Bayesian Sensor Calibration

The calibration of multisensor systems can cause significant costs in terms time and resources, particular when cross-sensitivities to parasitic influences are be compensated. Successful ensures the trustworthy subsequent operation a sensor system, guaranteeing that one or several measurands interest inferred from its output signals with specified uncertainty. As shown present study, this goal reached by reduced procedures fewer conditions than parameters needed model device response. This is achieved using Bayesian inference combining data system statistical prior information about ensemble which it belongs. Optimal sets identified method experimental design. demonstrated on Hall–temperature whose nonlinear response requires seven temperature range between $\boldsymbol {-}30$ notation="LaTeX">$150 ^{\circ} \text{C}$ for magnetic field values notation="LaTeX">${B}$ −25 25 mT. For prior, multivariate normal distribution acquired 14 specimens ensemble. I-optimal at one, two, three temperatures reduces root-mean-square (rms) standard deviation notation="LaTeX">$203 \boldsymbol {\mu } \text{T}$ before down 78, 41, notation="LaTeX">$34 }\text{T}$ . Similar conclusions apply G-optimal calibration. article describes how implement acquisition, inference, proposed approach help save resources cut