中文摘要：

卟啉传感器阵列系统可以检测肺癌呼出气体中特定的标志性气体,不同标志性气体检测输出的差值图谱不一样。介绍了一种结合反向传播（BP）神经网络和主成分分析（PCA）的肺癌标志性气体种类识别算法,并将其应用在卟啉传感器阵列系统中。通过计算卟啉传感器阵列中各点的主成分得分选出敏感点,保留各气体敏感点的值,并组成识别模板作为BP神经网络的输入层,达到去除冗余数据的目的。通过实验对比聚类分析结果、未降维数据的BP神经网络识别结果及已经PCA降维后的数据作为输入的BP神经网络识别结果,证明提出的算法可以更加精确地识别不同的肺癌标志性气体。

英文摘要：

Porphyrin sensor array system can detect lung cancer specific markers in exhaled gases, output vary from marker to marker. A pattern recognition algorithm based on backpropagation （ BP ） neural network and principal component analysis （PCA）is proposed and is applied in porphrin chemical sensor array integrated system. The sensitive points is selected by calculating principal component scores in porphyrin sensor array, and reserve value of each gas sensitive points, and template of recognition is formed as the input layer of the BP neural network to achieve the goal of removing redundant data. Comparing with the result of clustering analysis and BP neural network identification without reducing dimension and data after reducing dimension as input, the result of the proposed algorithm can identify precisely for lung cancer specific markers.