Detection Of Electrical Characteristics Of Fresh Wolfberry Damage And Storage Period

The quality inspection and grading of the fresh wolfberry is a difficult problem for the development of the wolfberry industry,which is the most competitive industry in Ningxia in terms of local characteristics and market competitiveness.Traditional quality testing methods are difficult to provide a timely,rapid,scientific and comprehensive response to the quality of the fruit.Therefore,it is of practical research significance to carry out non-destructive testing techniques based on electrical characteristics to promote the high quality development of the wolfberry industry.In this study,using a test bench built for testing the electrical parameters of fresh wolfberry,eight electrical parameters of the fruit were firstly determined under different electrical excitation signals,and the optimum test conditions for the electrical parameters of fresh wolfberry were determined.Secondly,the electrical parameters of fruits with different degrees of damage and storage time were measured to verify the feasibility of using electrical properties to discriminate the degree of damage and storage time of fruits.Finally,the original data were analyzed by principal component analysis(PCA)and partial least squares(PLS),and the damage level and storage time discrimination models based on support vector machine(SVM),random forest algorithm(RF)and convolutional neural network(CNN)were developed respectively.The following conclusions were obtained:(1)The effect of electrical excitation signals on the electrical characteristics of fresh wolfberry.The effect of test voltage on the electrical characteristics of the fruit was small,while the test frequency was closely related to the eight electrical parameters such as impedance Z.At the same test voltage,with the increasing common logarithmic value of test frequency lg f,the common logarithmic values of fruit impedance Z,reactance X,parallel equivalent inductance Lp,parallel equivalent capacitance Cp,parallel equivalent resistance Rp showed a decreasing trend,the common logarithmic value of electronagogram B showed an increasing trend,and the impedance phase angle θ and loss coefficient D showed a decreasing and then increasing trend,and all reached their extreme values at 39.8 KHz.The coefficient of determination R2 of the regression equation between the test frequency and the fruit electrical parameters was greater than 0.93.The best test voltage for the fruit electrical parameters was 1.0 V,the best test frequency was 3.98 MHz,and the better test frequencies were 3.98 MHz,1 MHz and 100 KHz.(2)Study of the electrical characteristics of post-harvest fresh wolfberry.At the same test frequency,Z,X,Rp and Lp of the fruit showed an increasing trend in the full frequency range from 100 Hz to 5 MHz,while B and Cp showed a decreasing trend with increasing damage.D and θ decreased with increasing damage in the higher frequency range from 1 MHz to 5 MHz.Fruit Z,X,B,Lp,Cp and Rp can be used as characteristic electrical parameters to discriminate the degree of fruit damage from 100 Hz to 5 MHz,and D and θ can be used as characteristic electrical parameters to discriminate the degree of fruit damage from 1 MHz to 5 MHz.At the same test frequency,Z,X,Rp and Lp of the fruit showed a decreasing trend and B and Cp showed an increasing trend in the full frequency range from 100 Hz to 5 MHz as the storage time increased.Fruit Z,X,B,Lp,Cp and Rp can be used as characteristic electrical parameters for identifying fruit storage time from 100 Hz to 5 MHz.It is feasible to use the electrical characteristics to discriminate the degree of fruit damage and the storage time.(3)Construction of a model for the discrimination of the degree of damage and storage time of fresh wolfberry.The accuracy of the six fruit damage discrimination models was greater than 89.33%in both the training and test sets,of which:the PLS-RF model was the most effective,with 99.56%and 91.00%accuracy in the training and test sets respectively.The discrimination accuracies of the six fruit storage time discrimination models established were greater than 82.67%in both the training and test sets.Among them,the PLS-RF model was the most effective,with 98.44%and 87.67%accuracy in the training and test sets respectively.The PLS-RF model was determined to be the best model for discriminating the degree of damage and the storage time of wolfberries,based on the complexity and accuracy of the model.The results of the study can provide a reference for the electrical detection technology of fresh wolfberry and lay the theoretical foundation for the development of non-destructive testing instruments based on electrical characteristics.