**3.2 Case 2: Fault diagnosis of locomotive rolling bearings**

The test bench of locomotive rolling bearings is shown in Fig. 5. The test bench consists of a hydraulic motor, two supporting pillow blocks (mounting with normal bearing), a tested bearing (52732QT) which is loaded on the outer race by a hydraulic cylinder, a hydraulic radial load application system, and a tachometer for shaft speed measurement. The bearing is installed in a hydraulic motor driven mechanical system. 608A11-type ICP accelerometers with a bandwidth up to 5 kHz are mounted on the load module adjacent to the outer race of the tested bearing for measuring its vibrations. The Advanced Data Acquisition and Analysis System by Sony EX is used to capture the vibration data. Parameters in the experiment are listed in Table 4.

Fig. 5. Test bench of the locomotive rolling bearings.

A bearing data set containing nine subsets is obtained from the experimental system under the nine different health conditions. The nine conditions of bearings, shown in Table 5, involve not only incipient faults but also compound faults. The incipient faults are actually slight rub of outer races, inner races or rollers. These faulty bearings are shown in Fig. 6. Each data subset corresponds to one of the nine conditions and it consists of 50 samples.

The Hybrid Intelligent Method Based on

Slight rub on the outer race

Slight rub on the inner race

Fig. 6. Faults in the locomotive rolling bearings.

Fig. 7. Diagnosis results of the locomotive rolling bearings.

It is seen that the training accuracies of six individual classifiers are from 70.67% to 98.60% (average 85.69%) and the testing accuracies from 65.33% to 98.60% (average 83.39%). For the proposed hybrid intelligent method, however, both the training and testing accuracies are 100%. This result shows that the hybrid method achieves obvious improvements in

Fuzzy Inference System and Its Application to Fault Diagnosis 167

Roller rub fault

Serious flaking on the outer race

Each sample is a vibration signal containing 8192 sampling points. Thus, the bearing data set includes altogether 450 data samples, which is divided into 225 training and 225 testing samples. This data is used to demonstrate the effectiveness of the proposed hybrid intelligent method in simultaneously identifying incipient faults and compound ones.


Table 4. Parameters in the experiment


Table 5. Health conditions of the locomotive rolling bearings

Two band-pass (BP1 and BP2) and one high-pass (HP) filters are used to filter the vibration signals of locomotive rolling bearings. The band-pass frequencies (in kHz) of the BP1 and BP2 filters are chosen as 1.5–2.7 and 1.5–4, respectively. The cut-off frequency of the HP filter is chosen as 3kHz. Similar to case 1, four sensitive features are selected from each of the six feature sets using the improved distance evaluation technique and then input into the corresponding classifier. The classification results of the hybrid intelligent method and six individual classifiers are shown in Fig. 7.

Slight rub on the outer race

166 Fuzzy Inference System – Theory and Applications

Each sample is a vibration signal containing 8192 sampling points. Thus, the bearing data set includes altogether 450 data samples, which is divided into 225 training and 225 testing samples. This data is used to demonstrate the effectiveness of the proposed hybrid intelligent method in simultaneously identifying incipient faults and compound ones.

Parameter Value Bearing specs 52732QT Load 9800N Inner race diameter 160mm Outer race diameter 290mm Roller diameter 34mm

Roller number 17 Contact angle 0°

Table 5. Health conditions of the locomotive rolling bearings

six individual classifiers are shown in Fig. 7.

Table 4. Parameters in the experiment

Sampling frequency 12.8 kHz

Condition Rotating speed Label

Two band-pass (BP1 and BP2) and one high-pass (HP) filters are used to filter the vibration signals of locomotive rolling bearings. The band-pass frequencies (in kHz) of the BP1 and BP2 filters are chosen as 1.5–2.7 and 1.5–4, respectively. The cut-off frequency of the HP filter is chosen as 3kHz. Similar to case 1, four sensitive features are selected from each of the six feature sets using the improved distance evaluation technique and then input into the corresponding classifier. The classification results of the hybrid intelligent method and

Normal condition About 490 rpm 1 Slight rub fault in the outer race About 490 rpm 2 Serious flaking fault in the outer race About 480 rpm 3 Slight rub fault in the inner race About 500 rpm 4 Roller rub fault About 530 rpm 5 Compound faults in the outer and inner races About 520 rpm 6 Compound faults in the outer race and rollers About 520 rpm 7 Compound faults in the inner race and rollers About 640 rpm 8 Compound faults in the outer and inner races and rollers About 550 rpm 9

Slight rub on the inner race

Fig. 6. Faults in the locomotive rolling bearings.

