| The moisture content of fruits largely affects the physiological metabolism and taste and taste of fruits,and is one of the most important indicators in fruit transportation and long-term storage.Judging from the current research situation,various detection methods have been applied to the detection of the moisture content of fruits and other agricultural products,but the water content detection model and the moisture content detection method based on the dielectric properties of fruits have not been given a complete system..In this paper,Apple is used as a research object,and LCR 821 tester and other devices are used to measure the dielectric constantε and the dielectric loss factor ε" of apples,and a method for determining the moisture content of apples using these two dielectric characteristics parameters is proposed.Two mathematical models expressing the relationship between these two dielectric properties parameters and apple water content and measurement frequency were compared,and the performances of the two models were compared and analyzed.The main research contents and conclusions of this paper are as follows:(1)In this paper,we first collected the relevant data of apple’s dielectric properties and apple’s water content.In the measurement of apple dielectric constant and dielectric loss factor,the LCR821 digital measuring instrument was used to measure the 22 frequency points respectively.In the measurement of apple moisture content,the electric heating thermostatic blast drying oven was used in this paper to pass the dry weighing method.Get the moisture content of the apple.This article finally collected a total of 297(after removing invalid samples)apple samples of dielectric constant,dielectric loss factor,moisture content data.(2)This paper studies the theoretical relationship between the dielectric constant,dielectric loss factor,moisture content,and measurement frequency of apple.The results of the analysis show that as the frequency of the experimental measurement increases,both the dielectric constant ε and the dielectric loss factor ε’ of the apple show a decreasing state;at the same experimental frequency,the apple’s dielectric constant and dielectric loss factor all follow the apple.The decrease in the moisture content decreases,but the degree of change in the dielectric properties of ε and ε’ is different.Therefore,in this paper,we can use the three dielectric constants ε,dielectric loss factor ε,and their combination to build and analyze the moisture content model of apple.(3)Two kinds of linear and nonlinear moisture content detection models were established using the two dielectric properties of apple’s dielectric constant and dielectric loss factor respectively,as follows:The combination of stepwise regression and multiple linear regression was used to establish the linearity.For the apple moisture content model,the stepwise regression method was used to extract the feature frequency,and then the multivariate linear regression method was used to establish the detection model of apple moisture content;the combination of continuous projection method and support vector regression method was used to establish the non-linear apple moisture content detection.In the model,the continuous frequency projection method is used to extract the feature frequency,and then the support vector regression method is used to establish the apple moisture detection model.After the establishment of the two models,this article compares and analyzes the performance of the two models based on the performance parameters of the model mentioned in the article,and selects the optimum from the two nonlinear and linear apple moisture content models.Apple moisture detection model.Through comparison and analysis,it can be concluded that using the combination of continuous projection method and support vector regression method,apple’s dielectric constant and dielectric loss factor are fused together as modeling variables,and the apple moisture content model established has the best performance.The deterministic coefficient of its test set can reach 0.811,and the root mean square error can also be as low as 0.0196. |