Nowadays high qualities welds which have a very narrow form and high accuracy can be obtained with advanced monitor and control system. In this paper are presented the main components of the electron beam welding equipment CTW 5/60 and proposed two welding strategies with seam tracking and adapted to the possibilities of the experimental setup. The first strategy consists in dividing the surface of the material over several successive frames, while the second strategy involves performing the operation in a single pass. Aspects of machine vision, meaning the image reconstruction, seam detection and image processing methods are also summarized.
The design of controllers, using conventional techniques, for plants with nonlinear dynamics or model uncertainties can be often quite difficult. Model reference adaptive control (MRAC) is a modern alternative for classic control algorithms and a convenient method to updating the controller's parameters for slow processes with parameter variations. In this paper is presented a comparative study of two design procedures of MRAC for first order slow processes, particularly applied and tested on level process control. The first method it is based on the MIT rule and the second one it is based on the stability theory of Lyapunov. The theoretical results are obtained using Matlab simulation environment.
In control engineering, the mathematical model of the studied process plays an important part. In representing the mathematical model, the inherent errors due to parameter variations define the parametric (structured) uncertainties; therefore there are differences between the mathematical model used in the design process and the real plant. In most cases, the control system needs to stabilize the process and also needs to assure certain performances even in the presence of uncertainties, unmodeled dynamics, disturbance signals and measurement noise, which all make the process vulnerable. This paper shows a method for behavioral modeling and simulation of a gear system, in which the moment of inertia and the friction coefficient represent the parameters which are affected by uncertainties, and are expressed by percentages to the nominal values. For the modeled process, by considering a classical controller, which is tuned experimentally, Matlab is used to represent the sensitivity functions in both time and frequency domains for the nominal cases, for randomly generated samples and for the worst case scenarios.