A Virtual Sensor is the output of a digital asset; a real time model output (including uncertainty).
It is based on both dynamic and static inputs which can be used in :
Operational and financial decision making
Automated control
Machine learning
Empirical simulation-based prediction models.
They differentiate from the“soft sensor” as they are processed in the cloud,and so benefit from constant re-training and re-calibration. The accuracy and robustness of the outputs are cross verified using equivalent outputs of parallel models. Their deployment is non-intrusive.