LSTM Modeling and Optimization of Rice (Oryza sativa L.) Seedling Growth using Intelligent Chamber

An intelligent growth chamber was designed in ۲۰۲۱ to model and optimize rice seedlings' growth. According to this, an experiment was implemented at Sari University of Agricultural Sciences and Natural Resources, Iran, in March, April, and May ۲۰۲۱. The model inputs included radiation, temperature, carbon dioxide, and soil acidity. These growth factors were studied at ambient and incremental levels. The model outputs were seedlings' height, root length, chlorophyll content, CGR, RGR, the leaves number, and the shoot's dry weight. Rice seedlings' growth was modeled using LSTM neural networks and optimized by the Bayesian method. It concluded that the best parameter setting was at epoch=۱۰۰, learning rate=۰.۰۰۱, and iteration number=۵۰۰. The best performance during training was obtained when the validation RMSE=۰.۲۸۸۴.

کلیدواژه ها:

Artificial intelligence ، MATLAB ، Radiation ، Recurrent Neural Networks ، Temperature