Smart sensor node for freight wagon condition monitoring systems

Unlike methods such as replacing components when they fail or on a calendar time schedule, Condition Based Maintenance (CBM) consists of quantifying component degradation in real-time, allowing repairs to be made only when necessary and improving the overall efficiency of mechanical systems. An ideal CBM application for freight railway wagons would consist of a wireless self-powered electronic device installed on each vehicle, detecting and communicating parameters such as brake, bearing or wheel faults, and dynamic instabilities. Such a monitoring device has not been achieved yet, mainly because of the lack of electricity on-board the vehicles and the cost of instrumenting massive fleets. Recent advances and cost reductions in devices and technologies for the Internet of Things (IoT) open the possibility for developing a feasible on-wagon monitoring device. Wireless data transmission, acquisition, and digital signal processing are the most power demanding tasks in on-board condition monitoring sensor nodes. Traditional approaches use mostly digital signal processing and opt for reducing monitoring and communication events, limiting the types of parameters that can be measured and the associated analyses. On the other hand, the approach of the present project is to develop an innovative hardware architecture based on analogue computing for decreasing energy consumption with a reduced sensor node architecture. That is, achieving fault detection using analogue electronic circuitry which directly extracts relevant information from sensor signals, hence reducing digital system workload and complexity and thereby being able to handle higher frequency analogue signals with simple electronic components. This project develops an innovative sensor node hardware architecture and algorithms for a practical on-wagon monitoring device, with low power usage and sufficient on-board calculation capability to provide warning messages when a fault emerges. Developments regarding sensors, IoT, integrated systems and fault detection techniques were reviewed. A wheel flat defect was used as a case study to develop and investigate the proposed condition monitoring sensor node. Railway vehicle dynamic behaviour was simulated to determine operating conditions for the device and the nature of the signals to be monitored. The device concept was firstly proven by combining vehicle dynamic simulations with a physical prototype of the on-wagon fault detection analogue circuit. Subsequently, a hardware prototype version of the circuit was constructed and tested on a scaled bogie rig. The proposed sensor node hardware architecture effectively reduced power consumption and memory requirements for detecting a wheel flat defect using on-board acceleration signals.