Industry: Industrial Equipment Region: Germany Transaction Price: USD 1 million Transaction Method: Equity Financing
Description of Requirement:
The German research cluster combines nondestructive evaluation with industrial data analysis and artificial intelligence to improve the performance of condition monitoring systems. The classification performance is enhanced in distinguishing between various damage or fault states. Potential industrial applications include slowly rotating drives, pumps or other large industrial equipment.
details:
Two German research institutes are collaborating to develop an improved condition monitoring system. The institutes combine their knowledge in non-destructive evaluation (such as ultrasonic testing) with industrial data analysis and artificial intelligence for sensor data fusion.
Condition monitoring systems should provide high sensitivity to emerging damage for predictable maintenance, robustness and defect status classification. A cost-effective system implementation is essential to the economics of condition monitoring. The system should be able to differentiate between different types and locations of damage to trigger maintenance operations. However, by varying operating conditions or uncertainties in system parameters, it becomes difficult to reliably assign sensor signal characteristics to associated damages.
In the case of slowly rotating drives, damage detection in vibration sensor signals is exacerbated by noise, which is always present in practical applications.
The application of heterogeneous sensors, such as ultrasonic transducers and low-frequency vibration sensors, can enhance classification and early damage detection. To this end, sensor signals are fused through neural network-based multi-level classification. Condition monitoring systems provide damage classification results for sensor patterns and combined methods. The combined classifier allows for higher sensitivity, while the output of the individual sensors can be used for verification and as a backup option in case one of the sensors fails.
The research institute has long-standing experience in condition monitoring, non-destructive evaluation and the development of machine learning systems for industrial applications, including both funded research and direct contract research. The collaboration resulted in a prototype implementation of a condition monitoring system tested on a laboratory scale.
In the next step, technical cooperation with pilot users from industry will be sought to implement feasibility studies to evaluate the technology in relevant industry use cases. If the feasibility phase is successful, the system can also be integrated into the existing condition monitoring infrastructure at the user's site, including a license agreement on the transmission data analysis algorithm.
Furthermore, the system can be enhanced by integrating more sensor modalities to explore other industrial applications in research collaborations.