Cai Jiaxiao, Tuo Suxing, Li Yanchun, Lu Hongbing, Wu Yizi, Zou You, Ma Zhen, Cui Yuqi, Kong Bo, Zhong Kejun
Technology Center of China Tobacco Hunan Industrial Co., Ltd., Changsha, 410007, China.
Information and Network Center of Central South University, Changsha, 410083, China.
Heliyon. 2024 Dec 3;10(24):e40873. doi: 10.1016/j.heliyon.2024.e40873. eCollection 2024 Dec 30.
This article proposes a novel approach for improving the efficiency of fragrance designing and the accuracy of automatic fragrance formula creation based on empirical fragrance formulas and graph traversal algorithms. By effectively extracting the composition information and further analyzing the combination of fragrance materials in 210 fragrance formulas, a relational network model was constructed in the form of a graph to illustrate the relationship between the ingredients used in the formulas. Additionally, a fragrance ingredients information database of 344 common ingredients was constructed and used as a reference for perfumers when setting algorithmic constraints based on their experience. Finally, an automatic fragrance formula creation algorithm was established by constructing a relational network subgraph and finding fragrance formula solutions with the help of depth-first search algorithm that satisfies the constraint conditions and combining with appropriate statistical strategy that could determine the usage of each component in the new fragrance formula. By testing the algorithm with the goal of creating a floral fragrance, the resulting formula well fulfilled our expectations and had practical application value.
本文基于经验性香精配方和图遍历算法，提出了一种提高香精设计效率和自动香精配方创建准确性的新方法。通过有效提取210种香精配方中的成分信息，并进一步分析香精原料的组合情况，以图的形式构建了一个关系网络模型，来说明配方中所用成分之间的关系。此外，还构建了一个包含344种常见成分的香精成分信息数据库，供调香师在根据经验设置算法约束时作为参考。最后，通过构建关系网络子图，并借助深度优先搜索算法找到满足约束条件的香精配方解决方案，结合适当的统计策略来确定新香精配方中各成分的用量，从而建立了一种自动香精配方创建算法。通过以创建花香型香精为目标对该算法进行测试，所得配方达到了预期效果，具有实际应用价值。
