Navigation | Page content | Additional information

Page content

InnoTrans Blog

IoT and artificial intelligence for predictive maintenance

March 2020

Small component – big effect
Photo: Konux GmbH

At InnoTrans 2018, KONUX GmbH presented the Predictive Maintenance System for turnouts – an end-to-end solution that uses IoT sensors and artificial intelligence (AI) to improve network availability, extend plant life and reduce costs.

The KONUX system continuously monitors and analyses the condition of key turnout components and delivers practicable findings. It thus enables better maintenance planning by helping infrastructure operators to detect failures at an early stage and plan the optimum time and type of maintenance required. Since then, further important milestones have been achieved.

New components

The switch frog has been added to the system as a new monitored component. Due to its characteristics, the frog is the component with the highest wear and tear. The KONUX Predictive Maintenance System now detects anomalies of the vibration data in the area of the frog, so that measures can be planned in time to avoid more serious and costly damages. The market launch of the solution for monitoring the condition of the turnout drive is planned for summer 2020. This will address 58 percent of the delays caused by this switch component.

New functionalities

Thanks to advanced algorithms in machine learning, a prediction can be made of how the condition of the track bed will develop over a period of 90 days; the hit rate is 90 percent. This allows the actual effect of maintenance measures such as tamping, grinding/ planing or welding to be validated. 100 percent of the trains running over the points around the clock are detected. This allows infrastructure operators to optimise maintenance planning, avoid downtime and extend the service life of their equipment - without sending anyone to the track or disrupting normal operations.

Next generation IoT sensor

The battery life of the IoT sensor in the field has been doubled from two to over four years and the range of acceleration measurement has been quadrupled without compromising the measurement resolution. In addition, it becomes possible to optimise the device management system, the process of device activation, the setup and firmware deployment Over The Air. All this helps to make major rollouts fast and efficient.

Konux Hall 23 | 470