Mathematical formulation of knitted fabric spirality using genetic programming

This paper proposes the use of genetic programming for the mathematical formulation of knitted fabric spirality. Both dry relaxed and wash-and-dry relaxed states of fabric spirality are studied. In total, six parameters are investigated, in which three parameters are derived from yarn and fabric, and the other three parameters are from the knitted condition. The three yarn and fabric parameters used are yarn twist liveliness, tightness factor and dyeing method, and the three knitting parameters are the number of feeders, rotational direction and gauge. Genetic programming is adopted to formulize the mathematical relationships between above the six parameters and two states of fabric spirality, respectively. For a comparison, a multiple linear regression approach is studied as well. The formulas generated by genetic programming and multiple regression for two states of spirality are comprehensively investigated and compared. Experimental results show that genetic programming, which can model non-linear mathematical relationships, obtains more accurate expressions than multiple regression for both dry relaxed and wash-and-dry relaxed states of spirality, demonstrating that genetic programming is a promising alternative for the mathematical formulation of fabric spirality.