1 of 2
China increasingly uses technological innovations in cigarette manufacturing.
2 of 2
AI plays an important role in different stages of the manufacturing process.
In recent years, the Chinese tobacco industry vigorously integrated scientific and technological innovation into smart manufacturing of cigarettes to accelerate digital transformation. All units in the industry actively explore the innovative application of “Internet+” in multi-levels and fields.
The industry practices new manufacturing patterns and new traceability of tobacco leaves, which activated the data and application advantages for the whole industry chain, and released new impetus to promote the high-quality development of the industry.
China Tobacco (Anhui) carries out smart testing based on AI visual technology
Hefei Cigarette Factory under China Tobacco (Anhui) Industrial Corporation promoted the construction of an intelligent factory to enhance cigarette production. Based on AI vision technology, the factory took the lead in the r&d of an intelligent detection device for cut tobacco width and an intelligent identification device for cigarette pack appearance defects.
Version 1.0 of the intelligent detection device for cut tobacco width used a high-pixel industrial camera to take online pictures of cut tobacco and calibrate the typical cut tobacco shape structure through a deep learning algorithm. Updated to version 3.0, the device now can realize second-level detection speed and micron-level detection accuracy. It greatly improves the sample size and coverage of cut tobacco width detection, and provides data support for cut process quality, timely correction, and equipment fault diagnosis.
To reduce the number of defect cigarette packs released into the market and the operation intensity of manual sampling inspection, Hefei Cigarette Factory developed a new cigarette pack defect identification device based on a deep learning algorithm. The device performs modeling training on visual-defect samples of cigarette packs through a convolution neural network, accumulates the feature recognition algorithms of different defects for online recognition and compares visual defects, and continuously optimizes recognition accuracy. At present, the recognition rate of the device for 14 typical defects reaches 95%.
China Tobacco (Hunan) develops digital flavoring platform
Cigarette flavoring is a core technology supporting the development of Chinese cigarette brands. With more than two decades of technological accumulation and innovative research, China Tobacco (Hunan) Industrial Corporation creatively put forward the concept of “digital flavoring” and integrated and developed a digital flavoring technology platform to meet the research demands of STMA.
The platform integrates the industry’s generally recognized physical database of aroma raw materials and the digital aroma flavoring expert system. Figuratively speaking, the digital flavoring technology platform is an “intelligent flavoring robot” with “data + algorithm + computing power” based on the interactive engine of model algorithm and empirical theory. The achievement won first prize for the 2019 science and technol-ogy progress award of China National Tobacco Corporation as it realizes the independent flavoring of single aroma raw materials of cigarette products, innovates and develops cigarette style characteristics, and stably improves the quality of cigarette products. The platform covers the whole industry and has grown to be one of the technology platforms with the widest application and the highest recognition in the industry.
Shanghai Tobacco Group promotes digital design of cigarette products
In recent years, Shanghai Tobacco Group Co., Ltd. has worked with many units in the industry to deeply focus on the digital transformation of cigarette product design. On the one hand, the data access mechanism for cigarette product design has been established systematically for the first time. The evaluation methods and requirements for sensory quality evaluation, chemical composition, flavor and fragrance characteristics, physical detection, spectrum, and other data have been formed. A dual channel data acquisition mode of the main-stream near-infrared spectrometer has been set up, and the whole process data link of product design and maintenance has been bridged.
On the other hand, a series of digitally aided design models and algorithms have been developed in such key design areas as leaf blending, flavors and fragrance, and tobacco materials, etc. The interaction algorithm and evaluation algorithm among various modules are constructed, and the digitally aided design technology of “free combination of design logic and linkage and interaction of design process” is formed, which promotes the transformation of cigarette product design from experience-driven to data-driven.