A Gear Chain Fault Detection Method Using an Adaptive Interference Canceling

A fault diagnosis method based on wavelet and adaptive interference canceling is presented for the identification of a damaged gear tooth. A damaged tooth of a certain gear chain generates impulsive signals that could be informative to fault detections. Many publications are available not only for the impulsive vibration signal analysis but the application of signal processing techniques to the impulsive signal detections. However, most of the studies about the gear fault detection using the impulsive vibration signals of a driving gear chain are limited to the verification of damage existence on a gear pair. Requirements for more advanced method locating damaged tooth in a driving gear chain should be a motivation of further studies. In this work an adaptive interference canceling combined with wavelet method is used for a successful identification of the damaged tooth location. An application of the wavelet technique provides a superior resolution for the damage detection to the traditional frequency spectrum based methods. An analysis and experiment with three pair gear chain show the feasibility of this study yielding a precise location of the damaged gear tooth. Introduction A gear pair is one of the most popular components of mechanical systems thanks to its efficient power transmission and adjustable velocity ratios. Vibrations usually could be generated from the tooth contact that propagates to the entire gearbox through shafts and bearings. Excessive manufacturing tolerances or design errors could end up with gear faults that generate severe wear and crack damage on its teeth. These faults on teeth not only generate audible noise but also shorten life of the machinery. It is therefore required to suggest a successful way detecting the fault in its early stage in order to prevent the costly system shut down. Since a power train is composed of many gear pairs, it is very difficult to sort out damaged gear pairs. The misalignment of shafts or unbalanced mass could also be another source of faults as well as damaged gear tooth. Even worse, the vibration signal generated from the fault pairs can be modulated as it propagates that hampers to locate a certain faulted tooth during the normal operation. Many studies have been published for the detection of fault gear pairs. A two stage adaptive algorithm that could successfully filter out impulsive signals from noise and a moving window procedure were suggested [1, 2]. A cracked gear was identified by using an adaptive amplitude and phase modulation [3]. A study applying a time-frequency analysis combined with wavelet was tried for the purposes [4]. The beta kurtosis and the continuous wavelet transform were used to improve doubling problem in multistage printing presses owing to imperfections in gear system [5]. Recently, an autoregressive based linear prediction fault detection technique was proposed to detect localized faults in gears [6]. This paper suggests an efficient method for the successful identification of a damaged gear pair and classification of the fault types by analyzing vibration signals measured from an experimental set-up. Three pairs of gears were adopted for the experiment. An adaptive interference canceling was used for the identification of faults. In addition it is shown that the fault could be well located by the wavelet technique. Key Engineering Materials Vols. 345-346 (2007) pp 1303-1306 online at http://www.scientific.net © (2007) Trans Tech Publications, Switzerland Online available since 2007/Aug/15 All rights reserved. No part of contents of this paper may be reproduced or transmitted in any form or by any means without the written permission of the publisher: Trans Tech Publications Ltd, Switzerland, www.ttp.net. (ID: 130.203.133.33-14/04/08,17:47:00) Figure 1. Gear driving experiment: (a) geared motor and gear chain; and (b) experimental set-up. Gear driving experiment Signals tapped from the running geared motor are analyzed in this work. Fig. 1(a) shows schematics for the geared motor assembly and detailed gear pairs in it. The assembly is incorporating three gear pairs providing 40 to 1 velocity scale. Fig. 1 (b) shows a schematic diagram of the experimental set-up used in this work. Table 1 shows the theoretical calculations of geared motor meshing frequencies used in this work for the rated input 90V at 75rpm. An accelerometer attached on the top surface of the gearbox is used to measure vibration signals and the rotational velocity at the output end is measured by a magnetic pick-up that generates a pulse on each revolution. Besides a DC controller is used at the DC motor power supply for the variable rotating input speed. A simulated load also attached as a form of coupled power brake. The measured signal by the accelerometer is fed into a digital computer through an A/D converter and it is analyzed with MATLAB. Diagnosis of gear pair faults First vibration patterns of the DC geared motor are characterized with various loading conditions. Second a tooth of a new gear pair is cracked in order to produce an impulsive signal when the tooth gets engaged in. The adaptive interference canceling is applied for the steady state vibration signals coming from a new and a cracked gear pair in order to determine geared motor faults. In Fig. 2 (a, b) vibration signals are measured from new gears and cracked one, respectively. Increased level of impulsive signal is of course monitored in Fig. 2(b). In addition, small peaks neighboring to the impulsive signals imply that more teeth might be damaged other than the one intentionally cracked. From the visual inspection a tooth located twelfth in backward from the initially cracked is found damaged. It corresponds to the calculated gear position by using the information retrieved from Fig. 2(b). Since the time interval for every tooth is 20ms that is calculated by time for one revolution (75rpm) divided by the number of teeth (40), the time duration for the twelve teeth is 140ms. Table 1 Gear meshing frequency (90V-75rpm) Z Gear Set Gear Pinion Fmesh (Hz) Connection 1st 2nd 3rd 37 42 42 10 13 13 627.5 169.6 52.5 Motor Output Casing Gear chain DC motor (a) (b) 1304 The Mechanical Behavior of Materials X