Optimization of the microwave determination of water in foods using principal component analysis

In this paper a method for determining the composition of foodstuffs from the measured microwave dielectric spectra using an open ended coaxial probe is presented. The dielectric spectra of several kinds of meat were collected in the range of 200 MHz to 12 GHz. The data are used to predict the composition of the materials applying principal component analysis and regression. Effects of the reduction of the used bandwidth on the performance are investigated. Optimal frequency ranges with respect to measurement accuracy and hardware effort for an instrument are defined.