In spite of a great deal of effort to prevent mixing foreign bodies in food materials, food manufacturers have still not been able to exterminate them. Nonmetal foreign bodies are very difficult to detect while metal foreign bodies can be detected easily. In this study, we developed a new nonmetal foreign body detection method and instrument. At first, we showed that the phases of the reflected waves from minced fish containing and not containing a few kinds of nonmetal foreign bodies were measured by setting the minced fish between two pieces of electrodes with high-frequency voltages. The results showed that the optimal frequency of the voltage supplied to the electrodes is 10 MHz, and the phase detector needs a precision of 0.1 deg. However, if floating the minced fish fast through the detector like the site in the factory, existing measuring instruments cannot be used because of their sampling speed and of the distortion of the signal. So, we developed a new digital phase-measuring instrument, which had the DSP's (digital signal processors) for calculation of the FFT (fast Fourier transform), are tangent and so on, and which used an adaptive sampling method. The developed instrument made it possible to measure the phase every 5 ms with a precision of 0.1 deg. However, the judgment of the foreign bodies was difficult, because the phase value of the minced fish itself is changeable in the floating experimental condition. So, we dealt effectively with the problem by using a neural network with a backpropagation learning method. The computer simulation results showed that a glass bead (2 mm in diameter), a styrene foam (3/spl times/3/spl times/3 mm/sup 3/) and paper (15/spl times/15 mm/sup 2/) could be detected.
展开▼