Soft Sensor

A soft sensor (virtual sensor) estimates a quantity that is hard or expensive to measure directly — like product quality or an internal temperature — from other available measurements using a model. It gives a real-time reading where a physical sensor is impractical.

Soft sensors use process data and a model (statistical or machine-learning) to infer a value continuously, instead of waiting for a lab test or installing a costly probe. Common uses include predicting product quality, emissions or composition from temperatures, pressures and flows. They enable tighter control and earlier intervention, and are a building block of advanced process control and predictive quality.