IJAV site directory

Published Articles

The Volume 10, No 3, September 2005

«« Back to the List of the Articles

Prediction of Probability Distribution for the Psychological Evaluation of Noise in the Environment Based on Fuzzy Theory

Akira Ikuta, Mitsuo Ohta, M.N.H. Siddique


It is necessary for the statistical evaluation of noise in the environment to find the relationship between the sound level fluctuation of environmental noise and its human response. In this paper, the psychological methods for evaluating noise in the environment are discussed theoretically by introducing fuzzy theory. More specifically, two types of methods for predicting the probability distributions of the sound level and noise annoyance are proposed. Furthermore, by applying the proposed theory to actual road traffic noise data, the effectiveness of the theory is confirmed experimentally.

Display the first page:

Dynamic Behaviour Analysis of Linear Rotor-Bearing Systems using the Complex Transfer Matrix Technique

B.B. Maharathi


A complex transfer matrix method is developed in the present work for analysing the steady-state response of linear rotor-bearing systems in the frequency domain. The transfer matrix of the shaft segment is derived by considering the state variables of the shaft in, a continuous system sense, to give the most general formulation. In this analysis, all the influencing parameters of the shaft, disk, and bearings are included. A three-disks-rotorbearings system has been used as the physical model to demonstrate the effectiveness of this matrix formulation for the evaluation of the dynamic characteristics of any rotor-bearing system. In order to establish the accuracy of this technique, experiments are also conducted on the same rotor-disks model under various bearing conditions as used in the numerical analysis. The numerical results evaluated by various researchers are also compared with these experimental results.

Display the first page:

Psychoacoustic Approach to Machine Fault Diagnosis

Primoz Potocnik, Edvard Govekar, Igor Grabec and Peter Muzic


A method for machine diagnosis, based on acoustic signals, is proposed. The method is based on psychoacoustic modelling, which simulates the ability of human aural analysis. The solution approach combines psychoacoustic pre-processing, feature-extraction modules, and feature-evaluation modules. Feature-extraction methods are defined for the detection of stationary and dynamic acoustic phenomena. Stationary features are appropriate for diagnosing machines that emit continuous sounds, and defects are reflected in altered sound patterns. Dynamic features are suitable for the detection of transient phenomena, such as collisions, etc. Statistical procedures, based on asymmetrically defined fault-detection margins, are proposed for an evaluation of extracted features. An application of the method for the diagnosis of industrially produced compressors is discussed. Noxious space defect and lubrication defect are diagnosed by a stationary approach during the normal operation of compressors. Various mechanical defects in the supporting spring system are diagnosed by extracting dynamic features for collision detection. Compressor defects are successfully recognised through the proposed method.

Display the first page:

Minimax Linear Quadratic Gaussian Control of Longitudinal Vibration for Cable Transporter Systems with Multiplicative Nonparametric Uncertainties

Yuhong Zhang, Sunil K. Agrawal, Hemanshu R. Pota and Ian R. Petersen


This paper demonstrates a new application of robust control to the longitudinal vibration suppression of cable transporter systems with multiplicative nonparametric uncertainties. Both the infinite-dimensional theoretical model for the system and the robust controller are presented. Experimental results have verified the effectiveness of the proposed robust controller. The infinite-dimensional model of the transporter system is truncated to obtain a finite-dimensional model for robust controller design. A minimax Linear Quadratic Gaussian (LQG) controller is designed for multiplicative nonparametric uncertainties. The minimax controller design involves the solution of two simultaneous Riccati equations. The robust controller is implemented on a laboratory cable connected linear transporter system using dSPACE 1103 systems. System identification theory is applied to obtain a statespace representation in the 5-20 Hz frequency range for the experiment. The experimental results show reductions of 9 dB in the fundamental resonant peak of the cable transporter system.

Display the first page:

«« Back to the List of the Articles