Data Analytics and Prediction
Written by Timo Latvala
CSO & Director, Embedded Solutions
Gathering data has become easier and more effortless than before. The benefits of data are multiple: it allows you to find more efficient solutions and make smarter choices. Unfortunately, it often happens that the benefits of data are drowned under its large mass. Data itself doesn’t tell you what to look for – data analytics is needed to dig a hole in the data mountain.
VR FleetCare and data utilization
Our client, VR FleetCare, is a railway maintenance company that provides technological solutions for the proactive maintenance of railway infrastructure with SmartCare services. One important factor in the smooth function of rail traffic is the use of functional switches and crossings, with which trains and trams are directed from rail to rail. Currently, switches and crossings are maintained within a predetermined schedule to avoid problems and malfunctions well before they arise. If a switch malfunctions, the continuity of services and the safety of rail traffic will suffer. On the other hand, excessive care also costs time, effort and resources that may be needed elsewhere.
By anticipating malfunctions, maintenance can be timed to the very moment when it is actually needed. The key to predictive maintenance is found in the data collected by VR FleetCare. The power data of the switching devices that control the crossings has been measured over the past couple of years: the turning of the switch leaves behind a power signal. From this power signal it is possible to determine the condition of the switches and crossings. Once these features are identified and isolated, maintenance can be carried out when it is actually needed.
Data Analytics to help
The keys to the solution must be excavated from the data mountain of measured electric curves, and the task is not easy. On top, the data looks like one large mass, but each part has its own individual characteristics. Variable factors for the switch are, for example, load, life cycle, usage, or environmental and weather variations. A switch that has functioned five winters on a bustling track gives very different data than a newly installed switch in a quieter region.
The combinations of different features produce an innumerable number of outcomes. Among millions of electric curves the human eye is not able to trace the changes that make predictions possible. For this, there is data analytics.
With the help of Huld’s Design Puzzle, and together with VR FleetCare, we brainstormed to start solving the challenges of tangled data and proactive maintenance. In the next blog post, we will go deeper into the world of data analytics: How did Design Puzzle work through this mountain of data? How does solving the tangle of data serve data management? Read the next blog post here.
If you are interested in new technologies for railway infrastructure, join us at VR FleetCare’s SmartCare seminar on 30 March 2022. Timo Latvala, our Business Unit Director, will participate in the closing panel discussion on new technologies as part of rail infrastructure maintenance. The discussion will focus on the challenges and opportunities brought by technology as part of the processes. Read more and sign up for the virtual event here. The event is in Finnish.