| Wheat is one of important agricultural products and basic materials of people’s life.Wheat provides 20%of the energy needed by humans,and about 35%of the world’s population uses wheat as its main source of food.The moisture content in wheat is an important indicator for evaluating its quality and processing.Currently,there are direct method and indirect method for detecting the moisture content of wheat.The direct method has high precision but is time-consuming and laborious.The indirect method includes resistance method,capacitance method,ray method and infrared method.However,these methods have their respective limitations.Compared with the traditional wheat moisture content detection method,the wheat moisture content measurement method based on the free-space,resulted in advantages such as low cost,less time-consuming and easy to implement continuous measurement.This paper,mainly through the study on dielectric properties of wheat and their relationships with influencing factors,will provide basis reference with exploitation of moisture meter for wheat.In this thesis,based on microwave free-space measurement,the effects of frequency(1~4GHz),temperature(10~40℃),moisture content(11.35%~17.79%)and bulk density(low and high)on the relative dielectric constants of wheat were studied.The factors causing the effect of changes in the relative permittivity of wheat were analyzed,and the prediction models of wheat moisture content with different frequencies were established.(1)At the measurement frequency of 1~4GHz,the relative dielectric constant of wheat showed a monotonous decreasing trend with the increase of frequency.The relative dielectric constant of wheat showed a monotonous increasing trend with the increase of temperature,water content and bulk density(2)At the measurement frequency of 1~4GHz,the dielectric loss factor of wheat decreases monotonously with the increase of frequency.The dielectric loss factor of wheat increases monotonously with the increase of temperature,water content and bulk density(3)Using the Design-Expert8.0.5b established the mathematical model of relative dielectric constant and dielectric loss factor and moisture content,bulk density and temperature of wheat.The significance of wheat dielectric property value model as less than0.0001,and the mathematical model was extremely obviously.The model variance analysis table shows that the most important factors affecting the dielectric properties are moisture content,temperature and bulk density(4)The verification results of wheat dielectric properties show that there is a very significant linear relationship between the predicted values of the dielectric properties of wheat and the measured values,and the R~2>0.95(5)At the single-frequency condition,the ELM prediction model has the best prediction effect.The best frequency at single-frequency is 3.9GHz,the predicted model has R_C and R_P of 0.9662 and 0.9784,respectively.At the multi-frequency conditions,the SVM prediction model has the best prediction effect.The best frequency band in multi-frequency is1.0~1.5GHz.The R_C and R_P of the prediction model are 0.9856 and 0.9736,respectively.At the full-frequency condition,the prediction effect in the SVM model is the best.The R_C and R_P of the prediction model are 0.9834 and 0.9838.(6)At the PLS,SVM and ELM prediction models used,the prediction effect under single-frequency is not as good as the multi-frequency and full-frequency prediction.At the PLS prediction model and the ELM prediction model,the prediction model results under multi-frequency conditions are better than the prediction model results under full-frequency conditions.Under the SVM prediction model,the prediction model results at full-frequency conditions are better than the multi-frequency prediction models. |
