V. Kalaichelvi

Analysis of gas metal arc welding process using GA tuned fuzzy rule based system

The weld quality is generally controlled by the welding parameters. In gas metal arc welding process, the welding parameters are inter related and the adjustment of one parameter may affect another parameter and hence it is considered as a non-linear process. The non-linear nature of the welding system makes it difficult to implement a conventional control method. Fuzzy logic control is an attractive alternative approach. The performance of fuzzy controller will be very much dependent on the knowledge provided to the system. Therefore in recent years, more research has been devoted to augment the approximate reasoning method of fuzzy systems with genetic algorithms. In the present work, genetic algorithm tuned conventional controller is implemented for gas metal arc welding system. Its performance is compared with that of genetic algorithm based fuzzy logic controller.

Application of fuzzy logic for charging control of lead-acid battery in stand-alone solar photovoltaic system

Overcharging in lead acid battery which is widely used in solar systems is a result of improper charging control. Overcharging can be prevented by designing a suitable control system for charging of batteries. A good charging control system will decrease the storage capacity and service time for power supply. In the current study, an attempt has been made to design a PI algorithm based charging control system for a first-order dynamic model of the lead acid battery system. To track the set-point response and to reject the disturbances due to external factors such as change in intensity of the solar radiation, a feedback control system has been designed. As the problem approaches to a highly non linear process, conventional control theory is not an appropriate choice. A fuzzy logic controller with a specially chosen triangular membership function has been suggested as an effective alternative approach. It is proved that by employing fuzzy logic technique, the voltage of the battery can be controlled effectively than with a conventional controller.

Characterisitcs of Al2O3 Nano-Particle Filled GFRP Composites Using Wear Maps:

In this study, glass fiber reinforced plastic (GFRP) filled with different volume fraction (1, 2 & 3%) of nano-Al2O3 particles are tested for their tribological behavior. The wear tests are conducted for different sliding velocities ranging from 0.312 to 2.512 m/s and normal force ranging from 4.91 to 24.71 N. Different wear mechanisms are identified through scanning electron microscope (SEM). Wear maps are drawn by taking sliding velocity on x-axis and normal force on y-axis. The wear maps are utilized to study the dominance of particular wear mechanism that dominates a particular wear regime. Out of three materials discussed, GFRP filled with 3% nano alumina (GFRP3) exhibits better wear resistance than the other materials used in this study.

Application of Model Predictive Controller for 2-DOF robot manipulator

Many of the industrial processes are difficult to model because of their complex behavior, influent characteristics and operational conditions. Control of robot manipulator for industrial applications is considered as one of the challenging tasks. In this paper, Model Predictive Controller (MPC), a class of advanced control technique is proposed in order to control the motion of the revolute joints. MPC is an optimal control strategy based on numerical optimization. Future control inputs and future plant responses are predicted using a system model and optimized at regular intervals with respect to a performance index. Prediction is to determine the future value of the output variables based on available information. This prediction can be used in the design of control laws for better performance of the control systems. The suggested predictive control approach uses an objective function centered on output estimates over a prediction horizon, and hence error is decreased by a selection of operated variable over a control horizon. The performance of the angular motion of the 2-DOF robot link is analyzed with various set point changes in terms of torque. The proposed controller has been verified and validated using satisfactory simulation results of a model of an industrial robot manipulator.

A novel fuzzy logic model for multiple gas sensor array

Gas sensors have the issue of non linearity, low selectivity and cross sensitivity to other gases which cause a huge aberration from the expected results. These can be alleviated if sensors are integrated and studied. While Artificial Neural Network models are not accurate in identification of complex mixtures of gases, this is improved by using a fuzzy logic model for an array of gas sensors which identifies the presence and the concentration of gases efficiently.

