Virtual analyzer / Soft sensors
The terms soft sensors and virtual analyzers describe “software-supported sensors” or “virtual online analytics”. No sensor is able to directly measure a physical variable. Instead, substitute variables are measured, which are then mapped onto the variables which are actually measured, using characteristics. For example, simple temperature sensors are usually based on materials which change their electrical resistance depending on their temperature. An electrical voltage drop is measured via the measurement sensor. This makes it easy to calculate the temperature.
While this principle also applies in the case of a real soft sensor, it does not make sense to describe these trivial measurement techniques as “soft sensor technology”. A real soft sensor is a system which calculates a measurement variable / laboratory variable, which is cumbersome to determine, from multiple simple measurement variables. For this reason, soft sensors can replace complex online measurement techniques or laboratory analyses for the calculation of quality variables.
NeuroModel® is the optimal path to soft sensors! NeuroModel® can learn how laboratory variables and simple measurement data are related. NeuroModel® calculates the laboratory variables online from easy-to-measure parameters (temperatures, pressures). This procedure makes it possible to reduce some of the elaborate laboratory methods, and “exact lab results” are ready in a few seconds!
NeuroModel® also can also (partly) replace a lab or expensive analyzers, and delivers results in real time! This makes it possible to achieve significant cost savings and quality improvements.