TIG Weld Width Parameter Self-Adjustment Fuzzy and Integral Hybrid Control

Literature Overview

This 1995 paper by researchers from South China University of Technology presents a hybrid control strategy combining fuzzy logic and integral control for automatic regulation of TIG weld width. The work addresses the challenge of maintaining consistent weld bead geometry during production welding when process parameters such as joint fit-up, material thickness, and travel speed may vary from the nominal conditions. The proposed control system uses real-time weld width measurement as the feedback variable and adjusts the welding current to maintain the target bead width.

Control Strategy Architecture

The hybrid control system integrates two complementary control mechanisms. The fuzzy logic controller handles the nonlinear aspects of the welding process and provides rapid response to large deviations in weld width. The integral controller ensures zero steady-state error and smooths the control output during stable operating conditions. The transition between fuzzy and integral control modes is managed by a switching mechanism that activates fuzzy control when the error exceeds a predefined threshold and switches to integral control when the error falls within acceptable limits.

The fuzzy logic component employs a Mamdani-type inference system with triangular membership functions. The input variables are the weld width error and the rate of change of error, while the output is the adjustment to welding current. The rule base contains 25 fuzzy rules that map the input combinations to appropriate control actions. The linguistic variables used include negative large, negative medium, negative small, zero, positive small, positive medium, and positive large for both inputs and output.

Weld Width Measurement and Feedback

The system employs an optical sensor to measure the weld bead width in real time. The sensor is mounted on the welding torch and captures the width of the solidified weld bead immediately behind the arc. Signal processing converts the optical intensity profile into a digital width measurement with a resolution of approximately 0.1 mm. The sampling rate is synchronized with the travel speed to ensure consistent spatial sampling intervals.

The measurement delay between the arc position and the sensor position introduces a time lag in the feedback loop. This delay is compensated in the controller design by incorporating a Smith predictor structure. The effective delay depends on the travel speed and the physical offset between the torch and the sensor, typically ranging from 10 to 30 mm. At travel speeds of 200 mm/min, this corresponds to a delay of 3–9 seconds, which significantly impacts controller design and stability margins.

Process Model and Parameter Sensitivity

The weld width in TIG welding is primarily influenced by current, travel speed, arc voltage, and electrode-workpiece distance. Among these, current has the most significant effect, with weld width increasing approximately linearly with current over the range of 80–200 A. Travel speed has an inverse relationship with weld width, but the sensitivity decreases at higher speeds due to the increased arc force at higher currents.

The process gain, defined as the change in weld width per unit change in current, varies with operating conditions. At low travel speeds, the process gain is higher because the heat accumulation effect amplifies the width increase. At high travel speeds, the gain decreases as the heat input per unit length becomes less dependent on current variations. This nonlinearity is precisely what the fuzzy controller is designed to accommodate.

Performance Evaluation

Experimental validation demonstrated that the hybrid control system achieved weld width regulation within ±0.3 mm under varying joint fit-up conditions, compared to ±0.8 mm for open-loop operation. The system successfully compensated for thickness variations of up to 0.5 mm and travel speed fluctuations of ±20%. The response time to a step disturbance in weld width was approximately 2–3 sampling periods, with no overshoot in the steady-state condition.

The integral controller contribution was most evident during stable welding conditions, where it maintained the current at the precise value needed to achieve the target width. The fuzzy controller dominated during transitions and disturbances, providing aggressive correction that prevented the integral controller from winding up. The anti-windup mechanism in the integral controller prevented excessive current commands during prolonged disturbances.

Engineering Practice Considerations

Implementing this control strategy in a production environment requires careful consideration of sensor reliability, torch maintenance, and operator training. The optical sensor must be periodically cleaned and calibrated to maintain measurement accuracy. Torch wear affects the arc characteristics and may require periodic recalibration of the control system parameters. The system should include diagnostic capabilities to detect sensor failure and revert to open-loop operation with an alarm.

For automated welding cells, this control strategy can be integrated with CNC torch motion control to provide comprehensive weld quality assurance. The control system can also serve as a process monitoring tool, flagging abnormal conditions such as excessive gas consumption or electrode contamination that may not be immediately visible in the weld width measurement.

Study Insights

This research represents an early but sophisticated application of intelligent control to welding process automation. The hybrid fuzzy-integral approach demonstrates that combining rule-based reasoning with classical control theory can achieve superior performance to either approach alone. The methodology established here can be extended to other welding quality parameters such as penetration depth, bead profile, and undercuts. The fundamental challenge remains the reliable acquisition of real-time process data, which continues to be an active area of research in welding automation.