No abstract

The use of LEDs as a light source in plant factories with artificial lighting is expected to reduce costs, but it has been reported that some crops do not grow as expected due to differences in the wavelength of the light source. Therefore, it is necessary to design LEDs with the appropriate wavelength for each crop to be grown. On the other hand, there is research toward the realization of smart plant factories that utilize artificial intelligence, and artificial intelligence may contribute to the design of the light environment in plant factories. In this study, we selected mini-tomatoes as a model crop, and prepared fluorescent lamps and LEDs as the light environment during seedling growth, respectively, and searched for suitable light source wavelengths while investigating the relationship with growth conditions using statistical analysis methods, one of the artificial intelligence techniques. We investigated the relationship between multiple light environments, PPFD, R/B ratio, and spectrum of wavelengths respectively, using LEDs, fluorescent lamps, and dimming filters, and growth indices stem diameter in order to clarify the light environments that contribute to growth. The correlation between the measured light environment and crop growth results was objectively shown by PLS regression analysis, and the contributing wavelengths were explored by calculating the selectivity ratio and regression coefficient. As a result, it was suggested that stem diameter was promoted at around 550 nm and 630 nm, and suppressed at around 460 nm.

A method to discriminate between green peaberry beans and normal beans was developed using an excitation and emission spectrometer that can be used to construct a simple optical system to save farm workers' labor. The excitation and emission matrix were obtained in the wavelength range of 300 - 800 nm using a hand-held spectrophotometer combined with a diffraction grating and a diode array. Light sources of 300, 375 nm were used to excite the samples by light-emitting diodes. The acquired spectral information was used to classify the coffee samples into peaberry and normal coffee using the chemometric method of partial least squares discriminant analysis (PLS-DA). The results showed that all coffee samples were discriminated as corresponding classes by PLS-DA. In the PLS-DA model, investigation of the main wavelengths contributing to classification using all sample VIP showed that excitation/emission wavelengths at 300/470, 300/670 375/470, 375/670 nm are important for coffee type determination. These wavelengths were closely related to the excitation-emission wavelengths of several important chemical components in raw coffee beans (caffeine, caffeic acid, and chlorogenic acid (CGA), tocopherol). These results will be useful as basic knowledge to contribute to the labor-saving and automation of coffee bean sorting in the future.