THE PREDICTION OF COAL PROPERTIES USING COMPRESSED INFRARED DATA FROMOSCULATING POLYNOMIALS

Reduction of the representation of infrared spectra from coal samples by osculating polynomials of degree nine is discussed. The reduced rep resentation contains polynomial coefficients of order zero to four. Ma thematical models of the original spectra are obtained by linear combi nation of the coefficients. These compressed models are statistically correlated to coal properties, namely, volatile matter, fixed carbon, ash content, heating value, hydrogen, carbon, sulphur, nitrogen, and m aximum vitrinite reflectance, and the results are compared with those previously obtained from second derivatives of the same spectra. The u se of compressed data, while giving slightly better correlations for s ome of the properties, has the advantage of requiring less computation al time.