The first big step has been taken: Our OnboardEU measurement systems are ready and the first test installation has been done successfully.

Damage to the track network often leads to squeaking and shaking streetcars as well as costly damage to the vehicles. In order to avoid these damages, we are researching together with our project partners German Aerospace Center e.V. and AIT Austrian Institute of Technology GmbH how to enable condition-based and cost-efficient maintenance of the track infrastructure using suitable AI methods for automatic data analysis.

Since the start of the project, we at i4M have realized a robust and powerful measurement system. On each train, one axle is equipped with two microphones and two acceleration sensors. This allows sound waves and vibrations to be recorded. The data is collected by a central base station and provided via LTE. With powerful inferencing capabilities at the Edge, we can implement complex AI algorithms that describe the current track condition.

Key features of the base station:

• Robust cabled connection for cabled sensors
• Wireless connection to our nemi Link 2400 network
• High precision GPS (<0,5m uncertainty)
• Integrated accelerometer, gyrometer and magnetometer
• x86 based Linux system
• LTE Cat 4 Modem
• Gbit Ethernet
• Robust High Speed SSD (500 GB)
• M.2 expansion for AI inferencing modules
• IP65, passively cooled

Key features of the acceleration sensor:

• 40 g MEMS Tri-Axial Sensor (sampling rate 4 kHz)
• 200 g MEMS Uni-Axial Sensor (sampling rate 8-16 kHz)
• Integrated temperature sensor
• sensor signals directly digitized and transmitted digitally and robustly

Key features of the microphone:

• 44.1 kHz 16 bit Mono
• high acoustic overload point (AOP): 128 dBSPL
• Ultra-low self-noise / ultra-high SNR (72 dB)

The project is funded by the German Federal Ministry of Digital Affairs and Transport through the mFUND funding program.