DVS® (Direct Virtual Sensor) is an innovative methodology for designing virtual sensor directly from experimental data, without requiring model construction.


The designed DVS-X estimates in real time an unmeasured variable X from measured variables.


Features of DVS® technology

  • The direct design of the virtual sensor from experimental data reduces the realization time and cost.
  • It accurately estimates variables of complex non-linear systems without requiring expensive model building phases.
  • It allows real sensors substitution when required by physical or economical constraints.
  • Robustness performances can be achieved versus variability of system operating conditions.



Milanese, M., Novara, C., Hsu, K. and Poolla, K. “The filter design from data FD2 problem: nonlinear set membership approach.” Automatica (2009), vol. 45, no. 10, pp. 2350-2357.

Novara, C., Ruiz, F., and Milanese, M. “Direct filtering: a new approach to optimal filter design for nonlinear systems.” IEEE Trans. Autom. Control (2012), vol. 58, no. 1, pp. 86-99. ico-download