Research And Development Of Chip Moisture Content Meter Based On BP Neural Net And Capacitance

In the production of the wood industry, there are several kinds of methods to measure moisture content, usually including electrical conductivity, infrared inspection, microwave irradiation, capacitance method and so on. The common capacitance sensor is composed of two-electrode planks which vary in shape and can be fixed in different plane of space. The driving and inductive electrode pole placed in the same plane of the uniplanar scattering-field capacitive sensor. The electric field presents scattering. As the shape of power line in the electric fields is like semicircular arc or ellipse, there is some limits in the measurement depth of uniplanar scattering-field capacitive sensor, and its penetration depth is equal to 2/3 of center distance of adjacent driving and irritable electrode pole. When the change in the component of the tested material measured the uniplanar scattering-field capacitive sensor causes the variation of the total dielectric constant ε_x, the measured capacitance C_X will also change. Therefore, capacitive sensor can be used to measure the specific component of the tested material indirectly.This paper designed and made moisture content meters of both high frequency and low frequency with frequency demodulation as its principle, chip as its measuring material and uniplanar scattering-field capacitor as its sensor. With the premise of that valid chip forming area (the valid uniplanar inductive area of capacitance sensor) keeps stable, experiment results showed that with the definite forming height and density of chip when the moisture content varied in ranges of 5% to 25% were definite, the moisture content was exponent function as measuring frequency; when the moisture ranged in 25% to 45%, there was linear relationship between measuring frequency and moisture content of chip. When the moisture content and forming density were certain, there was linear relationship between measuring frequency and forming height. When the moisture content and forming height were determinate, there was no function relationship between measuring frequency and forming density.The analysis showed when the measuring frequency was low , there existed relative little error between the frequency predictive value and the actual measuring value from stepwise regress equation and the data reached certain precision. While the measuring frequency was high, there was relative huge error between them. The variation tendency of the predictive value was inconsistent with the changing tendency of actual measuring value and theory analysis.Neural Net is called the sixth generation of computer as well as the second generation of artificial intelligence. It has the characteristic of intelligent and tremendous self-adaptation and self-study, reasonable and prompt judging thoughts and highly scattering information storage. Today BP Neural Net has become the most widely used study arithmetic. It is statistical that...