Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

Vibration diagnosis analysis system (VANDA)

Vibration monitoring is a long established technology, as is the use of accelerometers for vibration detection.

 

Train doors are a subsystem of trains which are relatively vulnerable to failure.  When train doors cease to function correctly, it can result in discomfort to passengers and impact negatively on operations leading to increased costs.  Therefore, determining the root cause of failures, and reliably analysing train door systems, has an important significance and is necessary for maintenance and design optimisation.

The vibration signals of the mechanism are separated into two cases, namely: door opening and closing.  For each of them, a signature of N openings, or closings, respectively is defined.  All subsequent acquisitions would be then stored and compared to the original signature.

Proposed solution

The VANDA system is a train door health inspection and monitoring system that utilises vibration analysis to assess the condition of train doors.

The methodology is based on capturing a vibration profile or signature, emitted by any piece of mechanical equipment in motion.  The signature reflects its operating, or specific health, condition at a certain point in time.  It is considered that the mechanism would either continue its operation in the same health condition, or have noticeable degradation when compared to its signature.

System description

The main features of the VANDA system include:

  • User friendly controls and touch screen

  • Compatibility with different sensors, i.e., 325, 326, 354, 1001, 1002

  • Sampling rate of 4000 data points per second

  • Parameter setting interface

  • Manually set measuring duration

  • Recorded vibration data display

  • Signal processing capabilities, i.e., relevant features of the vibration signal, allowing time-domain and frequency-domain analysis

  • Recorded data and results streamed to computer for post analysis

Validation

An experimental rotating rig was setup in TWI’s laboratory and used to test the sensitivity of VANDA to changes in the vibration of the mechanism.  During the algorithm validation process, the DEWESoft commercial system was used, proving that different levels of vibration can be distinguished.

Figure 1. VANDA screen
Figure 1. VANDA screen
Figure 2. VANDA system
Figure 2. VANDA system
Figure 3. Experimental setup in TWI's laboratory
Figure 3. Experimental setup in TWI's laboratory

Site trials and results

The VANDA system has been assessed on site for its capability to detect abnormalities in train door operation.  Once the system was installed on the train door mechanism, the inspection was carried out in its normal, unaltered state at various locations.

In the unhealthy door scenario, the train door mechanism was altered to simulate different practical and realistic defects that can occur during the life of the train door.  These measurements were gathered as a baseline for the healthy state of the door, and served as a reference for comparison with the six induced defects.

The parameters used for differentiating between the two states, as well as opening and closing, were the RMS Amplitude, the Crest Factor and the Euclidean Distance, such that:

  • The RMS Amplitude is defined as the variance of the signal magnitude, being an indication of the amount of vibration energy

  • The Euclidean Distance is used for quantifying all significant deviations from the vibrational spectra of a healthy door mechanism, during the opening and/or closing states

  • The Crest Factor is often used as a measure of discrete impulses which are larger in amplitude than the background signal but which do not occur frequently enough to increase the RMS level of the signal.

The vibration was higher in certain areas of the door and, therefore, easier to detect from different locations of the accelerometers.  It was observed whilst processing the acceleration signals that some parameters were more relevant than others, depending on the type of the damage.  Furthermore, the trials showed that some of the damage created higher vibration levels in the closing case of the door, and were more difficult to detect in the opening case.

The VANDA system was able to identify deviations from the baseline for all induced defects, with results comparative to commercially available systems.

For more information, please email contactus@twi.co.uk

Figure 4. Defect identification – data processing: comparison between baseline and signals after the defect was induced. Top: RMS Amplitude for baseline (first 15 measurements) and unhealthy state of the door (last 10 measurements). Left: Baseline raw data. Right: Raw data after the defect was induced in the door mechanism.
Figure 4. Defect identification – data processing: comparison between baseline and signals after the defect was induced. Top: RMS Amplitude for baseline (first 15 measurements) and unhealthy state of the door (last 10 measurements). Left: Baseline raw data. Right: Raw data after the defect was induced in the door mechanism.
Figure 5. Raw data from baseline signals (first 15 measurements) and signals gathered after inducing a defect in the door mechanism (measurements 16-25). Euclidean Distance used as the parameter in signal processing, showing the change in vibration signature.
Figure 5. Raw data from baseline signals (first 15 measurements) and signals gathered after inducing a defect in the door mechanism (measurements 16-25). Euclidean Distance used as the parameter in signal processing, showing the change in vibration signature.
Avatar Andra Stancu Project Leader – Condition and Structural Monitoring

Andra joined TWI in 2017, managing projects on vibration analysis and de-icing technologies for wind turbines.