Gi Joon Jeon

A parallel neuro-controller for DC motors containing nonlinear friction

Abstract This paper presents an application of a parallel neuro-controller for compensating the effects induced by the friction in a DC motor system. A back-propagation neural network based on a gradient descent algorithm is employed, and a bound on the tracking error is derived from the analysis of the tracking error dynamics. The parallel neuro-controller is a combination of a linear controller and a neural network controller which compensates for nonlinear friction. The proposed scheme is implemented and tested on an IBM PC-based DC motor control system. The algorithm, simulations, and experimental results are described. The results are relevant for many precision drives, such as those found in industrial robots.

A neuro-controller for synchronization of two motion axes

The coordinated and synchronized control of the motion of multiple axes is a challenging problem in motion control fields. In most multiaxis applications, controllers are usually designed for each of the motion axes, which results in a collection of decoupled single input and single output systems. For coordinated motion, however, decoupling sometimes causes damage to the overall performance objective. Therefore, a better way to control multiaxis systems is to introduce intelligent control actions in the controller so that the coordination objective of the desired motion is maintained. A method for achieving the synchronization of two motion axes using a neural network is described. We introduce a new cost function for better synchronization performance. Also, we derive a learning law to adjust the weights of the neural network, based on the gradient algorithm. The derived learning law guarantees good synchronization performance of two motion axes. Simulation and experimental results demonstrate the usefulness of the proposed scheme to synchronize the motion of multiple axes.

An index of applicability for the decomposition method of multivariable fuzzy systems

Research on the application of fuzzy set theory to the design of control systems has led to interest in decomposition of multivariable fuzzy systems. Decomposition of multivariable control rules is preferable since it alleviates the complexity of the problem, but the inference error is inevitable because of its approximate nature. In this paper we show that a large inference error is generated when the Guptas decomposition method (1986) is applied to Exclusive-nor (ENOR) gate model which is used as a counter example. We define an index of applicability which can classify whether the decomposition method can be applied to a multivariable fuzzy system or not. >

Real-time compensation of two-dimensional contour error in CNC machine tools

A real-time contour error compensation (RTCEC) system to improve contouring performance in CNC machine tools is proposed. The RTCEC system evaluates the contour error and modifies the reference position commands based on the computed contour error. Since the proposed compensation scheme can be executed on an external computer, it can be used in conventional CNC machine tools without modifying the interpolation and control algorithms of the CNC system. The compensation system, which introduces a new method for calculating the contour error in real-time without modifying the existing interpolation algorithm, has been implemented by integrating a PC with the existing CNC in a machining center. Experimental results on linear, circular and spiral contours demonstrate the performance of the proposed system.

Neural network indirect adaptive control with fast learning algorithm

Abstract A fast learning algorithm based on a new cost function and a linearized error signal is proposed. The proposed learning algorithm is applied to indirect adaptive control of nonlinear plants. In the proposed method, we use the identification error and the control error to train the NNI and the NNC, respectively. In addition, we introduce a linearized error signal in order to improve the learning speed. Computer simulation results show that the rate of convergence increases, and that the NNC based on the proposed method is insensitive to variations of the plant parameters.

Wireless electronic nose network for real-time gas monitoring system

We present a study on the development and testing of a wireless electronic nose network (WENn) for monitoring real-time gas mixture, NH3 and H2S, main malodors in various environments. The proposed WENn is based on an embedded PC, an electronic olfactory system and wireless sensor network (WSN) technology and neuro-fuzzy network algorithms. The WENn used in this work takes advantage of recent advances in low power wireless communication platforms and uses micro-gas sensors with SnO2-CuO and SnO{in2-Pt sensing films for detecting the presence of target gases. Each node in the network real-timely performs classification and concentration estimation of the binary gas mixtures using the fuzzy ART and ARTMAP neural networks and calculation of the measured humidity and temperature in a located point and then transmits the computed results from the measured data set to a sink node via a Zigbeeready RF transceiver. In addition, a monitoring manager virtual instrument (MMVI) is developed using LabVIEW to monitor efficiently the analyzed gas information from the sensor node. To test the reproducibility and reliability of the WENn, on-line experiments are conducted with the gas monitoring system.

Implementation of indirect neuro-control for nonlinear dynamic systems

Abstract This paper represents identification and control designs using neural networks for a class of nonlinear dynamic systems. The proposed neuro-controller is a combination of a linear controller and a neural network trained by an indirect neuro-control scheme. The proposed neuro-controller is implemented and tested on an IBM PC-based bar system, and is applicable to many dc-motor-driven precision servo mechanisms. The algorithm and experimental results are described. The experimental results are shown to be superior to those of conventional control.

Backlash compensation of nonlinear systems using fuzzy logic

A backlash compensator is designed for nonlinear systems using fuzzy logic. The classification property of the fuzzy logic systems makes them a natural candidate for the rejection of errors induced by the backlash, which has regions in which it behaves differently. A tuning algorithm is given for the fuzzy logic parameters, so that the backlash compensation scheme becomes adaptive, guaranteeing small tracking errors and bounded parameter estimates. Formal nonlinear stability proofs are given to show that the tracking error is small. The fuzzy logic backlash compensator is simulated on a nonlinear system to show its efficacy.

An Intelligent Wireless Electronic Nose Node for Monitoring Gas Mixtures Using Neuro-Fuzzy Networks Implemented on a Microcontroller

This paper presents an intelligent wireless electronic nose node (WENN) that has been designed to classify and quantify binary gas mixtures, NH3 and H2S, the main malodors in various environments. The proposed WENN is based on embedded PC technology and neuro-fuzzy network algorithms. The hardware part of the designed system consists of a microcontroller for processing the measured data set obtained from a micro-gas sensor array, and a Zigbee-ready RF transceiver for transmitting the processed data to a base node. The main program embedded on the designed hardware performs real-lime classification and concentration estimation of the binary gas mixtures using the fuzzy ART and ARTMAP neural networks. To verify performance of the designed intelligent WENN, the measured data from the experiments for the binary gas mixtures have been executed using the WENN. The results show the reproducibility of the measured data and the verification of real-time classification and concentration estimation for the target gas.

Saturation and deadzone compensation of systems using neural network and fuzzy logic

A saturation and deadzone compensator is designed for systems by the fuzzy logic (FL) and the neural network (NN). The classification property of the FL system and the function approximation ability of the NN make them the natural candidate for the rejection of errors induced by the saturation and deadzone. The tuning algorithms are given for the fuzzy logic parameters and the NN weights, so that the saturation and deadzone compensation scheme becomes adaptive, guaranteeing small tracking errors and bounded parameter estimates. Formal nonlinear stability proofs are given to show that the tracking error is small. The NN saturation and FL deadzone compensator is implemented on a system to show its efficacy.