Fuzzy logic based tool condition monitoring for end-milling

- Monitoring the cutting condition of the tool in end-milling is a very complicated and prone to error task.
- Filtering and processing the data can be an extremely tedious process because of the relatively high sampling rates and because of the very high levels of noise that are generated during machining operations.
- An adequate number of experiments is required in order to well define the fuzzy inference system with its membership functions and rule base.
- The fuzzy inference system is very convenient to update in the end of every machining process.
- The sensors have to be evaluated based on certain objective criteria and then the weight of their signal should be defined.
- The microphone signal gave quiet promising results and this is because of its low Overlapping Effect and high sampling rates.
- The dynamometer signals can be of great importance as they can provide a three dimensional view of the machining process (x, y and z-axis). Also, the dynamometer could detect which part of the tool is the most deteriorated since the flank is related to the z-axis and the rake is related to the x and y-axis of the sensor.
- The current sensor gave the same results as the dynamometer and because of their high correlation value, these two sensors can be used to validate their respective results.
- The accelerometer signal was very difficult to filter and it was prone to error.
- A software was developed based on the knowledge obtained from the conducted research.