Prediction performance analysis of spatial interpolation methods for indoor environmental factors in the plant factory

This study aims to predict indoor environmental factors at unobserved points in a plant factory with the use of spatial interpolation and analyze their prediction performance. The research process consists of four steps. (i) Step 1: Data & Setting: Plant factory and sensor monitoring; (ii) Step 2: Predict indoor environmental factors at unobserved points using spatial interpolation; (iii) Step 3: Validate the prediction performance of spatial interpolation; (iv) Step 4: Visualize spatial interpolation results. As the best result of validating the prediction performance of the spatial interpolation methods, the CV(RMSE) of the inverse distance weighting method (IDW) was 2.06%, and that of the radial basis function interpolation (RBF) was 1.90%. These results indicate the following two trends: (i) As the maximum and minimum deviations of sensor measurements decrease, IDW exhibits excellent prediction performance, while RBF does not show consistent tendency; and (ii) both IDW and RBF tend to show better prediction performance in indoor air temperature and CO₂ concentration as the distance between sensors decreases. Furthermore, as the distance between sensors increased, both IDW and RBF showed better prediction performance at relative humidity. In addition, the uneven distribution of indoor environmental factors was intuitively confirmed via visualization of the spatial interpolation results. It is expected that this study will be able to establish facility operation strategies to create a uniform growing environment through visualization of spatial interpolation results in the plant factory.