Abstract: This review summarizes the application of machine learning in the hemp flavor substances of Zanthoxylum bungeanum. Different varieties of Zanthoxylum bungeanum on the market contain different hemp flavor substances and their contents are also different. At the same time, the traditional detection and analysis methods of the composition and content of Zanthoxylum bungeanum hemp flavor substances have many limitations, so the introduction of machine learning algorithm has brought new possibilities to this field. It is of great significance for the genetic breeding of Zanthoxylum bungeanum to establish the prediction model and drying model of Zanthoxylum bungeanum quality, comprehensively evaluate the sensory evaluation of Zanthoxylum bungeanum, and establish the Germplasm Resource Bank of Zanthoxylum bungeanum using machine learning algorithms. This review systematically reviewed the composition and content of hemp flavor molecules in different breeds of Zanthoxylum bungeanum, and analyzed the application of machine learning algorithms in quality prediction model, drying model and data analysis of hemp flavor substances. By integrating machine learning technology, researchers can have a deeper understanding of the currently established models, and provide support for the optimization of the yield and quality of Zanthoxylum bungeanum based on hemp flavor substances.

Key words: machine learning, Zanthoxylum bungeanum, numb-tasted substance, model