Data analysis electronic sensor output

The data processing of the multivariate output data generated by the gas sensor array signals represents another essential part of the electronic nose concept. The statistical techniques used are based on commercial or specially designed software using pattern recognition routines like principal component analysis (PCA), cluster analysis (CA), partial least squares (PLSs) and linear discriminant analysis (LDA). 

The AD-interface connecting the analyser and the data unit is an 8-channel serial AD-interface with 12-bit resolution. The ADC is based on a Linear Technology chip (LTC1290) and requires only a few additional components (a print board, complete with all components can be purchased for about 50 US from Conrad Electronic, D-92240 Hirschau). This interface is connected to a maximum of 8 spectrophotometer analogue outputs (or other instruments, e.g., an attached conductivity sensor) and to a serial port of the PC. One of the analogue inputs reads the sampler status. A serial ADC allows longer distance between wet analysis and dry data treatment (about 15 m). If parallel ADC interface cards mounted in the computer are used, the wet-analyser and computer have to be close together (less than 5 m). No additional hardware is required. 

Seven types of espresso coffee were classified by Pardo and Sberveglieri with a system composed of an electronic nose and an SVM with polynomial and Gaussian RBF kernels.For each coffee type, 36 measurements were performed with an electronic nose equipped with five thin-film semiconductor sensors based on SnOi and Ti-Fe. The output signal from sensors was submitted to a PGA analysis whose principal components (between 2 and 5) represented the input data for the SVM classifier. The error surface corresponding to various kernel parameters and number of input principal components was investigated. 

---