Application of Complex Indices for Cotton Fiber Quality Characterization

Cotton fiber complex quality index can be evaluated by proper combination of individual HVI properties according to their influence on spinning processes and yarn quality. This index can be simply created based on the complex quality indices. The degree of quality (complex criterion) is here expressed as utility value U . Evidently, general quality of cotton fibers is characterized by various utility properties R_i (i = 1,…m) obtained from HVI measurements (fiber length-expressed as upper half mean UHM [mm], fiber length uniformity-expressed as uniformity index UI [%], fiber strength-as bundle strength STR [cN/tex], fiber elongation ant break-EL [%], fiber fineness and maturity-expressed by micronaire reading (MIC [-], short fiber content-SF [%] and thrash content-TR [%]).

When forming the aggregating function U from experimentally determined values of cotton fiber properties, the statistical character of the R_j quantities should be considered and the corresponding variance D(U) should be also determined besides the U.

One of methods for improving distribution of cotton imput data is power type data transformation. Combination of this transformation with graph of first two principal components can be used for identification of clusters of cottons as well.

The main aim of this work is detailed description of power type data transformation to approximate normality and creation of statistical characteristics by using of parametric Bootstrap.  The various clustering techniques are used for identification of cotton types. The program QCOTTON written in MATLAB is briefly mentioned. The applicability of this approach is demonstrated on the real data of crop study, which includes 33 sets of cottons.