| Maize is an important grain crop in China,with 36.4%of the domestic grain crop area under cultivation.In recent years,maize seed harvesting has received more and more attention,and seed dehydration rate is an important criterion to determine whether maize is eligible for seed harvesting.Maize grain moisture detection is a key component of physiological research to optimize the dehydration rate of maize.Traditional seed moisture detection methods are tedious,have long detection cycles and damaged samples.Online in-situ detection of maize cob seed moisture omits the intermediate tedious process,improves detection time and reduces sample damage.In response to the above problems,this paper developed a maize cob seed testing system to obtain maize seed moisture online and in situ in real time,which simplifies the testing process,improves the testing time,provides a basis for agricultural departments and farmers to control the seed harvesting time,and facilitates researchers to optimize the speed of maize dehydration.The main research work is as follows:(1)To address the problem of low accuracy of detecting maize cob seed moisture,the capacitive detection circuit was optimized and a random forest linear regression model was constructed to improve the accuracy of maize seed moisture detection.In the laboratory,a probe radially inserted inside the cob is connected to a high-precision impedance analyzer to determine the optimal detection capacitance frequency of 1MHz;according to this frequency,the circuit of the capacitance acquisition module for detecting maize cob is optimized.The size of the optimized capacitance detection module PCB circuit board is only one-fourth of the original PCB circuit board,and the capacitance measurement error is below 5%.The impedance analyzer was used to obtain a large number of cob capacitance values at different temperatures and water contents,establish the corresponding data sets of seed water content,cob temperature and capacitance values,construct a random forest linear regression model,optimize the combination of hyperparameters,and its detection value RMSE was reduced from 5.1 to 3.1,and R~2 was improved from 0.930 to 0.977,and the prediction effect was significantly improved.(2)In order to realize online and in-situ detection of maize cob seed moisture,a maize seed moisture detection system was developed in this paper to realize real-time analysis and visualization of maize cob seed moisture data.The system mainly includes a real-time collection terminal,a server terminal and a client terminal APP,which can collect capacitance,temperature and equipment status in real time online.The server side mainly realizes data storage,archiving and analysis,and the client terminal APP realizes data visualization display and collection terminal status monitoring.(3)In order to verify the accuracy,stability and reliability of the system,twenty maize ears of different maturity levels were selected in the field and field tests were conducted.The data collection terminal was installed on the cob in the field and the seed water content was checked by logging into the client terminal APP.The test samples were labeled and the seed moisture content of this cob was measured in the laboratory using the drying method,and the maximum measurement error of the system was calculated to be 2.95%,with a minimum error of 0.1%and an average error of 1.5%.To test the effect of the in situ,continuous probe moisture detection method on cob growth,the NIR light detection method was used as a control.The NIR light detection was used because it required stripping the bracts and large areas of the upper part of the cob at harvest time were rotted or pecked by birds.The spikelets tested using the probe method were overall well developed at harvest with only two damaged seeds and the rest of the seeds were full and shiny.The measurement results are within acceptable limits and can meet the needs of practical applications. |
PDF Full Text Request