The role of intelligent systems in weld process control

Abstract

In materials processing, the process parameters are generally controlled based on knowledge of the relationships between the parameters and the desired material properties. Post-processing quality control is then used to determine if the processing was successful. A primary goal of intelligent sensing and control is to bring quality control into the process by incorporating sensing capability, knowledge of process physics, control capability, and process engineering such that the intelligent processing system is aware of the state of the process and knows how to make a good product based on that awareness. Methods of incorporating intelligent systems, such as fuzzy expert systems and artificial neural networks, into control schemes are discussed along with standard classical and modern control theory. The methods of intelligent systems such as neural networks and fuzzy systems are often means of generating an input-output mapping function. An example of an intelligent control system employs both fuzzy logic and a neural network to control heat input and mass input in gas metal arc welding (GMAW). Two cameras are used to measure the area of the joint to be filled and the temperature gradient in the solidified weld metal. This information is processed by a fuzzy logic system tomore » determine the required mass to fill the joint and the heat input to the weld to maintain a specified cooling rate. This information is processed by neural network which maps this to the welding parameters, travel speed and electrode speed. Hardware, actuators on the weld machine then implement the require values.« less