Fabric dyeing formulation prediction algorithm based on particle filter

Different strength and interaction(hydrolysis and association) varing from different batches of dyes causes the first spline results of computer color matching system inaccurately,it needs to manually adjust formula time and again to achieve customer requirements.For these issues,according to the orthogonal experiment,dyeing was performed and the result was taken as a standard sample,then the error between standard sample and prediction one which corresponds to the initial formulation was analyzed by CCMS(Computer color matching system) forcasting,and a non-linear model between them was established.Based on the model and error analysis,and by combining CCMS prediction,particle filter was chosen to estimate the actual recipe,thus a PF algorithm prediction formulation was obtained.Experiments show that the color error(CIELAB) which uses PF algorithm to predict will be reduced by more than 50% compared with the CCMS prediction.Thus,the nonlinear model is accurate and the method of using PF algorithm to predict the actual formulation is feasible.