Research On Genetic Algorithm Based On RBF Neural Network For Dyeing Color Matching

Dyeing color matching is a very important process in textile dyeing industry. Traditional way for dyeing color matching is not only time-consuming, but also incorrect because it depends on the man's experience. In 1939, Kubelka and Munk proposed a simple theory from the integrated radiative theory and put the simple theory into practice. But the result produced by the theory is inaccurate and needs amending repeatedly because the theory is based on some pre-determined assumptions. Thus a model for dyeing color matching with Genetic Algorithm based on RBF neural network is proposed and realized by using VC and Matlab in this paper.A conclusion is drawn that there is a relation between dye concentration and the Lab value of dyeing by making use of the non-linear relation between dye concentration and dyeing reflectivity, and the non-linear relation between dyeing reflectivity and the Lab value of dyeing. In addition, the relation is approximated in the RBF(radial basis function) neural network. The dye concentration is the neural network's input value and the Lab value of dyeing is the output.The Genetic Algorithm is also involved in the model because it has the ability to solve the complicated problem and non-linear problem. In this model, let the dye concentration be the individual. The individual serves as the RBF's input, and we use the absolute value of the deviations between the value of the target color and RBF's output value as the objective value.Two conclusions can be drawn by experiments as following:1 The model for dyeing color matching with Genetic Algorithm based on RBF neural network is usable;2 It's very important to train a RBF network for the precision of the formula;...