2. Condition-monitoring techniques

Condition-monitoring (CM) techniques are required to acquire the operation condition data of a rotating machine. The CM data can then be processed and analyzed using appropriate signal analysis techniques to obtain the most relevant characteristic parameters before being used in a diagnosis or prognosis algorithm to evaluate/predict the health state of the machine. The data employed in a condition-monitoring program can be vibration, noise, electric current, oil and grease, temperature or a combination of these data. The CM data can be analyzed using either time-domain techniques, frequency-domain techniques, or time-frequency techniques according to the data properties such as linearity or nonlinearity, stationary or nonstationary, to extract the most useful and effective features for bearing fault diagnosis. This section summarizes some of the most commonly employed techniques in condition-monitoring applications of rolling element bearings.

#### 2.1. Vibration technique

Vibration technique is the most frequently employed technique for machine condition monitoring. A change of the vibration signal in a machine without changing the operation condition can imply a change of health state of the machine. Vibration signals are typically generated by defects in the moving components of the machine such as defects in a bearing, gearboxes, reciprocating components, and so on. For example, when a rolling element bearing operates under a faulty condition, an impulse signal will be generated as other bearing components passing through the faulty position. This will lead to an increase in overall vibration amplitude of the machine. The defect component of the bearing and its severity can be determined by the characteristic defect frequency component contained in the condition monitoring signal (see Figure 2). The frequency range of vibration measurement can be as low as in the infrasonic region (below 20 Hz) and span across to the upper limit of the audible frequency (20 kHz). Though the vibration technique can be problematic in acquiring useful signals when it is deployed for condition monitoring of large low-speed rotating machine where the energy of an incipient defective signal is usually weak and often submerges under the background noise. High-frequency techniques such as acoustic emission (AE) technique can be employed to overcome this limitation.

#### 2.2. Acoustic emission technique

Acoustic emission is a transient elastic wave generated by the sudden release or redistribution of stress in a material. For example, in bearing condition-monitoring applications, AE is generated by the sudden release of energy caused by the material deformation as other bearing components passing through a defect part. The signal then propagates to the bearing house to be detected by a monitoring AE sensor. Compared to vibration signals, the AE signal is less likely to be affected by the dominating noise and vibration generated by the moving mechanical components of the monitored system due to the high-frequency nature of the signal (typically above 100 kHz). Care should be taken in choosing the mounting locations of AE sensors to minimize the energy loss along the AE propagation path for better signal clarity. Furthermore, AE also comes with inherited problems such as calibration, nonlinearity, data storage and transfer, data processing and interpretation. Recently, Lin et al. [1–3] developed a number of signal-processing algorithms to overcome the problems of nonlinearity and large AE data. Nowadays, a large industry deployment of AE technique in bearing condition monitoring is still restricted by the expensive highly specialized AE data acquisition devices.

#### 2.3. Current signal technique

2. Condition-monitoring techniques

cations of rolling element bearings.

can be employed to overcome this limitation.

2.2. Acoustic emission technique

2.1. Vibration technique

42 Bearing Technology

Condition-monitoring (CM) techniques are required to acquire the operation condition data of a rotating machine. The CM data can then be processed and analyzed using appropriate signal analysis techniques to obtain the most relevant characteristic parameters before being used in a diagnosis or prognosis algorithm to evaluate/predict the health state of the machine. The data employed in a condition-monitoring program can be vibration, noise, electric current, oil and grease, temperature or a combination of these data. The CM data can be analyzed using either time-domain techniques, frequency-domain techniques, or time-frequency techniques according to the data properties such as linearity or nonlinearity, stationary or nonstationary, to extract the most useful and effective features for bearing fault diagnosis. This section summarizes some of the most commonly employed techniques in condition-monitoring appli-

Vibration technique is the most frequently employed technique for machine condition monitoring. A change of the vibration signal in a machine without changing the operation condition can imply a change of health state of the machine. Vibration signals are typically generated by defects in the moving components of the machine such as defects in a bearing, gearboxes, reciprocating components, and so on. For example, when a rolling element bearing operates under a faulty condition, an impulse signal will be generated as other bearing components passing through the faulty position. This will lead to an increase in overall vibration amplitude of the machine. The defect component of the bearing and its severity can be determined by the characteristic defect frequency component contained in the condition monitoring signal (see Figure 2). The frequency range of vibration measurement can be as low as in the infrasonic region (below 20 Hz) and span across to the upper limit of the audible frequency (20 kHz). Though the vibration technique can be problematic in acquiring useful signals when it is deployed for condition monitoring of large low-speed rotating machine where the energy of an incipient defective signal is usually weak and often submerges under the background noise. High-frequency techniques such as acoustic emission (AE) technique

Acoustic emission is a transient elastic wave generated by the sudden release or redistribution of stress in a material. For example, in bearing condition-monitoring applications, AE is generated by the sudden release of energy caused by the material deformation as other bearing components passing through a defect part. The signal then propagates to the bearing house to be detected by a monitoring AE sensor. Compared to vibration signals, the AE signal is less likely to be affected by the dominating noise and vibration generated by the moving mechanical components of the monitored system due to the high-frequency nature of the signal (typically above 100 kHz). Care should be taken in choosing the mounting locations of AE sensors to minimize the energy loss along the AE propagation path for better signal clarity. Furthermore, AE also comes with inherited problems such as calibration, nonlinearity, data Current signal technique is mainly used to monitor the bearing condition of electric motors. The technique is based on the principle that the vibration signal generated in a motor is closely related to the change of magnetic flux density in the motor. Stator current technique is the frequently employed current signal technique. When a motor bearing operates on a faulty condition, the radial motion of the motor axis would lead to a small shift of the rotor causing a change in the magnetic flux density between the stator and the rotor. The induced voltage will then cause the variation of the stator current. The stator current technique employs noninvasive sensors to monitor the variation of the stator current; therefore, it is easy to implement and simple to operate.

#### 2.4. Oil and debris-monitoring technique

The application of oil and debris-monitoring technique is a frequently employed tribology approach for machine condition monitoring. The health condition of a roller element bearing can be monitored by analysis of the properties and particle contents of the lubrication oil. The application of oil and debris-monitoring technique is rather straightforward. Though such tribology analysis is normally undertaken at laboratories using spectrometers and scanning electron microscopes rather than in situ. The technique is also limited to condition-monitoring applications of lubrication-related or wear-related problems.

#### 2.5. Thermography technique

Thermography technique detects faults in a bearing by measuring the emission of infrared energy using thermo-infrared devices during bearing operation process. Laser thermography devices or thermo-infrared cameras are the most common instruments used for such measurement. This technique can be applied to monitor the heat variation caused by the change of bearing lubrication, load, and operation speed. It is particularly sensitive in monitoring the overheating phenomenon caused by improper lubrication but less sensitive in detecting incipient fault developed in a bearing such as early pitting, slight wear, or peeling.
