Comprehensive Analysis of Metal Laser Welding Defect Detection Technologies: A Multi-Dimensional Quality Closed-Loop Approach from Process Monitoring to Internal Imaging

With the growing prevalence of laser welding technology, the continuous diversification of laser types, and the refinement of welding processes, laser welding has become a standard manufacturing technique widely employed in high-end sectors such as new energy vehicles, precision electronics, and medical devices. However, a critical issue has become increasingly prominent:

How can we "visualize" the internal quality of welds to meet the long-term functional requirements of the components?

Compared to plastic welding, metal welding involves complex phenomena such as severe melt pool fluctuations, spatter, and plasma interference, making defect detection more challenging. For comprehensive welding quality inspection, the industry has gradually established three major technical approaches:

· Welding Process Monitoring (Online)

· Weld surface inspection (post-inspection)

· In-process inspection (non-destructive testing)

This article systematically reviews mainstream defect types and detection methods, with a focused analysis of the application principles of cutting-edge technologies such as spectral monitoring, OCT, and ultrasound modules in industrial settings.

I. Typical Defect Types in Metal Laser Welding

From a mechanistic perspective, welding defects can primarily be classified into the following categories:

1. Geometric defects

· Not fully penetrated, not properly fused

· Weld collapse/burring

· Abnormal melt depth

Essence: Insufficient energy input or abnormal distribution

2 Porosity defects

· Single pore

· Continuous pores (honeycomb-like)

· Surface pinholes and internal blind holes

source ：

· Source: Poor protection introduces hydrogen; periodic collapse of the keyhole traps gas; oil contamination, moisture, or coating volatilization on the material surface; rapid cooling of the molten pool prevents gas escape.

3. Crack-type defects

· ·  Thermal cracks (crystalline cracks, liquefaction cracks)

· ·  Cold crack (delayed crack)

Essence: Stress + metallurgical microstructure issues

4 Spraying and Surface Defects

· Sputtered Adhesion

· Surface oxidation

Affects subsequent assembly and sealing performance

II. Limitations of Traditional Detection Methods

 The industry's key challenges are very clear:

Lack of detection methods that are "online, lossless, and quantifiable"

III. Process Monitoring Techniques: External Spectral and Power Signal Analysis

1. External spectral monitoring (Plasma/melt pool radiation)

During the laser welding process:

· Melting of metal → Generation of thermal radiation

· Keyhole formation → Generation of plasma

· Sputtering → Accompanied by fluctuations in light intensity

 This information is released in the form of spectral signals

Core Principle:

By collecting data during the welding process:

· Spectral intensity (Intensity)

· Wavelength Distribution

· Temporal fluctuation

Establish the following relationship:

 Spectral Characteristic = Direct Melting Depth Value

 Spectral Characteristics = Process State Mapping

⚠ Key Restrictions (must be emphasized)

The essence of this technology: it cannot directly measure the melting depth.

Only via:

· Early Process Calibration (Golden Sample)

· Create a feature database

· Perform differentiated comparison

realize ：

· Weld Penetration/Non-Penetration Determination

· Stability Assessment

· Abnormal Alarm

Engineering Implementation Method:

· Photoelectric sensor + Spectrometer

· Coaxial/Paraxial acquisition structure

· Analysis linked in conjunction with the laser power signal

2 Laser power feedback signal

Weld joint integrates real-time detection of power fluctuations:

· output power

· Reflection Light (Back Reflection)

Can be used to determine:

· Change in coupling efficiency

· Keyhole stability

Joint Analysis (Key Trends)

The industry is gradually shifting from relying on a single signal to adopting multiple signals:

 Multisignal fusion (spectral + power + image)

IV. OCT Technology: A Breakthrough Direction for Welding "Depth Perception"

what is OCT？

OCT (Optical Coherence Tomography) is a type of:

Micrometer-level depth measurement technology based on the interference principle

Applications in laser welding

OCT might-have-been ：

· Real-time measurement of melt pool depth

· Monitoring of changes in hole depth

· Weld contour reconstruction

superiority ：

Enables quasi-real-time monitoring of melting depth

High resolution (at the μm level)

Contactless

throw down the gauntlet ：

· The reflection of high-temperature metals is highly complex.

· Plasma interference

· High system costs

· The requirements for optical path design are extremely stringent.

Industry Positioning:

current generation ：

OCT = Popularization Program

OCT = Breakthrough technology for high-end applications

compliant ：

· Battery tab welding

· Precision Medical Devices

· Semiconductor Packaging

V. Ultrasonic Detection Module: Supplementary optimization of the external spectral detection scheme.

Technical Principles

· During the welding process, ultrasonic monitoring captures stress wave signals generated by the weld itself to analyze internal conditions such as keyhole oscillations, molten pool fluctuations, porosity, and crack initiation; this method does not fall under traditional ultrasonic echo testing.

· It can effectively compensate for the minute internal fluctuations that are difficult to detect by optical sensors.

Detectable defects:

· Small pores that cannot be detected by external spectroscopy.

· Abnormal melt depth

· Abnormal welding quality

VI. Multi-Technology Integration: The Future Mainstream Architecture

No single technology can solve all problems; the industry is moving toward:

Multimodal Detection System

Typical Architecture:

1 Process Monitoring Layer:

Spectrum, power signal, high-speed camera, ultrasound (general-purpose)

Structure Detection Layer:

-OCT (High-end)

Data Layer ：

-AI modeling

-Process Database

Core Capability Upgrade:

1 From "Detection" to "Prediction"

· Early identification of welding instability

2. From "human judgment" to "model-driven"

· AI Training Defect Recognition Model

3 Transition from "single-point testing" to a "full-process closed-loop system"

· Before welding → During welding → After welding

VII. Summary: How to select the appropriate solution?

epilogue

epilogue

Quality control in metal laser welding is transitioning comprehensively from experience-driven approaches to data-driven methods.

· Spectral technology enables us to accurately monitor process conditions.

· Ultrasonic technology enables us to capture intrinsic, subtle internal vibrations.

· OCT technology enables precise online measurement of keyhole depth.

· Multimodal fusion ultimately establishes a comprehensive, online, non-destructive, and quantifiable quality control loop throughout the entire process.