烟丝物理质量的制丝工艺神经网络模型及其评价

发布时间：2018-11-20 08:19

【摘要】：为研究制丝过程工艺加工强度对烟丝物理质量的影响,针对某品牌烟制丝工艺过程的松散回潮、一次加料、叶丝增温增湿和叶丝干燥4个加工过程,选取其中关键的5个工艺参数进行均匀试验设计;根据各试验组检测的烟丝填充值、整丝率、碎丝率,对工艺参数和质量指标进行BP神经网络建模;应用建立的模型预测不同工艺参数组合对应的物理质量,通过3组试验验证了模型的准确适用。整丝率检测结果为80.34%、82.73%、85.42%,对应的预测结果为80.89%、82.12%、84.86%;碎丝率检测结果为1.87%、1.61%、1.74%,对应的预测结果为1.6%、1.58%、2.2%;填充值的检测结果为4.57、4.73、4.60 cm3/g,对应的预测结果为4.48、4.51、4.62 cm3/g。此方法为制丝过程工艺参数优化提供了新思路,可辅助于生产指导、降低试验研究成本。
[Abstract]:In order to study the effect of processing strength on the physical quality of tobacco during the process of silk making, four processing processes, namely, loose moisture, single feeding, heating and humidifying of leaf silk and drying of leaf silk, were carried out in a certain brand of tobacco making process. Five key process parameters are selected for uniform test design. According to the filling value, the whole rate and the broken rate of tobacco, the BP neural network was used to model the process parameters and quality indexes. The established model is used to predict the physical quality corresponding to different process parameter combinations. The model is proved to be accurate and applicable through three groups of experiments. The whole silk rate was 80.34 and 82.73 and 85.42, and the corresponding prediction result was 80.89 and 82.120.86; The test result of broken silk rate is 1.87 and 1.61 and 1.74, and the corresponding forecast result is 1.58 and 2.2; The test result of filling value is 4.57 ~ 4.73 ~ 4.60 cm3/g, and the predicted result is 4.48 ~ 4.51 ~ 4.62 cm3/g.. This method provides a new idea for optimizing the process parameters of filament making process, which can be used to guide production and reduce the cost of test and research.
【作者单位】：云南中烟工业有限责任公司技术中心;昆明理工大学机电工程学院;
【基金】：中国烟草总公司2015年科技重点项目(110201402027)
【分类号】：TS452

【参考文献】