Holistic Approach for Condition Monitoring and Prediction of Machine Axis
The increasing demands made on operational availability and reliability of machine tools require the adoption of innovative methods to assess the machine’s current condition and predict its condition in future. The progression of degradation caused by wear and tear depends strongly on users’ product range and the manner of machine operation. Therefore, it is necessary to continuously collect information about the machine’s “health status”. Currently the condition of the machine axis is typically ascertained using special test routines under defined ambient conditions. This paper demonstrates a holistic approach for monitoring the condition of the machine axis with simple test routines and additional statistical information. To do this, additional wear relevant data available from the machine control must be logged and included in the computations. This procedure has the advantage that the “load history” of the machine axis can be taken into account for the prediction of future wear progression.
Thank you for your interest in our work. We have chosen a holistic approach, because of the existing interdependencies which have great impact on the degradation process. Let me give an example: If you want to buy a second-hand car, you'll surely be interested in the driving behaviour of the former car owner. Together with some test runs and concrete checks of the condition of wearing parts like tires and breaks you'll be able to evaluate the car's "health" condition and the probability of loss in the near future much better than just knowing the current "health" condition.
In our approach, many of the used information is fuzzy and the interdependies often can't be described in an easy way using physical formulas. Neuro-Fuzzy-Methods can help to close these gaps, because they enable you to combine fuzzy and crispy data for condition monitoring and have the ability to learn. Another advantage of fuzzy inference systems is the possibility to act as an universal approximator to uncover and map unknown interdependencies.
It is now clear to me as to why you have adopted a holistic approach and how a neuro-fuzzy system might help. Thank you. However, in what way is your approach holistic? Have you considered all possible relevant factors?
Holism is the idea that the properties of a system cannot be determined or explained by the sum of its components alone. In the context of machine diagnostics, this means that you should consider information about the machine's usage during its life cycle. I think that it is nearly unpossible to consider all possible relevant factors. So we try to limit to the kind of information that is used by service experts.
Thank you for your very interesting contribution to IPROMS 2005. Could you please explain why you consider your approach holistic? How does it improve on other approaches to condition monitoring? How will neuro-fuzzy data interpretation methods help?