Worsted spinning process parameters inversion based on a mixed population genetic-ANN algorithm

摘要

Demand diversity and individuation, make the textile production process is complicated. To solve the problem of worsted spinning process parameters inversion accuracy, the hybrid population genetic neural network algorithm is presented in this paper (mixed population based - artificial neural network, MPG - ANN), MPG - ANN's advantage lies in three distinct advantages. First, itimprove the premature problem of traditional genetic algorithm. Second, predict generalization performance is enhanced and the inversion model. Third, the results of the calculation of stability was improved. Based on the quality index of yarn CV value of worsted spinning the key process parameters for inversion in the process of production, and compared with traditional genetic algorithm is applied to the inversion results, verify the feasibility and effectiveness of MPG - ANN algorithm, the inversion accuracy of 97%, the method not only has an important guiding role in the textile production process quality control, but also has a very good reference for enterprises rapid process development of new product design decision.