Researchers at A*STAR’s Singapore Institute of Manufacturing Technology (SIMTech) have developed acoustic sensors that can identify wear and tear in industrial machinery. By catching problems before they even occur, this technique promises to speed up manufacturing.

Manufacturers of precisely engineered products such as engine components are increasingly monitoring their high-speed machinery tools online to ensure product quality and reliability.

Jun-Hong Zhou and her co-workers at SIMTech have developed a technique that gives online systems the ability to assess how well a tool is working. Using small, inexpensive sensors they analyse sound waves to determine the integrity of machinery.

So how do you interpret the state of a machine based on sound? Inside machines, a change in the acoustic signal can be detected in the form of pulses of sound energy that occur when, for example, a tool is chipped or worn down. A software program is then used to scan the sound patterns and determine whether a machine needs to be shut down for maintenance. Acoustic sensors offer rapid monitoring, but analysing the complex signals they generate can be tricky, as the use of too many variables makes computation slower and less accurate.

Wear and tear in industrial machinery can be monitored using acoustic sensors. © Creative Commons Sinead Fenton

Wear and tear in industrial machinery can be monitored using acoustic sensors. © Creative Commons Sinead Fenton

Zhou’s team have managed to resolve these issues by developing a new dominant-feature identification (DFI) algorithm. In this approach, the acoustic signals from tools are collected using embedded acoustic sensors and converted into a low-dimensional mathematical matrix. A procedure called ‘singular value decomposition’ is then applied which ultimately reveals the variables that dominate the acoustic signal. As only a fraction of the full data set needs to be processed, DFI can analyse signals 80 per cent faster than typical methods.

The team also based their algorithm on a new model or theoretical framework called ARAMX, which allows them to dynamically update the decision software with previously predicted values. Experiments showed that this method could predict the cutting tool lifetime for ball nose cutters in a milling machine with 93 per cent accuracy, a significant improvement on other processing systems.

“DFI is very efficient for identifying key input parameters, and combining it with the ARAMX model provides accurate predictions for online machine condition monitoring,” says Zhou.

The team plans to apply their acoustic-based system to the manufacture of function-critical devices such as aircraft gearboxes and wind turbine generators in the near future.

References:
DOI: Zhou, J.-H., Pang, C. K., Zhong, Z.-W. & Lewis, F. L. Tool wear monitoring using acoustic emissions by dominant-feature analysis. IEEE Transactions on Instrumentation and Measurement 60, 547–559 (2011).

For further information contact:

Ms Jun-Hong Zhou
Singapore Institute of Manufacturing Technology
Agency for Science, Technology and Research (A*STAR)
Singapore
Email: jzhou@SIMTech.a-star.edu.sg.