Sensors provide a variety of information to microprocessors which in turn process the information and provide useful information to the recipient (1). A wide variety of information is processed by computers and microprocessors and transferred to recipients and users such as humans or machines.
However, sensor technology is basically aimed at sensing and detecting basic information used by such computers, is far behind computer or communication technologies because of its higher complexity. As such, sensors have become a main hindrance to functional improvement in various systems.
Multivariate data from an analytical instrument can be processed in a variety of ways (2). If the instrument provides sufficiently resolved data then the problem is reduced to an univariate calibration.
For example, a chromatographic peak that is fully resolved from all the other components is sufficient for qualitative and quantitative analysis. Furthermore, in mass spectrometry, a peak with a characteristic mass-to-charge ratio is sufficient to detect a certain species without interference. In less favorable cases, a series of different responses must be processed by multivariate pattern recognition techniques. This method is useful for devices which have a non selective response.
In spectroscopy, the spectra of the individual species must be available. A further prerequisite is that spectral intensities are adding up linearly. Then, from ...