| Straw-based nutrient seedling-growing bowl tray is the core of straw-based nutrient seedling-growing bowl tray seedling and full mechanized planting technology system,which is prepared by processing crop straw as the main raw material,adding nutrient additives,sterilizing agents,biological glue,and pulp through processes such as pressure molding,drying and shaping.Due to the high moisture content and easy breakage of straw-based nutrient seedling-growing bowl tray after forming,it is not easy to store.Therefore,it is necessary to dry the formed straw-based nutrient seedling-growing bowl tray to reduce its moisture content and achieve safe moisture content for shaping and storage.Real-time monitoring and studying the changes in moisture content during the drying process of Straw-based nutrient seedling-growing bowl tray is a crucial approach to ensuring drying quality.Currently,the measurement methods for moisture content of drying materials,including straw-based nutrient seedling-growing bowl tray,mainly include oven drying and weighing method,alcohol combustion method,and pan drying method.However,these methods are not only time-consuming,labor-intensive,and operationally complex,but also lack real-time capability.This poses significant challenges to studying the moisture change patterns during the drying process of straw-based nutrient seedling-growing bowl tray and achieving automatic adjustment of drying parameters through real-time moisture content detection,thus ensuring drying quality.In some cases,these challenges may even render it unattainable.Therefore,in this study,a combination of hot air-assisted microwave drying technology was employed to dry straw-based nutrient seedling-growing bowl tray,and the drying characteristics were investigated.Furthermore,a drying kinetics model was established to study the moisture change patterns during the drying process.Additionally,hyperspectral imaging technology was utilized to detect the moisture content in Straw-based nutrient seedling-growing bowl tray during drying,and a predictive model based on hyperspectral data was developed to forecast the moisture change patterns during the drying process.Therefore,the main research contents of this study are as follows:(1)The impact of microwave power,hot air temperature,and hot air velocity on the moisture content,drying rate,and effective moisture diffusion coefficient of straw-based nutrient seedling-growing bowl tray during the drying process were investigated in this study.The results showed that the hot air-assisted microwave drying process of rice straw nutrition tray consisted of four stages:the rising rate drying stage,the falling rate drying stage,the rising rate drying stage again,and the falling rate drying stage again,without a distinct constant rate drying stage.The variation trend of effective moisture diffusion coefficient was consistent with that of moisture content.(2)Thirteen commonly used mathematical models from classical drying kinetics were selected as the research objects in this study.The experimental data was used for fitting,and R~2,χ~2,and RMSE were used as criteria for evaluation.The results showed that the Weibull distribution model performed the best among the 13 models,with the highest R~2 value and the smallestχ~2 and RMSE values.The R~2 values ranged from 0.99484 to 0.99963,theχ~2 values ranged from 0.000112977 to0.0008701,and the RMSE values ranged from 0.00789 to 0.0295,indicating that the Weibull distribution model could be used for predicting the moisture content of rice straw nutrition tray during drying process.To investigate the predictive performance of neural network for moisture content prediction of rice straw nutrition tray,a multilayer perceptron(MLP)model was established.The R~2 and RMSE values for the test set were 0.9804 and 0.0017,respectively,indicating good predictive performance,and it could be used for predicting the moisture content during the hot air-assisted microwave drying process of rice straw nutrition tray.A comparison of the performance between the Weibull distribution model and the MLP model showed that the overall error of the Weibull distribution model was smaller than that of the MLP model in the test set,indicating better predictive performance.(3)A moisture content prediction model for rice straw nutrition tray during the drying process was established based on hyperspectral imaging technology.Spectral data in the range of 400-1000 nm from 204 rice straw nutrition tray samples were extracted,and spectral data preprocessing techniques including multiple scatter correction(MSC),standard normal variate(SNV)transformation,and Savitzky-Golay convolutional smoothing(SG)were applied.Principal component analysis(PCA)and competitive adaptive reweighted sampling(CARS)were used for dimensionality reduction of the spectral data.Subsequently,six different prediction models for rice straw nutrition tray moisture content were constructed using the reduced spectral data,including partial least squares regression(PLSR),random forest regression(RF),particle swarm optimization-support vector regression(PSO-SVR),extreme gradient boosting(XGBoost),multilayer perceptron(MLP),and convolutional neural network(CNN)models.The performance of the models was evaluated using the coefficient of determination(R~2)and root mean square error(RMSE)as evaluation metrics.The research results showed that the CARS-XGBoost model,built using spectral data preprocessed with the SNV method,exhibited the best predictive performance and generalization ability.The R~2 value of the training set(R~2-C)was 0.9924,and the RMSE value of the training set(RMSE-C)was 1.7857.The R~2 value of the testing set(R~2-P)was 0.9675,and the RMSE value of the testing set(RMSE-P)was 4.3598.These results indicated that the model achieved high accuracy in predicting the moisture content of straw-based nutrient seedling-growing bowl tray during the drying process. |
PDF Full Text Request