Skip to main content
ARS Home » Research » Publications at this Location » Publication #100608
Title: PERMITTIVITY DETERMINATION BY USING AN ARTIFICIAL NEURAL NETWORK

Author
item BARTLEY, PHILIP - UNIVERSITY OF GEORGIA
item MCCLENDON, RONALD - UNIVERSITY OF GEORGIA
item Nelson, Stuart

Submitted to: Institute of Electrical Electronics Engineers Instrumentation Technology
Publication Type: Proceedings
Publication Acceptance Date: 6/12/1999
Publication Date: N/A
Citation: N/A

Interpretive Summary: The dielectric properties, or permittivities, of materials are those properties that determine the interaction of the materials with electromagnetic fields. For example, in microwave heating, these properties determine how rapidly the material will warm up in a microwave oven. These same properties are also useful in sensing the moisture content of materials such as grain and other food materials, because there are useful relationships between their dielectric properties and their moisture content. Since moisture content is very important in determining selling price, the potential for safe storage, and the optimum conditions for processing cereal grains, rapid means of testing moisture content have been developed based on instruments that use electromagnetic fields to sense the moisture content. Many different techniques are needed for measuring the dielectric properties of agricultural products in the development of moisture sensing instruments. One useful technique for broad-frequency-range measurements of these properties is the use of an open-ended coaxial-line probe. However the computations yielding reliable measurements are very complex, requiring numerous error corrections. This paper reports the use of artificial neural networks, whereby the measurement of dielectric properties can be materially simplified. This new technique shows promise in developing more efficient measurement methods that will be useful in sensing moisture and other quality factors in agricultural products, which will be helpful in maintaining competitive advantages for American agriculture.

Technical Abstract: An Artificial Neural Network (ANN) was trained to determine the dielectric properties of materials from reflection coefficient measurements of an open ended coaxial probe. The ANN was trained by using measurements made on eleven water/isopropyl alcohol solutions. Coefficient of determination values of 0.999 were obtained. This approach has the potential of greatly simplifying the characterization of new dielectric probe designs.