| 摘要: |
| 针对目前蜂蜜检测技术存在的无法快速、准确识别的问题，提出了一种基于激光诱导荧光(LIF)与学习向量量化(LVQ)有机结合对蜂蜜进行快速识别的方法；采用LIF获取蜂蜜的光谱数据，利用主成分分析(PCA)对光谱数据处理，将处理后的数据输入已建立好LVQ分类学习模型中进行分类识别；实验将挑选4种不同的蜂蜜，每种采集50组蜂蜜光谱数据，随机抽取120组蜂蜜光谱数据用于LVQ神经网络模型的训练，其余80组蜂蜜数据将输入训练好的LVQ模型进行测试；LVQ分类学习模型用于蜂蜜分类鉴定需要的时间为0.8 s，LVQ分类学习模型用于蜂蜜分类鉴定的准确率达到99.45%；实验结果表明：将基于LIF与LVQ有机结合，可以满足蜂蜜快速、准确识别的要求。 |
| 关键词: LVQ神经网络算法 激光诱导荧光技术 主成分分析法 数据处理 蜂蜜识别 |
| DOI： |
| 分类号: |
| 基金项目: |
|
| Application of Learning Vector Quantization to Honey LIF Spectral Image Recognition |
|
TANG Chao，ZHOU Meng-ran
|
|
School of Electrical and Information Engineering， Anhui University of Science and Technology， Anhui Huainan 232001,China
|
| Abstract: |
| Aiming at the problem that the current honey detection technology cannot identify the honey quickly and accurately， a method based on the organic combination of laser-induced fluorescence (LIF) and learning vector quantification (LVQ) to quickly identify honey is proposed. The author uses LIF to obtain the spectral data of honey， uses principal component analysis (PCA) to process the spectral data， and inputs the processed data into the established LVQ classification learning model for classification and recognition. The experiment selects 4 different types of honey， each collects 50 groups of honey spectrum data， 120 groups of honey spectrum data are randomly selected for the training of the LVQ neural network model for 4 types of honey， and the remaining 80 groups of honey data are input for training in the trained LVQ model. In this experiment， the time required for the LVQ classification learning model for honey classification and identification is 0.8s， and the accuracy of the LVQ classification learning model for honey classification and identification reaches 99.45%. The experimental results show that the organic combination of LIF and LVQ can meet the requirements of rapid and accurate identification of honey. |
| Key words: LVQ neural network algorithm laser induced fluorescence technology principal component analysis data processing honey identification |
