| 刘袆帆，杜卉妍，钟玉鸣，马路凯，李小敏，朱立学.基于香蕉、粉蕉成熟过程中硬度变化建立成熟度随机森林模型[J].广东农业科学,2020,47(6):106-115 |
|
查看全文
HTML
基于香蕉、粉蕉成熟过程中硬度变化建立成熟度随机森林模型 |
| Construction of Maturity Random Forest Models Based on the Changes of Fruit Firmness of Musa AAA Group Banana and Musa ABB Group Banana During Ripening |
| |
| DOI：10.16768/j.issn.1004-874X.2020.06.015 |
| 中文关键词: 香蕉 粉蕉 部位 果实硬度 成熟预测模型 随机森林 |
| 英文关键词: Musa AAA group banana Musa ABB group banana part fruit firmness maturity prediction model Random Forest |
| 基金项目:广东省重点领域研发计划项目（2019B020223003）；广州市科技计划项目（201804010480）；广东省现代农业产业技术体系创新团队项目（2019KJ139） |
|
| 摘要点击次数: 1588 |
| 全文下载次数: 587 |
| 中文摘要: |
| 【目的】研究香蕉和粉蕉成熟过程中果实柄部、中部和端部的硬度变化，并据此建立随机森林模型预测其成熟度。【方法】通过 GY-4 型果实硬度计分别测量香蕉和粉蕉成熟过程中果实柄部、中部、端部 3个部位的硬度，并用相关性分析研究果实硬度与其他相关成熟指标之间的关系。使用相似性分析（ANOSIM）分别对香蕉与粉蕉成熟过程中果实 3 个部位的硬度进行分级后，通过随机森林算法（Random Forest）构建模型预测其成熟度，再比较果实各部位硬度对模型的重要性。【结果】香蕉和粉蕉成熟过程中果实硬度变化趋势相似，果实 3 个部位的硬度变化不一致。利用相关性分析发现香蕉与各种酶类活性和乙烯释放量呈显著负相关关系。香蕉和粉蕉成熟度随机森林模型的错误率分别为 4.94% 和 0%，说明此模型能准确预测香蕉与粉蕉的成熟度，且香蕉果实中部与粉蕉果实柄部的硬度对模型的影响最大。【结论】利用随机森林模型，香蕉和粉蕉成熟过程中果实硬度能准确预测其成熟度，且果实 3 个部位的硬度变化不一致，根据果实不同部位的成熟度差异可提高利用率，香蕉果实中部与粉蕉果实柄部的硬度更能反映自身的成熟度，为鲜食蕉成熟度量化与快速监测分级提供参考，促进食品加工工业的发展。 |
| 英文摘要: |
| 【Objective】The study was to explore the changes of fruit firmness of the stalk, the middle and the tip of Musa AAA group banana and Musa ABB group banana during ripening. Based on this, Random Forest models were established to predict the maturity of Musa AAA group banana and Musa ABB group banana.【Methods】Use Type GY-4 fruit durometer was used to measure the fruit firmness of the stalk, the middle and the tip of Musa AAA group banana and Musa ABB group banana, respectively. Correlation analysis was used to study the relationship between fruit firmness and
other indexes related to maturity of banana. ANOSIM was used to classify the firmness of three parts during the ripening process of Musa AAA group banana and Musa ABB group banana, the Random Forest was used to construct models for their maturity prediction, and the importance of the fruit firmness of three parts on the model was compared.【Results】During the ripening process, the fruit firmness of Musa AAA group banana and Musa ABB group banana had a similar change trend, while the changes of fruit firmness of three parts were inconsistent. Through correlation analysis, it was found that banana had significant negative correlation with the activity of various enzymes and ethylene release. In addition, the error rates of the maturity prediction by Random Forest models were 4.94% and 0%, respectively,which indicated that these models could accurately predict the maturity of Musa AAA group banana and Musa ABB group banana, and the fruit firmness of the middle of Musa AAA group banana and the stalk of Musa ABB group banana had the greatest influence on their models.【Conclusion】By using Random Forest, the fruit firmness of Musa AAA group banana and Musa ABB group banana can predict their maturity accurately. The changes in fruit firmness of three parts are not consistent during the ripening process, and the utilization of bananas can be improved according to the maturity differences in various parts of the fruit. The fruit firmness of the middle of Musa AAA group banana and the stalk of Musa ABB group banana can better represent their maturity, which can provide a reference for quantification and rapid monitoring and classification of fresh bananas, thus promoting the development of food
processing industry. |
|
查看/发表评论 下载PDF阅读器 |
