Moving from reactive maintenance and preventive maintenance to predictive maintenance has been a big trend regarding IoT for the past years.
In this article, we will look at how Siemens, the world’s largest producer of energy-efficient and resource-saving technologies, is using sensor data to shift to predictive maintenance for their train component branch.
Siemens started a couple of years ago to innovate in the area of using mobility data. Siemens’ mobility data services team believed that Big Data analytics and IoT could enable them to forecast component faults weeks in advance.
Sensors connected to their trains’ components measure the current situation of the elements. The sensor data is then collected and analyzed in near real-time. Any anomalies potentially indicate a component that is likely to fail. Thus, preventive measures can be taken accordingly. Siemens’ mobility data is retrieved from tens of thousands of sensors from trains and rails, repair process data, weather data, as well as from all stages of the supply chain.
Machine learning using sensor data enables Siemens’ data scientist and engineers to quickly identify false positives (predicting a failure that does not actually happen) and gives a precise prediction of actual part failures. Because there are more false alarms than real ones, the organization is looking at work orders, serial numbers, repair processes and supply chain data to help identify and resolve genuine part failures. By incorporating weather data, Siemens can differentiate potential error modes connected to harsh environments from Russian cold to Spanish heat.
How does this work in real-life situations?
Spanish train operator RENFE uses Siemens’ high-speed trains, “Velaro E”, which are continuously monitored by Siemens. If the patterns of collected sensor data are abnormal, a team is dispatched to inspect these components, thus preventing the failure of trains on the tracks. This positive outcome of this is that only one in 2300 trains has been noticeably late.
With the implementation of sensor data, Siemens was able to enhance the reliability of its trains and, in turn, improve the on-time performance of multiple train operators.
By employing sensors that generate large amounts and varieties of data and merging with other data sources such as weather and historical information, a company can build a better picture of how its products or equipment operate in a real-time environment.
Further, analysis of such data can help a customer perform maintenance when it is needed rather than on a timed schedule. Employing such analytic services, such as Datumize’s Data Collector (DDC), has become a real asset in all leading industrial equipment makers.
What do we take from this?
There are endless possibilities with sensor data, and Siemens’ case shows one way of how you can implement this information to your benefit. In Siemens’ case, it becomes clear that the implementation of sensor data has helped prevent delays and critical errors by performing predictive maintenance. This, in turn, helped decrease their liability, improved client relationships, and the overall profitability.
Combining sensor data with environmental information and event history provides the tools for preventive maintenance, improved customer satisfaction and better financial performance. The constantly increasing performance history helps companies to streamline their operations to match their particular use case via continuous improvement.
