| dc.description.abstract | Acoustic sensors are used to monitor components in hydropower plants. Research and development of new solutions are of interest for plant maintenance, but requires processing of data sampled with a high frequency. Performing necessary transformations and calculations on a device before transmitting data can reduce the cost of operating the monitoring system. Industrial devices for general computation are available for a low investment cost. Additionally, compatible edge frameworks are available.
The goal of the project was to reduce the cost of future acoustic monitoring development and answer the research question: "How can an industrial PC be utilized as a platform for capturing and analyzing a high-frequency signal, while communicating with cloud services for fleet management, anomaly detection and data storage?". Different methods were used to answer the research question. Previous work with monitoring relevant components and data treatment methods was discovered with a literature study. Requirements for an edge architecture was gathered, based on the literature study. Documentation from cloud service vendors was used to design the architecture. The functionality of the architecture was tested by implementing a prototype system on an industrial PC.
An industrial PC can be used both to develop monitoring systems and implement them in plants. In order to reduce the operational cost, a general architecture for edge frameworks was proposed. The implementation of the architecture demonstrated the necessary functionality for the prototype system. However, the prototype system did not perform as required on the low-cost hardware used. | |