Type Identification And Implementation Of Respiratory Gases Based On Principal Component Analysis

Blood test is used commonly to detect diseases in the hospital, it is an invasive traumatic type detection means, it not only gives the patient pain, but also to detect a long time. Scientific research workers know that everyone exhaled gas composition is different,just like human fingerprints, and therefore can be called "gas fingerprints." Gas detection has advantages, such as non-invasive and non-traumatic, it is expected to replace blood tests to become a new way of disease detection. Respiratory diagnosis has become an important research area in today’s society.This paper focuses on the respiratory gas extraction and ionization signal after study after the main data processing method to work. The respiratory gas analysis comprises four components: respiratory gas gathering, respiratory gas smoothing,respiratory gas feature extraction,classification of virtual yin and yang.In the course of respiratory diagnostic studies, people continue to explore new gas detection technology to get reflected in the respiratory disease markers. In this paper, a new respiratory gas detection technology ——extractive electrospray ionization mass spectrometry(EESI-MS).The respiratory gas of yin,yang,yin-yang and normal are processed by EESI-MS technology, they are collected and set up data model. According to the mass to charge ratio of near principles,ionic strength of the original respiratory gas is to add up,the strength of the ion beomes discrete. Total intensity value of respiratory gas ionic which mass to charge is between 50 and 350 are taken as the data analysis target.The processing of the respiratory gas is gaussian smoothing to eliminate noise, and extract the characteristic. In the study of respiratory gas feature extraction method,principal component analysis(PCA)can better respiratory gas extraction characteristics by experiment. Finally,it stablish training model for four groups SIMCA classification,calculate the similarity of test sample in these four traing model.The type identificationresult is the gas type that the maximum extent belongs to.Experiment four respiratory gas sample size are 42,12,13 and 19 respectively,principal component scores are 43,7,8 and 13 respectively. Recognition rate and rejection rate of the four training models are more than 80%.Through experiments, the respiratory gas is processed by the PCA algorithm and the SIMCA classification, the virtual identify effect of the yin and yang is obvious.