测定了原始灵芝及其水提物、 95%乙醇提取物和石油醚提取物的红外光谱, 发现红外光谱能承载系统的化学信息和反映被测物中各组分分布情况。 依3 416～3 279, 1 541和723 cm-1各吸收峰面积与2 935～2 852 cm-1吸收峰面积之比可将四者区分开来。 提出了用欧氏距离计算样本集信息熵的方法, 并探讨了信息熵与样本集所承载的化学信息量之间的关系, 得出原始灵芝样本集承载化学信息量最丰富的结论。 对4个样本集分别做谱系聚类分析, 发现只有原始灵芝红外光谱集对黒芝、 青芝、 黄芝和赤芝有较好的聚类作用。 研究结果表明: 红外光谱承载着物质结构的化学信息, 与系统的化学成分密切相关; 红外光谱样本集的信息熵越大, 化学信息量越丰富, 越有利于模式识别。 本研究对模式识别样本集的构建具有指导作用。

From the determination of the infrared spectra of four substances (original ganoderma lucidum and ganoderma lucidum water extract, 95% ethanol extract and petroleum ether extract), it was found that the infrared spectrum can carry systematic chemical information and basically reflects the distribution of each component of the analyte. Ganoderma lucidum and its extracts can be distinguished according to the absorption peak area ratio of 3 416~3 279, 1 541 and 723 cm-1 to 2 935～2 852 cm-1. A method of calculating the information entropy of the sample set with Euclidean distance was proposed, the relationship between the information entropy and the amount of chemical information carried by the sample set was discussed, and the authors come to a conclusion that sample set of original ganoderma lucidum carry the most abundant chemical information. The infrared spectrum set of original ganoderma lucidum has better clustering effect on ganoderma atrum, Cyan ganoderma, ganoderma multiplicatum and ganoderma lucidum when making hierarchical cluster analysis of 4 sample set. The results show that infrared spectrum carries the chemical information of the material structure and closely relates to the chemical composition of the system. The higher the value of information entropy, the much richer the chemical information and the more the benefit for pattern recognition. This study has a guidance function to the construction of the sample set in pattern recognition.