Cardiac analysis can diagnose milling tools

Jari Repo has managed to extract useful process information from existing sensors in a multitask machine.

Jari Repo, researcher at University West, recently defended his doctoral thesis at KTH Royal Institute of Technology. His research demonstrates a simple way to automatically monitor machining processes, such as milling. The monitoring would result in fewer discarded products and equipment breakdowns and as a consequence cost savings.

Vibration patterns during machining can tell a lot about how well the process works, but this requires that you can interpret the patterns. Jari Repo has shown that a method for analysis of heart rate is useful for monitoring metal cutting operations.

Jari Repo's monitoring model utilizes the existing position encoders in a multitask machine. This means that there is no need to install additional measurement equipment. The position encoders can provide a wealth of information that is not used today. They are really designed to accurately position the tool, so that the workpiece gets the intended geometry. However, signals from them, also reflect the vibrations of the machine.

Cost savings
The automatic monitoring Jari Repo envisions would be important in several ways. It would make the machining less dependent on operator monitoring. It would increase the robustness of the processes and make them self-correcting. There would be automatic stops when necessary and automatic adjustments of process parameters.

"Perhaps we can adjust the process automatically in less than a second, compared to today when the operator may need minutes to correct something," says Jari Repo.

This would result in fewer breakdowns and fewer discarded products and consequently considerable savings. In addition, you would dare to use tools longer if there is a system that warns you when the tool is beginning to get worn.

"Today you change the tool before it is likely to be worn to avoid the risk of tool breakage, a damaged spindle of the machine or defective products," says Jari Repo.

A worn tooth
Jari Repo has investigated experimentally how signals from position encoders behave under different conditions. Among other things, he used a milling tool where one tooth was worn. This is a very small change, but the vibration pattern clearly changed. Jari Repo has tried many different methods to analyse the vibration patterns of the experiments. He found that a method used to analyse heart rate variability worked very well.

"This way I was able to quantify the wear by measuring the level of disorder in the vibration pattern," says Jari Repo.

When it comes to human heart rate more order indicates that something is wrong, while disorder indicates a healthy heart. For the milling tool the opposite was true. Disorder in the vibration pattern showed that the tool was worn. Regularity showed that it was intact. Not only worn tools can be detected in this way. In all likelihood you can also get information about problems like defects in individual machine elements, or incipient regenerative chatter.

For an automatic monitoring system based on information from position encoders to become reality, further development is needed. The analysis of the vibration patterns must be possible to perform in real time during the process.

Reference: Condition Monitoring in Machining Using Internal Sensor Signals, Jari Repo, KTH 2012