Ultra-Low Power Sensor Node for On-Board Railway Wagon Monitoring

Continuous real-time monitoring of unpowered railway vehicles such as general freight and heavy haul wagons is still limited by hardware scalability and power consumption challenges. This paper proposes an innovative sensor node hardware architecture that reduces power consumption and hardware costs by introducing an Analogue Fault Detector based on analogue signal processing. This technique allows data intensive fault detection and condition monitoring algorithms to be run by simple microcontrollers, reducing memory, execution time and computational requirements. An on-board wheel flat detection sensor node was tested in laboratory conditions using a hardware-in-the-loop setup, to quantify the improvements and explore the viability of the proposed sensor node architecture approach. The power required to detect a wheel flat defect in a simulated acceleration signal was reduced by one order of magnitude and the memory requirement was reduced by three orders of magnitude for the data acquisition and processing stage, compared to traditional sensor node hardware architectures. This is particularly relevant in data intensive applications using accelerometers for monitoring railway vehicles. The improvements delivered by the introduction of an Analogue Fault Detector in a sensor node hardware architecture are promising for further development of on-board ultra-low power condition monitoring sensor nodes for railway applications.