Sensor Rotational Measurement with GAN-Based Virtual Sensor

Quan Nguyen Thanh, Nguyen Quang-Hung, Nam Thoai

Abstract: Recent advances in sensor technology have increased human abilities to measure a wide range of phenomena and events. Nonetheless, due to a number of constraints, only a few sensors may be placed at a specific site in some circumstances. As a result, installing enough sensors in the proper areas to ensure consistent monitoring may be challenging. Furthermore, virtual sensing, which is a set of strategies for replacing a portion of physical sensors with virtual ones, has recently emerged. As a result, this work takes advantage of PGAIN-VS's imputation capabilities to create a black-box virtual sensing approach called Sensor Rotational Measurement with the aim of lowering the number of physical devices required in reality while maintaining monitoring accuracy. The method employs the PGAIN-VS and Borda voting methods to identify the subset of actual sensors that can take turns monitoring data within a certain time interval. The method is viewed as a black-box objective optimization problem with constraints, and solved by a generalized black-box optimization tool. We tested our strategy on real-world datasets and saw promising results, with around 20 % reduction in the overall number of physical sensors.