Showing papers by "Ricardo A. Ramirez-Mendoza published in 2014"

Experimental ANN-based modeling of an adjustable damper

Abstract: A model for a Magneto-Rheological (MR) damper based on Artificial Neural Networks (ANN) is proposed. The design of the ANN model is focused to get the best architecture that manages the trade-off between computing cost and performance. Experimental data provided from two commercial MR dampers with different properties have been used to validate the performance of the proposed ANN model in comparison with the classical parametric model of Bingham. Based on the Root Mean Square Error index, an average error of 7.2 % is obtained by the ANN model, by taking into account 5 experiments with 10 replicas each one; while the Bingham model has 13.8 % of error.

AI-Based Design of a Parallel Robot Used as a Laser Tracker System: Intelligent vs. Nonlinear Classical Controllers

Abstract: Classical ways for coordinate measuring devices are manual theodolites, photogrammetry-based systems, total stations and a recently-introduced device referred to as laser tracker systems. Basically, a laser tracker system is a more accurate and reliable 3D measurement tool that allows to increase and maintain accuracy as time goes by. Laser tracker systems deals with industry-based measuring problems which can be alignment, reverse engineering, tool building, part inspection, installation, and manufacturing and assembly integration. A very interesting case of the latter is robot-tracking calibration in an welding line. In a welding line, robots are controlled in order to keep a prescribed trajectory to accomplish its welding task properly. Nevertheless, in spite of a good control algorithm design, as time goes by, deviations appear and some maintenance has to be done on the robotic unit. So, robot calibration can be done with a laser tracker. Although laser tracker systems are made by very well established and serious companies, their laser products may be very expensive for small or medium size industries. Our contribution is to offer a parallel robot-based laser tracker system model whose implementation would result cheaper than sophisticated laser devices and takes advantage of the parallel robot bondages as high payload. As a first step, simulations of the controlled systems are done here. This parallel robot-based laser tracker is designed to help in the calibration process which consists in repeating some specified trajectory for the serial (welding) robot. The laser tracker system tracks the welding robot trajectory in a day-by-day period of time (for instance) in order to identify the moment when a deviation of the reference trajectory happens. Hence, corrections can be done avoiding greater problems in the welding line. In order to design the parallel robot-based tracker system, a kinematic analysis and a dynamical modeling have to be done in order to design a set of controllers which will be assessed. All of it assisted by AI (artificial intelligence) algorithms. The laser tracker kinematic analysis was done assisted by ANN (artificial neural networks) and by GA (genetic algorithms). This fact allowed to compute numerically/graphically the laser tracker workspace in order to warrant the right accessibility of the corresponding 3D (three dimensional) space. A dynamical model which represents the parallel robot-based laser tracker system was also obtained. This model was used by our set of controllers. The controller design is split into two groups: One considers AI-based algorithms and the second one, classical design-based controllers. A comparison between the two groups is done and advantages/disadvantages are shown in terms of performance in the presence of a persistent perturbation which models ground vibrations in the factory the welding robots are. Such vibrations are endlessly present because they are produced by other assembling machines which disturb the welding process. So, in spite of this perturbation our parallel robot-based laser tracker system showed to behave well with Intelligent Control keeping good tracking of a sinusoidal welding calibration trajectory in the serial robot. In this work it is assumed that a laser device is mounted in the parallel robot with inertial dynamical effect on the parallel robot. Analytical developments are provided as well as numerical/graphical solutions done in MATLAB/SIMULINK to deal with this complex dynamical system. An integral viewpoint with ANN, GA, and Fuzzy Logic was used in this study.