Serious flaking on the outer race

Roller rub fault

Fig. 7. Diagnosis results of the locomotive rolling bearings.

It is seen that the training accuracies of six individual classifiers are from 70.67% to 98.60% (average 85.69%) and the testing accuracies from 65.33% to 98.60% (average 83.39%). For the proposed hybrid intelligent method, however, both the training and testing accuracies are 100%. This result shows that the hybrid method achieves obvious improvements in

The Hybrid Intelligent Method Based on

Processing 25 (2011) 1765–1772.

Applications 37 (2010) 4040–4049.

Processing 22 (2008) 419–435.

(2008) 1–9.

2280-2294.

(2009) 594–600.

131 (2009) 1-6.

19 (2005) 427–445.

**7. References** 

Fuzzy Inference System and Its Application to Fault Diagnosis 169

[1] V. Indira, R. Vasanthakumari, V. Sugumaran, Minimum sample size determination of

[2] E. P. de Moura, C. R. Souto, A. A. Silva, et al., Evaluation of principal component

[3] N. R. Sakthivel, V. Sugumaran, S. Babudevasenapati, Vibration based fault diagnosis of

[4] Y. G. Lei, Z. J. He, Y. Y. Zi, et al., New clustering algorithm-based fault diagnosis using

[5] S. J. Loutridis, Self-similarity in vibration time series: application to gear fault

[6] Y. G. Lei, Z. J. He, Y. Y. Zi, et al., Fault diagnosis of rotating machinery based on multiple

[7] Y. G. Lei, Z. J. He, Y. Y. Zi, A new approach to intelligent fault diagnosis of rotating

[8] K. Mollazade, H. Ahmadi, M. Omid, et al., An intelligent model based on data mining

[9] Y. G. Lei, Z. J. He, Y. Y. Zi, Application of an intelligent classification method to mechanical fault diagnosis, Expert Systems with Applications 36 (2009) 9941-9948.

[10] B. Samanta, C. Nataraj, Use of particle swarm optimization for machinery fault detection, Engineering Applications of Artificial Intelligence, 22 (2009) 308–316. [11] S. Rasheed, D. Stashuk, M. Kamel, Fusion of multiple classifiers for motor unit potential sorting, Biomedical Signal Processing and Control 3 (2008) 229–243. [12] Y. Bi, J. Guan, D. Bell, The combination of multiple classifiers using an evidential

[13] F. Huenupán, N. B. Yoma, C. Molina, Confidence based multiple classifier fusion in

[14] Y. G. Lei, Z. J. He, Y. Y. Zi, A combination of W*K*NN to fault diagnosis of rolling

[15] A. K. J. Mohammad, R. Langari, A hybrid intelligent system for fault detection and

[16] Y. G. Lei, M. J. Zuo, Z. J. He, et al., A multidimensional hybrid intelligent method for gear fault diagnosis, Expert Systems with Applications 37 (2010) 1419-1430. [17] U. Benko, J. Petrovčič, D. Juričić, et al., An approach to fault diagnosis of vacuum

element bearings, Transactions of the ASME, Journal of Vibration and Acoustics

cleaner motors based on sound analysis, Mechanical Systems and Signal Processing

machinery, Expert Systems with Applications 35 (2008) 1593–1600.

reasoning approach, Artificial Intelligence 172 (2008) 1731–1751.

sensor fusion, Applied Soft Computing, 9 (2009) 415–422.

speaker verification, Pattern Recognition Letters 29 (2008) 957–966.

analysis, Expert Systems with Applications 37 (2010) 8650–8658.

vibration signals in machine learning approach to fault diagnosis using power

analysis and neural network performance for bearing fault diagnosis from vibration signal processed by RS and DF analyses, Mechanical Systems and Signal

monoblock centrifugal pump using decision tree, Expert Systems with

compensation distance evaluation technique, Mechanical Systems and Signal

diagnostics, Transactions of the ASME, Journal of vibration and acoustics 130

ANFIS combination with GAs, Mechanical Systems and Signal Processing 21 (2007)

and fuzzy logic for fault diagnosis of external gear hydraulic pumps, Insight 51

recognition accuracy and provides a better generalization capability compared to the individual classifiers. Therefore, it is able to effectively identify not only incipient faults but also compound faults of locomotive rolling bearings.
