupling stability index comprises: calculating a cross covariance of the high-frequency fluctuation component of the reflected light signal and the high-frequency fluctuation component of the plasma radiation signal within the sliding time window; Calculating the variance of the high-frequency fluctuation component of the reflected light signal and the variance of the high-frequency fluctuation component of the plasma radiation signal, and taking the square root of the product of the variances of the high-frequency fluctuation components of the reflected light signal and the plasma radiation signal as a normalization factor; And normalizing the cross covariance by using the normalization factor to obtain a normalized cross correlation function.
5. The method for real-time assessment of laser welding quality based on data analysis according to claim 4, wherein the coupling stability index further comprises: Determining a search interval and setting a sliding window of a plurality of lag times of the plasma radiation signal relative to the reflected light signal, the search interval covering a predicted maximum physical response delay between the reflected light signal and the plasma radiation signal; Calculating the normalized cross-correlation functions corresponding to different lag times in the search interval to form a cross-correlation function sequence; searching a maximum value in the cross-correlation function sequence, determining the maximum value as the coupling stability index, and defining a lag time corresponding to the maximum value as an instantaneous physical delay.
6. The method for real-time assessment of laser welding quality based on data analysis according to claim 1, wherein the energy fluctuation variance is inversely related to a dynamic decision threshold.
7. The method for evaluating the quality of laser welding based on data analysis according to claim 1, wherein the method for calculating the variance of energy fluctuation comprises: and calculating the statistical variance of the instantaneous power sequence of the plasma radiation signal in the current time window, and determining the statistical variance as the energy fluctuation variance.
8. The method for evaluating the quality of laser welding based on data analysis according to claim 1, wherein said determining occurrence of key Kong Shiwen defects comprises: comparing the coupling stability index with the dynamic decision threshold in real time; If the coupling stability index is smaller than the dynamic judgment threshold value, starting a timer or a counter; And outputting a defect judging signal when the state duration time smaller than the dynamic judging threshold value exceeds a preset time window threshold value, and judging the state duration time smaller than the dynamic judging threshold value as instantaneous disturbance if the state duration time does not exceed the time window threshold value.
9. The method for evaluating the quality of laser welding based on data analysis according to claim 8, wherein the keyhole instability defect comprises: At least one of a key Kong Tanta defect, a blow hole defect, an unfused defect, or a hump bead defect.
10. The method for evaluating the quality of laser welding based on data analysis according to claim 1, further comprising: calculating the integral area of the coupling stability index lower than the dynamic judgment threshold value, and defining the integral area as a destabilizing energy accumulation value; and classifying the current key Kong Shiwen defects into three warning grades of light, medium and heavy according to the instability energy accumulation value and the duration time below a threshold value.

Description

Laser welding quality real-time evaluation method based on data analysis Technical Field The application relates to the technical field of laser welding quality monitoring. More particularly, the application relates to a method for real-time assessment of laser welding quality based on data analysis. Background As an advanced connection technology with high precision and high efficiency, laser welding has been widely used in the fields of automobile manufacturing, aerospace, electronic appliances, new energy batteries, and the like. With the development of intelligent manufacturing and industrial digitization, higher requirements are put on real-time monitoring and evaluation of laser welding quality. However, existing laser welding quality monitoring techniques still face many challenges in practical applications. The prior art scheme mainly adopts a multi-sensor fusion scheme or a deep learning method. For example, the multi-mode fusion scheme utilizes a deep learning method to perform multi-mode data fusion by collecting various sensor data in the laser welding process, including visual signals, acoustic signals, thermal imaging data and spectrum information, so as to realize on-line monitoring of welding quality. The method based on molten pool image analysis is characterized in that a high-speed camera is used for collecting an image of a welding molten pool, and a convolutional neural network is used for analyzing and predicting the welding quality of the molten pool. The method based on the acoustic signals extracts time sequence features and frequency domain features to classify the quality by collecting the acoustic emission signals generated in the welding process. In summary, the existing laser welding quality monitoring technology has the following problems that a multi-sensor scheme is complex in structure, high in equipment cost, technical difficulties exist in synchronous acquisition and alignment of multi-source data, a deep learning method needs a large amount of labeled training data, a model training period is long, calculation complexity is high, a single signal characteristic is difficult to comprehensively reflect a welding quality state, strong light radiation of a molten pool area has extremely high requirements on image acquisition equipment, acoustic signals are easily interfered by environmental noise, adaptability of the existing method to different welding process parameters is limited, and model generalization capability is insufficient. Therefore, development of a laser welding quality real-time evaluation method capable of realizing low-cost deployment, efficient real-time processing and good process adaptability on the premise of ensuring evaluation accuracy is needed. Disclosure of Invention The application aims to provide a laser welding quality real-time evaluation method based on data analysis, which is used for solving the problems of complex and expensive multi-sensor scheme, poor real-time performance and insufficient generalization capability in the prior art. The laser welding quality real-time evaluation method based on data analysis comprises the steps of synchronously collecting a reflected light signal from a welding keyhole area and a plasma radiation signal from above a molten pool in a laser welding process, respectively performing trending treatment on the reflected light signal and the plasma radiation signal to extract a high-frequency fluctuation component representing a welding transient process, calculating a cross-correlation function between the high-frequency fluctuation component of the reflected light signal and the high-frequency fluctuation component of the plasma radiation signal based on a sliding time window to generate a coupling stability index, generating a dynamic judgment threshold based on the energy fluctuation variance of the plasma radiation signal in the current time window, wherein the dynamic judgment threshold is reduced along with the increase of the energy fluctuation variance, and comparing the coupling stability index with the dynamic judgment threshold to judge whether keyhole instability defect occurs in current welding. According to the application, the coupling stability index is generated by synchronously collecting the reflected light signal and the plasma radiation signal and calculating the cross correlation function, and the key hole instability defect can be accurately identified by combining a dynamic threshold mechanism based on the energy fluctuation variance. The method utilizes the physical causal relationship between keyhole opening and closing and plasma spraying, adopts signal correlation analysis to replace the traditional amplitude threshold comparison, effectively solves the problem of misjudgment caused by sensor aging or lens pollution, and realizes low-cost deployment and high-efficiency real-time processing on the premise of ensuring evaluation accuracy. Optionally, the collecting of the reflected ligh