关键词: 大数据, 叶丝干燥, 内在规律, 数据挖掘, 预测模型

Abstract: In order to mine and analyze potential values and rules of the data of cigarette production process, the production data of cut tobacco drying process were excavated and analyzed based on the analytic technology and method of large data. The stabilities of the key quality indexes and process parameters of cut tobacco drying process were mostly analyzed, and the internal relationship between quality indexes and process parameters as well. Results showed that: ① The stability control level of key quality indexes was determined as from high to low, discharge moisture content after cooling≥discharge temperature≥discharge moisture content, and there was great fluctuation in June. ② The stability control level of key process parameters was determined as from high to low, wall temperature of I area≥hot air temperature≥wall temperature of II area, and the fluctuation of wall temperature of II area was mainly caused by feedback control mode. ③ There were strong positive correlations between the discharge moisture content after cooling and negative pressure, wall temperature of I area, steam valve opening of I area, and strong negative correlations between the discharge moisture content after cooling and the steam flow of expansion unit, steam valve opening of SX, wall temperature of II area, steam valve opening of II area, opening of moisture exhaust valve. ④ There were no obvious correlation between discharge temperature and these process parameters considered in present study. ⑤ The prediction model of discharge moisture content after cooling was established, which had good prediction accuracy. It could be predicted that the large data analysis method would have a good application prospect in tobacco technology field.

Key words: large data, cut tobacco drying, inherent law, data mining, prediction model