Imagen de Portada

Detection of pest infestation in stored grain using an electronic nose system optimized for sensor arrays

Guardado en:
Publicado en:Journal of Food Measurement & Characterization vol. 19, no. 1 (Jan 2025), p. 439
Publicado:
Springer Nature B.V.
Materias:
Grain
Humidity
Data processing
Matrices (mathematics)
Regression analysis
Regression models
Electronic noses
Correlation analysis
Nondestructive testing
Least squares method
Food quality
Statistical methods
Statistical analysis
Prediction models
Error detection
Correlation coefficients
Correlation coefficient
Pattern recognition
Grain storage
Eigenvalues
Data analysis
Coefficient of variation
Variance analysis
Pests
Support vector machines
Root-mean-square errors
Sensors
Optimization
Multivariate analysis
Rice
Sensor arrays
Gaussian process
Infestation
Food safety
Methods
Mathematical models
Environmental
Acceso en línea:Citation/Abstract
Full Text - PDF
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
Resumen:Pest infestation during grain storage reduces the weight and quality of the grain, which poses a risk to food safety. It’s important to have a reliable, quick, and intelligent approach for spotting pests in grain storage. In this study, an electronic nose (e-nose) was designed to detect the different densities of Tribolium castaneum (Herbst) in stored wheat. To avoid the phenomenon of “dimensional disaster” caused by the large amount of data in the e-nose data processing, the eigenvalues of the e-nose response curve were extracted to form the original feature matrix for data analysis. Then, to obtain the optimal feature matrix, the initial feature matrix was gradually refined using multivariate statistical methods such as response strength analysis, analysis of variance, coefficient of variation analysis, and correlation analysis. Finally, the feature matrix was regressed using partial least squares regression (PLSR), principal component regression (PCR), support vector machine regression (SVR), and Gaussian process regression (GPR) to establish various prediction models. The GPR model presented the best prediction effect among the four regression models, and its correlation coefficient (R), root mean square error (RMSE), and relative analysis error (RPD) were 0.96, 9.08, and 2.24, respectively. This work provides a feasible optimization method by which the e-nose can be used to detect stored grain pest density within a very small error margin and promotes the development of intelligent agriculture.
ISSN:2193-4126
2193-4134
1932-7587
1932-9954
DOI:10.1007/s11694-024-02980-2
Fuente:Agriculture Science Database