Identification of TIG Weld Penetration Status Based on Acoustic Emission Signal Short-Time Energy

Literature Overview and Research Background

This study by Wang Qisheng, Gao Yanfeng, Gong Yanfeng, Huang Linran, and Zhao Jiamin from Nanchang Hangkong University presents an innovative approach to real-time monitoring of TIG weld penetration using acoustic emission (AE) signals. Published in Heat Processing Technology in 2020 and supported by multiple funding agencies including the National Natural Science Foundation of China (51465043) and Jiangxi Provincial Science and Technology Programs, this work addresses a fundamental challenge in welding quality assurance: the need for non-destructive, real-time assessment of weld penetration without interrupting the welding process.

In the context of pressure vessel fabrication and clad plate welding, ensuring complete penetration is critical. Incomplete penetration leads to stress concentration, premature fatigue failure, and potential catastrophic rupture under pressure loading. Traditional post-weld inspection methods such as radiographic testing (RT) and ultrasonic testing (UT) are destructive in terms of production flow — defects can only be identified after welding is complete, requiring costly rework.

Core Technical Methodology

Acoustic Emission Signal Acquisition

The researchers developed an AE signal acquisition system integrated with a TIG welding setup. The system employs piezoelectric AE sensors (typically R15α or R15β type) mounted on the workpiece surface, connected through pre-amplifiers to a dynamic signal analyzer. The key innovation lies in the use of short-time energy (STE) as the primary feature parameter for characterizing the penetration state.

The short-time energy is calculated using the following approach:

The AE signal is divided into sequential time windows (typically 5–20 ms).
Within each window, the energy is computed as the integral of the squared signal amplitude.
The STE feature vector is then compared against established thresholds to classify the penetration state.

Signal Feature Extraction and Classification

The study identifies three distinct penetration states:

Penetration State	AE Signal Characteristic	STE Feature Behavior
Under-penetration	Low amplitude, narrow bandwidth	STE values below threshold, low variance
Full penetration	Moderate amplitude, broadband	STE values within optimal range, stable variance
Over-penetration	High amplitude, sharp peaks	STE values exceed threshold, high variance

The classification algorithm employs statistical analysis of STE features, including mean value, standard deviation, and peak-to-peak ratio, to distinguish between the three states. The recognition accuracy achieved in the study exceeds 90% under controlled welding conditions.

Process Parameters and Signal Response

The study systematically varied key TIG welding parameters and recorded the corresponding AE signal responses:

Parameter	Under-Penetration Range	Full Penetration Range	Over-Penetration Range
Current (A)	60–90	90–120	120–150
Travel Speed (mm/min)	100–150	150–200	200–250
Arc Length (mm)	2.0–3.5	1.5–2.0	1.0–1.5
Joint Gap (mm)	0.2–0.5	0.5–1.0	1.0–1.5

The findings demonstrate that the AE signal intensity is strongly correlated with the arc stability and molten pool dynamics. Under-penetration conditions produce weak AE signals due to limited molten pool activity, while over-penetration generates intense signals from violent fluid flow and keyhole formation.

Engineering Practice Applications

Integration with Pressure Vessel Fabrication

In pressure vessel manufacturing, particularly for clad plate vessels and weld-overlay equipment, this AE-based penetration monitoring technology offers several advantages:

Real-time quality feedback: Welders can adjust parameters immediately when under-penetration is detected, preventing the need for post-weld rework.
Reduced inspection burden: By ensuring consistent penetration during welding, the number of welds requiring destructive testing can be reduced.
Process documentation: Continuous AE data recording provides a digital record of weld quality for traceability and audit purposes.

FMEA Analysis of AE Monitoring System

Applying Failure Mode and Effects Analysis (FMEA) to the AE-based monitoring system:

Failure Mode	Effect	Severity	Detection Method	Countermeasure
Sensor detachment	Loss of signal	High	Signal dropout alarm	Secure mounting with conductive epoxy
Signal noise from arc	False classification	Medium	Bandpass filtering	Optimized filter settings (100–500 kHz)
Threshold drift	Misclassification	Medium	Regular calibration	Automated baseline correction
Environmental interference	Signal degradation	Low	Signal-to-noise ratio monitoring	Shielded cable routing

Study Insights and Reflections

The acoustic emission approach represents a paradigm shift from post-weld inspection to in-process quality control. The short-time energy feature is particularly attractive because it captures the transient nature of weld pool dynamics — a single momentary AE burst can indicate a change in penetration state that may not be visible in the weld bead appearance.

However, several practical challenges remain. The sensitivity of AE signals to environmental noise, the need for sensor recalibration between welds, and the dependence on consistent welding conditions all limit the immediate industrial adoption of this technology. Furthermore, the study was conducted under laboratory conditions with controlled parameters; real-world production environments with varying joint fit-up, surface contamination, and operator technique present additional challenges.

The integration of AE monitoring with automated welding systems — such as robotic TIG welding with adaptive control — represents a promising direction for future development. By closing the loop between signal detection and parameter adjustment, the system could achieve truly intelligent penetration control, significantly improving first-pass quality rates in pressure vessel fabrication.