Prediction method of cigarette draw resistance based on correlation analysis

The cigarette draw resistance monitoring method is incomplete and single, the lacks correlation analysis preventive modeling, resulting in substandard cigarettes market. To address this problem without increasing hardware cost, paper, multi-indicator used to predict resistance. First, process of analyzed based on existing quality control framework, optimization ideas are proposed. In addition, for three production units, cut tobacco supply (VE), rolling (SE), cigarette-forming (MAX), direct potential factors associated with explored, linear non-linear analysis. Then, correlates as inputs machine learning model, predicted values outputs. Finally, research also innovatively verifies practical application value prediction: distribution characteristics prediction results, time interval being produced determined, probability model sampled derived, reliability result further verified by example. results show that has good performance months actual production.