Many critical processes in transportation, industry, and our daily lives rely on sensor measurements. Over time, however, these measurements can gradually become distorted and their accuracy decreases. This increases the risk of major disruptions due to incorrect sensor measurements. Individual sensor measurements are always inaccurate and deviate from the actual value. In order to detect faulty sensors at an early stage, a series of current measurements from each sensor must be constantly compared with the measurements from a certain number of other sensors – in other words, the sensors should form a diagnosable network.
Alena Otto, Professor of Operations & Technology at TUM Campus Heilbronn, has been working on safety-related sensors in railroad tracks that measure the wheels of passing trains and can detect possible damage. The researcher has investigated how these sensors can be positioned so that they can constantly check each other and their number is optimized. Otto's findings: In most cases, sensors cannot adequately check each other if their diagnosability is not taken into account. The good news is that only a few — sometimes even none — additional sensors are needed to guarantee the diagnosability of the network.