Using radio frequency spectral measurements for determination of corn mechanical damage

Dielectric measurements between 10 Hz to 13 MHz were obtained using an HP 4192A Impedance Analyzer for artificially damaged and combine damaged corn samples. For medium and severe artificial damage, the measurements were obtained at two moisture contents (11 and 19 percent), five damage levels (0, 10, 25, 50, and 100 percent), and two bulk densities. The results showed that dielectric properties were able to measure moisture content, bulk density, and mechanical damage level. Using a four-variable model, medium damage level was predicted with R2 = 0.95 and RMSE = 8.8 percent, while severe damage level was predicted with R2 = 0.98 and RMSE = 5.97 percent using a four-variable model. However, when the actual moisture content and bulk density were used along with dielectric variables, the severe damage level prediction was slightly reduced with R2 = 0.97 and RMSE = 6.96 percent although the model used only two dielectric variables in addition to actual moisture content and bulk density of the corn samples. Medium damage level prediction was improved with R2 = 0.98 and RMSE = 5.77 percent, using four-dielectric variable and the actual corn moisture and bulk density.;Similar analysis was performed on combine damaged corn samples. Four different damage levels were produced. Damage levels were evaluated using Chowdhury\u27s method and visual inspection method. The results from visual inspection were: 0, 10.6, 17.4, and 31.2 percent mechanical damage. Three moisture content levels were prepared (12.5, 18, and 23 percent), and two bulk densities were produced. The damage calibration model was developed for each moisture level separately. For low moisture samples, damage level was predicted with an R 2 = 0.95 and RMSE = 3.36 percent using a six-variable model. For medium moisture samples, damage level was predicted with R2 = 0.95 and RMSE = 3.27 percent, using a five-variable model. Finally, for high moisture samples, damage level was predicted with R2 = 0.97 and RMSE = 2.29 percent using a three-variable model.;The technology shows a good potential for developing a mechanical damage sensor. Further testing on additional varieties, types of grains, and more moisture contents is needed