Prediction and optimization of chemical fiber spinning tension based on grey system theory

Based on the grey prediction model, this paper studied the effect of the chemical fiber spinning process parameters on the winding tension. Suitable process parameters were selected to carry out grey incidence analysis with winding tension, and the feasibility of the grey prediction model in spinning tension prediction was validated by the designed experiments. The corresponding algorithm routines of various grey prediction models were written in MATLAB. The single-variable grey prediction model of GM(1,1) showed a higher prediction accuracy in the effect of the single process parameter changing on spinning tension; when multiple process parameters changed at the same time, the average modeling error of the MGM(1, n) multi-variable grey prediction model was 7.70%, and the maximum error was as high as 32.99%. The original MGM(1, n) model was optimized and the model background value was adjusted by using the auto-optimization and weighting method. The average modeling error of the improved model was reduced to 2.02%, which could meet the general accuracy requirement of tension prediction. Further combining fractional-order accumulation and adjusting the background value coefficient α and the cumulative order r jointly, the smallest prediction error was found among the 100,000 combinations, and the final error was further reduced to 1.30%. The results show that the grey prediction model is suitable and effective for predicting the spinning tension based on the process parameters. Appropriate model improvement mechanisms will increase the prediction accuracy significantly. This application provides a suitable method for spinning tension prediction, which has great significance for the tension control of chemical fiber products.