Artificial neural networks have been studied to determine their applicability to modeling and control of physical processes. Some basic concepts relating to neural networks and how they can be used to model weld-bead geometry in terms of the equipment parameters selected to produce the weld are explained. Approaches to utilizing neural networks in process control are discussed. The need for modeling transient as well as static characteristics of physical systems for closed-loop control is pointed out, and an approach to achieving this is presented. The performance of neural networks for modeling is evaluated using actual welding data. It is concluded that the accuracy of neural network modeling is fully comparable with the accuracy achieved by more traditional modeling schemes. >

Artificial neural networks applied to arc welding process modeling and control

Weld pool geometry is a crucial factor in determining welding quality, especially in the case of sheet welding. Its feedback control should be a fundamental requirement for automated welding. However, the real-time precise measurement of pool geometry is a difficult procedure. It has been shown that vision sensing is a promising approach for monitoring the weld pool geometry. Quality images that can be processed in real-time to detect the pool geometry are acquired by using a high shutter speed camera assisted with nitrogen laser as an illumination source. However, during practical welding, impurities or oxides existing on the pool surface complicate image processing. The image features are analyzed and utilized for effectively processing the image. It is shown that the proposed algorithm can always detect the pool boundary with sufficient accuracy in less than 100 ms. Based on this measuring technique, a robust adaptive system has been developed to control the pool area. Experiments show that the proposed control system can overcome the influence caused by various disturbances.

Sensing and Control of Weld Pool Geometry for Automated GTA Welding

A lumped-parameter, analytical model of material and thermal transfer is established in this paper for metal deposition by a moving, concentrated source, This is a dynamic description of the distinct width, height, length, and temperature of the ellipsoidal molten puddle, expressed with respect to the torch power, material feed and angle, and the source motion. This is established through scalar mass, momentum and energy balances of the puddle control volume, and thermal conduction in the substrate. The model is validated by GMA welding experiments, through measurements by an infrared camera and a laser profilometry scanner. These sensors are next employed for real-time identification of the model efficiency parameters, and for output feedback in a closed-loop geometry control system. Because of sensor delays, the model is run in-process to provide substitute estimates to the controller in a Smith predictor scheme. Closed-loop testing was conducted for control of the bead cross section profile through the torch velocity, and the applicability of such geometry regulation to solid freeform fabrication is discussed.

Geometry Modeling and Control by Infrared and Laser Sensing in Thermal Manufacturing with Material Deposition

This paper addresses intelligent techniques for the quality control of the pulsed gas tungsten arc welding process for butt joints, and it is a development to Ref. 1. Because there exist some important differences in butt joint welding and bead-on-plate welding, the modeling and control scheme in Ref. 1 does not completely fit for butt joint welding. In this paper, the differences be tween the two were investigated, The shape and size parameters for the weld pool were used to describe the weld pool geometry. A new real-time algorithm was developed for the size and shape para meters. A size and shape neural network model (SSNNM) was established to predict the maximum backside width. The model accuracy was verified. Furthermore, a self-learning fuzzy neural net work controller (FNNC) was designed for control of the maximum backside width and the fuzzy rules were modified online. Based on the FNNC, and combined with an expert system, a double-input and double-output (DIDO) intelligent controller was developed for controlling the maximum backside width and the shape of the weld pool. Experiment results showed the DIDO intelligent controller could form a better butt joint weld.

Intelligent methodology for sensing, modeling and control of pulsed GTAW : Part 2: Butt joint welding

This application paper addresses the use of adaptive predictive control on arc welding trailing centerline temperature control. For tackling the high level of uncertainty in the process the multivariable multipredictive adaptive regulator (MUSMAR) adaptive algorithm, relying on separate estimation of predictive models is used. Experimental results presented include characterization of plant uncertainty and of the effect in control performance of various available knobs, in particular in the presence of plates with variable geometry.

/pdf/multipredictive-adaptive-control-of-arc-welding-trailing-41vvaa1he1.pdf

J. B. Weinert

Papers

A Model of Full Penetration Arc-Welding for Control System Design