Abstract —Soybean phenology is strongly influenced by temperature and day length, and phenological records clearly reflect the changes in climatic conditions. A model including three artificial neural networks was designed to predict the time intervals between sowing, emergence, flowering, and maturity as dependent on climatic factors. Ensemble regression models were constructed to predict the yield, seed protein, and oil content in soybean. Data on maturation were analyzed for early-maturing soybean accessions phenotyped at two experimental stations of Vavilov Institute of Plant Genetic Resources in the North-Caucasian and Northwestern regions of Russia. The model was implemented in Python using the Keras and TensorFlow packages.

中文翻译：

---

从气候因素预测早熟大豆品种物候的人工神经网络模型

摘要——大豆物候受温度和昼长的影响很大，物候记录清楚地反映了气候条件的变化。设计了一个包含三个人工神经网络的模型来预测播种、出苗、开花和成熟之间的时间间隔，这取决于气候因素。构建了集成回归模型来预测大豆的产量、种子蛋白质和油含量。在俄罗斯北高加索和西北部地区瓦维洛夫植物遗传资源研究所的两个实验站对早熟大豆种质进行了表型分析，分析了成熟数据。该模型是使用 Keras 和 TensorFlow 包在 Python 中实现的。