| Moisture content of tobacco has an important influfence on it’s storage, transportation, processing performance and product quality. Silk production techniques is the key link in cigarette products processing, the moisture content of tobacco directly affects the machining performance, and have a major impact on the quality of the final product. In order to make the moisture content of tobacco meet the process requirementsin each working procedure link, guarantee the stability of the moisture content of the final product, and achieve the purpose of cigarette products stable quality, this paper studied the on-line monitoring, water content of slices tobacco near infrared spectrum analysis technique is used to establish the test model of moisture content in tobacco leaves, tobacco industry can get the moisture content of processing production more quickly and more safely, and combined with the actual process of silk in Sichuan tobacco industry, online spectrometer application tests have been carried out. Cylinder dryer is one of the key equipment in the cigarette production line, because of its simple algorithm and high reliability, PID control strategyis widely used in cigarette processing enterprises, Sichuan tobacco cigarette factory in Shifang is the use of this strategy for process control. But the control has a certain lag, resulting in the dryer outlet moisture of cigarette rate fluctuations, the control effect is poor. Therefore, in order to improve the stability of the outlet moisture content of tobacco is proposed in this paper, a control method based on BP artificial neural network. The main results are as follows:Near infrared spectroscopy in addition to sample information, also includes other independent of interference and noise, using CARS method to screen out the optimal wavelength, the final choice of two wavelengths, respectively 1908nm and 2108nm, two wavelengths are and O-H absorption related.PLS forecast model was established based on the two wavelengths, wavelength and not choose before the PLS model were compared, the training of the PLS model error is 0.09558, the prediction error is 0.14449, and the training of the CARS-PLS error is 0.03047, the prediction error is 0.04522, the prediction error is reduced, and reduce the workload of data processing, effectively simplifies the model and improve the prediction ability of the calibration model.In the process of silk, only monitoring and feedback continuous, rapid, real-time, accurate so that can we control the quality of the machining processmore effectively, improve product quality.Therefore, this article established the tobacco moisture content near infrared rapid detection method,Through the establishment of PLS model and the LS-SVM model and comparative analysis, the training of the PLS model error is 0.21118, the prediction error is 0.30865, and the LS-SVM training of the model error is 0.05054, the prediction error is 0.06342, performance of the model was compared with PLS. The results show that NIRS analysis technology combined with LS-SVM modeling method of tobacco moisture content detection is feasible.On cigarette processing production line, the moisture content is the key point of each working procedure, the assessment index water qualified rate has a very important influence on products processing performance and quality of products. In order to guarantee the stability of the moisture content in products better,must ensure the stable operation of the near infrared moisture meter, improve the detection accuracy is an important approach to ensure the data accurate and reliable. In this paper, through research and analysis concluded that the calibration on-line near infrared moisture meter sampling number should not be less than 5, you should use the dynamic sampling, moisture content measurement parallel measurement should be adopted. This method can be more accurate to correction of near infrared moisture meter, improve the detection accuracy.This paper uses the modeling method based on BP algorithm of artificial neural network, to material parameters for the model input, process indicators for the model output, baking machine parameter prediction, make the silk dryer wall temperature prediction correlation coefficient is 0.9672, the hot blast temperature prediction correlation coefficient is 0.9703, meet the application requirements of the silk dryer export of tobacco moisture content, better guarantee the stability of the silk dryer export of tobacco moisture content., the research In this paper has far-reaching significance on improving the quality of cigarette products for tobacco industry. |
PDF Full Text Request