Ensuring the reliability and availability of complex systems such as safe mechatronic systems or cost-sensitive machine components is of increasing importance. Besides the availability of problem-specific sensors and filtering techniques three major issues are of interest:
i) Preparation of the measured data (filtering),
ii) Interpretation of the data with respect to the machines state as well as to the machines’ remaining lifetime or guaranteed functionality, and
iii) establishing the required knowledge behind from available measurements and data.
Core of this contribution is the development and application of easy to apply and easy to interpret algorithms to be used directly with industrial data or measurements from technical systems in operation.
The three approaches applied are
AI: Acoustic Emission (AE) characterized by measurements in combination using a suitable filter to be designed,
AII: data analysis using operating system data feature capturing technique, and
AIII: Adaptive fuzzy-based filtering [1] with training and classification modules.
The approaches are developed in detail due to former research work [2], here they are applied and compared using the same experiment, the shown results are based on experiments. The three approaches use different algorithms but partially different signal sources (from the same experiment). Each of the approaches allows a different and distinguished problem-oriented insight to the complex wear process of the considered system, typical for mechanical engineering-related machines. The comparison between three different approaches for wear diagnosis can be considered as the main idea of this paper which allows insights into the advantages and disadvantages of each of these approaches.
