Based On The GA-BP Neural Network, The Color Prediction Of The Dope Colored Cellulose Yarn

The dope-dyed technology is different from traditional fabric dyeing.In the process of spinning(including color spinning and plain spinning)and weaving,the color of the product will change visually.Therefore,how to control the color of the finished product of dope-dyed fiber becomes a big problem.With the rapid development of science and technology,industrial production has gradually entered the stage of intellectualization.As an excellent representative of artificial intelligence,neural network has gradually matured with the progress of science and technology,and has been applied in various fields.In this subject,optical lens measurement method and yarn plate measurement method are used to obtain the color data of fibers and yarns respectively and the regularity of the two methods is analyzed.It is found that the color of yarns can not be characterized by establishing common linear functions.Therefore,GA-BP prediction model is established by using genetic algorithm to optimize BP neural network,combining the global search ability of genetic algorithm with the local search ability of BP neural network,the aim is to obtain the optimal color prediction result of plain yam.Based on the finite number of samples,the CIEL,CIEa and CIEb values of fiber and yarn were selected as input neurons and output neurons of BP neural network to establish a reasonable BP neural network.In the 28 groups of samples,25 groups were selected as training samples,and 3 groups were used as validation samples.In order to ensure the stability and scientific of the prediction results,the cluster analysis was used to divide the 28 groups of fiber samples into 3 categories,there are 7 samples in the first category,4 samples in the second category,and 17 samples in the third category.The most representative three groups of samples(golden,rose,and iron ash)are selected from each category as validation samples to construct the BP neural network structure of 3-5-3.The optimal weight and threshold obtained by genetic algorithm are given to BP neural network,the results are obtained after training:the neural network(GA-BP)optimized by genetic algorithm can converge at twenty-eighth times of iteration and eighth steps of training step.The correlation coefficient R of the fitting curve is above 0.99,which has a good linear relationship.The CIELAB color differences of the three groups of validation samples of rose,golden and iron ash were obtained by BP neural network:3.26,2.87 and 1.32.The CIELAB color differences of the above three sets of verification samples were obtained by GA-BP network:1.22,0.93 and 0.61,It can be seen that the prediction accuracy of GA-BP network is higher,which shows that the BP neural network optimized by genetic algorithm is more effective to predict the color of monochromatic yarn.Combining the results of cluster analysis with GA-BP prediction results,the smaller the number of samples in the same category,the larger the color difference of the validation sample,otherwise the smaller the color difference,the color difference between the predicted value and the true value of the golden and iron ash is less than 1,which can meet the production requirements of the enterprise,there are only 4 groups of samples in the same category as rose,and the color difference of rose is large.It is concluded that the BP neural network optimized by genetic algorithm can be applied to predict the color of monochromatic yarn made from raw liquid colored fibers in the case of sufficient sample size and uniform color gamut distribution,and the ideal prediction results is obtained.This topic provides a certain basis for solving the color prediction problem for the color of the solid color yarn of dope-dyed fiber,and will effectively promote the application and development of dope-dyed fiber.