A hybrid modeling using clustering algorithm for textile slashing process

The slashing is a very important procedure in textile manufacturing process which can improve warp quality, loom efficiency and reduce warp break. A hybrid modeling method is proposed for textile slashing process. Data are divided to multiple subsets by clustering algorithm, and then artificial neural networks (ANN) and partial least square (PLS) regression are used to model multiple sub-models respectively according to size of subset. The weight coefficient of sub-model is obtained by Lagrange multiplier method, and the whole model is established by combining multiple sub-models. The simulation result shows that the proposed hybrid modeling method has a better predictive accuracy and robustness.