论文部分内容阅读
将温度限制串联相关网络与红外光谱分析技术相结合,对大黄样品的真伪进行分类。采用小波变换对原始数据进行压缩,将原来的775个数据点压缩到49个数据点,既提高了网络的训练速度又保持了原来的特征谱峰。对45种样品进行了测定和鉴别,正确率可以达到84．4％。对影响分类结果的网络参数,进行了讨论。红外光谱法作为中药鉴别的一种方法与神经网络相结合,使中药鉴别更加快速、方便。
The temperature-limited series correlation network was combined with infrared spectrum analysis technology to classify the authenticity of rhubarb samples. Using wavelet transform to compress the original data, the original 775 data points are compressed to 49 data points, which not only improves the training speed of the network but also maintains the original characteristic peaks. 45 kinds of samples were measured and identified, the accuracy rate can reach 84.4%. The network parameters affecting the classification result are discussed. Infrared spectroscopy, as a method of identification of traditional Chinese medicine, is combined with neural network to make Chinese medicine identification more rapid and convenient.