Design for a preview control of semi-active suspension system using fuzzy-logic and image processing techniques

The paper presents a fuzzy logic model for a semi-active car suspension with preview control assisted by Image processing techniques and also emphasizing its benefits over passive suspension system. The proposed system determines and operates over the range of damping constant required for the acquired road profile. An Inertial Measurement Unit (IMU) sensor has been suggested to determine suspension motion. This approach greatly reduces the rugged usage of a Magneto-Rheological damper while providing optimum vehicle stability and ride comfort.

Vision-Based Forward Kinematics Using ANN for Weld Line Detection with a 5-DOF Robot Manipulator

While robotic manipulators are becoming a common sight in today’s industries and fast paced production lines, it is becoming difficult to develop foolproof methods for automation of these manipulators, owing to their geometric and structural variety. Creating a common algorithm for these manipulators would help in setting a base standard for their automation. Trio Motion coordinators are most widely used for robotic manipulators in recent times. The objective of this paper is to create a simple interface based on Visual Basic programming language to coordinate directly with the robot’s motion coordinator by bypassing all other programming methods which are otherwise needed for sending commands to the robot. This interface can be easily adapted for further tuning methods and also for more or lesser degrees-of-freedom robotic manipulators. MATLAB has been used for detecting the weld line in the image using image processing techniques. A suitable artificial neural network has been used to give forward kinematic solutions with image coordinates as the input.

Analysis of a proposed algorithm for point to point control of a 3 DOF robot manipulator

Industrial robots exhibit how human labor can be automated due to the evolution of automation. The control of robotic manipulators for industrial applications is a booming field of research. Precision is one of the major factors that is taken into consideration for the evaluation of performance in such manipulators. In this paper, an algorithm has been developed for a point to point controlled 3 DOF model of a spatial robot. Further, simulation studies were carried out and also tested in real life. A set of inverse kinematic equations have been developed which enables to derive a relationship between joint angles, length of links and the desired position in relation to the initial position (Homing position). This is then compared to an algorithm that is designed using rotation matrices with small incremental prediction to achieve minimum error. An analysis between inverse kinematics and the simplified proposed algorithm is done to compare performance characteristics, primarily focusing on accuracy on repeated testing. The two results are compared and it is observed that the proposed algorithm gives better accuracy than the conventional method.

Comparison of GA tuned fuzzy logic and NARMA-L2 controllers for motion control in 5-DOF robot

Abstract Robots are useful in industries in many ways. In today’s economy, the manufacturing industry needs to be efficient to cope with the competition. Installing robots in the industry is often a step to be more competitive because robots can do certain tasks more efficiently than humans. Some of the manufacturing tasks in which robots perform better are assembling products, polishing, and cutting. In order to accurately perform these industrial operations, an effective controller needs to be implemented instead of conventional Proportional plus Integral (PI) controller. When a fuzzy controller is implemented directly, there will be a problem of computational complexity. Therefore, soft computing-based approach namely, genetic-fuzzy systems are proposed in this paper. Fuzzy systems have been integrated with Genetic Algorithm (GA) to optimize the scaling factors that define the fuzzy systems. GAs inspired by the process of biological evolution, are adaptive search and optimization algorithms. A system to be optimized is represented by a binary string which encodes the parameters of the system. This methodology is highly robust and imprecision tolerant. If a unique optimum exists, the procedure approaches it through gradual improvement of the fitness and if the optimum is not unique, the method will approach one of the optimum solutions. Hence, GA tuned fuzzy system is proposed and compared with NARMA-L2 controller for controlling the motion of a 5DOF robotic manipulator. This analysis has been performed using SOLIDWORKS and MATLAB/SIMULINK environments.

Image classification using bag of visual words model with FAST and FREAK

This paper presents a novel technique of image classification using BOVW model. The entire process first involves feature detection of images using FAST, the choice made in order to speed up the process of detection. Then comes the stage of feature extraction for which FREAK, a binary feature descriptor is employed. K-means clustering is then applied in order to make the bag of visual words. Every image, expressed as a histogram of visual words is fed to a supervised learning model, SVM for training. SVM is then tested for classification of images into respective classes. The maximum accuracy obtained by the method proposed is 90.8%.