A vector hybrid triboelectric sensor (HTS) for motion identification via machine learning

Rapidly responding and cost-effective sensors played a crucial role in industrial detection. However, the lack of versatile strategies for identifying and classifying operating states on various practical behaviors has limited the rapid development of monitoring technology. This study developed a vector hybrid triboelectric sensor (HTS) with surface nanocrystalline containing triboelectric vibration and rotation units (triboelectric vibration unit (TVU), triboelectric rotation unit (TRU)) capable of detecting the vibrational and rotary states of the device. The synchronous detection of two sensing signals can be achieved due to the hierarchical structure as the basic unit of the HTS, which contributed to reducing the volume and spatial distribution of the HTS. Based on the voltage/current/charge (U–I–Q) signal amplitudes and phase features generated by the TVU, the vibration frequency and orientation of the device can be identified by using a double-layer neural network (D-LNN), in which the accuracy reaches 96.5% and 95.5% respectively. Additionally, by combining logistic regression, D-LNN, and linear regression, the accuracy of the TRU for rotary classification exceeds 93.5% in practical application. In this study, the great potential application of the HTS combined with the machine learning methods was successfully explored and exhibited and it might speed up the development of industrial detection in the near future.