Aiming at the key process section of silk workshop, the export process quality control of drying silk is mainly based on the experience setting, which can only realize a single control of the export process quality, and in the face of many process parameters, it is not possible to solve the problem of accurately obtaining the optimal combination in order to improve the quality of the export process of drying silk. This paper proposes the multi-objective optimization of silk drying process parameters by BP neural network combined with NSGA-II. Taking the moisture export temperature of silk export as the research objective, collecting the actual production data set, and comparing the advantages and disadvantages of the three types of models, the BP neural network is used as the fitness function of the optimization objective. Based on the actual production situation, the constraint range of process parameters is established, and NSGA-II is used for the multi-objective optimization of the process parameters of silk drying, which shows that the prediction accuracy of BP neural network is higher than that of multivariate regression and random forest, and the mean square error is lower than that of the other two, and its R2 and RMSE are 90.72% and 0.38 respectively, and the optimal process parameter combinations are verified, and the comparison reveals that the error of the model and the actual value is 1.5 times that of export moisture. The error between the model and the actual value is 1.1% for the export moisture and 2.4% for the export temperature, which indicates the feasibility and accuracy of the model, and provides a certain reference for the optimization of the parameters of the silk drying process.