| Melt blowing method is an important method to process microfiber non-woven.The fiber diameter is the key index that affects the heat insulation,sound insulation and filtration.In the actual production process,microfiber is formed by the stretching force of the flow field,and the air velocity and temperature in the flow field are important factors affecting the stretching force.Therefore,it is essential to optimize the structure of melt blowing die to improve the air velocity and temperature in the flow field.This thesis mainly does the following:(1)The conventional melt blowing die is improved to the concave-bottom melt blowing die,and the mathematical models of the two flow fields are established and simulated.Then,the conclusion that the velocity distribution and temperature distribution of the concave-bottom melt blowing die is better and determines the zone division according to the change of air velocity and temperature.The influence of nozzle parameters such as slot width,slot inclination and arc radius on air flow field is analyzed by using the control variate method.It is found that with the increase of slot width and slot inclination,the peak values of air velocity and air temperature increase significantly;high temperature decay rate is more affected by slot width;there is no significant linear change between arc radius and air velocity and temperature.Finally,the air velocity peak value,air temperature peak value and high temperature decay rate are determined as the performance indexes of the air flow field.(2)Orthogonal design and genetic algorithms are used to optimize the structural parameters of the concave-bottom melt blowing die.Based on the results of orthogonal simulation,the influence of other parameters on the flow field performance is analyzed when slot width or slot inclination is close to the boundary,and the possibility of "air flow threshold" is proposed.Then,the stagnation temperature was selected as the optimization target to optimize the air flow field and the influence of the concave-bottom melt blowing die's parameters on the velocity and temperature distribution of the air flow field was analyzed by applying the impact map and analysis of variance,and the optimal parameters were obtained according to the regression equation.Finally,the optimal structural parameters of melt blowing die are obtained by using genetic algorithm and CV value,and the optimization result is better than that of orthogonal experiment.(3)Two machine learning models,SVR and random forest,are used to solve the problems of insufficient speed weight and airflow threshold in the process of optimization.The machine learning model for the die parameters optimization is established.This paper describes the data preprocessing process,evaluation index and the method of adjusting the super parameter required by the training model.Then the two machine learning models are trained and the corresponding super parameters are adjusted,before the optimal model is obtained by comparing the MSE and r2_score.After considering the accuracy,stability and operation time of the models,the random forest regression model was selected.This model is used to make predictions and calculates the average deviation difference of each generation.Then,based on the results,predict the boundary of airflow threshold can be predicted and size of the optimization area can be determined.In this area,the multi-objective programming approach is used to get the optimization results.The results show that the optimization of random forest based on air air flow threshold is more conducive to reduce energy consumption.In this paper,machine learning is used to optimize the parameters of the concave-bottom melt blowing die.The results have positive significance for improving the air flow field and reducing energy consumption. |
PDF Full Text Request