摘要
基于免疫算法的基本思想 ,提出了新的免疫主成分分析法 (IPCA) ,该方法将免疫算法中抗体对抗原的消除运算应用于二维数据矩阵的正交分解 ,可得到矩阵的特征值和特征向量 .结果表明 ,IPCA与传统的主成分分析法比较 ,对 HPLC-DAD模拟信号的计算结果基本一致 .对 HPLC-DAD实验信号的解析结果表明 ,将 IPCA与窗口因子分析技术结合比传统的 WFA具有更强的解析能力 .
Based on the principal of immune algorithm, a novel algorithm for chemical factor analysis was proposed, and called immune principal component analysis(IPCA). The basic idea of the proposed method is that it takes a data matrix as antigen, the retrieving eigen vector as antibody, and the orthogonal decomposition of the matrix can be achieved by an iteration of subtracting the principal component represented by the eigen vector, simulating the process of the interaction between antibody and antigen in an immune system. Comparing with the conventional principal component analysis, similar results were obtained. But if we combine the IPCA algorithm with window factor analysis(WFA) technique, it will be more suitable for resolution of overlapping HPLC DAD signals than the conventional WFA technique. Both simulated and experimental data sets were investigated. Similar results are obtained by the IPCA and PCA for principal component analysis, but the IPCA WFA is superior to the conventional WFA in resolution of the multicomponent overlapping HPLC DAD signals.
出处
《高等学校化学学报》
SCIE
EI
CAS
CSCD
北大核心
2002年第7期1269-1272,共4页
Chemical Journal of Chinese Universities
基金
国家自然科学基金 (批准号 :2 9975 0 2 7)