Analytische Chemie
Filtern
Dokumenttyp
- Vortrag (1)
- Posterpräsentation (1)
Sprache
- Englisch (2)
Referierte Publikation
- nein (2) (entfernen)
Schlagworte
- Sensor Network (2) (entfernen)
Organisationseinheit der BAM
Eingeladener Vortrag
- nein (1)
Occupational health is an important topic, especially in industry, where workers are exposed to airborne by-products (e.g., dust particles and gases). Therefore, continuous monitoring of the air quality in industrial environments is crucial to meet safety standards. For practical and economic reasons, high-quality, costly measurements are currently only carried out sparsely, both in time and space, i.e., measurement data are collected in single day campaigns at selected locations only.
The project “Robot-assisted Environmental Monitoring for Air Quality Assessment in Industrial Scenarios” (RASEM) addresses this issue by bringing together the benefits of both – low- and high-cost – measuring technologies enabling cost-efficient long-term air quality monitoring in realtime: A stationary network of low-cost sensors that is augmented by mobile units carrying high-quality sensors. By mapping the distribution of gases and particles in industrial environments with the proposed RASEM system, measures can be identified to improve on-site working conditions much faster than using traditional methods.
In this presentation, we detail the technical aspects of RASEM and introduce the mobile platforms used.
Gas distribution mapping is important to have an accurate understanding of gas concentration levels in hazardous environments. A major problem is that in-situ gas sensors are only able to measure concentrations at their specific location. The gas distribution in-between the sampling locations must therefore be modeled. In this research, we interpret the task of spatial interpolation between sparsely distributed sensors as a task of enhancing an image's resolution, namely super-resolution. Because autoencoders are proven to perform well for this super-resolution task, we trained a convolutional encoder-decoder neural network to map the gas distribution over a spatially sparse sensor network. Due to the difficulty to collect real-world gas distribution data and missing ground truth, we used synthetic data generated with a gas distribution simulator for training and evaluation of the model. Our results show that the neural network was able to learn the behavior of gas plumes and outperforms simpler interpolation techniques.