Condition monitoring of machines using sensor technology has long been an important topic in mechanical engineering. Trends such as digitalization, cyber-physical systems, Industry 4.0, remanufacturing, and artificial intelligence (AI) give this topic a new boost. The focus in these trends is rather on networking, application of AI, or presentation of data. The quality and significance of sensor data are often given little consideration. However, they are elementary for condition monitoring and intelligent machines .
Challenges in Current Measurement Methods
Machines as systems consist of their elements. Gear wheels in particular, as key elements in most drive trains, are at the heart of the action. Changes such as damage or wear usually occur first in the teeth. This causes changes in the gear mesh vibrations, which is the base for many condition monitoring approaches. Current vibration-based condition monitoring measure mostly at the gearbox housing . This results in long signal paths of the vibration. Therefore, this type of measurement is susceptible to interferences (Figure 1). For example, studies have shown that the gear shaft bearings in the signal path transform or superimpose vibrations . This impairs the interpretability of the measurement.
Figure 1: Transmission Path of Vibration Signal with Possible Interferences, source: IPEK - KIT
In practice, interpretability is improved through signal filtering. Vibrations specific to components in the powertrain are filtered, based on their frequency. However, this filtering approach is model based and therefore adds uncertainty.
Solution Proposal: In-Situ Measurement
One approach to improve the vibration measurement quality is to shorten the signal path between the cause of the vibrations and the sensor. This can be achieved through sensor integration in the gear wheel. The measurement closer to the process reduces uncertainties and provides data of superior quality, as some other studies revealed [1, 4, 5]. However, these existing approaches are not yet available as a compact integratable solution and have not yet been validated for general condition monitoring.
The challenges of integrating sensors for acceleration or other measureands into gearwheels are numerous. Installation space, environmental influences, mountability, and costs must be considered. Oftentimes, only trade-offs between different requirements are possible . Regarding the sensor, requirements for sensitivity and measuring range must be considered which are opposing goals. Moreover, requirements are often insufficiently known, for example the actual level of vibration amplitudes at the gear wheel. Tests to gather internal vibration data for sensor dimensioning can only be carried out with considerable effort. To meet these challenges we at the Institute of Product Engineering at Karlsruhe Institute of Technology research sensor integrated gears, as a case in our wider research on sensor-integrated machine elements.
We integrated commonly available digital MEMS acceleration sensors into a gear. A widely available, low-cost microcontroller development board and a battery are mounted on the gear shaft for data acquisition and energy supply. The setup is shown in Figure 2. The integrated sensors on the gear show better signal-to-noise ratios compared to identical sensors mounted on the bearing block, see Figure 3 . Additionally, wear detection with state-of-the-art condition metrics, as well as regression-based machine-learning, were applicable.
Figure 2: Sensor System on Gear, source: IPEK - KIT
Figure 3: Vibration Measurement Comparison in Frequency Spectrum of In-Situ vs. Ex-Situ, showing the first three orders of the gear mesh frequencies and sidebands 
Based on the first measurement results, we are now optimizing the sensor selection, power supply, and add wireless transmission via Bluetooth Low Energy (BLE). Together with a sensor electronics company, we develop an optimized sensor-integrated gearwheel, see Figure 4. The unit integrates sensors, data acquisition, memory, energy management, and wireless interface on the integrated pcb. Two sorts of acceleration sensors are integrated enabling a high sensitivity for various operating conditions. Temperature sensors round up the sensing capabilities of the unit.
Figure 4: Optimized Sensor-integrated Gear, source: IPEK - KIT
Sensor-integrated machine elements offer the potential to generate a high-quality database for condition monitoring and control application. Thereby, providing valuable information without costly data post-processing. The high data quality also enables e.g. AI based methods for wear detection.
Therefore, sensor-integrated machine elements can be the basis for successful digitization in mechanical engineering. Sensor networks, sensors distributed on relevant machine elements, can increase service life through adapted operating strategies. It can also improve remanufacturing, by providing information about the condition of individual machine elements.
The challenges to be met in order to bring sensor-integrated machine elements to market need a cross domain approach. The interfaces and dependencies between mechanical, electrical, and information technology parts need to be considered from the very beginning to ensure a working product. Especially the information technology part is important when sensor-integrated machine elements are embedded in existing systems. Communication standards have to be met and updateability of the software part of the machine element to be enabled.
The current studies on sensor-integrated gears, focussing the MEMS sensors and the wear detection by standard condition metrics and machine-learning, will be presented in detail at the International Conference on Gears 2021.
Julian Peters, M.Sc.
Doctoral Researcher in Department Mechatronic Machine Elements
Lorenz Ott, B.Sc.
Master Student in Department Mechatronic Machine Elements
Matthias Dörr, M.Sc.
Doctoral Researcher in Department Mechatronic Machine Elements
Thomas Gwosch, Dr.-Ing.
Head of Research Department for Mechatronic Machine Elements and System Reliability
Sven Matthiesen, Univ.-Prof. Dr.-Ing.
Chair of Power Tools and Machine Elements
All authors are members of IPEK – Institute of Product Engineering at the Karlsruhe Institute of Technology (KIT)
IPEK Institut für Produktentwicklung, Campus Süd, Kaiserstr. 10, 76131 Karlsruhe
 E. Kirchner, G. Martin, and S. Vogel, “Sensor Integrating Machine Elements: Key to In-Situ Measurements in Mechanical Engineering,” in 23° Seminário Internacional de Alta Tecnologia, 2018.
 R. B. Randall, Vibration-based Condition Monitoring: Industrial, Automotive and Aerospace: John Wiley & Sons, 2011.
 G. Martin, S. Vogel, T. Schirra, G. Vorwerk-Handing, and E. Kirchner, “Methodical Evaluation of Sensor Positions for Condition Monitoring of Gears,” in NordDesign, Linköping, Sweden, 2018.
 T. Utakapan, B. Kohn, M. Fromberger, M. Otto, and K. Stahl, “Evaluation of gear noise behaviour with application force level,” Forsch Ingenieurwes, vol. 81, 2-3, pp. 59–64, 2017, doi: 10.1007/s10010-017-0228-y.
 D. Lewicki, N. Lambert, and R. Wagoner, “Evaluation of MEMS-Based Wireless Accelerometer Sensors in Detecting Gear Tooth Faults in Helicopter Transmissions,” NASA, Cleveland, USA, 2015.
 H. B. Juul-Nyholm, M. Ebro, and T. Eifler, “Barriers for Industrial Sensor Integration Design—An Exploratory Interview Study,” Journal of Mechanical Design, vol. 143, no. 7, 2021, doi: 10.1115/1.4050078.
 J. Peters, L. Ott, M. Dörr, T. Gwosch, and S. Matthiesen, “Design of sensor integrating gears: methodical development, integration and verification of an in-Situ MEMS sensor system,” Procedia CIRP, vol. 100, 672-677, 2021, doi: 10.1016/j.procir.2021.05.142.